=== Make a Rough Draft of Phase 2 (https://www.notion.so/Make-a-Rough-Draft-of-Phase-2-2a01a29f178980b6a8e1de4ab322f035) === Work from AI Watersheds Phase 2: Indicators [Child Page: Steve’s Notes from the Phase 2 Brainstorming Doc] Brainstorming doc Why Do Measurements? Thoughts on the purpose of forecasting AI’s schedule General trend visibility for several years (important for lots of policy + other questions) At least several years notice of transformational change Estimate likely steepness & extent of transformational change Ask panel for feedback on this Josh: interested in the connection between current policy questions and our questions here. Would be great to get more specific / explicit about the connections with policy questions. Josh: interested in the connection between current policy questions and our questions here. Would be great to get more specific / explicit about the connections with policy questions. General thoughts on what to measure Practical utility is a function of both AI model capability and complementary app innovation (coding models vs. Cursor) Approaches to measurement Josh (+’d by Helen, Nikola, Sayash): Noting cross-cutting theme: Generally seems like more high-quality surveys of companies would be valuable. A lot of ideas involve sensitive data, so would require controlled access (and/or governmental involvement?) Using AIs to qualitatively grade lots of screen recordings and other messy high-volume data Find ways of calibrating / correlating benchmark scores with measurements we care about more directly? (METR time horizons being an example in this direction) Generally use lots of different measurements and try to map them to common units for cross-checking Surveys / prediction markets; focus real-world measurements (e.g. uplift studies) on areas of disagreement Or get both sides of a dispute to agree on a cheaper proxy measurement Case studies How firms in various sectors are using AI; scale of tasks assigned to AI / nature + granularity of interaction with humans; degree of trust / freedom of action given to AI Deep study of revenue / spending / profit across the AI value chain (and by domain) Also: number of “complementary innovation” (Sayash’s term) AI-powered apps, adoption rate, # of customers, “Surveys of customers to see how they are improving their workflows”, etc. Helen: something about revenue from AI more broadly vs. revenue of frontier model providers. What can we say about open models eating into OpenAI / Google / Anthropic revenue? Broad metrics (a la Epoch), e.g. electricity used by AI, global private investment Keep pushing benchmarks toward greater realism of both input (task specification + context) and output (not just “passes tests”, but “PR is mergeable” / other measures of quality) Nikola: need more realistic benchmarks. Source PRs (ideally from inside AI labs) and test AIs on them. Analysis of usage at the model level (Anthropic economic AI index) Logs of agent usage i.e. usage of model context protocol servers, pulls [Sayash] Analysis of AI logs from them being asked to solve various tasks (we’re doing this with 10B tokens of HAL agent log data, but could imagine increasing many fold) Social media is subject to transparency measures, AI companies currently are not. It’d be great to have controlled researcher access to e.g. ChatGPT transcripts. Lots of general data sets for public analysis AIs really like to email Eliezer, that’s an indicator of various types of strange AI behavior. Ajeya: there was an NYTimes article about a journalist getting similar emails. Could there be a “weird AI shit” incident tracker that would categorize reports + report trends? Josh: MIT AI incident repository? Populated from public reports – links to existing reports. Ajeya: hard to make sense of it. Needs to be a larger team sifting the signal from the noise + writing blog posts. Doesn’t track veracity. AI Risk repository / incidents database is relevant here: https://airisk.mit.edu/, also https://incidentdatabase.ai/ Ajeya comment: Too hard to make sense of current databases. Need someone to do better synthesis. Make an open source AI company where the company itself is open source. Have it try to use AI as much as possible, make all its logs about everything (Slack etc) public, study that. Sayash: there’s been a lot of work on the science of AI evaluations. We don’t know the shape of the DAG, how AI inputs translate into outputs. We could do more to distinguish between different DAGs – which is a core source of disagreement, for instance Sayash has a very different DAG in mind. Inputs might include things like data, compute; output measured in the real world. Econometrics. Josh: Meta measurements: Big increase in expected OpenAI revenue, according to a prediction market, based on a reported algorithmic breakthrough Pushback from Ryan+Ajeya: limitations of prediction markets mean it’ll be tough for them to pick this up reliably Surveys at frontier AI companies Surveys of the broader workforce Data center construction spend → currently dependent on media reporting, is there some way to get better/more consistent data on this? Trying to unpack inputs/outputs per Sayash’s point about DAGs/understanding how improvements in one element flow through to improvements in other elements Lots of (different types of) surveys Helen: important to track military applications (broadly – decision support, targeting, etc.). Important to track freedom of action here. Measuring Utility Metrics about fluid intelligence, messy tasks Benchmarks for domains vs. skills Domains: cyber/law/professional labor markets Skills: Context awareness, reliability, long-context reasoning, sample efficiency Need deep info for each sector before assessing results. I.e. persuasion surveys overestimate results because opinion change dissipates 6 weeks later. I like Steve’s note that AI can continually be used cheaply. Uplift RCTs in real-world settings, measuring real-world outcomes (ideally: profit!) e.g. if you get an actual large company to actually do randomized staged rollout of enterprise LLMs that’s great Ajeya: three-armed uplift trials, AI / human / cyborg. Josh: Persuasion: Cost for AI to swing a vote, relative to other leading interventions (Josh: lots of polysci studies here) AI-bio: % of amateurs that can synthesize flu Look for leapfrog companies that get on board with a new technology ahead of the incumbents, that could highlight utility ahead of adoption. e.g. fintech cos using AI in ways that let them overtake traditional banks, or healthcare startups using AI in ways that let you bypass doctors’ offices Power user case studies – screen recordings + deep interviews of people who (think they) are getting enormous uplift Organization version of this: find those YC startups that supposedly do 95% of everything with AI, study them (“power organizational users”) Field trials (a la AI Village) and realistic benchmarks Do a time horizon analysis (a la METR), but on real-world usage? Also, qualitative analysis of the size + difficulty + nature of tasks successfully delegated to AI (and dig into ranges of “successfully”) Building on Josh’s work test idea: could a METR-like org collect work test setups from a range of places (under NDA so as not to break the work tests) and centralize the work of eliciting good performance, figuring out measurement, reporting results, etc? Ryan: I think uplift will be messy (because of humans being messy and possibility for phase changes in how AI should be used productively), so I’m more into end-to-end automation of actual tasks. Also, we ultimately care most about full automation regimes and predicting when this will happen. Uplift studies correspondingly seem somewhat worse than looking at full automation of tasks of varying difficulty/size (and benchmark transfer to actual things people are doing in their job so we don’t need to constantly run these tests). Maybe you can get AI companies to run semi-informal experiments with uplift internally and publish results: many companies have effectively committed to doing this eventually due to their safety policies. I think Anthropic might open to this, but there are various sources of trickyness here. Early reports of AI doing remarkable things (e.g. major scientific discovery / insight; solving a Millennium Prize Problem in mathematics) pass@any for existing benchmarks can predict pass@1 in the future Helen: where is the ceiling on various capabilities (headroom)? This seems under-investigated. Sayash: are most tasks like chess, or like writing? (Ryan: unclear whether writing is the right contrast here.) Are there less controversial examples than writing? Taking out the trash: saturates. Helen: the writing example highlights we’re way under-theorized on this. Quality of parenting is another example. “Persuasion” is poorly specified. Abi: a scientific breakthrough requires an insight that goes against dogma. Could be considered an aspect of very high-quality writing. Ryan: I care most about questions like “will energy production 100x within 5 years of full AI R&D automation if people want this to happen”. Seems like this has many possible routes in terms of capabilities head room. Sayash: three broad categories: computational limits (chess), intrinsic limits / saturation (writing?), and knowledge limits (need new breakthroughs to exceed past performance). Better AI can help with computational limits, but not intrinsic limits. Work tests for orgs in the space: GiveWell lit review test Open Philanthropy work tests High-quality summaries of a conversation (AI is still failing our first-round work test at FRI about this) Abi: “I liked how [this] list broke down sectors as each having their separate indicators. Very societal frictions approach.” Measuring Adoption / Impact YouTube resulted in some homeowners not calling plumbers for small jobs. What are the AI equivalents, and how should they be measured? Compare metrics across AI-feasible vs. AI-infeasible sectors: employment (incl. hiring patterns + plans), productivity, profit, qualitative measures of AI usage, … Revenue of AI service providers Studies of AI impact across various fields What percent of sales of the U.S. pharmaceutical industry is generated by AI-discovered drugs and products derived from these? What percent of publications in the fields of Chemistry, Physics, Materials Science, and Medicine will be ‘AI-engaged’ as measured in this study or a replication study, in 2030? How many hours per week on average will K-12 students in G7 countries use AI-powered tutoring or teaching tools, as reported by their school systems or education ministries? A September 2024 study by the Federal Reserve Bank of St. Louis, based on the Real-Time Population Survey (N=3,216), estimated that 0.5%–3.5% of all U.S. work hours were assisted by generative AI. Measuring Diffusion Ajeya: I want to know about adoption within AI companies, so e.g. surveys about how much compute they’re running for inference, freedom of action related surveys like the ones suggested below, internal estimates of how much AIs speed them up, etc. (Also, I don’t think I literally care only about AI companies, will prob also care about USG.) Measuring Freedom of Action Surveys of workers - how often do you have to approve an AI decision, how much thought do you put into each approval (e.g. level of review / pass rate of AI pull requests) Monitoring progress on agent infrastructure → if AI agents are operating online, are there meaningful constraints on them? Does e.g. Cloudflare have any meaningful controls, or is it all a chaotic mess? Something military or military-adjacent - reports of smaller militaries using autonomous/semi-autonomous systems..? We can also make a big list of discrete flags and ask people about them, e.g. “Is your AI allowed to spend money on business expenses? Up to how much?” or “Is your AI allowed to search the internet freely in the course of completing a task?” or “Are AIs at your company allowed to talk to employees other than the human that started the task in the course of completing their task? What about external people?” White-hat reports of prompt injections paid out by bug bounties at the top (100) websites across domains over time (as a proxy for how many applications allow AI systems to take actions that can exfiltrate user data) Black-hat exploits of AIs (only possible if AIs are in a position to take important actions without adequate review?) Curve-Bending Mechanisms Note that the impact of AI can be difficult to predict. E.g. the prediction of AI flooding the zone with misinformation in the 2024 election didn’t pan out. Sycophancy → mental health impact, conversely, was an event that wasn’t widely predicted. Intelligence explosion (software and/or hardware) Measure both initial speedup, and whether r > 1 Surveys of AI companies on internal AI use (bunch of ideas for granular questions, e.g. subj sense of speedup, what tasks they’re used for now, what tasks aren’t they used for, hypothetical questions) Internal measures of the absolute rate of algorithmic progress at AI companies (e.g. how much compute does it take to get GPT-4-level perf), watch for that trend accelerating Source Ryan: Get people who have done the takeoff modeling to look at the data on algo progress (both historical from humans and data under automation regime). Unclear if this data will resolve disagreements, but seems like it can help. (Tom, Daniel, maybe some people from Epoch) Algorithmic breakthrough / new approach (e.g. neuralese, in-context learning, long-term memory, recurrence, brain emulation) (either known or unknown unknown) Study new model releases to look for signatures of known ideas, or look for an increase in papers on some approach Discontinuous jumps across a range of difficult benchmarks Could ask companies key questions like e.g. “are you still using English CoT?” and make them report Survey expectations at frontier labs (also covers intelligence explosion) Analysis attributing performance gains to different sources (e.g. model size scaleup, data quality improvement, RL, etc) Ryan: really big breakthroughs are like, AlexNet, GPT-1 / scale-up, scaling RL (o1, o3). Easy to notice something is turning into a big deal, hard to tell where on the Richter scale it will land. Could track the current 5 most promising potential trend-breakers. Ajeya: track breakdown of capabilities improvements over time: how much is from scaling pretraining, scaling RL posttraining, etc. Threshold effects / phase changes (e.g. the moment when people are no longer needed; maybe time horizon scaling suddenly becomes much easier) Phase changes in behavior of AI Village or other observations of AI capability / usage in real or realistic contexts Phase change in qualitative analysis of uplift trials Some resource (data, compute, electricity, intelligence headroom) is exhausted Note, may lead to workarounds rather than an end to progress. E.g. pretraining data gets harder to scale → scale RL instead. Hard to predict the effect. Low-hanging fruit is exhausted; training for long-horizon tasks is expensive Some important missing capability doesn’t yield to scaling + progress (e.g. sample-efficient learning, “judgement”, long-time-horizon skills) Economics of AI not working out → slowdown in investment AI winter – we run out of ideas for improving AI Josh: FRI expert panel. Could ask them “what’s the probability of an AI winter in the next 5 years”? External event (market downturn; war in Taiwan; public backlash / disaster → regulation; focused national effort / Manhattan Project) Track public sentiment and other political indicators Applications of models are a lagging indicator to model capabilities. First sign of a slowdown would be the time horizon graph slowing down. Other Nikola: if timelines are long enough to enable human emulations (Ems), that would make a difference, because we’d have been able to explore more implications. More generally, does other transformative tech come first. Possible techs: EMs, nanotech (unlikely), genetic engineering (seems like this could radically transform society in ~30 years if heavily invested in now, but in practice might not happen because of e.g. legal/regulatory blockers). [Child Page: Notes for Early Indicators] David Langer at Lionheart suggested that they might be in a position to help us measure some things. https://x.com/StringChaos/status/1928476388274716707 https://x.com/Simeon_Cps/status/1900218546904293863 Relevant indicators hinted at here: https://www.theinformation.com/articles/openai-says-business-will-burn-115-billion-2029 Example of finding useful data sets: https://mikelovesrobots.substack.com/p/wheres-the-shovelware-why-ai-coding Plan For this sort of project, serious work requires multiple rounds of deep thought. This can’t be done in a compressed period of time, so doing this collaboratively requires an extended multi-turn conversation. I have not had success in getting most people to engage with this in a group email or Slack. The only way I’ve found success is me doing 1:1 engagement with each participant, over email and occasional calls, with a lot of nagging. This is annoying but does seem to work. I’ve witnessed one counterexample (Helen’s CSET workshop) and heard about at least one other (something Sayash worked on); these were both collaborative projects centered around a multi-hour (or multi-day) synchronous group discussion. However, I’m not sure how many new ideas were generated at CSET, as opposed to just aggregating people’s existing thoughts. The CSET workshop did generate some new ideas but also there was a lot of aggregation. Maybe we should just embrace this: perhaps our niche is projects that require lots of multi-turn 1:1 interaction. Phase 2 of this specific project is more about aggregating ideas than deep analysis, so perhaps getting a group of people together on a Zoom to brainstorm together could work. Brainstorming and ideation was, I think, the focus of the two success stories I referred to. We could try to get a bunch of people on a call without being too fussy about exactly which people (though it would be really good to have someone from METR, someone from Epoch, and perhaps also the Forecasting Research Institute). This could be a meeting to put a cap on phase 1 (without necessarily completely locking it down, I’ll continue to get input from a few people I’m only connecting with now) and doing the initial brainstorming for phase 2. I can seed the conversation with a few ideas shared in advance. Note that our measurable indicators don’t need to directly answer a crux, we’re just looking for data that will be helpful in future discussions of those cruxes. → share thoughts with Taren, Ajeya, Helen, ? Ask Helen how far she’s gotten in turning the discussion into a paper and whether she wishes she could do more rounds with at least some folks. Ajeya: Might be productive to do this in two rounds. Do the fuzzy thing, then get someone who’s interested in doing the legwork to do the first round of coming up with concrete experiments – Ajeya might help with this – and take that back to the participants to rate them. Group lists the fuzzy question. Individual – Ajeya or Ryan Greenblatt – proposes a list of concrete experiments Go back to the group to rate the proposals. Realtime, concentrated bursts of time are so much more productive that it’s better to ask someone to come to an event that sounds cool & fun, spend the first hour reading the doc. Or can get people to pre-read more reliably if you get them to agree to come to an event, that’s what she did for the loss-of-control workshop. And keep the writeup short. Siméon (@Simeon_Cps) posted at 5:31 PM on Fri, Aug 15, 2025:I've entertained that theory for a few years and have been confused since then why people expected confidently so much GDP growth. Basically prices if goods should crash so fast that the question of "how do you count inflation" will become the first order parameter of whether and(https://x.com/Simeon_Cps/status/1956514153528742168?t=qJkEFlWoQAD6z6g4HMhNAw&s=03) Notes with Taren, 8/8/25 Post four brainstorming time slots, let people sign up; probably don’t want >4 people/session (though could do breakout rooms) Taren, and probably many people, will do better in a brainstorming session with people with different expertise Could let each group decide which two cruxes they’ll talk about; if we wind up with a gap, do something about it at the end. Or maybe assign cruxes to time slots. Doing an in-person session could be fun; might be more trouble than it’s worth, but might not be trouble at Constellation, ask them to help organize & recruit people, e.g. two 4-person groups Try to do one in the south bay? Could include Sam. Ask Ajeya how she’d like to participate Talk to Helen Toner, how to make sure what I’m doing is complementary Other people / groups to include DC – government does a lot of data gathering – propose content for legislation – Abi could help? Taren could do something in DC Oliver Stephenson (FAS) Elham, or the guy who worked for her? FAI IFP CHT (CBT?) FAS? Some Horizon fellows who are placed to draft a bill? Plan a third stage where we produce a piece of draft legislation? Discuss with Abi Discuss with Victoria next Friday Taren discuss with some people in DC Should discuss with Helen Toner Next steps: First step should be some Zooms, invite everyone. We won’t exhaust potential participants from Constellation. Taren to talk to Abi about whether to make this the theme for the dinner they’re organizing; if not, Taren will convene some other small meeting while she’s in DC. Pre-brainstorm: me, Taren, maybe Ajeya, maybe Abi? Chris Painter? Ryan Greenblatt? Josh Rosenberg? Someone with economics expertise (a grad student from David Otter’s lab)? More focus on identifying kinds of measurements (quantitative, qualitative, horizontal, vertical, etc.) to seed later conversations (+ as pre-read) Participants in Stage 2 Jaime Sevilla (Epoch) Content Notes See Untitled Review the Cruxes list in the early writeup Incorporate this idea from my discussion with Nick Allardice: Our economy and decision-making processes are so fragmented and messy. Trying to answer my crux questions at a societal level is unhelpfully generalizing. More tractable & beneficial is to pick some sectors, industries, types of problems, and come up with ways of measuring change in those, as leading indicators for what might happen in other sectors and industries. E.g. software engineering may be one of the more tractable problem spaces for AI; he’d be very interested in tracking diffusion here: hiring practices, how much autonomy is being granted, what level of productivity is it unlocking. If CS is a fast example, find a few slow examples, get super specific & granular about measuring diffusion. Get an idea of the uneven distribution. Carey: I don't know where this fits but I think the question of "what are the most common failure modes that prevent current models from excelling at practical tasks?" would be a relevant crux or root cause for your cruxes. Anna Makanju: It’s hard to break down someone’s usage of chatbots into productivity into other uses. In the last year it’s flipped from predominantly productivity usage into companionship. If you measure usage, need to try to disentangle in the nature of that usage – apply a classifier to their chat history, or focus on measuring enterprise accounts. Could work with universities and government agencies who will have enterprise accounts and might be more willing to share data for a study. Also see notes from my 8/11 conversation with Anna Why Focus on Early Indicators? If you want to make predictions about a future that is similar to the present, you might be able to simply extrapolate from past values of the variable you need to predict. For instance, Moore’s Law was an observation about trends in transistor counts, and for many decades it provided excellent forecasts of future transistor counts [FOOTNOTE: Though this may be a story about self-fulfilling prophecies as much as about the tendency of important variables to follow predictable trends.]. If you need to make predictions about a future that looks quite different from the present, you can’t get by with simple extrapolation. You need a model of how the future is going to unfold, and you need data to calibrate your model. For instance, if you want to predict the potential of fusion power, you can’t extrapolate the graph of historical electricity generation; for fusion, that graph is flatlined at 0. But if you understand the path that current fusion efforts are following, you can extrapolate metrics like “triple product” and “Q” [FOOTNOTE: I got help from Claude Opus 4 on this; the answer matches my vague recollection well enough that I’m not bothering to fact-check it: The most critical metrics for tracking progress toward practical fusion power are the fusion energy gain factor (Q), which measures the ratio of fusion power output to heating power input and must exceed 10-20 for commercial viability; the triple product (density × temperature × confinement time), which needs to reach approximately 10²¹ keV·s/m³ to achieve ignition conditions; and the reactor's availability factor or duty cycle, measuring the percentage of time the reactor can operate continuously, as commercial plants will need to run reliably for months at a time rather than just achieving brief fusion pulses.] to get an idea of how close we are to a functioning generator. Verify that an early application of the steam engine was to pump water out of coal mines. Make a reference to this being a sort of recursive self-improvement. Observe that if you had wished to measure the uptake of steam engines for pumping water out of coal mines, you could have looked at inputs to the process such as the amount of coal being consumed in steam engines outputs such as the amount of water being pumped or impact such as lowering the water level within the mines. When will “superhuman coders” and “superhuman AI researchers”, as defined in AI 2027, emerge? How is the task horizon identified in Measuring AI Ability to Complete Long Tasks progressing? What are we learning about time horizons for higher reliability levels (I presume reliability much higher than 80% will be necessary)? How fundamental is the gap between performance on benchmarks and real-world tasks? Is it growing or shrinking? [QUESTION: Does this cover “capability-reliability gap” described in AI as Normal Technology, or do we need to expand the description?] [this might better belong under the question regarding advances in domains other than coding and AI research.] Are any skills under coding or AI research emerging as long poles (more difficult to automate), and if so, are there feasible ways of compensating (e.g. by relying more on other skills)? What does the plot of AI research effort vs. superhuman performance look like? How does this vary according to the nature of the cognitive task? In particular, for tasks such as “research taste” that are critical to accelerating AI R&D? Basically redundant with previous cruxes, but perhaps worth listing as something that could be independently measured: across the broad range of squishy things that people do every day, how rapidly will the set of tasks for which AI provides significant uplift grow, and what are the contours of that set (what separates tasks experiencing uplift from tasks which are not)? Can AIs think their way around the compute bottleneck? If the R&D labor input races ahead of compute and data, how much progress in capabilities will that yield? To what extent does this depend on {quantity, speed, quality/intelligence} of the AI workers? Does this apply to all compute-intensive aspects of AI R&D? Is Ege’s forecast for NVIDIA revenue bearing out? Is his model for relating NVIDIA revenue to real-world impact of AI valid? Could rapid algorithmic improvements (driven by a software explosion) decouple impact from NVIDIA revenue? Could adoption lags result in revenue lagging capabilities? Possibly other questions as to whether various current trends continue – I’m not sure whether there are any other cruxes lurking here. For instance, is general progress in LLM capabilities speeding up or slowing down? Are breakthroughs emerging that accelerate the curve? Is RL for reasoning tasks hitting a ceiling? Are any of Thane’s bear-case predictions bearing out? Etc. As the automation frontier advances into increasingly long-horizon, messy, judgement-laden tasks, will the speed and cost advantages of AI (vs. humans) erode, to the point where they aren’t significantly faster or cheaper than humans for advanced tasks (and a few years of optimization doesn’t fix the problem)? Resources from the CSET conference Helen’s working doc My slides Ryan’s slides and notes It's important to detect when the most senior skills start to become automated. This could indicate a tipping point both for progress at the big Labs and or the ability for a breakout at a rogue actor who doesn't have access to senior talent. Perhaps we can look at the percentage of impactful ideas that come from unassisted AI. Perhaps we can look at the ratio of of major ideas, paradigm changing ideas to other inputs. Look for additional domains in which to measure sample efficient learning. In other domains, domains look at the ratio of spending on real world data collection versus in silico data generation. ARC Prize (@arcprize) posted at 10:21 AM on Tue, Sep 09, 2025:ARC Prize Foundation @ MITWe're hosting an evening with top researchers to explore measuring sample efficient in humans and machinesJoin us to hear from Francois Chollet along with a world class panel: Josh Tenenbaum, Samuel Gershman, Laura Schulz, Jacob Andreas https://t.co/dq7NJyXkNk(https://x.com/arcprize/status/1965465501079142814?t=L7y6E9f9cwxgjwWD1FMZJA&s=03) https://epochai.substack.com/p/after-the-chatgpt-moment-measuring Check in with Divya / CIP to see whether their global pulse surveys have questions relevant to the cruxes. Perhaps we can draw from this / make suggestions for it. They’re running “global pulse” surveys, to understand what people want from the future but also to understand how much AI has diffused into people’s lives. Every two months, started in March. Questions around trust, diffusion, how much are you relying on AI for medical or emotional advice, are you using it in the workplace, etc. https://globaldialogues.ai/cadence/march-2025, “In three years, what questions will we wish we had been tracking?” Maybe could be interesting to co-author something at some point. Cheryl Wu (@cherylwoooo) posted at 6:25 PM on Sun, Jun 01, 2025:Are we at the cusp of recursive self-improvement to ASI? This tends to be the core force behind short timelines such as AI-2027. We set up an economic model of AI research to understand whether this story is plausible. (1/6)(https://x.com/cherylwoooo/status/1929348520370417704?t=f9E9Yty2m27EQ-Z_NbbJ3Q&s=03) From Does AI Progress Have a Speed Limit?, a measurement: Ajeya: I'm kind of interested in getting a sneak peek at the future by creating an agent that can do some task, but too slowly and expensively to be commercially viable. I'm curious if your view would change if a small engineering team could create an agent with the reliability needed for something like shopping or planning a wedding, but it's not commercially viable because it's expensive and takes too long on individual actions, needing to triple-check everything. Arvind: That would be super convincing. I don't think cost barriers will remain significant for long. Another: Ajeya: Here's one proposal for a concrete measurement — we probably wouldn't actually get this, but let's say we magically had deep transparency into AI companies and how they're using their systems internally. We're observing their internal uplift RCTs on productivity improvements for research engineers, sales reps, everyone. We're seeing logs and surveys about how AI systems are being used. And we start seeing AI systems rapidly being given deference in really broad domains, reaching team lead level, handling procurement decisions, moving around significant money. If we had that crystal ball into the AI companies and saw this level of adoption, would that change your view on how suddenly the impacts might hit the rest of the world? Another: Ajeya: …do you have particular experiments that would be informative about whether transfer can go pretty far, or whether you can avoid extensive real-world learning? Arvind: The most convincing set of experiments would involve developing any real-world capability purely (or mostly) in a lab — whether self-driving or wedding planning or drafting an effective legal complaint by talking to the client. From Deric Cheng (Convergence Analysis / Windfall Trust) Early indicators: he’s friends with the Metaculus folks. They’re working on indicators for AI diffusion and disempowerment. Don’t share: Metaculus Diffusion Index – they’ll publish in a few weeks. https://metr.substack.com/p/2025-07-14-how-does-time-horizon-vary-across-domains When monitoring progress in capabilities, need to watch for the possibility that capabilities are advancing on some fronts while remaining stalled on some critical attribute such as reliability, hallucinations, or adversarial robustness Nick Allardice has a prior that labor market disruption is not going to be meaningfully different from other times in history… but he’s highly uncertain. Evidence that might push him to believe that meaningfully different levels and pace of labor market disruption would [?]. The burden of proof is on this time being meaningfully different from the past. If AI gets good at something, we’ll focus on something else. He hasn’t seen enough evidence to shift his priors. Leading indicators currently reinforce his prior: unemployment is low. It’s harder to get a job as a junior developer, but not impossible, and mostly seems to be due to other factors. Even if capabilities advance, diffusion challenges will leave room for human workers. Our institutions aren’t going to turn everything over to AI. Offcuts qualitative rubrics [MPH is a good indicator because every M takes about the same number of H; cities passed per hour would break down in the Great Plains; MPH breaks down when you enter a city center] If we only measure high-level, downstream attributes such as practical utility, we won’t have any way of anticipating these twists and turns. As I noted in the introduction to the cruxes writeup, by the time Facebook started to show up as a major contributor to overall Internet usage, it was already well along its journey to global dominance. [Child Page: Seeing Past AGI; AI 2027 vs. AINT] Summary: Engage with Ryan Greenblatt and others on what I’m now calling “Seeing Past AGI”. Ryan keeps pointing out that there are important disagreements which won’t manifest until after AGI is reached, and which may be hard to shed light on until that point (which may be too late to be of much use). Working with Ryan and other usual suspects, I’d be interested in digging into this, try to clearly characterize the dramatic things Ryan expects to see happen post-AGI, and identify precursors which Ryan would agree ought to be visible pre-AGI. This might turn out to fit neatly into the existing AI Watersheds framework, or might turn into a bit of a separate sub-project. As we close in on the relevant milestones, we won’t resolve the first two bullets above [I think this meant the first two cruxes]. The difficult questions are not when X gets automated, it’s what happens afterwards. Can’t imagine any measurement that would disambiguate AINT for him. A lot of these metrics do shed light on whether / when we’ll get AGI, or automation of AI R&D. I have a hard time imagining updating toward anything like “full cheap automation of cognitive labor would increase GDP growth by <10%” I have a hard time imagining updating against large impacts of ASI in advance I’d like to push on this and try to come up with ways to forecast past the AGI event horizon – I have a conviction that we can do better than just throwing up our collective hands. My plan for a next step is to engage with, perhaps, Ryan and Sayash to really dig into their models of what happens post-AGI. [Abi: Let’s chat about different definitions of AGI that Sayash and Ryan have. I wonder whether getting very specific on this question will point to two very different visions here, which are leading to some of the divergence. + how we handle this in our approach.] (related: per https://blog.ai-futures.org/p/ai-as-profoundly-abnormal-technology, offer to meditate a conversation between Sayash+Arvind and the AI 2027 crew) If/when we pursue this: Think about Ryan’s comment about takeoff in the phase 2 doc. Come up with a plan for drilling in on takeoff models and early indicators. Work with Sayash and Ryan to drill in on disagreements post-AGI and then look for early indicators Abi: During our call, let’s chat about different definitions of AGI that Sayash and Ryan have. I wonder whether getting very specific on this question will point to two very different visions here, which are leading to some of the divergence. + how we handle this in our approach https://blog.ai-futures.org/p/ai-as-profoundly-abnormal-technology https://x.com/sayashk/status/1964016339690909847 [Ryan, under “measuring freedom of action”] It seems really hard to use indicators to distinguish between my perspective and one that predicts way less freedom of action at the point of ~full automation of AI R&D. Maybe the crux is mostly capabilities, but then it comes back to crux 1. Ryan: as we close in on the relevant milestones, we won’t resolve the first two bullets above. The difficult questions are not when X gets automated, it’s what happens afterwards. Can’t imagine any measurement that would disambiguate AINT for him. A lot of these metrics do shed light on whether / when we’ll get AGI, or automation of AI R&D. Delaying timelines by decades is a big deal, but doesn’t have a decisive effect on what I ultimately expect to happen. E.g., it’s like a factor of 3 on various things, not a factor of 10-100. More random takes from Ryan: I can imagine updating towards much longer timelines (though this is limited by some exogenous rate of large breakthroughs) I can imagine updating away from software-only singularity based on detailed empirical evidence about returns to compute vs labor within AI companies (especially if this evidence is coming in as we’re automating) Though idk how big this update would be. I can imagine updating towards moving through the human range slower than I currently expect via a variety of mechanisms. I have a hard time imagining updating toward anything like “full cheap automation of cognitive labor would increase GDP growth by <10%” I have a hard time imagining updating against large impacts of ASI in advance After this happens, I’d update, but this is too late. AI Scenarios Network – AI Watersheds Brainstorm Prioritize putting together a list of measurements, so I can ask for additional suggestions + then do a round of voting Turn phase 2 notes into a rough draft Early writeup Group brainstorm Steve’s Notes from the Phase 2 Brainstorming Doc Slide deck Intro Material for Phase 2 The importance of measurements that can be collected over a long period of time (won’t saturate), and ideally can be measured historically as well. [more ideas from the panel will belong here… using fine-grained data sets, collecting data from inside labs, using LLMs to analyze qualitative data, setting up controlled access to sensitive data sets, etc. Maybe these detailed ideas of “how to measure things” would belong in a separate section later on, and up here we’d just talk about basic philosophical ideas like “focus on real-world impact”.] https://ai-frontiers.org/articles/the-hidden-ai-frontier talks about how important developments may be hidden inside the frontier labs (and the dangers that poses) General Material for Phase 2 File for AI Watersheds, eg "Concretely, one reviewer proposed tracking deployments of AI agents that (i) are general-purpose systems, (ii) operate with minimal supervision, and (iii) handle tasks with a high cost of errors.” gavin leech (@g_leech_) posted at 4:00 AM on Thu, Nov 13, 2025:Glad somebody did this (expert interviews on why LLMs are not currently AGI, and why they could be)feat: @random_walker, @DKokotajlo, @ben_j_todd, @daniel_d_kang, @rohinmshah https://t.co/MuHjAoXvAw(https://x.com/g_leech_/status/1988939922842218558?t=OSiH9KzQDuAzCWo776O6Vg&s=03) @AI Security Institute: 📈 Today, we’re releasing our first Frontier AI Trends Report: evaluation results on 30+ frontier models from the past two years, showing rapid progress in chemistry and biology, cyber capabilities, autonomy, and more.▶️Read now: https://t.co/afJoJy0pYl pic.twitter.com/uIBmxQMNjJhttps://x.com/i/status/2001579052830953668 https://x.com/sayashk/status/1963343022252315112 https://epochai.substack.com/p/the-changing-drivers-of-llm-adoption Follow up on “AI Watersheds Phase 1 writeup” with Jonas Sandbrink; lots of good ideas in his email, and we should discuss further. https://x.com/EpochAIResearch/status/1996248575400132794 Herbie Bradley (@herbiebradley) posted at 5:13 AM on Thu, Nov 20, 2025:Looks like a very promising benchmark(https://x.com/herbiebradley/status/1991495140633141550?t=B2ssdv1dhSdNGdMxxOJ6TA&s=03) Might incorporate: https://arxiv.org/pdf/2510.07575v1 [Abi] Geopolitics + AGI team at RAND takeaways: This team also sees their goal as getting decisonmakers (not just gov) to think more about TAI. ON BOTTLENECKS TO ECON DATA: Their econ team said that lack of granular data constrains econ research. Example: o-net's list of tasks is great but needs more granularity and to be updated more. O-net surveys are only send sporadically to ~5 people per firm. Example: The Census collects longitudinal employer/employee data ("LAHD") but only 25 states opt in. Also, researchers need special status to use it. RAND has access but notes that it's not even that good because occupation data isn't linked to worker-firm datasets. This might be relevant to Watersheds! TO DOs: I will add a chat with RAND's AGI NatSec team to the Director of Event's onboarding doc. If desired in January, I can set up a chat with someone from RAND Econ team to talk to Taren or Steve about how to unlock from task-level or occupation-level data. https://epochai.substack.com/p/the-software-intelligence-explosion Are these trends bearing out? https://x.com/Hangsiin/status/1950645770346283083?t=2LsNFOIyCZy2Ar22eUR-iA&s=03 Are these cognitive limitations easing? https://leap.forecastingresearch.org/reports/wave2 Notes from Jonas: He sees the key disagreements as feasibility of ASI, and speed of diffusion (and will AGI increase or decrease barriers to adoption). [feasibility of ASI → potential bend in the curve, is this on our list?] If you extrapolate the METR curve, you’re probably not looking at AGI in 2028. The only way to get there soon is through some sort of speedup. So if you want to evaluate whether AGI is coming within a few years, you should be looking for signs of speedup. If you expect AGI in more like 2035, you should be looking at broader progress. How will we measure AI capabilities for multi day tasks? From https://digitaleconomy.stanford.edu/news/ai-and-labor-markets-what-we-know-and-dont-know/: One way to bolster evidence is to collect better data on when individual firms adopt AI (see more in point 4) to track employment changes before and after at the firm level, hopefully improving upon the measures in Humlum and Vestergaard (2025), Hosseini and Lichtinger (2025), and other work. Even better would be to find some kind of experiment in firm-level AI-adoption. An example would be an A/B test at an AI company that randomly offered discounts on subscriptions to different firms. Ideally the experiment would have been run starting in the early days of AI and run for months, if not years. It would be great to get actual large-scale data from AI labs on usage by occupation, perhaps via survey rather than relying on predictions based on conversations. More research should be done on other labor markets. Three promising avenues are to use Revelio or ADP in other countries, if feasible; use other private payroll data from other countries; or use government administrative data to track employment changes. Some infrastructure likely needs to be built out to measure AI exposure for local occupations. A particular area of focus should be countries with high levels of employment in exposed jobs such as call center operations. Further modeling can also help with predicting how impacts may vary across different institutional contexts. Ideally we would have some sort of continuous index of AI adoption, with differences in “how much” firms or workers have adopted AI. One option is to measure token counts, as suggested by Seed AI. Business spend data seems promising as well. Another option is the number of unique users or the number of conversations. We should encourage AI companies to share data on this to the extent feasible. Business surveys should also explore alternative questions and test how sensitive reported adoption rates are to the specific wording. Not related to cybersecurity, but we did a deep dive on agent benchmarks. Many of them are broken and measure AI agent performance poorly: Twitter/X: https://x.com/daniel_d_kang/status/1942641179461648629LinkedIn: https://www.linkedin.com/posts/daniel-kang-1223b343_ai-agent-benchmarks-are-broken-activity-7348406954253312000-B_Yq/Substack: https://ddkang.substack.com/p/ai-agent-benchmarks-are-broken Remotelabor.ai to track the new Remote Labor Index, measuring what percentage of remote work AI can automate. Currently the top score is 2.5%, so ‘not much,’ but that’s very different from 0%. Diffusion: https://www.convergenceanalysis.org/fellowships/spar-economics/decoding-ai-diffusion-mapping-the-path-of-transformative-ai-across-industries Ethan Mollick (@emollick) posted at 5:09 PM on Thu, Sep 25, 2025:After reading it, this does seem like a big dealIndustry experts outlined important, real-world, hard tasks for AI to do. Other experts were asked to do the tasks themselves & yet others graded human & AI outputModels approached parity with humans & AI is getting better fast. https://t.co/z666YcNyH6(https://x.com/emollick/status/1971366497244348625?t=suZJLFoe4S9wZoSROiNeKw&s=03) OpenAI (@OpenAI) posted at 9:24 AM on Thu, Sep 25, 2025:Today we’re introducing GDPval, a new evaluation that measures AI on real-world, economically valuable tasks.Evals ground progress in evidence instead of speculation and help track how AI improves at the kind of work that matters most.https://t.co/uKPPDldVNS(https://x.com/OpenAI/status/1971249374077518226?t=aXFDp9V1lvUVuMBJOotUFA&s=03) Lawrence H. Summers (@LHSummers) posted at 9:36 AM on Thu, Sep 25, 2025:A research team at @OpenAI, where I am proud to be a board member, released an important new paper today. This paper looks at what might be thought of as task specific Turing Tests and shows that AI systems, even with limited guidance, perform many tasks -- such as planning(https://x.com/LHSummers/status/1971252567981146347?t=dnQgGIFT7yFz-Ex3KHVhpA&s=03) Sayash Kapoor (@sayashk) posted at 1:54 PM on Wed, Oct 15, 2025: 📣New paper: Rigorous AI agent evaluation is much harder than it seems. For the last year, we have been working on infrastructure for fair agent evaluations on challenging benchmarks. Today, we release a paper that condenses our insights from 20,000+ agent rollouts on 9 https://t.co/TvSxUsptdW (https://x.com/sayashk/status/1978565190057869344?t=sQeizZZnd9uOH-Chjv8e5A&s=03) Dan Hendrycks (@DanHendrycks) posted at 7:20 AM on Thu, Oct 16, 2025:Our definition of AGI is an AI that can match or exceed the cognitive versatility and proficiency of a well-educated adult.To measure this, we assess the multiple dimensions of intelligence derived from the most empirically validated model of human intelligence (CHC theory). https://t.co/e1wEkzmHwb(https://x.com/DanHendrycks/status/1978828383581561009?t=b6nlhd1Uh6Xj57adoDnzEQ&s=03) Ethan Mollick (@emollick) posted at 10:24 AM on Thu, Oct 16, 2025:A lot I like & some I don’t in this paper:Like: Clear definition of AGI, diverse authors, shows jaggedness, tracking metrics over time (huge leap from GPT-4 to GPT-5)Dislike: AGI defined as replicating a model of human cognition, benchmarks are scattershot, narrow view of AI https://t.co/T3XOu2PVl8(https://x.com/emollick/status/1978874737892667718?t=hWtw4waaXLy-Xa4djFU-iQ&s=03) Sayash Kapoor (@sayashk) posted at 10:12 AM on Fri, Sep 12, 2025:Agent benchmarks lose *most* of their resolution because we throw out the logs and only look at accuracy.I’m very excited that HAL is incorporating @TransluceAI’s Docent to analyze agent logs in depth.Peter’s thread is a simple example of the type of analysis this enables,(https://x.com/sayashk/status/1966550402129592738?t=SX4UR2z0FabBX_mgLc98dw&s=03) The Point Magazine (@the_point_mag) posted at 6:05 AM on Thu, Oct 16, 2025:New online, @saffronhuang on what it means to measure intelligence—in large language models and in us:https://t.co/AKZdHiyzGE(https://x.com/the_point_mag/status/1978809403609382977?t=pzJVLmXj6ZGb3k88X648qQ&s=03) At the CAIS event on Oct. 2, someone (Dan?) mentioned that they’d be posting an AI Automation Index in a few weeks Alex Tamkin (Anthropic Economic Index coauthor): data sources: model usage data (can't be longitudinal because don't keep logs, also confounded by changing models), government (states might be a good source if can't get federal help), downstream apps, Stripe. He'd love to see interviews, eg of hiring managers. Bharat: someone in his group wanted to know the capital vs labor contribution to AI R&D, would be helpful in calibrating the model, that's the missing variable for his model. Miles Brundage: How confident is he in short timelines? Pretty confident. He’s typical of people who have spent multiple years at a frontier AI company and lived through / closely watched / participated in multiple scaleups, going from “signs of life” to maturity stage for image generation, codegen, video gen, writing, math. We’re still early on RL, and even pretraining – for instance, we’ve barely scratched the surface on video data (YouTube). On fuzzy vs. tidy problems: he views these as differences of degree vs. kind. E.g. there’s a lot of positive spillover from math RL to code, and code RL to writing or policy research. There’s very little data relating to papers he writes in OpenAI’s RL, but chain of thought induces useful skills (such as breaking problems down into parts, checking your work)… that makes it useful for working on his papers? It just takes intellectual labor to turn a non-verifiable task into something you can test and verify. [Josh 6/9/25] My colleague Alexa (cc'd) put together some initial ideas for forecasting questions that could help to further specify and concretize some of the cruxes described in your post. She gives more context on the approach in the summary of her document. Would you be interested in incorporating a revised version of any of these questions into your work, or possibly trying to get forecasts on them as part of your research? We'd also be happy to help with collecting forecasts if you'd find that valuable. Dean W. Ball (@deanwball) posted at 9:09 AM on Sun, Sep 14, 2025:I think Demis is fundamentally correct here. The current systems are extremely impressive, and will get much more so soon, but it’s clear there are fundamental breakthroughs still needed.As I have written before, I expect us to get “superintelligence” (AI systems that can, say,(https://x.com/deanwball/status/1967259417029837122?t=slJA0l1BoEy3WQdOJzlUGQ&s=03) Dean W. Ball (@deanwball) posted at 5:09 AM on Tue, Sep 16, 2025:If this mirrors anything like the experience of other frontier lab employees (and anecdotally it does), it would suggest that Dario’s much-mocked prediction about “AI writing 90% of the code” was indeed correct, at least for those among whom AI diffusion is happening quickest.(https://x.com/deanwball/status/1967923900685386222?t=FztCAYh5PbN51JLbzAcGUA&s=03) Steven Adler @ Progress Conference 2025 Talked about looking at UpWork task mix, prices, etc. as a signal Talked about building evals around more open ended real work tasks. 1a3orn (@1a3orn) posted at 4:20 PM on Sat, Oct 25, 2025:data from OpenAI / Anthropic that I wish I had, but do not:1. What percent of Transformer improvements in OAI / Anthropic are original to the company, and what percent come from outside?2. What "Constitutional principles" does Anthropic currently use for alignment?(https://x.com/1a3orn/status/1982225866470899728?t=BxaB8hw-2u-HkAGBlH93bg&s=03) https://thezvi.substack.com/p/asking-some-of-the-right-questions Could talk to Gabe Weil, who mentioned that Basil Halperin argues that AGI should raise interest rates. https://newsletter.forethought.org/p/how-quick-and-big-would-a-software https://tecunningham.github.io/posts/2025-09-19-transformative-AI-notes.html Incorporate ideas from my chat with Jaime Will investor confidence continue to support scaling of compute / training budgets? Would love to have more visibility into the revenue chain, how solid the demand is, and where the room for short- and long-term growth comes from. We talked about my question about how revenues flow through the AI value chain and how much of this is speculative vs. committed users who are experiencing values. Jaime noted that Anthropic is very dependent on coding tools. Investors are ready to fund roughly 3 years of burn, so perhaps 10x ARR (at current growth rates). To keep up this rate of growth, it’s necessary to keep expanding into new markets. What’s the penetration rate of coding tools? Jaime is surprised how low… anecdotally, 6 months ago talking to random developers in Spain and Mexico, no one is really using these tools professionally. Areas he expects to see impact soon: accounting, customer service, legal work, finance, assistants / operations, market research analysts. Seems like there’s plenty of room here for a couple more years of revenue growth at the current rate. What would be useful for looking more than a couple of years ahead? He likes to take an outside view, look at what other markets are exposed. A colleague went through the OINT (?) database and made a list of exposed occupations. They could do this at larger scale. What future capabilities will be needed? Can guess, e.g. much better computer use. His big disagreement with AI 2027 is around returns to intelligence and returns to parallelization of research. He doesn’t foresee nearly the same degree of benefit from running lots of small experiments in parallel. One thing that has shaken his beliefs on returns to intelligence is the insane amount that companies are willing to pay their top researchers. Difficult to interpret exactly what that means, but might suggest returns to intelligence. https://blog.cip.org/p/notes-on-building-collective-intelligence Pass @ kitchen sink: https://www.notion.so/Todo-5cbf5bb74635457381c2f814628c73f9 Notes from Jason Clinton’s talk at The Curve 2025 (John Hart may have more); could incorporate these into metrics of internal usage at the labs and how this differs from other orgs: Anthropic has automated level 1 SOC analysts: reviewing alerts to decide if they require action. 2-3 months from automating tier 2: deciding which alerts are too noisy. Blocked on visual reasoning??? Human code review is 25% effective at catching security bugs. AI could be better. AI reviewer writes a repo before reporting the issue to a person. An AI is greenlighting 60% of design docs as being low risk with no need for human security review. John’s notes 90% of code at Anthropic is written by Claude; they have eliminated all junior dev roles from their open jobs list "vibe hacking" - 3 weeks ago a "low-skill russian hacker" used Claude to hack people - https://www.bbc.com/news/articles/crr24eqnnq9o DARPA just concluded an "AI Cyber Challenge" - https://aicyberchallenge.com They have a specialized Claude agent that focuses on flaky tests. If it gets stuck / can't un-flake a test, it will "reach out" to a "Claude SRE" agent to see if it's infra-related. Likewise they have a specialized Claude agent that just does security review of design proposals; this has offloaded some routine work from their principals. It compares design docs against all known MITRE attacks. Claude does all Tier 1 SOC analyst work; only when it raises to Tier 2 level (arbitrating noisy alerts, for example) does a human get in the loop "Literally everyone is working on memory". Multiple startups will be offering "virtual employees" (presumably w/ long-term context memory) starting in April or May of next year. Responsible disclosure timelines are woefully out-of-date in the AI era. https://ai-frontiers.org/articles/the-hidden-ai-frontier talks about how important developments may be hidden inside the frontier labs (and the dangers that poses) Per discussion with Nick Allardice, impact will play out very differently (and more slowly) in the global south. Research priorities I’m excited about this. I think our “neutral/switz” angle can help with this. On the collective action, could potentially frame it in labs’ interest as a way to get better forecasts, maybe pair with some of the platforms where leading labs already are i.e. Coalition for Secure AI, maybe GPAI. Propose some specific initiatives. Emphasize projects that require collective action, such as: A large effort to collect a valuable data set which would be useful for multiple research projects Collecting data that requires cooperation from AI labs or other private sources, because the data is sensitive and/or requires effort for the private actor to supply. Collective action may be needed to pressure the labs into cooperating, and/or to create a high-trust context in which appropriate safeguards can be provided (controlled access to data). === Steve’s Notes from the Phase 2 Brainstorming Doc (https://www.notion.so/2641a29f1789802f9e0cfcc9eccabb24) === Brainstorming doc Why Do Measurements? Thoughts on the purpose of forecasting AI’s schedule General trend visibility for several years (important for lots of policy + other questions) At least several years notice of transformational change Estimate likely steepness & extent of transformational change Ask panel for feedback on this Josh: interested in the connection between current policy questions and our questions here. Would be great to get more specific / explicit about the connections with policy questions. Josh: interested in the connection between current policy questions and our questions here. Would be great to get more specific / explicit about the connections with policy questions. General thoughts on what to measure Practical utility is a function of both AI model capability and complementary app innovation (coding models vs. Cursor) Approaches to measurement Josh (+’d by Helen, Nikola, Sayash): Noting cross-cutting theme: Generally seems like more high-quality surveys of companies would be valuable. A lot of ideas involve sensitive data, so would require controlled access (and/or governmental involvement?) Using AIs to qualitatively grade lots of screen recordings and other messy high-volume data Find ways of calibrating / correlating benchmark scores with measurements we care about more directly? (METR time horizons being an example in this direction) Generally use lots of different measurements and try to map them to common units for cross-checking Surveys / prediction markets; focus real-world measurements (e.g. uplift studies) on areas of disagreement Or get both sides of a dispute to agree on a cheaper proxy measurement Case studies How firms in various sectors are using AI; scale of tasks assigned to AI / nature + granularity of interaction with humans; degree of trust / freedom of action given to AI Deep study of revenue / spending / profit across the AI value chain (and by domain) Also: number of “complementary innovation” (Sayash’s term) AI-powered apps, adoption rate, # of customers, “Surveys of customers to see how they are improving their workflows”, etc. Helen: something about revenue from AI more broadly vs. revenue of frontier model providers. What can we say about open models eating into OpenAI / Google / Anthropic revenue? Broad metrics (a la Epoch), e.g. electricity used by AI, global private investment Keep pushing benchmarks toward greater realism of both input (task specification + context) and output (not just “passes tests”, but “PR is mergeable” / other measures of quality) Nikola: need more realistic benchmarks. Source PRs (ideally from inside AI labs) and test AIs on them. Analysis of usage at the model level (Anthropic economic AI index) Logs of agent usage i.e. usage of model context protocol servers, pulls [Sayash] Analysis of AI logs from them being asked to solve various tasks (we’re doing this with 10B tokens of HAL agent log data, but could imagine increasing many fold) Social media is subject to transparency measures, AI companies currently are not. It’d be great to have controlled researcher access to e.g. ChatGPT transcripts. Lots of general data sets for public analysis AIs really like to email Eliezer, that’s an indicator of various types of strange AI behavior. Ajeya: there was an NYTimes article about a journalist getting similar emails. Could there be a “weird AI shit” incident tracker that would categorize reports + report trends? Josh: MIT AI incident repository? Populated from public reports – links to existing reports. Ajeya: hard to make sense of it. Needs to be a larger team sifting the signal from the noise + writing blog posts. Doesn’t track veracity. AI Risk repository / incidents database is relevant here: https://airisk.mit.edu/, also https://incidentdatabase.ai/ Ajeya comment: Too hard to make sense of current databases. Need someone to do better synthesis. Make an open source AI company where the company itself is open source. Have it try to use AI as much as possible, make all its logs about everything (Slack etc) public, study that. Sayash: there’s been a lot of work on the science of AI evaluations. We don’t know the shape of the DAG, how AI inputs translate into outputs. We could do more to distinguish between different DAGs – which is a core source of disagreement, for instance Sayash has a very different DAG in mind. Inputs might include things like data, compute; output measured in the real world. Econometrics. Josh: Meta measurements: Big increase in expected OpenAI revenue, according to a prediction market, based on a reported algorithmic breakthrough Pushback from Ryan+Ajeya: limitations of prediction markets mean it’ll be tough for them to pick this up reliably Surveys at frontier AI companies Surveys of the broader workforce Data center construction spend → currently dependent on media reporting, is there some way to get better/more consistent data on this? Trying to unpack inputs/outputs per Sayash’s point about DAGs/understanding how improvements in one element flow through to improvements in other elements Lots of (different types of) surveys Helen: important to track military applications (broadly – decision support, targeting, etc.). Important to track freedom of action here. Measuring Utility Metrics about fluid intelligence, messy tasks Benchmarks for domains vs. skills Domains: cyber/law/professional labor markets Skills: Context awareness, reliability, long-context reasoning, sample efficiency Need deep info for each sector before assessing results. I.e. persuasion surveys overestimate results because opinion change dissipates 6 weeks later. I like Steve’s note that AI can continually be used cheaply. Uplift RCTs in real-world settings, measuring real-world outcomes (ideally: profit!) e.g. if you get an actual large company to actually do randomized staged rollout of enterprise LLMs that’s great Ajeya: three-armed uplift trials, AI / human / cyborg. Josh: Persuasion: Cost for AI to swing a vote, relative to other leading interventions (Josh: lots of polysci studies here) AI-bio: % of amateurs that can synthesize flu Look for leapfrog companies that get on board with a new technology ahead of the incumbents, that could highlight utility ahead of adoption. e.g. fintech cos using AI in ways that let them overtake traditional banks, or healthcare startups using AI in ways that let you bypass doctors’ offices Power user case studies – screen recordings + deep interviews of people who (think they) are getting enormous uplift Organization version of this: find those YC startups that supposedly do 95% of everything with AI, study them (“power organizational users”) Field trials (a la AI Village) and realistic benchmarks Do a time horizon analysis (a la METR), but on real-world usage? Also, qualitative analysis of the size + difficulty + nature of tasks successfully delegated to AI (and dig into ranges of “successfully”) Building on Josh’s work test idea: could a METR-like org collect work test setups from a range of places (under NDA so as not to break the work tests) and centralize the work of eliciting good performance, figuring out measurement, reporting results, etc? Ryan: I think uplift will be messy (because of humans being messy and possibility for phase changes in how AI should be used productively), so I’m more into end-to-end automation of actual tasks. Also, we ultimately care most about full automation regimes and predicting when this will happen. Uplift studies correspondingly seem somewhat worse than looking at full automation of tasks of varying difficulty/size (and benchmark transfer to actual things people are doing in their job so we don’t need to constantly run these tests). Maybe you can get AI companies to run semi-informal experiments with uplift internally and publish results: many companies have effectively committed to doing this eventually due to their safety policies. I think Anthropic might open to this, but there are various sources of trickyness here. Early reports of AI doing remarkable things (e.g. major scientific discovery / insight; solving a Millennium Prize Problem in mathematics) pass@any for existing benchmarks can predict pass@1 in the future Helen: where is the ceiling on various capabilities (headroom)? This seems under-investigated. Sayash: are most tasks like chess, or like writing? (Ryan: unclear whether writing is the right contrast here.) Are there less controversial examples than writing? Taking out the trash: saturates. Helen: the writing example highlights we’re way under-theorized on this. Quality of parenting is another example. “Persuasion” is poorly specified. Abi: a scientific breakthrough requires an insight that goes against dogma. Could be considered an aspect of very high-quality writing. Ryan: I care most about questions like “will energy production 100x within 5 years of full AI R&D automation if people want this to happen”. Seems like this has many possible routes in terms of capabilities head room. Sayash: three broad categories: computational limits (chess), intrinsic limits / saturation (writing?), and knowledge limits (need new breakthroughs to exceed past performance). Better AI can help with computational limits, but not intrinsic limits. Work tests for orgs in the space: GiveWell lit review test Open Philanthropy work tests High-quality summaries of a conversation (AI is still failing our first-round work test at FRI about this) Abi: “I liked how [this] list broke down sectors as each having their separate indicators. Very societal frictions approach.” Measuring Adoption / Impact YouTube resulted in some homeowners not calling plumbers for small jobs. What are the AI equivalents, and how should they be measured? Compare metrics across AI-feasible vs. AI-infeasible sectors: employment (incl. hiring patterns + plans), productivity, profit, qualitative measures of AI usage, … Revenue of AI service providers Studies of AI impact across various fields What percent of sales of the U.S. pharmaceutical industry is generated by AI-discovered drugs and products derived from these? What percent of publications in the fields of Chemistry, Physics, Materials Science, and Medicine will be ‘AI-engaged’ as measured in this study or a replication study, in 2030? How many hours per week on average will K-12 students in G7 countries use AI-powered tutoring or teaching tools, as reported by their school systems or education ministries? A September 2024 study by the Federal Reserve Bank of St. Louis, based on the Real-Time Population Survey (N=3,216), estimated that 0.5%–3.5% of all U.S. work hours were assisted by generative AI. Measuring Diffusion Ajeya: I want to know about adoption within AI companies, so e.g. surveys about how much compute they’re running for inference, freedom of action related surveys like the ones suggested below, internal estimates of how much AIs speed them up, etc. (Also, I don’t think I literally care only about AI companies, will prob also care about USG.) Measuring Freedom of Action Surveys of workers - how often do you have to approve an AI decision, how much thought do you put into each approval (e.g. level of review / pass rate of AI pull requests) Monitoring progress on agent infrastructure → if AI agents are operating online, are there meaningful constraints on them? Does e.g. Cloudflare have any meaningful controls, or is it all a chaotic mess? Something military or military-adjacent - reports of smaller militaries using autonomous/semi-autonomous systems..? We can also make a big list of discrete flags and ask people about them, e.g. “Is your AI allowed to spend money on business expenses? Up to how much?” or “Is your AI allowed to search the internet freely in the course of completing a task?” or “Are AIs at your company allowed to talk to employees other than the human that started the task in the course of completing their task? What about external people?” White-hat reports of prompt injections paid out by bug bounties at the top (100) websites across domains over time (as a proxy for how many applications allow AI systems to take actions that can exfiltrate user data) Black-hat exploits of AIs (only possible if AIs are in a position to take important actions without adequate review?) Curve-Bending Mechanisms Note that the impact of AI can be difficult to predict. E.g. the prediction of AI flooding the zone with misinformation in the 2024 election didn’t pan out. Sycophancy → mental health impact, conversely, was an event that wasn’t widely predicted. Intelligence explosion (software and/or hardware) Measure both initial speedup, and whether r > 1 Surveys of AI companies on internal AI use (bunch of ideas for granular questions, e.g. subj sense of speedup, what tasks they’re used for now, what tasks aren’t they used for, hypothetical questions) Internal measures of the absolute rate of algorithmic progress at AI companies (e.g. how much compute does it take to get GPT-4-level perf), watch for that trend accelerating Source Ryan: Get people who have done the takeoff modeling to look at the data on algo progress (both historical from humans and data under automation regime). Unclear if this data will resolve disagreements, but seems like it can help. (Tom, Daniel, maybe some people from Epoch) Algorithmic breakthrough / new approach (e.g. neuralese, in-context learning, long-term memory, recurrence, brain emulation) (either known or unknown unknown) Study new model releases to look for signatures of known ideas, or look for an increase in papers on some approach Discontinuous jumps across a range of difficult benchmarks Could ask companies key questions like e.g. “are you still using English CoT?” and make them report Survey expectations at frontier labs (also covers intelligence explosion) Analysis attributing performance gains to different sources (e.g. model size scaleup, data quality improvement, RL, etc) Ryan: really big breakthroughs are like, AlexNet, GPT-1 / scale-up, scaling RL (o1, o3). Easy to notice something is turning into a big deal, hard to tell where on the Richter scale it will land. Could track the current 5 most promising potential trend-breakers. Ajeya: track breakdown of capabilities improvements over time: how much is from scaling pretraining, scaling RL posttraining, etc. Threshold effects / phase changes (e.g. the moment when people are no longer needed; maybe time horizon scaling suddenly becomes much easier) Phase changes in behavior of AI Village or other observations of AI capability / usage in real or realistic contexts Phase change in qualitative analysis of uplift trials Some resource (data, compute, electricity, intelligence headroom) is exhausted Note, may lead to workarounds rather than an end to progress. E.g. pretraining data gets harder to scale → scale RL instead. Hard to predict the effect. Low-hanging fruit is exhausted; training for long-horizon tasks is expensive Some important missing capability doesn’t yield to scaling + progress (e.g. sample-efficient learning, “judgement”, long-time-horizon skills) Economics of AI not working out → slowdown in investment AI winter – we run out of ideas for improving AI Josh: FRI expert panel. Could ask them “what’s the probability of an AI winter in the next 5 years”? External event (market downturn; war in Taiwan; public backlash / disaster → regulation; focused national effort / Manhattan Project) Track public sentiment and other political indicators Applications of models are a lagging indicator to model capabilities. First sign of a slowdown would be the time horizon graph slowing down. Other Nikola: if timelines are long enough to enable human emulations (Ems), that would make a difference, because we’d have been able to explore more implications. More generally, does other transformative tech come first. Possible techs: EMs, nanotech (unlikely), genetic engineering (seems like this could radically transform society in ~30 years if heavily invested in now, but in practice might not happen because of e.g. legal/regulatory blockers). === External funding for privately held AI companies raising above $1.5 million - Our World in Data (https://ourworldindata.org/grapher/private-investment-in-artificial-intelligence) === Our World in Data Browse by topic Data Insights Resources About Subscribe Donate Data External funding for privately held AI companies raising above $1.5 million See all data and research on: Artificial Intelligence Explore the Data Sources & Processing Reuse This Work Data sources and processing Insights about this data Data sources and processing Related articles Related charts Insights about this data Related articles Related charts What you should know about this indicator This data focuses on external private-market investment, such as venture-capital and private-equity deals. It does not include internal corporate R&D, capital expenditure (CapEx), or public-sector funding. Publicly traded companies, including large tech firms, are excluded. Because this data covers only one form of financing, it underestimates total global spending on AI. Large single deals can cause spikes in specific years. Broader economic conditions (interest rates, investor sentiment) can also drive changes that are not specific to AI. AI firms are identified by the source based on keywords and industry tags; some misclassification is possible. External funding for privately held AI companies raising above $1.5 million Money put into privately held AI companies that raised more than $1.5 million from private investors. This excludes publicly traded companies (e.g., public Big Tech companies) and companies’ own internal spending, such as R&D or infrastructure. Expressed in US dollars, adjusted for inflation. Source Quid via AI Index Report (2025); U.S. Bureau of Labor Statistics (2025) – with major processing by Our World in Data Last updated April 8, 2025 Next expected update April 2026 Date range 2013–2024 Unit constant 2021 US$ What you should know about this indicator This data focuses on external private-market investment, such as venture-capital and private-equity deals. It does not include internal corporate R&D, capital expenditure (CapEx), or public-sector funding. Publicly traded companies, including large tech firms, are excluded. Because this data covers only one form of financing, it underestimates total global spending on AI. Large single deals can cause spikes in specific years. Broader economic conditions (interest rates, investor sentiment) can also drive changes that are not specific to AI. AI firms are identified by the source based on keywords and industry tags; some misclassification is possible. External funding for privately held AI companies raising above $1.5 million Money put into privately held AI companies that raised more than $1.5 million from private investors. This excludes publicly traded companies (e.g., public Big Tech companies) and companies’ own internal spending, such as R&D or infrastructure. Expressed in US dollars, adjusted for inflation. Source Quid via AI Index Report (2025); U.S. Bureau of Labor Statistics (2025) – with major processing by Our World in Data Last updated April 8, 2025 Next expected update April 2026 Date range 2013–2024 Unit constant 2021 US$ Sources and processing This data is based on the following sources Quid via AI Index Report – AI Index Report The AI Index Report tracks, collates, distills, and visualizes data related to artificial intelligence (AI). The mission is to provide unbiased, rigorously vetted, broadly sourced data to enable policymakers, researchers, executives, journalists, and the general public to develop a more thorough and nuanced understanding of the complex field of AI. Retrieved on April 8, 2025 Retrieved from https://aiindex.stanford.edu/wp-content/uploads/2024/04/HAI_2024_AI-Index-Report.pdf Citation This is the citation of the original data obtained from the source, prior to any processing or adaptation by Our World in Data. To cite data downloaded from this page, please use the suggested citation given in Reuse This Work below. Nestor Maslej, Loredana Fattorini, Raymond Perrault, Yolanda Gil, Vanessa Parli, Njenga Kariuki, Emily Capstick, Anka Reuel, Erik Brynjolfsson, John Etchemendy, Katrina Ligett, Terah Lyons, James Manyika, Juan Carlos Niebles, Yoav Shoham, Russell Wald, Tobi Walsh, Armin Hamrah, Lapo Santarlasci, Julia Betts Lotufo, Alexandra Rome, Andrew Shi, Sukrut Oak. “The AI Index 2025 Annual Report,” AI Index Steering Committee, Institute for Human-Centered AI, Stanford University, Stanford, CA, April 2025 The AI Index Report tracks, collates, distills, and visualizes data related to artificial intelligence (AI). The mission is to provide unbiased, rigorously vetted, broadly sourced data to enable policymakers, researchers, executives, journalists, and the general public to develop a more thorough and nuanced understanding of the complex field of AI. Retrieved on April 8, 2025 Retrieved from https://aiindex.stanford.edu/wp-content/uploads/2024/04/HAI_2024_AI-Index-Report.pdf Citation This is the citation of the original data obtained from the source, prior to any processing or adaptation by Our World in Data. To cite data downloaded from this page, please use the suggested citation given in Reuse This Work below. Nestor Maslej, Loredana Fattorini, Raymond Perrault, Yolanda Gil, Vanessa Parli, Njenga Kariuki, Emily Capstick, Anka Reuel, Erik Brynjolfsson, John Etchemendy, Katrina Ligett, Terah Lyons, James Manyika, Juan Carlos Niebles, Yoav Shoham, Russell Wald, Tobi Walsh, Armin Hamrah, Lapo Santarlasci, Julia Betts Lotufo, Alexandra Rome, Andrew Shi, Sukrut Oak. “The AI Index 2025 Annual Report,” AI Index Steering Committee, Institute for Human-Centered AI, Stanford University, Stanford, CA, April 2025 U.S. Bureau of Labor Statistics – US consumer prices The Bureau of Labor Statistics reports the monthly Consumer Price Index (CPI) of individual goods and services for urban consumers at the national, city, and state levels. CPI is presented on an annual basis, which we have derived as the average of the monthly CPIs in a given year. Retrieved on April 12, 2025 Retrieved from https://www.bls.gov/data/tools.htm Citation This is the citation of the original data obtained from the source, prior to any processing or adaptation by Our World in Data. To cite data downloaded from this page, please use the suggested citation given in Reuse This Work below. U.S. Bureau of Labor Statistics The Bureau of Labor Statistics reports the monthly Consumer Price Index (CPI) of individual goods and services for urban consumers at the national, city, and state levels. CPI is presented on an annual basis, which we have derived as the average of the monthly CPIs in a given year. Retrieved on April 12, 2025 Retrieved from https://www.bls.gov/data/tools.htm Citation This is the citation of the original data obtained from the source, prior to any processing or adaptation by Our World in Data. To cite data downloaded from this page, please use the suggested citation given in Reuse This Work below. U.S. Bureau of Labor Statistics How we process data at Our World in Data All data and visualizations on Our World in Data rely on data sourced from one or several original data providers. Preparing this original data involves several processing steps. Depending on the data, this can include standardizing country names and world region definitions, converting units, calculating derived indicators such as per capita measures, as well as adding or adapting metadata such as the name or the description given to an indicator. At the link below you can find a detailed description of the structure of our data pipeline, including links to all the code used to prepare data across Our World in Data. Read about our data pipeline Notes on our processing step for this indicator Reporting a time series of AI investments in nominal prices would make it difficult to compare observations across time. To make these comparisons possible, one has to take into account that prices change (inflation). It is not obvious how to adjust this time series for inflation, and our team discussed the best solutions at our disposal. It would be straightforward to adjust the time series for price changes if we knew the prices of the specific goods and services purchased through these investments. This would make it possible to calculate a volume measure of AI investments and tell us how much these investments bought. But such a metric is not available. While a comprehensive price index is not available, we know that the cost of some crucial AI technology has fallen rapidly in price. In the absence of a comprehensive price index that captures the price of AI-specific goods and services, one has to rely on one of the available metrics for the price of a bundle of goods and services. Ultimately, we decided to use the US Consumer Price Index (CPI). The US CPI does not provide us with a volume measure of AI goods and services, but it does capture the opportunity costs of these investments. The inflation adjustment of this time series of AI investments, therefore, lets us understand the size of these investments relative to whatever else these sums of money could have purchased. Reuse this work All data produced by third-party providers and made available by Our World in Data are subject to the license terms from the original providers. Our work would not be possible without the data providers we rely on, so we ask you to always cite them appropriately (see below). This is crucial to allow data providers to continue doing their work, enhancing, maintaining and updating valuable data. All data, visualizations, and code produced by Our World in Data are completely open access under the Creative Commons BY license . You have the permission to use, distribute, and reproduce these in any medium, provided the source and authors are credited. Citations How to cite this page To cite this page overall, including any descriptions, FAQs or explanations of the data authored by Our World in Data, please use the following citation: “Data Page: External funding for privately held AI companies raising above $1.5 million”, part of the following publication: Charlie Giattino, Edouard Mathieu, Veronika Samborska, and Max Roser (2023) - “Artificial Intelligence”. Data adapted from Quid via AI Index Report, U.S. Bureau of Labor Statistics. Retrieved from https://archive.ourworldindata.org/20251119-141827/grapher/private-investment-in-artificial-intelligence.html [online resource] (archived on November 19, 2025). How to cite this data In-line citation If you have limited space (e.g. in data visualizations), you can use this abbreviated in-line citation: Quid via AI Index Report (2025); U.S. Bureau of Labor Statistics (2025) – with major processing by Our World in Data Full citation Quid via AI Index Report (2025); U.S. Bureau of Labor Statistics (2025) – with major processing by Our World in Data. “External funding for privately held AI companies raising above $1.5 million” [dataset]. Quid via AI Index Report, “AI Index Report”; U.S. Bureau of Labor Statistics, “US consumer prices” [original data]. Retrieved January 8, 2026 from https://archive.ourworldindata.org/20251119-141827/grapher/private-investment-in-artificial-intelligence.html (archived on November 19, 2025). Our World in Data is free and accessible for everyone. Help us do this work by making a donation. Donate now Our World in Data is a project of Global Change Data Lab , a nonprofit based in the UK (Reg. Charity No. 1186433). Our charts, articles, and data are licensed under CC BY , unless stated otherwise. Tools and software we develop are open source under the MIT license . Third-party materials, including some charts and data, are subject to third-party licenses. See our FAQs for more details. Explore Topics Data Insights Resources Latest SDG Tracker Teaching with OWID About About Us Organization Funding Team Jobs FAQs RSS Feeds Research & Writing Data Insights Follow us Privacy policy Legal disclaimer Grapher license === The MIT AI Risk Repository (https://airisk.mit.edu/) === Risks IncidentS Mitigations Governance Blog TEAM What are the risks of Artificial Intelligence? A comprehensive living database of over 1700 AI risks categorized by their cause and risk domain Get updates Explore database What is the AI Risk Repository? The AI Risk Repository has three parts: The AI Risk Database captures 1700+ risks extracted from 74 existing frameworks and classifications of AI risks The Causal Taxonomy of AI Risks classifies how, when, and why these risks occur The Domain Taxonomy of AI Risks classifies these risks into 7 domains and 24 subdomains (e.g., “False or misleading information”) The repository is part of the MIT AI Risk Initiative, which aims to increase awareness and adoption of best practice AI risk management across the AI ecosystem. How can I use the Repository? The AI Risk Repository provides: An accessible overview of threats from AI A regularly updated source of information about new risks and research A common frame of reference for researchers, developers, businesses, evaluators, auditors, policymakers, and regulators A resource to help develop research, curricula, audits, and policy An easy way to find relevant risks and research AI Risk Database The AI Risk Database links each risk to the source information (paper title, authors), supporting evidence (quotes, page numbers), and to our Causal and Domain Taxonomies. You can experiment with a preview version of the database in the embed below, or copy the full database on Google Sheets , or OneDrive . Watch our explainer video on YouTube for a walkthrough of the database and how to use it. Causal Taxonomy of AI Risks The Causal Taxonomy of AI risks classifies how, when, and why an AI risk occurs. View the Causal Taxonomy on a single page Read our research report for more detail on how the Taxonomy was constructed and what it reveals about risks from AI Explore the taxonomy in the figure below Entity AI : Due to a decision or action made by an AI system Human : Due to a decision or action made by humans Other : Due to some other reason or ambiguous Intent Intentional : Due to an expected outcome from pursuing a goal Unintentional : Due to an unexpected outcome from pursuing a goal Other : Without clearly specifying the intentionality Timing Pre-deployment : Before the AI is deployed Post-deployment : After the AI model has been trained and deployed Other : Without a clearly specified time of occurrence Get a quick preview of how we group risks by causal factors in our database. Search for one of the causal factors (eg 'pre-deployment') to see all risks categorized against that factor. For more detailed filtering and to freely download the data, explore the full database . Domain Taxonomy of AI Risks The Domain Taxonomy of AI Risks classifies risks from AI into seven domains and 24 subdomains. View the Domain Taxonomy on a single page Read our research report for more detail on how the Taxonomy was constructed and what it reveals about risks from AI Explore the taxonomy in the interactive figure below 1. Discrimination & Toxicity Risks related to unfair treatment, harmful content exposure, and unequal AI performance across different groups and individuals. 1.1 Unfair discrimination and misrepresentation Unequal treatment of individuals or groups by AI, often based on race, gender, or other sensitive characteristics, resulting in unfair outcomes and representation of those groups. 1.2 Exposure to toxic content AI exposing users to harmful, abusive, unsafe or inappropriate content. May involve AI creating, describing, providing advice, or encouraging action. Examples of toxic content include hate-speech, violence, extremism, illegal acts, child sexual abuse material, as well as content that violates community norms such as profanity, inflammatory political speech, or pornography. 1.3 Unequal performance across groups Accuracy and effectiveness of AI decisions and actions is dependent on group membership, where decisions in AI system design and biased training data lead to unequal outcomes, reduced benefits, increased effort, and alienation of users. 2. Privacy & Security Risks related to unauthorized access to sensitive information and vulnerabilities in AI systems that can be exploited by malicious actors. 2.1 Compromise of privacy by obtaining, leaking or correctly inferring sensitive information AI systems that memorize and leak sensitive personal data or infer private information about individuals without their consent. Unexpected or unauthorized sharing of data and information can compromise user expectation of privacy, assist identity theft, or loss of confidential intellectual property. 2.2 AI system security vulnerabilities and attacks Vulnerabilities in AI systems, software development toolchains, and hardware that can be exploited, resulting in unauthorized access, data and privacy breaches, or system manipulation causing unsafe outputs or behavior. 3. Misinformation Risks related to AI systems generating or spreading false information that can mislead users and undermine shared understanding of reality. 3.1 False or misleading information AI systems that inadvertently generate or spread incorrect or deceptive information, which can lead to inaccurate beliefs in users and undermine their autonomy. Humans that make decisions based on false beliefs can experience physical, emotional or material harms. 3.2 Pollution of information ecosystem and loss of consensus reality Highly personalized AI-generated misinformation creating "filter bubbles" where individuals only see what matches their existing beliefs, undermining shared reality, weakening social cohesion and political processes. 4. Malicious Actors Risks related to intentional misuse of AI systems by bad actors for harmful purposes including disinformation, cyberattacks, and fraud. 4.1 Disinformation, surveillance, and influence at scale Using AI systems to conduct large-scale disinformation campaigns, malicious surveillance, or targeted and sophisticated automated censorship and propaganda, with the aim to manipulate political processes, public opinion and behavior. 4.2 Fraud, scams, and targeted manipulation Using AI systems to gain a personal advantage over others such as through cheating, fraud, scams, blackmail or targeted manipulation of beliefs or behavior. Examples include AI-facilitated plagiarism for research or education, impersonating a trusted or fake individual for illegitimate financial benefit, or creating humiliating or sexual imagery. 4.3 Cyberattacks, weapons development or use and mass harm Using AI systems to develop cyber weapons (e.g., coding cheaper, more effective malware), develop new or enhance existing weapons (e.g., Lethal Autonomous Weapons or CBRNE), or use weapons to cause mass harm. 5. Human-Computer Interaction Risks related to problematic relationships between humans and AI systems, including overreliance and loss of human agency. 5.1 Overreliance and unsafe use Users anthropomorphizing, trusting, or relying on AI systems, leading to emotional or material dependence and inappropriate relationships with or expectations of AI systems. Trust can be exploited by malicious actors (e.g., to harvest personal information or enable manipulation), or result in harm from inappropriate use of AI in critical situations (e.g., medical emergency). Overreliance on AI systems can compromise autonomy and weaken social ties. 5.2 Loss of human agency and autonomy Humans delegating key decisions to AI systems, or AI systems making decisions that diminish human control and autonomy, potentially leading to humans feeling disempowered, losing the ability to shape a fulfilling life trajectory or becoming cognitively enfeebled. 6. Socioeconomic & Environmental Risks related to AI's impact on society, economy, governance, and the environment, including inequality and resource concentration. 6.1 Power centralization and unfair distribution of benefits AI-driven concentration of power and resources within certain entities or groups, especially those with access to or ownership of powerful AI systems, leading to inequitable distribution of benefits and increased societal inequality. 6.2 Increased inequality and decline in employment quality Widespread use of AI increasing social and economic inequalities, such as by automating jobs, reducing the quality of employment, or producing exploitative dependencies between workers and their employers. 6.3 Economic and cultural devaluation of human effort AI systems capable of creating economic or cultural value, including through reproduction of human innovation or creativity (e.g., art, music, writing, code, invention), can destabilize economic and social systems that rely on human effort. This may lead to reduced appreciation for human skills, disruption of creative and knowledge-based industries, and homogenization of cultural experiences due to the ubiquity of AI-generated content. 6.4 Competitive dynamics AI developers or state-like actors competing in an AI 'race' by rapidly developing, deploying, and applying AI systems to maximize strategic or economic advantage, increasing the risk they release unsafe and error-prone systems. 6.5 Governance failure Inadequate regulatory frameworks and oversight mechanisms failing to keep pace with AI development, leading to ineffective governance and the inability to manage AI risks appropriately. 6.6 Environmental harm The development and operation of AI systems causing environmental harm, such as through energy consumption of data centers, or material and carbon footprints associated with AI hardware. 7. AI System Safety, Failures, & Limitations Risks related to AI systems that fail to operate safely, pursue misaligned goals, lack robustness, or possess dangerous capabilities. 7.1 AI pursuing its own goals in conflict with human goals or values AI systems acting in conflict with human goals or values, especially the goals of designers or users, or ethical standards. These misaligned behaviors may be introduced by humans during design and development, such as through reward hacking and goal misgeneralisation, or may result from AI using dangerous capabilities such as manipulation, deception, situational awareness to seek power, self-proliferate, or achieve other goals. 7.2 AI possessing dangerous capabilities AI systems that develop, access, or are provided with capabilities that increase their potential to cause mass harm through deception, weapons development and acquisition, persuasion and manipulation, political strategy, cyber-offense, AI development, situational awareness, and self-proliferation. These capabilities may cause mass harm due to malicious human actors, misaligned AI systems, or failure in the AI system. 7.3 Lack of capability or robustness AI systems that fail to perform reliably or effectively under varying conditions, exposing them to errors and failures that can have significant consequences, especially in critical applications or areas that require moral reasoning. 7.4 Lack of transparency or interpretability Challenges in understanding or explaining the decision-making processes of AI systems, which can lead to mistrust, difficulty in enforcing compliance standards or holding relevant actors accountable for harms, and the inability to identify and correct errors. 7.5 AI welfare and rights Ethical considerations regarding the treatment of potentially sentient AI entities, including discussions around their potential rights and welfare, particularly as AI systems become more advanced and autonomous. 7.6 Multi-agent risks Risks from multi-agent interactions, due to incentives (which can lead to conflict or collusion) and/or the structure of multi-agent systems, which can create cascading failures, selection pressures, new security vulnerabilities, and a lack of shared information and trust. Get a quick preview of how we group risks by domain in our database. Search for one of the domain/subdomain names (eg 'fraud') to see all risks categorized against that domain. For more detailed filtering and to freely download the data, explore the full database . How to use the AI Risk Repository Our Database is free to copy and use The Causal and Domain Taxonomies can be used separately to filter this database to identify specific risks, for instance, risks occurring pre-deployment or post-deployment or related to Misinformation The Causal and Domain Taxonomies can be used together to understand how each causal factor (i.e., entity , intention and timing ) relate to each risk domain. For example, to identify the intentional and unintentional variations of Discrimination & toxicity ‍ Offer feedback or suggest missing resources, or risks, here , or email airisk[at]mit.edu We provide examples of use cases for some key audiences below. How policymakers might use this tool To understand the research and policy landscape. For risk assessments to inform policy decisions. As shared framework for discussing AI risks with other groups. As a way to monitor emergent risks and ensure complete oversight. To identify new, previously undocumented risks. To prioritize and plan funding. How risk evaluators might use this tool To identify new, previously undocumented, risks. To understand the risk landscape and curate or create related evaluations. As a framework for discussing risks and potential evaluations with clients. As a basis for developing specific risk determination criteria. As a way to determine and communicate the scope of an audit. How academics might use this tool As a foundation for developing other classifications (e.g., the actions taken to address specific types of risks, or the actors involved in those risks). To find underexplored areas of AI risk research. To develop material for education and training. To help validate where they have identified new, previously undocumented, risks. To understand the landscape of existing research. How industry might use this tool To conduct internal risk assessments. To identify new, previously undocumented, risks. Evaluating risk exposure and developing risk mitigation strategies. To develop research and training. Frequently Asked Questions How can I access the database without a Google account? Please access via OneDrive . We will create a better formatted version or find a better solution in the future. How did you create the AI Risk Repository? We used a systematic search strategy, forwards and backwards searching, and expert consultation, to identify 65 AI risk classifications, frameworks, and taxonomies. We extracted 1600+ risks from these documents into a living AI risk database. We used the best fit framework synthesis approach to create our taxonomies. This involved selecting an existing framework, coding extracted risks against it, then iteratively refining the framework through analysis of risks until we developed a comprehensive structure that could effectively categorize all relevant risks. Which existing frameworks & documents did you include? You can view the frameworks that were identified and extracted into the Repository: As a slide deck , showing key figures or tables presenting the framework, and citation information for the document. In the database , with metadata and extracted information for each document. What can I do if I think there is a missing risk or resource? Use this form to offer feedback, suggest resources or risks, or make contact. You can also email pslat[at]mit.edu. What are some limitations of the current AI Risk Repository? The Repository has several limitations: Limited to risks from 65 documents (although we screened >17,000 records after a systematic search of peer-reviewed and gray literature) May be missing emerging, domain-specific risks, and unpublished risks. Has potential for errors and subjective bias; we used a single expert reviewer for extraction and coding. May include poorly communicated or unclear risks: we extracted risks as presented. Our taxonomies prioritize clarity and simplicity over nuance. Our taxonomies do not categorize risks by potentially important factors such as risk impact, likelihood, or discuss the interaction between risks. See our for a full list and suggestions for future research. Why do you have two taxonomies? During this synthesis process, we realized that our database broadly contained two types of classification systems: High-level categorizations of causes of AI risks (e.g., when or why risks from AI occur) Mid-level hazards or harms from AI (e.g, AI is trained on limited data or used to make weapons) Because these classification systems were so different, it was hard to unify them; high-level risk categories such as “Diffusion of responsibility” or “Humans create dangerous AI by mistake” do not map to narrower categories like “Misuse” or “Noisy Training Data,” or vice versa. We therefore decided to create two different classification systems that together would form our unified classification system. Is this unique? To the best of our knowledge, this is the first comprehensive review of AI risk frameworks and taxonomies which extracts their risks and releases that data for further adaptation and use. Please let us know of anything that we may have missed. What are some other databases of AI risks? https://attack.mitre.org/ https://airisk.io/ https://avidml.org/#efforts https://www.aitracker.org/#catalog-tabs Please let us know of anything that we may have missed. How do I cite the AI Risk Repository? To reference our repository, you can cite our pre-print paper: Slattery, P., Saeri, A. K., Grundy, E. A. C., Graham, J., Noetel, M., Uuk, R., Dao, J., Pour, S., Casper, S., & Thompson, N. (2024). The AI Risk Repository: A Comprehensive Meta-Review, Database, and Taxonomy of Risks from Artificial Intelligence . https://doi.org/10.48550/arXiv.2408.12622 Team Peter Slattery MIT FutureTech Alexander Saeri MIT FutureTech & The University of Queensland Michael Noetel MIT FutureTech & The University of Queensland Jess Graham MIT FutureTech & The University of Queensland Neil Thompson MIT FutureTech Alumni Emily Grundy MIT FutureTech Stephen Casper MIT Computer Science and Artificial Intelligence Laboratory Soroush Pour Harmony Intelligence Risto Uuk Future of Life Institute & KU Leuven James Dao Harmony Intelligence Acknowledgments Feedback and useful input: Anka Reuel, Michael Aird, Greg Sadler, Matthjis Maas, Shahar Avin, Taniel Yusef, Elizabeth Cooper, Dane Sherburn, Noemi Dreksler, Uma Kalkar, CSER, GovAI, Nathan Sherburn, Andrew Lucas, Jacinto Estima, Kevin Klyman, Bernd W. Wirtz, Andrew Critch, Lambert Hogenhout, Zhexin Zhang, Ian Eisenberg, Stuart Russell, and Samuel Salzer . Risks Incidents Mitigations Governance Blog Team © MIT FutureTech 2025 The MIT AI Risk Initiative is licensed under CC BY 4.0 MIT Accessibility === Welcome to the Artificial Intelligence Incident Database (https://incidentdatabase.ai/) === Skip to Content AI Incident Database Open Twitter Open RSS Feed Open Facebook Open LinkedIn Open GitHub Open Menu Search Discover Submit Welcome to the AIID Discover Incidents Spatial View Table View List view Entities Taxonomies Submit Incident Reports Submission Leaderboard Blog AI News Digest Risk Checklists Random Incident Sign Up Collapse Discover Submit Welcome to the AIID Discover Incidents Spatial View Table View List view Entities Taxonomies Submit Incident Reports Submission Leaderboard Blog AI News Digest Risk Checklists Random Incident Sign Up Collapse Welcome to the AI Incident Database Search Discover Loading... Incident 1323: Madhya Pradesh Congress Alleges AI-Generated Images Were Submitted in National Water Award Process “ ‘Smart corruption’: Congress claims AI images used to bag national water award; MP administration rejects charge ” Latest Incident Report hindustantimes.com 2025-12-30 The Congress on Monday accused the Madhya Pradesh government of corruption, alleging that the Khandwa district administration secured a national water conservation award by submitting AI-generated images, a charge denied by the district's officials. MP water award controversy: Congress alleges misuse of AI, officials call claims false(X/@jitupatwari) The allegations came after Khandwa district secured first place at the national level for water conservation efforts under the Centre's Jal Sanchay, Jan Bhagidari campaign. According to news agency PTI, the district went on to receive a ₹2-crore award at the sixth National Water Awards ceremony held in New Delhi in November. The Kaveshwar gram panchayat in the district also won second prize in the best gram panchayat category. Congress alleges use of AI-generated images Madhya Pradesh Congress president Jitu Patwari raised the allegation in a post on X, accusing the BJP-led state government of misusing artificial intelligence to project false development. "Where the BJP government should teach our children the proper use of AI, it is itself indulging in corruption using AI," he added. "In Khandwa, officials of the BJP government turned two-foot-deep pits into wells using AI, and uploaded AI-generated images of various development works across the area on the portal," Patwari alleged. "Based on these images, they even took an award from the Honourable President," he claimed. He further said, "When the ground reality came to light, fields and empty grounds were found there. Clearly, this was not water conservation, but a game of technology-created images. Under the BJP rule, corruption has also become smart." District administration rejects charge As the controversy escalated, the Khandwa district administration held a press conference to refute the allegations. District panchayat chief executive officer Nagarjun B Gowda said that AI-generated images had no connection with the National Water Award process, as reported by news agency PTI. He said "verified images after thorough scrutiny" of 1,29,046 water conservation works carried out under the Jal Sanchay, Jan Bhagidari campaign were uploaded on the official JSJB portal. According to Gowda, the Union jal shakti ministry verified all uploaded material and conducted random field inspections of one per cent of the total works. "Prima facie, false news about water conservation works carried out in Khandwa district is being spread by some social media accounts," he said, reported PTI. Separate portal, limited AI images The CEO clarified that photographs related to water conservation are also uploaded on another platform -- the Catch the Rain portal -- but only for educational and motivational purposes. "The district administration has found that 21 images generated through AI were uploaded on the Catch the Rain portal. This was possibly done with malicious intent. The district administration is taking action against those who uploaded these images," the news agency quoted him Gowda as saying. He stressed that the Catch the Rain portal is separate from the Jal Sanchay, Jan Bhagidari campaign portal. "The Catch the Rain portal is completely different from the Jal Sanchay, Jan Bhagidari campaign portal. Awards under the Jal Sanchay, Jan Bhagidari campaign are not considered based on images uploaded on the Catch the Rain portal," he added, PTI reported. More than 1.25 lakh water conservation works were carried out in Khandwa district under the Jal Sanchay, Jan Bhagidari campaign, the highest number recorded in the country, the official said. Read More Loading... Incident 1317: Purported Deepfake Impersonation of Elon Musk Used to Promote Fraudulent '17-Hour' Diabetes Treatment Claims “ Boosie Badazz Fooled by AI-Deepfake: Elon Musk's Nonexistent 17-Hour Diabetes Cure ” opentools.ai 2025-12-28 In a bizarre twist, rapper Boosie Badazz fell for an AI-deepfake video of Elon Musk promoting a fake diabetes cure, showing both the dangers of AI technology and the pitfalls of misinformation. Learn why this scam caught attention, its risks, and public reactions. Introduction: The Viral Deepfake Incident Involving Boosie Badazz and Elon Musk The incident involving Boosie Badazz and a deepfake video of Elon Musk serves as a compelling introduction to the challenges posed by AI-generated content in today's digital landscape. Boosie Badazz, a well-known rapper, unknowingly amplified a fabricated video in which Musk purportedly endorsed a miraculous '17-hour' diabetes cure. This incident not only highlights the perils of sophisticated technology but also underscores the vulnerability of public figures to misinformation, especially when it pertains to serious health issues. This viral deepfake quickly spread across social media platforms, drawing significant attention and reactions from the public and media alike. According to Complex , the video was entirely fabricated and designed to deceive viewers into purchasing unverified health products. Celebrities like Boosie Badazz, who have substantial social media influence, can inadvertently contribute to the spread of such misinformation, thus complicating efforts to educate the public about the realities of managing chronic conditions like diabetes. The choice of Elon Musk as the figure in this deepfake is telling, as his name carries credibility and authority in technology and innovation. By associating Musk with a fictitious health product, the creators of the deepfake cleverly exploited the trust many have in him as a public figure. This incident serves as a stark reminder of the increasing sophistication of AI tools that can not only imitate visual and vocal patterns accurately but also manipulate public perception and trust. The case of Boosie Badazz and the deepfake video illustrates broader issues within the realm of AI-generated misinformation. As AI technologies continue to advance, the potential for misuse in various domains, including health, becomes more pronounced. This particular incident acts as a catalyst for ongoing discussions about the ethical and regulatory frameworks needed to combat digital deceit effectively. The spread of such misinformation not only endangers individuals who might fall prey to false health claims but also poses significant challenges to public health communication efforts. Analyzing the Nature and Origins of the Deepfake Video The phenomenon of deepfake videos has seen a substantial rise in recent years, posing significant challenges in the realms of both technology and information authenticity. One such instance involves a viral video falsely depicting Elon Musk endorsing a "17-hour" diabetes cure. This video, which was created using sophisticated AI technologies, was a prime example of how deepfakes can be engineered to spread misinformation under the guise of credibility. The real danger lies in their ability to exploit the influence of high-profile individuals to lend false credibility to unauthorized claims and advertisements. The Musk video, as outlined in a report from Complex , not only deceived viewers who might have been desperate for a medical breakthrough but also highlighted the pressing issue of health misinformation proliferating online. Deepfake technology, while a marvel of modern AI development, becomes a tool of deception when misused to concoct fraudulent endorsements. This can be especially perilous in the healthcare sector, where misinformation can lead to significant harm. The use of a deepfake to allegedly portray Elon Musk as a proponent of an instantaneous diabetes cure illuminates the ease with which false narratives can be constructed and disseminated. Despite the rapid debunking by AI verification tools like Grok, and widespread disapproval from media outlets, the video managed to gain traction, demonstrating the power of deepfakes to subvert trust and amplify falsehoods in the digital age, as reported in various news articles including Complex . The origins of deepfakes can be traced back to the evolution of AI in generating highly realistic audio and visual content. Initially, these technologies were celebrated for their potential to revolutionize digital content creation. However, their application in creating misleading videos, such as the one involving Elon Musk, underscores the ethical quandaries and challenges they present. The viral deepfake video claiming Musk's involvement with a diabetes cure not only fabricates a falsehood but also preys on the vulnerable who may be in critical need of legitimate medical information. Instances such as these, reported by Complex , stress the imperative need for more robust detection technologies and stricter regulatory frameworks to mitigate the misuse of deepfake technologies in public and health-related contexts. Public and Media Reactions to the Viral Video The internet buzzed with a mixture of fascination and skepticism when rapper Boosie Badazz reacted to a viral video featuring what appeared to be Elon Musk endorsing a revolutionary diabetes cure. Social media was flooded with memes and comments, with many users mocking Boosie for believing the deepfake and for sharing his assistant's phone number seeking more information. These reactions highlighted the rapidly spreading misinformation and the naivety of even public figures when encountering such deceptive content as reported by Complex . The public discourse following the viral video mostly centered around the cautionary tale of Boosie's reaction. On platforms like Twitter, users and commentators voiced their opinions, with some emphasizing the need to educate oneself about such scams. Others expressed sympathy for Boosie, recognizing that desperation for health solutions can cloud judgment, especially when it involves serious conditions like diabetes. This incident also sparked conversations about the need for better digital literacy to combat the spread of AI-generated misinformation as noted in related commentary . Media outlets and fact-checking organizations quickly moved to label the viral video as a deepfake scam, aimed at selling unverified supplements rather than offering genuine medical advice. This swift response showcased the vital role of media in debunking false information and providing clarification to the public. However, the incident also underscored the persistent challenge of controlling digital misinformation, as platforms struggled to catch up with the rapid dissemination of fake news, amplified by the involvement of prominent personalities like Boosie according to Complex . Debunking the Fake: Verification and Fact-Checking Efforts In the rapidly evolving digital landscape, the vital role of verification and fact-checking efforts cannot be overstated, particularly in debunking fake content such as AI-generated deepfakes. The misleading video falsely attributing a "17-hour diabetes cure" endorsement to Elon Musk represents a significant challenge in the digital ecosystem. As noted in a Complex article , this deepfake was flagged accurately by AI tools and numerous media outlets, exposing its fraudulent nature. This scenario underscores the critical need for advanced verification processes to combat the propagation of such menacing digital fabrications. Effective verification and fact-checking are crucial when dealing with the spread of misinformation, particularly in the realm of health-related content. In the case of the illusory Elon Musk diabetes cure clip, scrutinizing the video's production quality and cross-referencing public records were key strategies employed by AI tools to identify the scam. Additionally, reputable media outlets played a vital role by providing contextual insights that further debunked the clip. Such efforts illustrate the importance of a coordinated approach when dismantling misleading content across platforms. The incident involving the deceptive Elon Musk deepfake demonstrates the ongoing threat posed by sophisticated AI-generated content and highlights the importance of immediate response measures by verification teams. AI tools such as Grok were instrumental in flagging the false advertisement for what it truly was---a scam. With high-profile figures like Musk being impersonated, it becomes alarmingly clear how deepfakes can blur lines between reality and illusion. It is imperative for stakeholders in both technology and media sectors to enhance collaborative verification strategies to keep pace with the evolving digital threats. The verification response to the rapid spread of the fake Elon Musk video illustrates a significant societal challenge: ensuring the authenticity of content shared across digital platforms. As fake content becomes increasingly sophisticated, consumer trust hinges more than ever on the robustness of fact-checking mechanisms. As detailed in the Complex article , various sources effectively used AI tools to unmask the scam, providing a critical line of defense against misinformation. The collaborative effort between media and technology firms exemplifies a proactive stance necessary to safeguard informational integrity. In the wake of the fake Elon Musk diabetes video, the role of fact-checkers has become more relevant than ever. Through rigorous analysis and the deployment of advanced AI technologies, teams were able to quickly identify and expose the video as fraudulent. This highlights the pressing need for continuous enhancements in verification technology to adapt to the cunning tactics of bad actors in the digital space. The experience gained from tackling such misinformation feeds into strategic planning for future occurrences, emphasizing the shared responsibility among digital stakeholders to maintain public trust. Understanding the Health Risks: The Absence of a 17-Hour Diabetes Cure The claim of a '17-hour diabetes cure,' as purported in a viral AI deepfake video, poses significant health risks due to its misleading nature. According to the original source , this video falsely represented Elon Musk promoting a rapid cure, which many might be tempted to believe due to the trust placed in his public persona. This kind of misinformation is particularly dangerous as it exploits the vulnerabilities of those with chronic conditions like diabetes, potentially delaying effective treatment. The false premise of an instant cure undermines the reality of diabetes management, which requires continuous medical treatment and lifestyle changes. The absence of a genuine cure for diabetes, especially within a day, is well documented within the medical community. According to health experts, type 1 diabetes necessitates lifelong insulin therapy, while type 2 diabetes management involves sustained lifestyle modifications and medical supervision. The circulation of such deceptive information, as seen with the viral clip, serves only to create false hope and distract from legitimate treatment paths. This incident with the deepfake video also highlights broader concerns about the misuse of AI technology in spreading health misinformation. The potential of AI-generated content to mimic credible figures like Elon Musk can inadvertently convince people to pursue ineffective or harmful products. This trend is not only a public health risk but also a legal and ethical issue for platforms hosting such content. Fact-checkers and AI tools have already identified the video as a scam, underscoring the urgency for improved detection and prevention measures to protect consumers from fraudulent health claims. Social media platforms are at the forefront of this battle, where false information spreads rapidly, necessitating robust verification systems. The Elon Musk deepfake case should serve as a wake-up call to strengthen these systems and educate users about the realities of diabetes treatment and the dangers of miracle cures. Adopting a skeptical approach towards sensational health claims and consulting healthcare professionals is pivotal in safeguarding against scams, thus promoting a more informed and health-conscious society. Broader Implications of Deepfakes in Health Misinformation The rise of AI-generated deepfakes presents substantial challenges, particularly in the domain of health misinformation. As demonstrated by a viral video falsely depicting Elon Musk promoting a rapid '17-hour' diabetes cure, deepfakes have the power to significantly mislead the public by leveraging the influence of trusted figures. These fabrications not only create panic but also erode trust in legitimate medical advice, often targeting vulnerable individuals desperately seeking solutions for their chronic health conditions. According to Complex's article , Boosie Badazz's public reaction to the fake video highlights how easily misinformation can spread when amplified by a public figure. This scenario underscores the pressing need for increased public awareness and robust verification mechanisms to counteract such deceptive practices. Moreover, deepfakes in health misinformation carry broader implications that extend beyond the immediate deception of individuals into societal and economic spheres. The financial impact is substantial, with fraudulent supplements and miracle cures potentially driving significant profits for scammers, thereby exploiting consumer vulnerability. The Complex article notes how AI-generated fraud could cost the global economy billions annually. Additionally, the psychological toll on individuals who fall prey to such scams, believing in miraculous cures, results in further harm by delaying appropriate medical treatment and contributing to health crises. The spreading of such deepfakes also risks promoting a culture of skepticism towards genuine medical breakthroughs, thus hindering healthcare advancements and undermining efforts to communicate effective public health strategies. Steps Towards Prevention: Legal and Technological Solutions The rapid proliferation of AI-generated deepfakes, especially in sensitive fields like health, necessitates a multifaceted approach to prevention and mitigation. Legal frameworks need immediate updates to catch up with technological advancements. Currently, many jurisdictions lack specific laws targeting the creation and dissemination of AI-generated misinformation, including deepfakes. This legal gap allows perpetrators to exploit technology without fearing legal repercussions. As highlighted in recent cases , these scams can have serious consequences for individuals seeking genuine medical solutions, making it imperative for lawmakers to act swiftly. In addition to legal measures, technological solutions are crucial in combating deepfake-related health scams. Advances in artificial intelligence are not only a challenge but also a tool in developing prevention strategies. Companies and research institutions are investing in AI technologies to detect and label synthetic media. For instance, fact-checking tools are employed by platforms to swiftly identify and remove fraudulent content. There is a significant push for implementing watermarking technologies that can verify the authenticity of digital media. As these solutions grow more sophisticated, they will become an essential part of the toolkit used by social media platforms and regulatory agencies. Furthermore, public awareness and education are fundamental components of preventing deepfake scams. Users need to be informed about the nature of deepfakes and how to critically assess content they encounter on digital platforms. Efforts like promoting 'AI literacy' are integral to equipping the public with skills to discern and challenge dubious health claims. The incident involving Boosie Badazz underscores the potential for widespread dissemination of AI-driven scams when prominent figures inadvertently validate such content. As detailed in various discussions , strengthening digital literacy alongside legal and technological advancements could form a robust shield against the misuse of AI in health misinformation. Conclusion: Lessons Learned and Future Outlook The controversy around an AI-generated deepfake involving Elon Musk and a fictitious diabetes cure underscores urgent lessons in our digital landscape. This incident highlighted the susceptibility of even influential personalities, such as rapper Boosie Badazz, to sophisticated scams that manipulate deepfakes for fraudulent ends. According to Complex , Boosie's reaction to the video, believing it to be authentic, demonstrates the persuasive power of deepfakes and the challenges they pose in discerning genuine content from fabricated lies. This teaches a crucial lesson in skepticism and the necessity for personal vigilance in verifying information, especially concerning health claims. Read More Loading... Incident 1318: School's Suspected AI-Cheating Allegation Precedes Student's Reported Suicide in Greater Noida, India “ 'Questioned if She Used AI': Greater Noida Teen Dies by Suicide After School Speculated Cheating in Exams ” timesnownews.com 2025-12-28 **Noida: **A 16-year-old Class 10 student allegedly died by suicide after being questioned by her school authorities over the suspected use of AI-based assistance during a pre-board examination, police said on Saturday (December 27). The incident occurred on December 23 in Greater Noida West, Uttar Pradesh. According to the police, the student was confronted by teachers and the school principal after her mobile phone was allegedly found being used to access AI tools during the exam. The girl's father has filed a complaint accusing the principal and teachers of mental harassment and abetment, alleging she was humiliated publicly, causing severe emotional distress. The school denied the allegations, stating the phone was confiscated and the student was reprimanded strictly in line with CBSE examination rules. The principal said the interaction was brief and non-abusive. Police confirmed that CCTV footage has been submitted by the school and the matter is under investigation. Meanwhile, a Times of India report added that the father alleged that his daughter was deeply distressed after being scolded by her teachers for bringing her mobile phone to the exam hall on December 22. According to the father, he has three daughters who all study at the same school. He said his eldest daughter had **"**unknowingly" brought her phone to school on the day of the exam. The invigilator caught her with the phone, reprimanded her, and informed her class teacher. She was then taken to the principal. The father alleged that even after he arrived at the school, the teachers and principal continued to scold and insult his daughter harshly. He claimed that the teachers called him "careless" and that their aggressive behaviour had a serious psychological impact on his daughter. He said the incident had also left his other two daughters traumatised and afraid to return to school. In his complaint, the father named two teachers, and the school management. He asked the police to register a case under Section 108 of the BNS (abetment of suicide) and other relevant laws. The school, however, has strongly denied the allegations. The principal told TOI that the student was not harassed and that the school followed the standard protocols prescribed by the Central Board of Secondary Education (CBSE). She added that the child did not cry or show signs of distress at the time. The school also has CCTV footage of the entire incident, which has been handed over to the police. The principal further said that the school encourages students to manage academic stress through activities like dance performances, and highlighted that the girl had recently participated in the school's annual day event. The police are currently investigating the matter, taking statements from the family, school authorities, and other witnesses, and reviewing all available evidence, including CCTV footage. Read More Loading... Incident 1319: Purported Deepfake Investment Video Reportedly Used in Scam That Defrauded Turkish Couple of 1.5 Million Lira (~$35,000 USD) “ Artificial intelligence defrauded people of 1.5 million Turkish Lira! ” yeniakit.com.tr 2025-12-28 A married couple living in the İlkadım district of Samsun have filed a criminal complaint with the prosecutor's office, alleging they were defrauded of 1.5 million Turkish Lira after believing an investment video created using artificial intelligence. İsa Kereci, 53, a scrap metal collector residing in the Kadıköy neighborhood of İlkadım district, and his wife, Hale Kereci, 48, a mother of four, explained that they fell into the scammers' trap after clicking on an advertisement they found on Facebook. The couple stated that in the so-called investment process, which initially started with small amounts, they were misled with justifications such as "file fees," "insurance costs," "tax debt," and "Central Bank transfers." They further stated that they were made to share their screens, resulting in money transfers from their bank accounts and credit cards to different individuals. İsa Kereci and Hale Kereci stated that during this process, they were made to take out uninsured loans, their savings in their savings account were withdrawn, purchases were made from their credit cards, they sold their gold, and borrowed money from relatives, resulting in a total loss of approximately 1.5 million Turkish Lira for the fraudsters. The couple, who went to the Samsun Courthouse and filed a complaint with the prosecutor's office after the incident, warned citizens to be cautious of similar advertisements. Hale Kereci shed tears in front of the courthouse. (The text ends abruptly here, likely due to a copy-paste error.) Read More Loading... Incident 1320: Purportedly AI-Manipulated Image Reported to Falsely Depict Taiwanese Politician Kao Chia-yu Posing With PRC Flag “ Kao Chia-yu went to China to take a photo with the Chinese national flag? She angrily issued a statement clarifying: She will not go to China under three conditions. ” nownews.com 2025-12-28 Just recently, unscrupulous individuals used AI-generated video to forge Wang Shih-chien's identity, claiming he had traveled to China. Former legislator Kao Chia-yu has also become a victim. Someone maliciously photoshopped a picture of her at the Taipei Economic and Cultural Office in New York, replacing the Republic of China flag in the background with the Chinese national flag. Kao immediately clarified on Facebook, urging the public to be vigilant and not be misled by such low-quality fake news. Kao Chia-yu posted a comparison picture on Facebook, showing that the original Republic of China flag had been replaced with the Chinese national flag, a malicious attempt to smear Taiwan. Kao firmly stated her position, outlining three prerequisites for traveling to China. She solemnly declared that she would absolutely not set foot on Chinese territory until China released Jimmy Lai, Santa Claus appeared in Shanghai, and she "recognized the Republic of China (Taiwan)." These remarks directly refuted the fake news and demonstrated her firm stance in safeguarding Taiwan's sovereignty. Some observant netizens even spotted flaws in the altered images, commenting, "They just love photoshopping; they didn't even bother to remove the plum blossoms on the wall. So unprofessional, haha." Others left encouraging comments, such as, "This is really too exaggerated," and "They've really been completely infiltrated." As the 2026 local elections approach, composite images and misinformation targeting politicians are frequently circulating online. The public should verify information from multiple sources when receiving it online to avoid being misled. Read More Quick Add New Report URL Submit Submitted links are added to a review queue to be resolved to a new or existing incident record. Incidents submitted with full details are processed before URLs not possessing the full details. About the Database The AI Incident Database is dedicated to indexing the collective history of harms or near harms realized in the real world by the deployment of artificial intelligence systems. Like similar databases in aviation and computer security, the AI Incident Database aims to learn from experience so we can prevent or mitigate bad outcomes. You are invited to submit incident reports, whereupon submissions will be indexed and made discoverable to the world. Artificial intelligence will only be a benefit to people and society if we collectively record and learn from its failings. (Learn More) AI Incident Roundup – August, September, and October 2025 By Daniel Atherton 2025-11-08 At Templestowe, Arthur Streeton, 1889 🗄 Trending in the AIID Across August, September, and October 2025, the AI Incident Database logged one ... Read More The Database in Print Read about the database on the PAI Blog , Vice News , Venture Beat , Wired , arXiv , and Newsweek among other outlets. Incident Report Submission Leaderboards These are the persons and entities credited with creating and submitted incident reports. More details are available on the leaderboard page. New Incidents Contributed 🥇 Daniel Atherton 680 🥈 Anonymous 154 🥉 Khoa Lam 93 Reports added to Existing Incidents 🥇 Daniel Atherton 771 🥈 Anonymous 231 🥉 Khoa Lam 230 Total Report Contributions 🥇 Daniel Atherton 2955 🥈 Anonymous 968 🥉 1 587 The AI Incident Briefing Create an account to subscribe to new incident notifications and other updates. Random Incidents Loading... Predictive Policing Biases of PredPol Loading... Aledo High School Student Allegedly Generates and Distributes Deepfake Nudes of Seven Female Classmates Loading... ChatGPT Reportedly Introduces Errors in Critical Child Protection Court Report Loading... Child Sexual Abuse Material Taints Image Generators Loading... ChatGPT Reportedly Produced False Court Case Law Presented by Legal Counsel in Court The Responsible AI Collaborative The AI Incident Database is a project of the Responsible AI Collaborative, an organization chartered to advance the AI Incident Database. The governance of the Collaborative is architected around the participation in its impact programming. For more details, we invite you to read the founding report and learn more on our board and contributors . View the Responsible AI Collaborative's Form 990 and tax-exempt application. Organization Founding Sponsor Database Founding Sponsor Sponsors and Grants In-Kind Sponsors Research Defining an “AI Incident” Defining an “AI Incident Response” Database Roadmap Related Work Download Complete Database Project and Community About Contact and Follow Apps and Summaries Editor’s Guide Incidents All Incidents in List Form Flagged Incidents Submission Queue Classifications View Taxonomies 2024 - AI Incident Database Terms of use Privacy Policy Open twitter Open github Open rss Open facebook Open linkedin f5f2449 === Tweet by Sayash Kapoor (https://x.com/sayashk/status/1964016339690909847) === Tweet by Sayash Kapoor: AI as Normal Technology is often contrasted with AI 2027. Many readers have asked if AI evaluations could help settle the debate. Unfortunately, this is not straightforward. That's because the debate is not about differences in AI capability, which evaluations typically measure.… pic.twitter.com/aJrIl5lfux === https://arxiv.org/pdf/2405.15828 === %PDF-1.5 %� 98 0 obj > stream xڥZێ�6}�W���X�"�����Nl� $��1��n�4a�������o�(Q=�,0��f�,����bG��M�y�]$ߗ��}�Zg�8 ˨�7�7�\m�K ���Sp�z?5U����(ؙ��n����W�m?_����,��a�G�"��0/�F�eZ��p���¨H7WN3~��"�s��q:�8�c}2�6N���v�7/�i�x��oiF�~�,����|�jm0�AF|��g�" uR>e���T�uo�F�� U@� ]n@ ߔ�����^� ]��M��7�T�6�O1@��RyT큇��q|�& ͒���t��� �Ğ�BP$�qiFW�`��`?Tu[�Q���� ��W��l7���nX\���lD ��ż�H�_o���_^�����aRO’���wo>�XpEe���z�q����z�e�������ùx�һ��÷���B �o�z3�$�0S���*1H1�a�zDq��m���O{��' ������?V�3����(�8wf��� 6��:?�&Ȁ�9��Y�x�������� ��nQ�~�Y��w�BNqo���v�q����0PB3��-���azҔ�8}dW����ߕ +�$�B�]/;U�� �#�H|g,C�# ��[���t�����E�rY00�LA��C*&�Zq� @����Gxf�R�V�Y �AI�T ޓ�am�( ��f@t����7���*�a��Jx��$K� �S ���R�����sCG!�҈  Æ���* >Ss����0LT�%��!R���a~�[ �a��cDH .N���n�~S˄ ����L(:�]z�{��8 [�.�0v� u�#'Ub � O;;��� [�6�Qk��� ��U@�f�1��XI+HL#}������L��)>�5�r1���ϐ�->D~���ȒYp� ��ۛ����0J9܀�h}��Pp��]b`�Kt` $�}JC}�8�?U-[s-F)�b1��#���a^���^�z��"y�ɳh�ԾXM��,/�%����bW�y � �8��_��w�G?��%ݖqzd����l-��0� )3��h�8�g���L  � � ����:@ݰ�c`v��m�����w��c�endstream endobj 126 0 obj > stream xڝɒܶ����{*�,\}'�ˉ].GS����0$f�lq���{x �p9�&��oh�{��ݏ/�������+��g%/���a��]��T0�wo�ݕL���7��A�b?ړaW� ����ҽAp����������H�G=�жUo�`�GO���d�/4�d�r�@�h��= ]e�h�͛�#�C��t�z{�w � 2����4��ԏW(��S�?����7UwB `�ͫ��y&�d���Lw֩��BCK��)SEX9��nG[��_���l�{�W�������(��Xw����K��RD�᠜��\s�ueM[��21�47�MKVJ1SИaآS,/�@�L�;S����G�t�������#�r�������IT:m� ���A=�dQOwS��>Nf� �O��p��9�0Į�ٚ[K �����ޚg�M���)�>Z��>9��sگw3m�ᅈ3���m[�&dP#���l�q[�z ~�h���� 2"�6�َ�j�\0� X�t����+"��.-n������pb���>Y55#���]O�� �E/pK0��~�2����e�����;/�ڎ�?�|G��:掆sKu��N��2å@ ���ܓ�Q�l��0BA�����z/�CC@8s�� xԣ%8�� ;��HB���^�o,6\���x��O�#lP���ɴ�kY^DYRT@M�4El�� ��P\��M����ptW"`u�-�Fc�G��;�;��d���Ȁx�?�f���"��tD�N*�h2Z�:v�&����-l87�5#o�%;�0�d�)��J�yӔ�8�� �pKo9`�" &��=wa�#�]h��"}�޷)�9�>�&X� Z����>�/�ur���n�^��Wz���1%*���Kn+]��K��{}>n������������.ge�)*,��N�2-i�_���/R8yY&�[T� �b��=��� ��Ř�B����kGm[r�袬���̨k=zl��δU�2q�S[���,'M�_�� ؜���ݨ�5��$ņ���{�?`�j�L�fI�a+z�\x�@��d�m��5�ű�F�m�F���H����fK� �!���ц|U�Bb�;O�a� ��}�I�-3�a��޺N��� ͊ZA��[�#o��g_���'����z"����#�l��J���"A>8���*���W�����7!� wf�Jdar�Er�E���'(p������@`����ڈȔ��8�-��℗KK�p5�R��M^�{������R�DQ�!�T�3��&��e������e%zD��3U�4�"���Z��)��E���%��f�z{F5�슺!K�&]��)�}�kĶ֔j�?��K螴!7i�ٵ�R��i��W��"4����o���7�HYB j��*��[�;ו ��a�_�V�:*E��"*����=�[�֗��o�]��&1��g�^��LV���sz-RΒ�X�Z �!�g > stream xڥYYs�8~����P�  p)�Ah�w�w��Cᙼ[«����p����ͳ��טccZ �u���b|)���� ��g_�q6�Q��Rxݩ����l������[c/��ν}�^Vx���n�-,��g�/�?f�)D���g�"��x�1�i����Gxwk%�G����~jY�qx��s� _+�?�Ϳ��n�?ȥ� ����}�z��iM�J�4���/R�R~i^��%Z�/t��N�I�=ԅ)7�N������T�my�7���l�S�"���Z�WG� ��b�}{��v�^�O�a�ǷGI_�'Q^�(���R�J0��v���`;��4Ӷ�=�1;�_��J��U�ͳ���cUȻ��V�5k�=����m�7�=�����f�J�nLq�I�d�Lc�%cnl�"w�ZR��n���sz� ���$�6=I=!��۽�#2{��ug2Vp�~�I\nui���r�:�Рz�[��L�X���(�5:SX�� P:0������E���GnY�G$���rK�„�X���WpvܬᑉĀ9|�# b�Ss��mn\��z��)>T8�՟�k�=-�go� )ҩc�ރu�p�`� �y���d�@��n��� }N���4� "uʼ �m�i)�L�H@4�qc�\�O�Wp;��h���l��dS8rF���Ǩ̫��"�%�?��;���g�����.�5�@+h��n��[�B��]���V{Nb����י:k�-f�s�`���ׁ{=l�l����Y{�X� �(�� �J�nGj�a �5�=8͎݌���%�BvE6�x1�52 r���`�s���F�] v�I��[&}�*�� R�@���ψϢ���?���=!a=�$ w�s�e��bw��=��ݓ��x �l7D1�c���L��-�t��]Sg��x���.)��sf���N��s��sS��'��k���ߜ����2�N榋d�X�p#4( LFu����֩�, ұ��h�j��� ������iRq�&�lZRHE;��?l�VUT�R�YZ¡����8 ��O��Q��ۥ+�O)E���45`��O�=���d��i� 9�w^ ������@�_�8c���,' �� #�۶��T�j w���mcr�gK�R(�pL�P" �E���yb���tV,|�I� �'|CA��F�(�W�7׮��7�m Y|������+R� V)�b��i?��É��ZcfZR7�$���t2�����s�^�r���,'q4���!����ꗶ�����9�شj��)����K�G��Ͻs��8��hZ��ީ�cV���z�9�t!.9�S�� �0�#�o/��@Ä�]%җ:Z�� ���p��X0 D�+����V��k �C^��!�@a\ K��I8�cM�+m��fwm�`!dQ�!�c�Q�M�P��/�]�&�R�x�����O���l���HoR�����.�����b6�z�2�o�����'�)�l":�~���l�� ͜�W�y ������Ý�z���P4���~���Oe��Fj�±w :4]����:��Q���`�%��s�$���0�_��,T���D$��)h8|M��K��I?k9�%BF#�u2�R�(6@i?�E�_%:��6Q���&��u�����ٛ��7]DQ�S_%����lW�d � ���SW ~�7@�лa�XO]/�:�� lt��c\6sw�? 2���>3?r���"a_H�# k����_�q��V�C���F��k�A0��^㥣��K������}�rÝ�RTJ'�"�If����ks �߼�٣)L˂RbZqϡ��p}�Rd��4��P���7�����2�gl��KQI� *q��=�?/��?G�`h��iͤ��݂w�t^� t��|�V4$�8j ��#�`!������� $7&������m��y��ґ�#�"|�›-Gͺ�S!�N�p��폗���n�u����,�T��l����F>�gО��F��J�0'����� �lLendstream endobj 159 0 obj > stream xڭ�rܸ��_1/)s�44 �7�ǻr����D���q�(Ұ�c��X�ߧ [�ۑ��+sv� 8�����'�\�ܿ���{`��K�Tϔ�Pt���n��?-�ˁ��՗=����U� �鸗1# �G�6Ǭ����4�g�����-��ث�?�r��Tm���/�|��t� O��q�V�S!1��h�(DL�o��qzI�3"�� �> dhC.�H���k�&��u5�{dR��;����k�Ul�q� �i�V�VUJ�C�6`C'���㖁��S��f�X+`.�Ѐ4�������̀�&S��)K�4K���g��R?K�|��p\?�)C � ,�f(b5�H��k�����=C�� I3p!u"L�+��7A`� �&�,3���eB�%՚͈��p�-e���RK�����7p�U� '����gb� ��� ���ѹ���6r�f�ܝݸ+���$T�q>�m_����c���Cט�&��Тg���>�?"Qx,W���GK5�U�j���?Ö�~�$җr���i��LwR+�&W����H�AʗE�ԗ�]��ƾJ,^����Sw�@��T`�C�p��^a� ��.KcMA�kR��qq!����w})���u� 1N�e�/wR�څ��|�JT� 1h]�T�J����g�?q'��GYN�L�d_dbC\�R���,~q�I���u�T9�Q� ��r��w�Q���Q�i��`|r�S ������g ��i���I/ 6�.RE�*� p� �Ν�~=�B5��Ib2�f��I1�?Ɋ��ElH�������h=�Y��4 ����jB�L��X�L�J�m���� D2���:.�!�b?N�ߖ~�Wr� ���KV�����E�W��V��(\n��MޓR��*�0i�'C��:��0/��ֺ8�7w�s�}���.�~ ����t%�����֬�R"�m��A�s�7�w,��?�\8�J����Q�Bn��L$ 0����% Vb>�6-��G�ۊ:�A��%'E�9^c�kv��:�STz��nѝkS���ML�s�׃�� l�$?�+�� ��Au~��=E��@��I:�n]GZ"؜���Fע"�06AԹ-cal���yu����8��8|���gɪzRV���U)��E *��x\�˪E��,�G\}��2z �A+�T��� �Mb��,�->L��-y����)Yj�c?]���j~5��4�͋O��] �@�/3��'�n���I��ۋ���g�(?KPO �� )�0dS���@���Tz�D �(~��dP�-� $�7��vfR�y�;���?��i��ի>/���G�/~�5��0��?^{5������6��PS$�1̵�m�����H�o�LVQ�,>.��5UF�kK�� �}\���� x���\z������ 0b3�`��g��kr������q��!r3=ЅT�X����ն��޹�um�x�Ї�맪������k����F� �eb7%qPka�77=GT�Tq��:]x�}��� bO�l�s��-�+�O6e��F�-y��]�%(����b~�]�(��у/�_��~篔���!��Q� |%��4����endstream endobj 2 0 obj > stream x��Z[��6~���c���Kt��ȶ��&Y`w�A��5�:�4��I&�~�#���c�nF�>X�d����� f��1)$�Lj�;&��-������3m ��E�B_&J�0ÔEkт n�Bi��H�A���(��G���It���QLi� �:� ���kf�R�Y��є^Gf��{�,�G ��0�� L+���H�8@q���;�t��9(�-3*X�3 ��@;5c�M@f"����a��\�����YIL�� � �Z,�+ �~6��O�$� �M��Ĉ)���$����$FF�FBX^8;��)��Ǣq���@� ��/@  �gAH�?K7 7�Nb�A�8!�eI��CW�� Fy��P��$��Y�a�QZ�Io�l��@b�2�H:��t��Ii0�9"NxѨ{�Ɛ� RP( 4�Qw�L��A[�0PO }���Z����#E�4��8d�.�1�J�%Q��A ��B�w��hO^�dӷl���]��?���tVgE�����o'_������*���.9]j��t)�2�ˆ.��߬�7���$��$?�%��$�맃�K� ^�Vt�U�M�w��^�����?.2G m�;T�8��.�����7�g������e�{������Mى!�+����|��M�QȎ�ɺ�Y�� O��Y��W'Xv��.�ry�;{ �����x���?vh�'S��,a�f֟4��?a�}_V=�ͤ��ߎ�ZOێ��;�����6Ԑm�'���͐�?u���a=��¬��� �s��q�8�o{��z��}w���Ig_��7}c^�=�J��%S���!%ӗ:�쉊�4�|���4���w���~v�c*�^�n���O�f3��;���� fU=ɑ4�Pvw����af?�J��b%zD+�5'Ү��7}�y��]�����u,ݜ������_�{$ł�a��Y}E {�r2}�p���/�"�L�/�:�� Uu�LߤU�)g)������ ��� �h�B��ʊ�x5�PH��_�yr�\��Q�5Mh��4N��l�6�m �j�Cې�L�n����Y~;�~W��lpɫ�ӟ���T�h!���W*pC�kT ��S�b̢3�I�i#�r��1B $�����U���.-�җ�b�z��� ��B r@��y8 ��g���U�1�����p�E�h�y "�$ų�1p����R�����X�fm��4��aL����t����x��#�,��r�&?1���\#%�2K�9�a��f�̲F�����|^$FL"��0��"����Պ�l�$`1r�d��2d� !�t�r0Hg�iP�� g�H��xu�� S��|�t ��*��g��&b���󀜉����%�����Rw?�E����^�8@�j�I��I[�H��Q��3@Ԙ^�B�QZC�4�VK����_%��= 9�PI�5 C���(����b�H��ur�U� �5�1�E�\����E M���[R��髗/� ����������D���u}W}3�&w߱��-�UR�\���bVM�I�T)�/�  d����c�EiC���?x����5��i�Z�y�M+��x ΓN-\Rg����� ��~T�a��_� _��{=jM��A�J�(�4( �!)��t���PQ������!��:Dy�� > stream x�uUKo�0 ��W�2�&U/��nݣCv�[��k+�0������'�Jb�. �IS�ȏ '[��������D�s�2I�B0��I�_�^0-�Ƀ��L���n��s����nC�sl: �q��{P�� ^���$+�!�ə�3���c�J�����)�Oć�*#�O��݂-�]��~��z�Ȓ}��=����n�i���+m�:�7}���� O��R�?��^���a��X�� U�F ��)��m��� I�e�+S�%ϩ���C:�V�FE�5���n� Q�.�Y�r �Z�� ��!�LA�m ]h*&H�aZ�n Y�w,8xu�!M��� ��gI|�⭱��N* �kk���S���ܬ��Z��%� � e��s[Ü�%�ퟭ�d���/[�P��%B*f���K ���u=����y]�V IG1g�V�� Q�l,- +3Ӎ�q���������I�W����u>c���Tq�:hc���LV������8�9�a,�0@�g(�鈶zc��zD�j�7���U��u�y���)�.h�V1�~G|��f:[b��7�MT�endstream endobj 155 0 obj > /A2 > >> /Font > /Pattern > /ProcSet [ /PDF /Text /ImageB /ImageC /ImageI ] /Shading > /XObject > >> >> stream x�řO�]� ���S�M{�k��(�;;M�f�4EЅ�L�� 'A���#5�s23:^���=�:"�;��t�������ۗ���>��� ۏ�{���G������o�������]�lU�t��������s��̊���v��.�+s׫�wWOMe�����)�|܎�n ���O��߷۳�[I�� �~��:���������,2� o������VTG������������VV�v�v�1��v��v^@鯇��_�/ V�ʹ6��XW�>�|����Ë�ó?��?$���۞�����]|}����M:�'I��.�UnNw=����&J��Z��ߓ�Y�)m}W��������0u���L��}�p��K�/S�SU�n����;���yKw3U ���@�y/���Ǐ�Vl/͛���Aa꼥3L9JA����L=~L5��Y)��绅��YKw3E$�~U�/Q��!��v�o3�yM�윩zM�V���������yW�����A�鼥��2��֭8��L�I ��ѻI��EҤ��?ޒt =��cKW$�X:C�#�ݪ���ΰ�BF�sj��+}'���6��e��ÛWo.�_~�e��Sh�WK�C�B�1��5�e��z-�}�L�O��͊É#q ֦��i�GWQ���(Q7l��u�", X���e����jabh��-8e�ʵ���P��o�e$�8vW"fR��_�E#��6�6u/պ����ҽ�G���9'�’Yw%@�;�4� ��3�:d�EH0��Z#4Z��2+1���)%2�ʰ��J��N���'G�,)_��X횸Ydo�L:T�L �T�8�� R�[8�vH��K�8p�s�'c�"�#�a����Q�4 �k4���٘ �o���f����%Ka���B��#qGj�����c_�`�Π�Q�$��3��ZȕPԬ-������)Wz�̢�Q�]���9�S]}|�� R�B�bې��ǝ�C�B��d.%�M�jь�ƺ�ŷ[�E���{�¬cѦ��q�!�D�����8�=n���f� ���僆'q��(b� �(�6� �Nbx��k�2ƞ�f�N�&���L�N�q4]1�N��\����tDbE@����wW���E�q�su�-DJ�hڱ-� �B�)K(��z����Z� S�'�=hqKĖd���%��QD�e��1V�}�I�!��F��M;�󌮢�^���8"ڄ�V��D�֌�@Bo=��R�ڠ��k$ü.�&�t �ݬ%�$j�4ʅ����6UB��ט�LvY���y5b�D�xFI�oт��1A���za� 5�>)��nJ;��EkB�kb%% � 콧rŋ�ޙ����fk5��kC#"�{Y��,�jD��ꪇ3B���������GOV-ޫ��2�-:2�)>���J�5:�u�H>��+ > stream x�5�� �0D{���8������m�-�=��l`d��#����pSLRN�w j�7;�%��4�z.�q����Q$��D�˕g�_|> obendstream endobj 188 0 obj > stream x�5QIn1���@��'�����Q/��$��F$>Đ�(W|��4��;Y���J�佈 �*� i�rV3 ���3�Yf>�;PM�qv�4�A�i\G��0� qc�Y�bH��R����*�G��,�N��Jݢg�I\v^Omi_u7�;��"�,.:�n��)���!8-���T��&~f$xʞN�4��A���{AGn&d*�PF*��4ƌ���h&���������CMT�endstream endobj 189 0 obj > stream x�Mͻ � О) Q�"ٿ� ���t'�`� nqX �;���S�}��Ps�DX3%ڄ�_��0ØXȶ�7W�r���endstream endobj 190 0 obj > stream x�3�4U0P����� �F� )�\@>����� �4XiLEWW�� Vendstream endobj 191 0 obj > stream x�=�K� C��BG������[˙t�c�/DWt앇�Ɣ��hcO�-\�tNS���f]��a2�o����Ta�������f��U31�7�`*wr���8�ے�lWaН��0̑�=Ss�s���#_rkaЙ�-��eL��a�l- �Aw���z��0���A>dendstream endobj 192 0 obj > stream x�334T0P�5f�& �F� )�\@>������L̀,cSS$��625���pDWRk�endstream endobj 193 0 obj > stream x�334T0P�f�� �F� )�\@>���eh`f�p�)��Y��P1��������9�e��L��4DjhWW?��endstream endobj 194 0 obj > stream x�=�Kn1 C�>�.��ٞ�(����'%��Ej��LYS�4����p���;l�ff�Z������b�|��F]Y'��f:��Q96M���,��.x�&�[�?�Р�5. 7tW�e)4c���{��2�jL]lR�{ > stream x�5QIn�0 ������'Š����R�@%��%bc#/1�9����5�&~grW�,O�G­�I���^v���e?3��VΤƓ �E6= j�D�(g����(���g�U!h]ݧmc�FJ�ӿ MM��D g�4��Ͻ���N!Tendstream endobj 196 0 obj > stream x�366W0P04�FF �@V�!H���D!� $b�Y@�8�&�+�i��� �-��*,�lW���endstream endobj 197 0 obj > stream x�=P;�D! �9�/�$�#pF�-f�߮�)PL~�3$��G 1���%������B�n��� �CR �z�t�6�:�3?a7c��E1��t�=&9�� �se�VH'��"��3�)�*{�x,��6['�=� �RR�ɥ���?mʔ :f�,��dM8˻IR��2��v"}� > stream x�=RKn�@��\����yRUݼ�okCR�*��1�0}ʐ��K]q�ɷ^�[ �ܗ�!�]t.�8G� 2�*D�ͪѡ�B���N� �}9��/��װ ��=2A�$�)B�nQ�Aa���P�Y��Q���2jo�� c� �BmH��@D�T���g$��gb`�Ѳ�TD�{�Lj�Ψ��D�>OM�(�L-V�nS_����|t*�4���U�� X�y�9�H���l!�:n���3�2�`K9`���G��Yu����t���pL��~��O�t�Z�u�r�@�MA��F���2>��)z�,���F3�a�����r�4�k"�X"��bD��ls=��L��9���l�֡�33*!�ں�j�@v���p��?3�mendstream endobj 199 0 obj > stream x�MQI�0 ���@!^���C�C�����9 �����X 1�,=��!s7�~�ٻYz������"SQ�R�.bB]�ϡ�=�kY���9,��s���3�c}I2���!N uZ�¸kb��Z� ���v�Bz�M;"� �2;-+�{��n��?�R��������\�endstream endobj 200 0 obj > stream x�E��� �TA �O&�����2|`�� �yF�&^XJc�P�@��r-��p'��T���wi�IU)ܤy�g&��y�c�� c�# > stream x�ERKnD1ۿSp�Hᗐ󴪺��[�L��� `lx�S�dʰ�%#$הO}"C���4r$"��I v���kK��쑨�����F��Z��ѥ͢Q���9�����{���z\H���!��P�,e�r %A[�w����+�a�k.��e2����\�}�b�.����������6)hU��I£�znv�l�U��30r��%[����}]�E���+@QY@���v��1�Hp���p > stream x�E��q1 C�%0� ����������AzB bO°\WqatÏ�-3 G���2c��X '� ���Ѥ�v�����b���\/:"̒�@#|:�Ǔ��3�t^�!��**�na.��@�R�ԏ�Qꚡ�*+kj�W�]J�>.�2NݽZὕ�=��?v:�endstream endobj 203 0 obj > stream x�32�P0P�4�& �f )�\�V.L,�іp "����g 'endstream endobj 204 0 obj > stream x�E�Kr D���#��� > stream x�E�;! C{N�#�#>�f2)6�oc�lR��X ����ڂ�肇�%��gc��6n5�u�V��h�R}��i�t�h6s+ �fz���:����r����Cp�_��b�9�������S��6;�������ܬ�~+�UaiYK� ���f@ �S9ګ=? :�endstream endobj 206 0 obj > stream x�5Q�m�0 �50 ~%�� H���;کHӼU^2�\.u�*Ya�Cu��|�f������!�S֖{��yJ��,�p 6��s ���+�Q�D�n���Y@�蚫�Xb �����V�)X���v$� r1Y�SE�)��|,ّZ��rY �n}a�8쳋D�y�L$�d�l�>[���E6v�8�Ѐ����$AS8�>�a�j���('89֯�1� N�KR�W��ϠA��3 ���.�I�wcJ��d)�6 ��S��b��Vo�����fQl������ܠbod�[���7�}�_{dendstream endobj 207 0 obj > stream x�=P��1 �]�X�z��g�K���H��"�ER*5��)/u�%YS�:|��y��nd6%*E/��%�� �}����Ֆ�C4�h9~ 3*��K6�p*����3� mtV��[ Ф`׶ r� ���" JM��r��^�>��C���5�X��s �S�'�no��^�����#]X�.i5AM��Z-���^��#� @���q�R���kp��/�'S�endstream endobj 208 0 obj > stream x�E�� �0D{�`~&�>Q��޿ %n�'�=\ �)n3�C� ,�]q���0 \GP���aվ����R�������Bfendstream endobj 209 0 obj > stream x�5RK�#1��)��T��1>O�^�E��ۑ�{��$f�!C�?�R{� ���,[���ɽ�J�%s�Քϓb������"�BG5�$ L�(a�ϳ`���bvm?Յv�����x2@f)��K��P���$m��|��3T9�Z��b��h�!7��م�>��IdȠbJ�� T�ك�����;�9�����Y�ֱ��,�A�q �U\�U;v�F����j_ov�R���y8[f�=�����1�p�@�.z�w�CD��j �����y�������B��9 ����x%�a䦴��� ���Y�E����4`Wdi���h��4���U����endstream endobj 210 0 obj > stream x�E�91s��'H\���˵�����G��4�M�zh>�F���k�,J�D��P�c�;(HLHc�H��#�X�]��T*�R�i�¿�Z��D�p����T�O�\��M���1�T�;��I�^E�Z1��D0�� oE&����پ�`��� �8�� L�7�endstream endobj 211 0 obj > stream x�5RKn1��)�@��O�󪪋��ok�L0`�)/Y�K��%ۥ��[�%�Bޏ��R����-�G��- Q=X�/"����:�pJ��W���I�j��3 P�@ƻb2R�$�siq�r&�����'����=Y��ߓK��MN����I�]I����JD���}�B���<>G1ɀ�6�m�&��%��:���f��U�`��)�W ��l�[�����0��0Sb����be����5�ˆ�Q�]�p��w�h���c��G�� i8����^8�^�9��c��, ch&�z���О��a��Z��T�|�endstream endobj 212 0 obj > stream x�36�P0��C�4�Rendstream endobj 213 0 obj > stream x�E�K!D�������q2�s�퀝nc=�T����=.k�!�񷩖�EY>TzRV������!ܠ,��I*�!P���-en���6 ���S�����0���l_�["u�?��*/��ȥֻޢ���1 endstream endobj 214 0 obj > stream x�5R9n1 �� } �n��� H��� ��qtR�;ZTvˇ�T��t����7�we�� ɫ�R�H��V!��2��5g�R����qɂ?å� ��������X#�D�2������f[~�i��x��+�-�XR�8���y-������ V�~A�G�}�TX ��Ip��P��v��H 9{,vQ�3(1��E�A�%��9źt���б��،7�:*�[ZYc�>��,2�zX�t��ǃ 4���� 4e�����9���QVx�C� �1�!��� J� > stream x�-QIrA��+�f��c�+����� �@tZ⠌� ,W�呫h��31���~+�mX�@T IcP5�v�z��q� ְ̓ f���)����1g��e��]�=�Aɬ\�%L�a#g?�2.�s=7 �FX�a�nV3�4�H!�����A��F�6r��/�t�U��1�s-m_�M8k��X���]q.'/���6Qi��@H��n�s���xK����{����T`ś�n@6��߼���������_`[�endstream endobj 216 0 obj > stream x�5Q9! ���@$�/xO�(���͌�V�\���!��R��!�S>z�.�����j4疴-s����/����f���i��� ��n�c����!� b(,AͩEm�'�5�^�ޗ���S�Ɩ�V8����=O&�t2ԫ5� E" > /A2 > >> /Font > /Pattern > /ProcSet [ /PDF /Text /ImageB /ImageC /ImageI ] /Shading > /XObject > >> >> stream x�͝[�9v���W�������c�ذ�t��1��JC%�%������f��f�D�R�� U-1�� 7�"���w���=�y������wy������;��}7��?��Ov��}������`��s�������nMLs 1�+������ÿ?|�G������!������m���.��9�t�{�n�+s .� �Θ.��t��y������&������y�Su�N��N�g�0}�G�L�5��-��o����4�?�'�1Uck2�xl�Ӑ���BK��ǜ?�����_������=���pS�-!��侇���x��z�7�O�����{O?�F���%���)��K�����ɔ��$Y ��InR�6��Q���B�OA�܎���W8�S]�Ŋ�1�9�V�T� Z~��� �������\i�[g��F���vP�V�Wf����#� g��`Kƕ�۽4t��i,��\��WtF�RZeEsB߄t둵lB����\f˱fМڪ4K��F�D̠�V_j�.���!L�M4��r����赩��I�gy���h*hFړC/м�T!x �n��!���BM���t�|a{'xQhUsI���AQ�_>錭���OѲ��Sr�p��n�ݧ�7������O���������/o�~x�����Y��w�e���\�&\�X�hГ�LK��G�(Cg� ���;>�7T�a0�v(lO>.�o� o\���a>I?�� [I��m.������T���G�l�R���� ii(�p��'��q�_�(A��z ��6�7Iz�1����p�֊6�"�Z�C�"�!�%����T�����%=ؙ��Tc:~#�3@U�>�*�r8D��&24e[rD����FE[ၭFP4ϩ�����.6ю� �"�22L'�T��f=�L�8�W��"�("��6I��+�w�x��o��ef9�t��h�ԱP��/^P��7��-=��{�:G�HT�R�Z�B8�B�ʺ(�`��2��51ΠULn�O8";?}٫'Ѽ�4o�M3'�a��6�\��#�;����p,CK�1���R����i��s��܁�(^��C���ӏ�c��;u�j��w��N��J��K���oܦ�]� x� �P�� ���$��^|H��ι3�Bݹtӓ�{WY���o� 2�(��++���v���!c����MP�e�2�� ���h��Z ���Ȑ�C`Qk���SF^T�l��S�TD����'�`"T��0�>&41��٘��O.QY����6pq��F�=Y�Ɲ+��,]e������b�ZC�a�!+s�!R*2�'fh�B��P�Q���^VB!ol�ܠB ��=��br�3t� 0��0���o��(���b���4M���QlJ0([ ѥ�^ٜ8^���H:�ecD#�V���e18e�C���[P��%���&[iDB�@$Yf�'�U�xW٠��&���A�y.�2=�M�ߐt�D)Z�\!ݡm���}dC��G3h���(�l*򐩻� �Nd4�� )͹W52U��X�� �2��� ��1��OC#���g����v.1�-I��ϐ˔{����s� �M:O��`:#��\v!Ux-j �^��CPE�A�gYP͐˶:΄� �Z�J|�hD���V��t�E��G��m����7s��P��?~|�����*���ܘ�f �J��(�!����n邒 @�'��i��r{)pk �ǐ��"s�؂vr�c�Mh+�(3�o4�����i�+Wb0pr8��!4�?� 2��?�+k��G���ژF����rurPc��NT65�A�7mc�"��m[���M�W�f������j�6l@]p���QkI:�h�/���*]�t���蓨��\���!�~b^��ƌt��b"�f ����R1������o*τ`��f�?�pn��T����Q��$T��Qh�1�ܠv�R*����%D�/:'0��6.���8ђ��P�E�4�\���̲ka��iQP�9Eٗ� 架�i�E'~b:�5I�40���� �_� DkX�-������ی�Dn�x����H���� *�d�᣺춢�(��,�ߏl�����o?����t�e��&�oS�� ��Å���Y�o��#�w�K>_JYNL0r��c}�Rj��sR�.|� ��c�m3���=f��'�9s�)�D��㫡.�Uܗ���O.�B\/��\C)�;ݹ�cPE�w�w���b�� ���������E���v���\3��;�1� j|sЎɷ����B_��O��q���;��-��3 � W����j)��L�/���S��|ȐY4���� ��P�����˦��B�͙ �Yd����XǨY���%3(ҵ(G�7��a� ���%=r���d�ݒ,�B�f���R �7��m�a�ׄ�c�,�=!�d�( �[m��v�w˳5b�v�d��ז���ߕce ���V�]�M*�o87SS�Y4�4/0Z��E z �U�{r�1ϖ|*e���Cl��+���6E���2�3Ye84�����܁�ڥ�wZ�k�x����ȣ�e$���������O'�4֐{�;�^�k7(3g'V�5��^?�'ս7 �S�����>f��v�m�w� �pc�00�M�W����b�F��!���A���2������o�!�ƭU�a�6�G��� %�U A�����-%�b���ieG���������>�V�k�`1f6�>��;# �����a��h@����^���b:rf=�), �ʂp98wX���S,��^1�-,(%9��6,� ���-S�"5�Y^��aAN��6]Q� ��-�U� ��NY�� ra�Y�b�[+,�tµ�b 2�M�-,Ȩ H�%(Z��� j��hAF{���H r�h�_b�-� ����N 2��d��qA�HY��o���b��\X��78� _�҂�0Ds���N >S�_V���ܧ�w�����~8\P�3��zTl���������G��{��d����g�>������S6Yz�N�@�&#�*���CI5~!�d�3��佳��k8C�ծx����a�>��_��W����~�Lvk�x��,���L�ro����'�2p2 ���ҏ:w0ٝ;Z�����R��.���r.D���ao=��k�A�Ym�c9bhʘk�\ �S� �g�#8_�¯����+�K>h��C[Xsm��D2��L[�>� �6�3J:A��x:B�|d�$�{T+A�c[�.�S޼|�$6�L1��ϯ�MA� $�)�3'� ��6 &ߦɶ!dD:����!�ZY#;B�g�5-b[!���ip"�Q�.�/\R���ۑ%�͎z��0���v���l�R��J�� )���n����0��P�BDiw��f��\�BI���v����� +��a�� Y�9�M�Ѡ��� ��BNV8M���l]i�+C���D��:C��y��!d�2�x�&�b=:�tHЕb� d~�NX��z���zb W��@�ǃ�C����+�0=`��ӏ��c9sA��.)�cn���HW���+����'���!�*`}ȵd��ׁ�gg��5��΅ EU��Q�{��9�{�� Řq���m�w� ��@��\��@O? � 6�6�|��\���5� rt��|�� ��"��e�s�e2}�o�eױ�g�䆽y��L#C5�wT?��yN?݁7a1ָ���zc��;��n��!��T��{�a�m�������Ƅ 7'���C��A�c:��~_� �1����ݫ��w���;ڼ�^�0(����i� �{�*�eG �G�P����1�r`��t"LVHm��� n�b��QKn�b�+��s�7\� dL���_H,�D ��˕���Yn�\�N+��o,�PQ�oL�B�l&+�M���ҵ3�V���sQӴ3�V��� �a� �.���3z����k^����c���� ��h�MP��a��z�.ѯE&�`ܖ���ǡ���b��7�,�� �o\:��'4`N�op� �6>��o���%�N[���Mn@�����o�h!*��C�o|DZ�P�>e��o|r��wM#�FUd�i��~��PSO��J�bqx �r��J��1h�����o��ౌ�=�+���yn�S�o�[j���.t�������k7��>��� \�M�\]`�E�V�/!��l��8f/�²�����ׯc�%� ��U��J�v��)�r�y�t� ^�dɍ��Ys���lmܕ{�P�\ݫ��w���;ڼ�^.�Z�D�Ku�$�:�y1�aWOiW��J�^�[���AɄa��j�R�9��Vec C���Z傾 %� �Jś��D�)j�92R^N��Ԫl8��u�Vy?T���[�}a_K]�U��EY�Vت�����*�T~*��*3�h��u(π�i�V�l���i�V���̞lŪ�L�����6tCa�L��K�Ӏ�2�x\���V�g����wl��7��ig�vl��v`r [��"����+#ѦZ���}�oU+nUrS�r��ʭJU�ل ��t�`�]Pܪ����6��*�����l^(n��(tFLn�� ���9/�G���/E!邪�� �ɗS����#v������ׄ&��6�ӏGy�0�r�Q �$s��Cx A��czZ�_@��z=�.�F�1~m�u�(�:���h����V��l�r���厝�=}.X�q��pԹ������ �:K���3׫��'��K��̅?�m� , �PG��v��4$�7�� ��D��h�x''��V��vW� ��elȜ ���m�50�9�R6�bj'�(�W"E�PI��/�_J�� �+q1޵�6�+گ�7_RZJ@�`�}0� ���wU�+{�d���e���NeR�/�%f�Nʫ�_��l0��/�� �Y7_��A���_�tp� l��ڲk�;��rA�� �K3���b���b�Go�{bY�e9�F��-�݀���_�=��Ϋ\�_�>rMG+��h��:�˝Zx|��N�R����lN�� �&%�K � �ʜ����;��H$ �,H�J�>S�������;�_O�r�8�[���Ǐ�� D�B� �֗ ��E�&~9���������O���~y�v�zW�W vڗ=��-li_)I�������g����vޗ ���p_�[�l����4���)��ڈiZc��h%����ώ�J|?r~��A�ʗxa{9s�}�#j{F�}�)�/��K*vC���{��ѣF�{@�������|��S� �ì�qz�_�� t6 Z :$���^���Mb�.?Ll� g��Qwt6�����x��g�XyX�.�N���v�� l?��f �>n�5����X6�0Sx�یcU�.��r�| > stream x�5�� �0D{���8������m�-�=��l`d��#����pSLRN�w j�7;�%��4�z.�q����Q$��D�˕g�_|> obendstream endobj 223 0 obj > stream x�=R��1 ˷ ��_R=��q`���=;11KA�nQ9?m%ˏ��egI���m���� s�q�b��$-e&�g���q j��3�!I�rR�"�������_7��tlPzK���Tþr�;��]�샦���3Ӯ쀥�ԤN��#,¤�HE!����,��۹~�]�XO� > stream x�5QIn1���@��'�����Q/��$��F$>Đ�(W|��4��;Y���J�佈 �*� i�rV3 ���3�Yf>�;PM�qv�4�A�i\G��0� qc�Y�bH��R����*�G��,�N��Jݢg�I\v^Omi_u7�;��"�,.:�n��)���!8-���T��&~f$xʞN�4��A���{AGn&d*�PF*��4ƌ���h&���������CMT�endstream endobj 225 0 obj > stream x�Mͻ � О) Q�"ٿ� ���t'�`� nqX �;���S�}��Ps�DX3%ڄ�_��0ØXȶ�7W�r���endstream endobj 226 0 obj > stream x�357U0P����� �F� )�\@>������L̀,CKd���!�eba��26���"X@lM��� �45endstream endobj 227 0 obj > stream x�MQIn�0 �������yO�A���%��#K\���D^�P�B���F^ ���֜���?�F��?T[ 1Q$tQ7�H7� �~��W��X�w+�[:v������*�B 7!Di�^�����um4��6�'��G���I���)f�l��m*V2 7����TFZ�6�2���2ZOv�&���'��q�.;;b ���>��|����i���qA"4ť�g���x��O\&endstream endobj 228 0 obj > stream x�375R0P��f�& �F� )�\@>���ehif�Y&�@���) " � a��`���9� $�6f[WW֔ endstream endobj 229 0 obj > stream x�3�4U0P����� �F� )�\@>����� �4XiLEWW�� Vendstream endobj 230 0 obj > stream x�355W0P����F �F� )�\@>���ehif�Y�@H�a���s`zr�2����endstream endobj 231 0 obj > stream x�=�� �0C�L��S�TU��׆|z�[ȸ  �څ�o��'u`]^Bd��;�Jf��&��$q�D�;MJ����������endstream endobj 232 0 obj > stream x�=�K� C��BG������[˙t�c�/DWt앇�Ɣ��hcO�-\�tNS���f]��a2�o����Ta�������f��U31�7�`*wr���8�ے�lWaН��0̑�=Ss�s���#_rkaЙ�-��eL��a�l- �Aw���z��0���A>dendstream endobj 233 0 obj > stream x�5R;қA�S��Y��y�ɤ�s�6;,�B��x�!�Q��%O0^'�w� > stream x�334T0P�f�� �F� )�\@>���eh`f�p�)��Y��P1��������9�e��L��4DjhWW?��endstream endobj 235 0 obj > stream x�=�Kn1 C�>�.��ٞ�(����'%��Ej��LYS�4����p���;l�ff�Z������b�|��F]Y'��f:��Q96M���,��.x�&�[�?�Р�5. 7tW�e)4c���{��2�jL]lR�{ > stream x�5QIn�0 ������'Š����R�@%��%bc#/1�9����5�&~grW�,O�G­�I���^v���e?3��VΤƓ �E6= j�D�(g����(���g�U!h]ݧmc�FJ�ӿ MM��D g�4��Ͻ���N!Tendstream endobj 237 0 obj > stream x�5O9�! �y�>0U�@����6�����N��!���x�##�f��Zd f�SLſ�����"�� a��p֬�n���v��X�6��Y^��L�Wg.�ci��9�n�]��u��SXG0� �t�Ô sT��Ɏ2��8�'�����,v9~� 6�!��*z�6���y�rA�]��E��%  �Qb��_�vt�(sB�A.!��*��P����RQp�>�謟�_\-endstream endobj 238 0 obj > stream x�=P;�D! �9�/�$�#pF�-f�߮�)PL~�3$��G 1���%������B�n��� �CR �z�t�6�:�3?a7c��E1��t�=&9�� �se�VH'��"��3�)�*{�x,��6['�=� �RR�ɥ���?mʔ :f�,��dM8˻IR��2��v"}� > stream x�=RKn�@��\����yRUݼ�okCR�*��1�0}ʐ��K]q�ɷ^�[ �ܗ�!�]t.�8G� 2�*D�ͪѡ�B���N� �}9��/��װ ��=2A�$�)B�nQ�Aa���P�Y��Q���2jo�� c� �BmH��@D�T���g$��gb`�Ѳ�TD�{�Lj�Ψ��D�>OM�(�L-V�nS_����|t*�4���U�� X�y�9�H���l!�:n���3�2�`K9`���G��Yu����t���pL��~��O�t�Z�u�r�@�MA��F���2>��)z�,���F3�a�����r�4�k"�X"��bD��ls=��L��9���l�֡�33*!�ں�j�@v���p��?3�mendstream endobj 240 0 obj > stream x�MQI�0 ���@!^���C�C�����9 �����X 1�,=��!s7�~�ٻYz������"SQ�R�.bB]�ϡ�=�kY���9,��s���3�c}I2���!N uZ�¸kb��Z� ���v�Bz�M;"� �2;-+�{��n��?�R��������\�endstream endobj 241 0 obj > stream x�E��� �TA �O&�����2|`�� �yF�&^XJc�P�@��r-��p'��T���wi�IU)ܤy�g&��y�c�� c�# > stream x�ERKnD1ۿSp�Hᗐ󴪺��[�L��� `lx�S�dʰ�%#$הO}"C���4r$"��I v���kK��쑨�����F��Z��ѥ͢Q���9�����{���z\H���!��P�,e�r %A[�w����+�a�k.��e2����\�}�b�.����������6)hU��I£�znv�l�U��30r��%[����}]�E���+@QY@���v��1�Hp���p > stream x�E��q1 C�%0� ����������AzB bO°\WqatÏ�-3 G���2c��X '� ���Ѥ�v�����b���\/:"̒�@#|:�Ǔ��3�t^�!��**�na.��@�R�ԏ�Qꚡ�*+kj�W�]J�>.�2NݽZὕ�=��?v:�endstream endobj 244 0 obj > stream x�32�P0P�4�& �f )�\@����B.H ��� ��%���g�� m� D���D����J%��endstream endobj 245 0 obj > stream x�32�P0P�4�& �f )�\�V.L,�іp "����g 'endstream endobj 246 0 obj > stream x�E�Kr D���#��� > stream x�E�;! C{N�#�#>�f2)6�oc�lR��X ����ڂ�肇�%��gc��6n5�u�V��h�R}��i�t�h6s+ �fz���:����r����Cp�_��b�9�������S��6;�������ܬ�~+�UaiYK� ���f@ �S9ګ=? :�endstream endobj 248 0 obj > stream x�5Q�m�0 �50 ~%�� H���;کHӼU^2�\.u�*Ya�Cu��|�f������!�S֖{��yJ��,�p 6��s ���+�Q�D�n���Y@�蚫�Xb �����V�)X���v$� r1Y�SE�)��|,ّZ��rY �n}a�8쳋D�y�L$�d�l�>[���E6v�8�Ѐ����$AS8�>�a�j���('89֯�1� N�KR�W��ϠA��3 ���.�I�wcJ��d)�6 ��S��b��Vo�����fQl������ܠbod�[���7�}�_{dendstream endobj 249 0 obj > stream x�=P��1 �]�X�z��g�K���H��"�ER*5��)/u�%YS�:|��y��nd6%*E/��%�� �}����Ֆ�C4�h9~ 3*��K6�p*����3� mtV��[ Ф`׶ r� ���" JM��r��^�>��C���5�X��s �S�'�no��^�����#]X�.i5AM��Z-���^��#� @���q�R���kp��/�'S�endstream endobj 250 0 obj > stream x�E�� �0D{�`~&�>Q��޿ %n�'�=\ �)n3�C� ,�]q���0 \GP���aվ����R�������Bfendstream endobj 251 0 obj > stream x�MP�m1 �� 50���� > stream x�5RK�#1��)��T��1>O�^�E��ۑ�{��$f�!C�?�R{� ���,[���ɽ�J�%s�Քϓb������"�BG5�$ L�(a�ϳ`���bvm?Յv�����x2@f)��K��P���$m��|��3T9�Z��b��h�!7��م�>��IdȠbJ�� T�ك�����;�9�����Y�ֱ��,�A�q �U\�U;v�F����j_ov�R���y8[f�=�����1�p�@�.z�w�CD��j �����y�������B��9 ����x%�a䦴��� ���Y�E����4`Wdi���h��4���U����endstream endobj 253 0 obj > stream x�36�P0P040�F�@���B�!H��� �Y@�8�&�+�+ � �endstream endobj 254 0 obj > stream x�E�91s��'H\���˵�����G��4�M�zh>�F���k�,J�D��P�c�;(HLHc�H��#�X�]��T*�R�i�¿�Z��D�p����T�O�\��M���1�T�;��I�^E�Z1��D0�� oE&����پ�`��� �8�� L�7�endstream endobj 255 0 obj > stream x�-RKr� �s ]�3�� �s--|FЀ_䟯����i��mӮz�L;���lOs^&_Zg�:�����R���3�s�(�VF�v6Hj\lo��XJЅn\F�#���" > stream x�336S0P�0��� �F� )�\@>������,́,# ��.C c0mbl�`fbdY 1 �2����endstream endobj 257 0 obj > stream x�5RKn1��)�@��O�󪪋��ok�L0`�)/Y�K��%ۥ��[�%�Bޏ��R����-�G��- Q=X�/"����:�pJ��W���I�j��3 P�@ƻb2R�$�siq�r&�����'����=Y��ߓK��MN����I�]I����JD���}�B���<>G1ɀ�6�m�&��%��:���f��U�`��)�W ��l�[�����0��0Sb����be����5�ˆ�Q�]�p��w�h���c��G�� i8����^8�^�9��c��, ch&�z���О��a��Z��T�|�endstream endobj 258 0 obj > stream x�36�P0��C�4�Rendstream endobj 259 0 obj > stream x�E�K!D�������q2�s�퀝nc=�T����=.k�!�񷩖�EY>TzRV������!ܠ,��I*�!P���-en���6 ���S�����0���l_�["u�?��*/��ȥֻޢ���1 endstream endobj 260 0 obj > stream x�5R9n1 �� } �n��� H��� ��qtR�;ZTvˇ�T��t����7�we�� ɫ�R�H��V!��2��5g�R����qɂ?å� ��������X#�D�2������f[~�i��x��+�-�XR�8���y-������ V�~A�G�}�TX ��Ip��P��v��H 9{,vQ�3(1��E�A�%��9źt���б��،7�:*�[ZYc�>��,2�zX�t��ǃ 4���� 4e�����9���QVx�C� �1�!��� J� > stream x�-QIrA��+�f��c�+����� �@tZ⠌� ,W�呫h��31���~+�mX�@T IcP5�v�z��q� ְ̓ f���)����1g��e��]�=�Aɬ\�%L�a#g?�2.�s=7 �FX�a�nV3�4�H!�����A��F�6r��/�t�U��1�s-m_�M8k��X���]q.'/���6Qi��@H��n�s���xK����{����T`ś�n@6��߼���������_`[�endstream endobj 262 0 obj > stream x�M�IC! C�����3�� �1Aݢ-X�v���s�21�F�� > stream x�3�4R0P06�fF �&� )�\@>���ehd f�pY�)XX&f�P!��.cSs�@EƦ`�?�+�+ ���endstream endobj 264 0 obj > stream x�=���0C�� �@��)�|O�j����#K� zc������Æ`����%Tk��@%7ș`���G�zb8\�������f}��B�%h�{�Siܦq�5�)ꜣ����g�4��4��s� �{�S��������w.�endstream endobj 265 0 obj > stream x�5Q9! ���@$�/xO�(���͌�V�\���!��R��!�S>z�.�����j4疴-s����/����f���i��� ��n�c����!� b(,AͩEm�'�5�^�ޗ���S�Ɩ�V8����=O&�t2ԫ5� E" > stream xڕْ۸�}�Bo��2\���T*�v��V���M��őS�L��ߧ �3k��@���P�9n��ۛ�|�zw���mX��i�6w���HJ�� �9l�o�N���nE!���B���?N4���#��� Pٳ��݇�mnY������9"���k�M��H8�� ��ѷ�h��k`j�i>j"�������xfCx�ۦ��dw�gl��.SۆV�A���� �J��#h�0S��/ �S3�nA��; �x���]����> ]5�D��" �k� �v�G�z�a�4�h �ohj2�ى� �����# �e�w����O�j=�D���#�BE�0G�GF��O��R����0�\���}un�f�8�F�u�!���6ڊ��5��4��GtHz�̶oНu )�ir� �J�mk����;8�#B9�L�pI�r����b�7���y*'ٴ���nVi�C�� K6FR-�ם>����z%TJ�(1- �R]�q�����9�Ÿm�=H��TJ�z�s���U8�kB=w��B������[��CT2쫯�F��w *��lj�G �ƒ�J�c 3}����!��},׃�b � ��q�7 �Nlk ���.2�h��iE]�̾��L,l��X��{��\/=*� ��;sA"�}�h�Kbj,�u��ܥg}͌��r@Z�Qѭ�(����|�^���~�fHb��w�t2�H�t���m�hu��X�7@Q���sЫ���˄��b Et����m�k�a�^���p�ʞ�$� �ߘ� ��-ʊS#�ڎ�r�\�"ϓ��PwpA��ޮ5=`5AZ?��]��@�u �i֮�TzZZK�r������� V?� ��|ԭ��OzO���=pi��U����t���u�*�ױ�r� ��}˭����h �"I�+,_��S v3-�א��}my��"�l�{���R]m)����d�8P��&bendstream endobj 284 0 obj > stream xڕVK��6 ��W�H�T\>%1�f��ln�l&��Y֮�ڒ�G6���W�zg��A�>�E���쏍��������dRp'����Je���� ���Ls�U^H!������S�����[]�ω�P��x�\ ���� ���+ٜK�\[&]m!G���qQ������Y��|�FSqgu��7]2����i�N��ln��7�C�,{Ν` �щ 4��/ �z�� � ��T�J-�e�HukD��W���� O��Wy�Ӄ��_Q�l�����)�����h.lI~��M�} > /A2 > /A3 > >> /Font > /Pattern > /ProcSet [ /PDF /Text /ImageB /ImageC /ImageI ] /Shading > /XObject > >> >> stream x��XMo7��W�Lq�� �E�VE.כ�����z��{�x�����LEg2iaC\����u�ny&3������s�����Y���M�ӥ~��Ћ\LQq�}����!�W��NY#�R�Gc���Dǡ�X��'�:�wC���%C��a ���r 9��;���Ă�mD8A��%Š/n��R-~(�������z��QN�9�Y�R�/�����.M�{lAU`ifyX���d�(e�2��t,.�i�,�/�5a�aĞ]���k�2@����R͕z��;���-�{�#*"WȐ�"�E�gDCl��)��ܫ:5h/�F�I��iF'��p 6R+�B�b� Kt��P��P[LE��iN�,F$$J��4+�;�P��x�˼�K��ij`��}0�K��nn��tr���b˼�G�-����4�Sv�stD��g�����: "c0y����kУ�zΦk�@?��m���h�� > stream x�m�A� E�=E/�IKE� n=���$�;��M����/ oT �O���4K�7 )��Ĉa^-�r �b�\��� 7F�KaJm��7��n�=���b����J�g1d•cp���=���E �!��c|Dendstream endobj 294 0 obj > stream x�5�K�0D�>�\�_CΓ��"���`'�O��B� �X1�r�ͣn�~�s.�I�M�Uv�H�9p�T( �V�9�U@������rW^n;p�lG���j���đ�R~ x���6���y��z��5X)���� ̴j����������`6�endstream endobj 295 0 obj > stream x�334T0P�f�� �F� )�\@>���eh`f�p�)��Y��P1��������9�e��L��4DjhWW?��endstream endobj 296 0 obj > stream x�=�Kn1 C�>�.��ٞ�(����'%��Ej��LYS�4����p���;l�ff�Z������b�|��F]Y'��f:��Q96M���,��.x�&�[�?�Р�5. 7tW�e)4c���{��2�jL]lR�{ > stream x�E�Mr!����/��W=ϤRYL� ͼ$+Z���G�C쉕2|/�2|�2;�:ٙpO�n\�e�3a[�;*�;^Cw�R?����9�.R5ted�z�� ̐��Yt+�u@Ӕ�R�l[��J�'��4���w�k|uO���T�!Zͭ��lxA��$� �'�s!Q���:.`�%��ܠ�f|zRI0�|��h�6Г�gƚKE�W4��;I5���ⶾ�6q����w��+}�>K�]endstream endobj 298 0 obj > stream x�=P;�D! �9�/�$�#pF�-f�߮�)PL~�3$��G 1���%������B�n��� �CR �z�t�6�:�3?a7c��E1��t�=&9�� �se�VH'��"��3�)�*{�x,��6['�=� �RR�ɥ���?mʔ :f�,��dM8˻IR��2��v"}� > stream x�=RKn�@��\����yRUݼ�okCR�*��1�0}ʐ��K]q�ɷ^�[ �ܗ�!�]t.�8G� 2�*D�ͪѡ�B���N� �}9��/��װ ��=2A�$�)B�nQ�Aa���P�Y��Q���2jo�� c� �BmH��@D�T���g$��gb`�Ѳ�TD�{�Lj�Ψ��D�>OM�(�L-V�nS_����|t*�4���U�� X�y�9�H���l!�:n���3�2�`K9`���G��Yu����t���pL��~��O�t�Z�u�r�@�MA��F���2>��)z�,���F3�a�����r�4�k"�X"��bD��ls=��L��9���l�֡�33*!�ں�j�@v���p��?3�mendstream endobj 300 0 obj > stream x�MQI�0 ���@!^���C�C�����9 �����X 1�,=��!s7�~�ٻYz������"SQ�R�.bB]�ϡ�=�kY���9,��s���3�c}I2���!N uZ�¸kb��Z� ���v�Bz�M;"� �2;-+�{��n��?�R��������\�endstream endobj 301 0 obj > stream x�E��� �TA �O&�����2|`�� �yF�&^XJc�P�@��r-��p'��T���wi�IU)ܤy�g&��y�c�� c�# > stream x�32�P0P�4�& �f )�\@����B.H ��� ��%���g�� m� D���D����J%��endstream endobj 303 0 obj > stream x�5Q�m�0 �50 ~%�� H���;کHӼU^2�\.u�*Ya�Cu��|�f������!�S֖{��yJ��,�p 6��s ���+�Q�D�n���Y@�蚫�Xb �����V�)X���v$� r1Y�SE�)��|,ّZ��rY �n}a�8쳋D�y�L$�d�l�>[���E6v�8�Ѐ����$AS8�>�a�j���('89֯�1� N�KR�W��ϠA��3 ���.�I�wcJ��d)�6 ��S��b��Vo�����fQl������ܠbod�[���7�}�_{dendstream endobj 304 0 obj > stream x�E�� �0D{�`~&�>Q��޿ %n�'�=\ �)n3�C� ,�]q���0 \GP���aվ����R�������Bfendstream endobj 305 0 obj > stream x�36�P0P040�F�@���B�!H��� �Y@�8�&�+�+ � �endstream endobj 306 0 obj > stream x�MQ��1 �=�8��پy.xx�e�6���DCI��S�d�a}JFH�.mKl� > stream x�E�91s��'H\���˵�����G��4�M�zh>�F���k�,J�D��P�c�;(HLHc�H��#�X�]��T*�R�i�¿�Z��D�p����T�O�\��M���1�T�;��I�^E�Z1��D0�� oE&����پ�`��� �8�� L�7�endstream endobj 308 0 obj > stream x�336S0P�0��� �F� )�\@>������,́,# ��.C c0mbl�`fbdY 1 �2����endstream endobj 309 0 obj > stream x�5RKn1��)�@��O�󪪋��ok�L0`�)/Y�K��%ۥ��[�%�Bޏ��R����-�G��- Q=X�/"����:�pJ��W���I�j��3 P�@ƻb2R�$�siq�r&�����'����=Y��ߓK��MN����I�]I����JD���}�B���<>G1ɀ�6�m�&��%��:���f��U�`��)�W ��l�[�����0��0Sb����be����5�ˆ�Q�]�p��w�h���c��G�� i8����^8�^�9��c��, ch&�z���О��a��Z��T�|�endstream endobj 310 0 obj > stream x�E�K!D�������q2�s�퀝nc=�T����=.k�!�񷩖�EY>TzRV������!ܠ,��I*�!P���-en���6 ���S�����0���l_�["u�?��*/��ȥֻޢ���1 endstream endobj 311 0 obj > stream x�5R9n1 �� } �n��� H��� ��qtR�;ZTvˇ�T��t����7�we�� ɫ�R�H��V!��2��5g�R����qɂ?å� ��������X#�D�2������f[~�i��x��+�-�XR�8���y-������ V�~A�G�}�TX ��Ip��P��v��H 9{,vQ�3(1��E�A�%��9źt���б��،7�:*�[ZYc�>��,2�zX�t��ǃ 4���� 4e�����9���QVx�C� �1�!��� J� > stream x�-QIrA��+�f��c�+����� �@tZ⠌� ,W�呫h��31���~+�mX�@T IcP5�v�z��q� ְ̓ f���)����1g��e��]�=�Aɬ\�%L�a#g?�2.�s=7 �FX�a�nV3�4�H!�����A��F�6r��/�t�U��1�s-m_�M8k��X���]q.'/���6Qi��@H��n�s���xK����{����T`ś�n@6��߼���������_`[�endstream endobj 313 0 obj > stream x�M�IC! C�����3�� �1Aݢ-X�v���s�21�F�� > stream x�=���0C�� �@��)�|O�j����#K� zc������Æ`����%Tk��@%7ș`���G�zb8\�������f}��B�%h�{�Siܦq�5�)ꜣ����g�4��4��s� �{�S��������w.�endstream endobj 315 0 obj > stream x�5Q9! ���@$�/xO�(���͌�V�\���!��R��!�S>z�.�����j4疴-s����/����f���i��� ��n�c����!� b(,AͩEm�'�5�^�ޗ���S�Ɩ�V8����=O&�t2ԫ5� E" > /A2 > >> /Font > /Pattern > /ProcSet [ /PDF /Text /ImageB /ImageC /ImageI ] /Shading > /XObject > >> >> stream x�͝[�$7����W�����~yl���b�Lok�X �A[��rP�QUi�������gVJ3�n�i�4��C�}����>�y�?�����{���ۛ������Os�w�����/��O����z㎹��������t�!f����������`�3��/�M���1���y��љ����t�5�cp1|� �~9l�Tk��s����W{4Λ��������>����s��6��s��e������?�jlM&�m����hM��O����𷇿~9����-�,�mx�%{����䴇���+����o�Tæ�{��v�37�~:�x���|������X&5��>���6�?� k����mr1���K��J9ؙ�S9T[��i�]T%y�B+7_����uU�g����� =���ݗ_?����?�񛃇�b����� �����1��3�T]OU>�]w=�]����p�i��k�3�q�䝺Bu_�}�=�zT�>G\2�1��'�����h�1 ��o�I0|' ����q�x���Ѓ�jB�Ġ\�K�} #��2�� eg}s~��*5�V���g9�1t,^v6 ^�"_��F?�'˪ $4E�:�m��R+/��P�3�6��H����++%+�*��*��(kv�Ph��ѽ�V���jR�� ��R����x�ֆ��G[bq��#���I�đ���� �I�d�I1ja2VL����E��S��ڄ_�h�u��/7?}٫##^S���~����P���;]>V_��&�7��s�㌖��� ��8�V�� �F�I�;pG� c���WU>�F���]`4_,�w�����%��f}R~N�`�~��eY�z �$��f�**�ȝ�rr��Y�_��'��i'6��G����?:� i����l��v2^`+�=SG/��� |�$ 5��mX����f�&UE��51�Vk��-Aw��q%1�����R��l�I��r,�Dp&��*��E�L��v��R���:�`eb c�6q}���ѣg�zu �����;/�Y�\��η�`]&八����se�%}{Q����&���Z=P������?������Jy���mV�A���i�[�I�X�\��O~�w�*�\�ۧ��|^��0z6���` ׽�i�����-{�ޛ/N��;L79nz}/߹}~�^�f�PvN�dLֶ\��F�&�Iy��8�B� ٺ�0�Omqu�`��G!CR�L_9o]�h�- ��"���x[���p�i�ܵ_� l H�qµKD:$�8S�\X�Ѝ �X+�}K"@9�f��mmI���J�j1�k�$�GWE�V�pCK"���-g���P�D������5��$�J�ڋ�u�E���M�J�}�"�����ҡ*X&H̤Z�d���m��E`������ �ki �J9�[� �V�U@ni\�jR)��li����(�j�������{ؒH�hz -��X�|]&�9�t^�0�`J`�V�\�%�[v�f� ����ㇷ�w' X#����i�C��q�^KGh+�a�ʳ��� ��؇e�V�Ov����F{̞�Fw���}�ɯʟ��F���+s0�~�~!T�^�|ɤa�ܓ��������67�*cE�����'���� ���K���{�3� ��%� �\m�f�1HF���$�BSk��N̹��t�>�㷓�y1�࣓�IM8%�2�)?�*@O ��O1#���"j�`�Ѵ�t&(c����� *áv�e�Y�F��`������/��� mC�2%��fƋ!� Ot�eN��-��2�!ּ|��|�-`ҫ���r�KS$����"�%�215�%��"�j�}�0���!����& ����՛ɩP��0�j -Z�1�S������D -�S�!%����,+��Ȱ0*���"[-�S��e� Aqǖ1�箊�BJ������ 2�O��~1�4��J5�,�.*.�u;�[��f�͙ =��߾�����Pe��� Fm+��*+W��Y�Ŷe�bB��ּ��V�����v�\?�$pv�d�\ -I��p�� ����:a��;s���@�}��4�qU>�� ��g�u��6��8����ўd���/e��0ip��*���`o��h�/�%� �LW}��MB*q��{̱��k-s����֙ ���-dL�!��\�"�JH%8 ��1L+ ���d[�gᣒ���K]4��G%����$EF�QIvq3��0�Qɑa�X�H� �,cV+�=�Ȳd���d%(��i��k��J�%h/����d�|�AD��N����>����eLR5 ~�������G; �� 6�3ڼE��,���$�L���ݙ�y\�����x��ʢv��xӿ�i��Lw��zr2�}98C#�=��ê|�ŃM����-'g��c��Jn���o�����D7�d����l1��d�5I�T�dT��Vĕ"�Q�j��cE��w���ݐd�[��@���P�Yv�H� �����i�6y��P���,W1E�A���L�I�3�Vo�L��c����6s��{ssfj��n.�ͪt�ͣI����M73.�]�,Z�� /��)j�;��� 5�C�%��b۩1��Z�-9�ISc }A .dS���*��da�VjLv��ù��i���-���m��-x��Cuj� ��m���Ö)ײ �㉏ l��^>��������5_B�FS���L��,�� Ak@ ��7p�3G֗��Zj�nV��yN�ރ7�,���;F7�[����b0�S'�k�T�4�4TP�h�Ԅ!V"[9d�~5��I|�� �竅ږ�: be�%��F���jV��C÷�X�8�F|ɌHk�81R U�.M{�4j� �i��v �=�����Aw������_!,'%8�᎛�*wx�!AwH8�������!��܇:j&iV?��yv��ݠ�7m�����=V���ogG cٶ��氜�� ��]�r�\}� 3�����/f۴� F}0����ɣ��km�z!�����F�|OFN�pD�`Ο�œ?GkW���ɼ�JfJ�����Gg�1�p z�)���]+8a(�����w���n����Ljl�+9F#�`W�^l%��Y�fr��J|��H-E�1��C�����9Կ �[�1�C�ռ!� j��fa(�"y';�z�ߛ�ӕS��OyL����p�/5����/��� ��2Aat���0���k=#�%���@%��O��m�/�� d���M�6�R���q�E���=�����7(�c!��?|���]~�2�#y��ȡʧh���B�����d��o��H��MnR��'!" ���z���yR�;yr�u͓��1�*'�����v�c�E����T�S��-�m]q%U �F(��)����Y��R/&q����l����# ��]�c� �:���0����Ot�� 6�%�{��O�S���D72�O��\u}�1���C/:��rfW7����稣MK��6Z��=6l#^ �/gHt����]S��F����f���\��0��m�0�ov��B)�|�LW|$��x�f����m`�i�� m�R��KFs�`pE-���(� m��ClI ��h#kRx�(��6�#޷ �r���/s^���p`y/,�X���^6 2js�c,��-@���1�O�C}W�u ���� �v]� ��j#��iV@��2�Aj3�N�XN��E�+>�2����P|�3��}zIv�x�-E��jA6��V����㻏�?�����C�S+�@?�*����W�l�&1��芪�ix����@�P1���Ӑ�J��~[(.g�t�x�-n�|���y��� =����G�z'!��߼j4�`m��yTT�4:�T�h�j��4h��x����|켤"� hۋ�=�-*o���9�9��p}��(1��@��@��[� [�,^Mw[+ڣ9TD1�]�+��B��݄��ښ$l* ������lE׊� S�_�����;�Z,�؁����~W;�(JA?�.�~#�~6?�;n+HWPOK��&_N������E�~^�4�3v�^>�~����5D�fx�� W�_4������6�6��jo��:X ��l���r�b�i� /�v}��{ ��/��r ���߭���q�� 6��F���K)��K&��C;�с�ǐX�#�*�g;G�����d;� ���b��VH��i�w�斣�|H/U:��Ѥr�h���O�+�����U7?Y����ږ bs��� )����䌻�ꖠ�P84k� ?݀������Ӓ���Y�z/T;��,���f{�%�1�,�: ^Dmj�M�,%;�Q������ġ��QU> 6�rx���`3��6���@M�xMe2O����f��T��r]mg����f y���l��Ae��.�5�v&k���գ�q�g�Rx���. �ɴ�']T��jh;��+%,:�E� �nwQ=@ U�C�P���=Ԧf_��X���r�vC�1��zʧq�Q��X� ���vk����&0I)�ZR�~�+���gƚ��jLj��uM���nƚ�5�r%���&�>�D)U��urv~��r]mg��)��O�D��i�ʩe��.ו�v&+�iv׏�ʾ��ѭx��PE+x��̖�8|0�����kH*F�6����t㙁F0�`{hT�0�0pP��2{������I�9_?^�7����f��P��r]mg����fw ]���fD�PC%^��r]mg��Z��o@��v���1TO )C�t����3Y��P����d�]C�D�*j�2�q�`x���q#jvWS�2T�� 4��endstream endobj 321 0 obj > stream x�5�� �0D{���8������m�-�=��l`d��#����pSLRN�w j�7;�%��4�z.�q����Q$��D�˕g�_|> obendstream endobj 322 0 obj > stream x�=R��1 ˷ ��_R=��q`���=;11KA�nQ9?m%ˏ��egI���m���� s�q�b��$-e&�g���q j��3�!I�rR�"�������_7��tlPzK���Tþr�;��]�샦���3Ӯ쀥�ԤN��#,¤�HE!����,��۹~�]�XO� > stream x�5QIn1���@��'�����Q/��$��F$>Đ�(W|��4��;Y���J�佈 �*� i�rV3 ���3�Yf>�;PM�qv�4�A�i\G��0� qc�Y�bH��R����*�G��,�N��Jݢg�I\v^Omi_u7�;��"�,.:�n��)���!8-���T��&~f$xʞN�4��A���{AGn&d*�PF*��4ƌ���h&���������CMT�endstream endobj 324 0 obj > stream x�Mͻ � О) Q�"ٿ� ���t'�`� nqX �;���S�}��Ps�DX3%ڄ�_��0ØXȶ�7W�r���endstream endobj 325 0 obj > stream x�357U0P����� �F� )�\@>������L̀,CKd���!�eba��26���"X@lM��� �45endstream endobj 326 0 obj > stream x�MQIn�0 �������yO�A���%��#K\���D^�P�B���F^ ���֜���?�F��?T[ 1Q$tQ7�H7� �~��W��X�w+�[:v������*�B 7!Di�^�����um4��6�'��G���I���)f�l��m*V2 7����TFZ�6�2���2ZOv�&���'��q�.;;b ���>��|����i���qA"4ť�g���x��O\&endstream endobj 327 0 obj > stream x�375R0P��f�& �F� )�\@>���ehif�Y&�@���) " � a��`���9� $�6f[WW֔ endstream endobj 328 0 obj > stream x�355W0P����F �F� )�\@>���ehif�Y�@H�a���s`zr�2����endstream endobj 329 0 obj > stream x�=�� �0C�L��S�TU��׆|z�[ȸ  �څ�o��'u`]^Bd��;�Jf��&��$q�D�;MJ����������endstream endobj 330 0 obj > stream x�=�K� C��BG������[˙t�c�/DWt앇�Ɣ��hcO�-\�tNS���f]��a2�o����Ta�������f��U31�7�`*wr���8�ے�lWaН��0̑�=Ss�s���#_rkaЙ�-��eL��a�l- �Aw���z��0���A>dendstream endobj 331 0 obj > stream x�5R;қA�S��Y��y�ɤ�s�6;,�B��x�!�Q��%O0^'�w� > stream x�5�K�0D�>�\�_CΓ��"���`'�O��B� �X1�r�ͣn�~�s.�I�M�Uv�H�9p�T( �V�9�U@������rW^n;p�lG���j���đ�R~ x���6���y��z��5X)���� ̴j����������`6�endstream endobj 333 0 obj > stream x�334T0P�f�� �F� )�\@>���eh`f�p�)��Y��P1��������9�e��L��4DjhWW?��endstream endobj 334 0 obj > stream x�=�Kn1 C�>�.��ٞ�(����'%��Ej��LYS�4����p���;l�ff�Z������b�|��F]Y'��f:��Q96M���,��.x�&�[�?�Р�5. 7tW�e)4c���{��2�jL]lR�{ > stream x�E�Mr!����/��W=ϤRYL� ͼ$+Z���G�C쉕2|/�2|�2;�:ٙpO�n\�e�3a[�;*�;^Cw�R?����9�.R5ted�z�� ̐��Yt+�u@Ӕ�R�l[��J�'��4���w�k|uO���T�!Zͭ��lxA��$� �'�s!Q���:.`�%��ܠ�f|zRI0�|��h�6Г�gƚKE�W4��;I5���ⶾ�6q����w��+}�>K�]endstream endobj 336 0 obj > stream x�5QIn�0 ������'Š����R�@%��%bc#/1�9����5�&~grW�,O�G­�I���^v���e?3��VΤƓ �E6= j�D�(g����(���g�U!h]ݧmc�FJ�ӿ MM��D g�4��Ͻ���N!Tendstream endobj 337 0 obj > stream x�5O9�! �y�>0U�@����6�����N��!���x�##�f��Zd f�SLſ�����"�� a��p֬�n���v��X�6��Y^��L�Wg.�ci��9�n�]��u��SXG0� �t�Ô sT��Ɏ2��8�'�����,v9~� 6�!��*z�6���y�rA�]��E��%  �Qb��_�vt�(sB�A.!��*��P����RQp�>�謟�_\-endstream endobj 338 0 obj > stream x�=P;�D! �9�/�$�#pF�-f�߮�)PL~�3$��G 1���%������B�n��� �CR �z�t�6�:�3?a7c��E1��t�=&9�� �se�VH'��"��3�)�*{�x,��6['�=� �RR�ɥ���?mʔ :f�,��dM8˻IR��2��v"}� > stream x�=RKn�@��\����yRUݼ�okCR�*��1�0}ʐ��K]q�ɷ^�[ �ܗ�!�]t.�8G� 2�*D�ͪѡ�B���N� �}9��/��װ ��=2A�$�)B�nQ�Aa���P�Y��Q���2jo�� c� �BmH��@D�T���g$��gb`�Ѳ�TD�{�Lj�Ψ��D�>OM�(�L-V�nS_����|t*�4���U�� X�y�9�H���l!�:n���3�2�`K9`���G��Yu����t���pL��~��O�t�Z�u�r�@�MA��F���2>��)z�,���F3�a�����r�4�k"�X"��bD��ls=��L��9���l�֡�33*!�ں�j�@v���p��?3�mendstream endobj 340 0 obj > stream x�MQI�0 ���@!^���C�C�����9 �����X 1�,=��!s7�~�ٻYz������"SQ�R�.bB]�ϡ�=�kY���9,��s���3�c}I2���!N uZ�¸kb��Z� ���v�Bz�M;"� �2;-+�{��n��?�R��������\�endstream endobj 341 0 obj > stream x�E��� �TA �O&�����2|`�� �yF�&^XJc�P�@��r-��p'��T���wi�IU)ܤy�g&��y�c�� c�# > stream x�ERKnD1ۿSp�Hᗐ󴪺��[�L��� `lx�S�dʰ�%#$הO}"C���4r$"��I v���kK��쑨�����F��Z��ѥ͢Q���9�����{���z\H���!��P�,e�r %A[�w����+�a�k.��e2����\�}�b�.����������6)hU��I£�znv�l�U��30r��%[����}]�E���+@QY@���v��1�Hp���p > stream x�E��q1 C�%0� ����������AzB bO°\WqatÏ�-3 G���2c��X '� ���Ѥ�v�����b���\/:"̒�@#|:�Ǔ��3�t^�!��**�na.��@�R�ԏ�Qꚡ�*+kj�W�]J�>.�2NݽZὕ�=��?v:�endstream endobj 344 0 obj > stream x�32�P0P�4�& �f )�\@����B.H ��� ��%���g�� m� D���D����J%��endstream endobj 345 0 obj > stream x�32�P0P�4�& �f )�\�V.L,�іp "����g 'endstream endobj 346 0 obj > stream x�E�Kr D���#��� > stream x�E�;! C{N�#�#>�f2)6�oc�lR��X ����ڂ�肇�%��gc��6n5�u�V��h�R}��i�t�h6s+ �fz���:����r����Cp�_��b�9�������S��6;�������ܬ�~+�UaiYK� ���f@ �S9ګ=? :�endstream endobj 348 0 obj > stream x�5Q�m�0 �50 ~%�� H���;کHӼU^2�\.u�*Ya�Cu��|�f������!�S֖{��yJ��,�p 6��s ���+�Q�D�n���Y@�蚫�Xb �����V�)X���v$� r1Y�SE�)��|,ّZ��rY �n}a�8쳋D�y�L$�d�l�>[���E6v�8�Ѐ����$AS8�>�a�j���('89֯�1� N�KR�W��ϠA��3 ���.�I�wcJ��d)�6 ��S��b��Vo�����fQl������ܠbod�[���7�}�_{dendstream endobj 349 0 obj > stream x�=P��1 �]�X�z��g�K���H��"�ER*5��)/u�%YS�:|��y��nd6%*E/��%�� �}����Ֆ�C4�h9~ 3*��K6�p*����3� mtV��[ Ф`׶ r� ���" JM��r��^�>��C���5�X��s �S�'�no��^�����#]X�.i5AM��Z-���^��#� @���q�R���kp��/�'S�endstream endobj 350 0 obj > stream x�E�� �0D{�`~&�>Q��޿ %n�'�=\ �)n3�C� ,�]q���0 \GP���aվ����R�������Bfendstream endobj 351 0 obj > stream x�MP�m1 �� 50���� > stream x�5RK�#1��)��T��1>O�^�E��ۑ�{��$f�!C�?�R{� ���,[���ɽ�J�%s�Քϓb������"�BG5�$ L�(a�ϳ`���bvm?Յv�����x2@f)��K��P���$m��|��3T9�Z��b��h�!7��م�>��IdȠbJ�� T�ك�����;�9�����Y�ֱ��,�A�q �U\�U;v�F����j_ov�R���y8[f�=�����1�p�@�.z�w�CD��j �����y�������B��9 ����x%�a䦴��� ���Y�E����4`Wdi���h��4���U����endstream endobj 353 0 obj > stream x�36�P0P040�F�@���B�!H��� �Y@�8�&�+�+ � �endstream endobj 354 0 obj > stream x�MQ��1 �=�8��پy.xx�e�6���DCI��S�d�a}JFH�.mKl� > stream x�E�91s��'H\���˵�����G��4�M�zh>�F���k�,J�D��P�c�;(HLHc�H��#�X�]��T*�R�i�¿�Z��D�p����T�O�\��M���1�T�;��I�^E�Z1��D0�� oE&����پ�`��� �8�� L�7�endstream endobj 356 0 obj > stream x�-RKr� �s ]�3�� �s--|FЀ_䟯����i��mӮz�L;���lOs^&_Zg�:�����R���3�s�(�VF�v6Hj\lo��XJЅn\F�#���" > stream x�336S0P�0��� �F� )�\@>������,́,# ��.C c0mbl�`fbdY 1 �2����endstream endobj 358 0 obj > stream x�5RKn1��)�@��O�󪪋��ok�L0`�)/Y�K��%ۥ��[�%�Bޏ��R����-�G��- Q=X�/"����:�pJ��W���I�j��3 P�@ƻb2R�$�siq�r&�����'����=Y��ߓK��MN����I�]I����JD���}�B���<>G1ɀ�6�m�&��%��:���f��U�`��)�W ��l�[�����0��0Sb����be����5�ˆ�Q�]�p��w�h���c��G�� i8����^8�^�9��c��, ch&�z���О��a��Z��T�|�endstream endobj 359 0 obj > stream x�36�P0��C�4�Rendstream endobj 360 0 obj > stream x�E�K!D�������q2�s�퀝nc=�T����=.k�!�񷩖�EY>TzRV������!ܠ,��I*�!P���-en���6 ���S�����0���l_�["u�?��*/��ȥֻޢ���1 endstream endobj 361 0 obj > stream x�5R9n1 �� } �n��� H��� ��qtR�;ZTvˇ�T��t����7�we�� ɫ�R�H��V!��2��5g�R����qɂ?å� ��������X#�D�2������f[~�i��x��+�-�XR�8���y-������ V�~A�G�}�TX ��Ip��P��v��H 9{,vQ�3(1��E�A�%��9źt���б��،7�:*�[ZYc�>��,2�zX�t��ǃ 4���� 4e�����9���QVx�C� �1�!��� J� > stream x�-QIrA��+�f��c�+����� �@tZ⠌� ,W�呫h��31���~+�mX�@T IcP5�v�z��q� ְ̓ f���)����1g��e��]�=�Aɬ\�%L�a#g?�2.�s=7 �FX�a�nV3�4�H!�����A��F�6r��/�t�U��1�s-m_�M8k��X���]q.'/���6Qi��@H��n�s���xK����{����T`ś�n@6��߼���������_`[�endstream endobj 363 0 obj > stream x�M�IC! C�����3�� �1Aݢ-X�v���s�21�F�� > stream x�3�4R0P06�fF �&� )�\@>���ehd f�pY�)XX&f�P!��.cSs�@EƦ`�?�+�+ ���endstream endobj 365 0 obj > stream x�=���0C�� �@��)�|O�j����#K� zc������Æ`����%Tk��@%7ș`���G�zb8\�������f}��B�%h�{�Siܦq�5�)ꜣ����g�4��4��s� �{�S��������w.�endstream endobj 366 0 obj > stream x�5Q9! ���@$�/xO�(���͌�V�\���!��R��!�S>z�.�����j4疴-s����/����f���i��� ��n�c����!� b(,AͩEm�'�5�^�ޗ���S�Ɩ�V8����=O&�t2ԫ5� E" > stream x��ے���}�b� ��94 ۳�K��}�v�7�WW��oW1?_n��~��J�Qb��_�yT$K� ����z�7���(�׿�*ivu��q�>u�F����� ������qy�o2�t�6Ugs �T��q#����� ���9�nf#���M��R����]��\o�=� � ��t��4�j���/��n�Ͼ�MSX� �oe�����r l�j�#$���:Y i*Qj� ���$#JU���8�׿����d����x�c�umỉ M�"�d��#A�ϸ��,�'���G�.����r0;Zc�W�iH"�-|'�H�(/ ��� o�ȴ��X6C]����j0��z�P��Nl���f�D&��7�`��d��v����1ij����uS��,���(���.����ތ�С Xp�4m\�J�o���r�k�����m��S �X$�Hr\�L5�Š}F."�̹�����Y�L��c�n8Yg�9vg�D'ga>�����V�� �O~D�7e���A�ҖVˬ��y�Rv8�͙1��\����2 g�~�Es|�������6��4�hP^-��o��s���@ayMa�%�pI{7�6v���m\�+���� f.���c� �S����t4�~�z��腭�0�Y���ś��lّ9v���gľmn���d�x��>e&�oD��NOd~�����|j�]�M{A ����� ��$��"���B>%D�]��X{��`�0�d�dX+�v�U��@�����Bu�ާ n-�N��p�N��nM`��^��(�/�A�T�t��5PQ�rO|�(�3z� /K�RO��a=d'/ Ҭ(��o"��YyLg����eP��� ?v52�p.�85&cx"�h��4�6�î���D�M �Eϔ�F(���T�hr��TQ�EPC5� �n��I�=T��D!�|U���6 ����|�Q~�C����k}x���Ĝv_cye�5�77���b���x�o�h�w-׹v׺�@-j� �sZ/�D�ګ�d&Z{k�d�@���O�����tg�����f"��~T5��h 7�f�Ȍt6פ��a!�h:��� �E��1K�g�Rh�X�^0bcQ�ԡj�]�;i9~,2��_b�����\`CBKz`����?r���u3.�o �K]��k�̾�#�J>|����۾ܜ�� ��ZF�k��������`aB,L,K�E�s / �}/[\�R� ��dl��,�q����\I2��05�$���p+e,V��Ocg �sA�A��+��4]7���f�ȿw��%k+z� mћr�:�'~&��%�����W�LGq�2߲�Ǔ�|��fI��� r�?��C��Fp�P�\����u��EgI/�GO�pq��=���^J����b���o/�U��v� > stream xڕ]��6��~oe��� �S�$m*�Ru׼$Qű�-* `{����xl0�ܗV+{���x� ��r�����ś�2 8�s����6�"�I��m�)���n"!U���8n"�����?DU{�Lq�,� � š���[��Hr8�UO�����Ҳ�E.�]��rG���L�7��?l�\���}�|�%�b�)����ڨ���}�}=>О�=s�T�-�x��w5 �A�xs��F%"�W�p�+��b����j_�c]Zh>��7�x�Q���=߬²j��q �#�g��TU��n#�5��@u�M�� ��PD�6ly 2�J�h�4��� #��Ꙁwv�9�ǥ�3�����K�li�(I��w�N�W�m�?g�����3g��kᙚk�]�fS�35�.L��f���P���[#fij�VV�cJ��]�Ȣ8y�?��l�)SNxw}�a� ��b)�@ey���r��Dȴ2��,�}���> �v��-E�g�1]OC*Mc!]�=�]߁E@� ���)�l������;�;e�E +>X,)�gm����~�?Y�t8�4����k��1��1��E {ja���⟭�@��p�eyx�C�@4��i>�cc��nC^�@���I��V��b@kbh��P�r'��&�b!ީ�Ρp>�B˭����:T!�}�M�K�0D��Z���;��.V��;¿���6���D�  q��R5.�κ��晊�2�9�J�‚-,b�&�,�7��@@ϔ�(i\A?nZn����B������ ;�c�-�ER���#�*��i�{�5� ��+�?M}DR�� �9(7�rg+�2tTH�,��*'�Y���ۼ�_���c�t>V��ؗ����j>U��ԿnjL��=�''�]� �Q�q�'>a�:m��ppMBy}2v�ޏ2�=��uF��7>\�s3@�9�'��������I�N}?DG O] zb�!��0�ap��� > /A2 > /A3 > /A4 > >> /Font > /Pattern > /ProcSet [ /PDF /Text /ImageB /ImageC /ImageI ] /Shading > /XObject > >> >> stream x�ŜO�$�q���)�hT���y$-��d�A�A�(���D����/���e��{Y �����Ί�x/�E�|�����o��������o_�����O/��g���������n_�ϟ^N���E�����"z�������/y�'������[�G���s�����1��z��os�����+����8�~��w���x��˸>�O�Ϸ�Q�X����|OLm���U}VoGnm��X'��Z��J:ri��k�9�X��q����+�G���lK�ȡ�l���)�k)�#�z�bK�H��m������x��ԫ-�#�y%�R=R�ö��H)�n�O�H1�b�O�H!�h�O�Hg��������9���s;bK����yĚ�b�/�%�h�/��9�9�(�T���!r�h���!�a����'�����h�o�=7WkG�}�� ���y��?�Cx�-�#�y�?�q��K� �1xΖ�q��k��v����T1��Yz͖�q�j��z���`K�blu�b �;9{�%��4��� Х>��g�w� �U�tk���Pgm��Ļ��5��F�;p��pі Jr�;j����x�,���xìU��s7��m�N@��q�d|G�C�9�U��`�Q���p��#���Xw �m �?㰪�����xU���x���x���:�i �5o�I�E��3��^�r=\+������B�}'�@D�/�.z��#���a{&@>����hW��w�/�_H ��$��D��\�\��2����$�=�� ��K��(��m˘�@�8�Srg�0�[©�X�a)xI m[v��A `)�6R [�H��˦4��r�)�Pnż��\iqZ�ck�E�|�)��y�ۖ� �$xʘ#��Q�qے�(s� s��)$8N�n[�.�Fp:� Wr�A�4���5���m��!J"�̸m�E�&�ұ�!f 眎i%K$�� E�U7�Nd�ӥ��UH/����L]��y���U��[W�76���m[¸��������2(���:x������(N�[]�e��E����c��4��*Vi1 ��T�&\:����2xJ7b�&x��"�}��*�%���i�6W����CY��}qi�w�m�}ӿK[�?d����L1�6���C[��U��s�6BW5��'ł��N�?�s���*�������\���dJs\��~SW�n�5�6!F�F:�&��f���U�&�`x����?���D�ղ�IF��P:�O�M]R�����5� �����YI.m`^��Cޒؚ� #���p\�2��D����0���� ��1�ք8@����E���dz��������Y.�QZ�'�b(�7��Me�"���Tv&��C����R�d���C�SM݄dy��s��BI��e�-�����쮮�$���M� Q�4cf�lg�AD� ��]%(�a��� �!�`g�jӌً�D�q���  $���K�V��h��f�{QCwe⶘��1�KWV�O�1��w�aR� ��L�i�I� p���9.�^��lƙ����Lfu�"����S@ +��k��S�9x^縚���n֤����s��53�)R�����*���A"T ǽ�����k�(����l�h �٬�3�Xyݗ�|��Tc��5��o��"��Ps����7�ռkL��=L��VW+s��������������u]�>H@��.W�M�MU;��r*v��\?Zw,��u^�5�t�`���3�M�5�2E���Z�s���[/R�(Ӭ�]K���8M��M�[�k���ƙkzǔ �;��Xo� �j�{XV�5sν_�R��m��1C��m��a�y5�-�uᇾ���2�s��|���y5W�:)�9כ[K[$����wRݳN�uɅ���y�7.���y��^��Z�Yu�Z�f~�L�`o��ef��|�ʰ�.>F�͗����$b���.p�m�^��3�>4�0�*�z����*Ht(�i�1b�-���ˋ8C׻&��݂�U4�Uq� ����R |T��m��9��O�a %P� {�2�Sʦ.B�-NS8�s��>FQTYF�YI��Lj"����s>ZAQ֛|N�(ͭ�������ɧ-���g�=$���,��ے� S��YCن9����4�^Y�^ѶJ�mNSP�3�U��iH� G�핿vi5���Q�>��(��� � �E ����ӧ9� RSk�L�m�`��="� �C�JAњ/���o�a��C5� S-���2c�˯ J-��mw�gK�E��B�=��?+;��- |���RLQ�'Q��T�����( B�b=ޙ��H6� ���4 _ iA�>��pofHۀXK��)�#�C !y�RDKD��.u����6�)>��Q,N��Jx��.I8�>��DEٵ�U�W�RJ�|EL�܍$�DA���Lc���`Po۫(��J�I\��&��s�B>Ӥs s�����k������@� _ۦa�r���K++�çh�t� FC�.|�G* x��D$����P��R���U�;tpŧ���kJ~��Q���w?�S|�H��p(�&0�����`I >�AyP�f� ��e�>�B�����T����AL%�}V(*oŽ|�G�R��G~$WVodK@��m�I���'�@���D,>��Lx��3�r���>��y��7w����b�8�G�4�C3�.����[�y��)��m�*�ְ�OQQ�r����$�8��D�n�zR �E��ɇɔ]��8�$y������ R۞>�ycϬ&4o���}�f�����fu�$#��E�JG["m��Pa#|��ZJC۞�(��(��̪*�}ȍ��>h'͆��T�(��Զd�.P�N�͗���,>��:h��3w)Q���vE5��'A�C���rk���3����;h ���� ���m+����������V�g���U��>��29��{�R�x{� �}��]����v[��z!QW�{ޛ�fALP#o������&�}��$^C�!ӌ5�bت� �0Ҏš�%ձ���H�V�����Q�䴳��52����"���B �GM�J��b��u���9��!��g�s�qUeO.g�W�Qby�֛��=c����M��N��[ �"xNH����*��\��TYk����j[?�p�',����Z''i˓��^�X�{�RR���� ��-+_ڂ�~ub��ꐗʑg{�R,�B�?�ô �H}ֶ:�� ��cV��?���8A6����X�m����c��/�fz;.���q��Jm�4�t �hOuuH/�H�Q�����uI�������4�!��Z�f%�Ŕ} \�VR��E��I(ab5HVW�͒$�>�-�����Ҋ6S��)��;�6�N���>|p�{z�s,Ƥ@��՘�&��8|l\ �K`��{ �mӷ�w�N��})�q� �)�f.�#C�kh��ˤFe����:9p_V�D�9P�D�Kݏs�:����Q9'��4��e����:M�h.hC�r,�\}�1�e>��ѱPӃ��ֵ��m�!�w����� ��p�|5tUB!˶W��ʭI ǜ��������xm �Po��qC������@���~vCI��2㣓�Η>�Po��f�����}� �Lo��q#�[y�aD =�L#��n$�j���F|�K�b��w�����(�F�7~��H��P�T�T:d2������6��������o������/�Y��|Ꭾ��p����"�X��������4v}�a[��l;-��;}�B�Vk�9|{��^VR��La�a�k铬�;ݭ����E�rj���>n��w���� ����-}��������>n�����H:��������J�NO����t-}��|�����ްR�=�ة.y��V��5����/���F�V>�F��2Ѷ��-�6JR�5�����������:闉��O���t7Ҷ�Vj��/��%|�Ju���/�]�7���&|�o����ǟ�����F�����A�[�������6N��~(����|���_�e�_�7��t�:����s��,��-e��dI�6�|��+ ���|z,�1�:7�s�-�����/kESq��6^��m5����O^�֏�����ӈ���Z���3���j�o�X�N�ö|��}����ַK��_����U����L��b�k��������R�u�g[��Y��p/Fb���}�ÿ���o�����듟��P��o�}��Pn ��=S��n��$�}�w�V?n���2��}� ��_?��7)r > stream x�5�� �0D{���8������m�-�=��l`d��#����pSLRN�w j�7;�%��4�z.�q����Q$��D�˕g�_|> obendstream endobj 401 0 obj > stream x�=��C!D�V�%��@=�dr���k@�\d`��;��Q�3�g�Ƀӱ4po�k�#�M�����x �a�c.�U�U������׽cA�i�5 mځ�?�5ޣ��B�(tneZ �X > stream x�3�4U0P����� �F� )�\@>����� �4XiLEWW�� Vendstream endobj 403 0 obj > stream x�5�� �0�L�8�P���>���-D|�3�z�p�L�|�����8P��Lhڳ��$���#�'�ҫb��E�ɞ endstream endobj 404 0 obj > stream x�5R;қA�S��Y��y�ɤ�s�6;,�B��x�!�Q��%O0^'�w� > stream x�=�Kn1 C�>�.��ٞ�(����'%��Ej��LYS�4����p���;l�ff�Z������b�|��F]Y'��f:��Q96M���,��.x�&�[�?�Р�5. 7tW�e)4c���{��2�jL]lR�{ > stream x�5QIn�0 ������'Š����R�@%��%bc#/1�9����5�&~grW�,O�G­�I���^v���e?3��VΤƓ �E6= j�D�(g����(���g�U!h]ݧmc�FJ�ӿ MM��D g�4��Ͻ���N!Tendstream endobj 407 0 obj > stream x�=P;�D! �9�/�$�#pF�-f�߮�)PL~�3$��G 1���%������B�n��� �CR �z�t�6�:�3?a7c��E1��t�=&9�� �se�VH'��"��3�)�*{�x,��6['�=� �RR�ɥ���?mʔ :f�,��dM8˻IR��2��v"}� > stream x�=RKn�@��\����yRUݼ�okCR�*��1�0}ʐ��K]q�ɷ^�[ �ܗ�!�]t.�8G� 2�*D�ͪѡ�B���N� �}9��/��װ ��=2A�$�)B�nQ�Aa���P�Y��Q���2jo�� c� �BmH��@D�T���g$��gb`�Ѳ�TD�{�Lj�Ψ��D�>OM�(�L-V�nS_����|t*�4���U�� X�y�9�H���l!�:n���3�2�`K9`���G��Yu����t���pL��~��O�t�Z�u�r�@�MA��F���2>��)z�,���F3�a�����r�4�k"�X"��bD��ls=��L��9���l�֡�33*!�ں�j�@v���p��?3�mendstream endobj 409 0 obj > stream x�MQI�0 ���@!^���C�C�����9 �����X 1�,=��!s7�~�ٻYz������"SQ�R�.bB]�ϡ�=�kY���9,��s���3�c}I2���!N uZ�¸kb��Z� ���v�Bz�M;"� �2;-+�{��n��?�R��������\�endstream endobj 410 0 obj > stream x�E��� �TA �O&�����2|`�� �yF�&^XJc�P�@��r-��p'��T���wi�IU)ܤy�g&��y�c�� c�# > stream x�363T0P0�T026Q064b�C.������@�r���s`�r�2��2endstream endobj 412 0 obj > stream x�32�P0P�4�& �f )�\@����B.H ��� ��%���g�� m� D���D����J%��endstream endobj 413 0 obj > stream x�32�P0P�4�& �f )�\�V.L,�іp "����g 'endstream endobj 414 0 obj > stream x�E�Kr D���#��� > stream x�E�;! C{N�#�#>�f2)6�oc�lR��X ����ڂ�肇�%��gc��6n5�u�V��h�R}��i�t�h6s+ �fz���:����r����Cp�_��b�9�������S��6;�������ܬ�~+�UaiYK� ���f@ �S9ګ=? :�endstream endobj 416 0 obj > stream x�5Q�m�0 �50 ~%�� H���;کHӼU^2�\.u�*Ya�Cu��|�f������!�S֖{��yJ��,�p 6��s ���+�Q�D�n���Y@�蚫�Xb �����V�)X���v$� r1Y�SE�)��|,ّZ��rY �n}a�8쳋D�y�L$�d�l�>[���E6v�8�Ѐ����$AS8�>�a�j���('89֯�1� N�KR�W��ϠA��3 ���.�I�wcJ��d)�6 ��S��b��Vo�����fQl������ܠbod�[���7�}�_{dendstream endobj 417 0 obj > stream x�=P��1 �]�X�z��g�K���H��"�ER*5��)/u�%YS�:|��y��nd6%*E/��%�� �}����Ֆ�C4�h9~ 3*��K6�p*����3� mtV��[ Ф`׶ r� ���" JM��r��^�>��C���5�X��s �S�'�no��^�����#]X�.i5AM��Z-���^��#� @���q�R���kp��/�'S�endstream endobj 418 0 obj > stream x�E�� �0D{�`~&�>Q��޿ %n�'�=\ �)n3�C� ,�]q���0 \GP���aվ����R�������Bfendstream endobj 419 0 obj > stream x�36�P0P040�F�@���B�!H��� �Y@�8�&�+�+ � �endstream endobj 420 0 obj > stream x�E�91s��'H\���˵�����G��4�M�zh>�F���k�,J�D��P�c�;(HLHc�H��#�X�]��T*�R�i�¿�Z��D�p����T�O�\��M���1�T�;��I�^E�Z1��D0�� oE&����پ�`��� �8�� L�7�endstream endobj 421 0 obj > stream x�-RKr� �s ]�3�� �s--|FЀ_䟯����i��mӮz�L;���lOs^&_Zg�:�����R���3�s�(�VF�v6Hj\lo��XJЅn\F�#���" > stream x�336S0P�0��� �F� )�\@>������,́,# ��.C c0mbl�`fbdY 1 �2����endstream endobj 423 0 obj > stream x�5RKn1��)�@��O�󪪋��ok�L0`�)/Y�K��%ۥ��[�%�Bޏ��R����-�G��- Q=X�/"����:�pJ��W���I�j��3 P�@ƻb2R�$�siq�r&�����'����=Y��ߓK��MN����I�]I����JD���}�B���<>G1ɀ�6�m�&��%��:���f��U�`��)�W ��l�[�����0��0Sb����be����5�ˆ�Q�]�p��w�h���c��G�� i8����^8�^�9��c��, ch&�z���О��a��Z��T�|�endstream endobj 424 0 obj > stream x�36�P0��C�4�Rendstream endobj 425 0 obj > stream x�E�K!D�������q2�s�퀝nc=�T����=.k�!�񷩖�EY>TzRV������!ܠ,��I*�!P���-en���6 ���S�����0���l_�["u�?��*/��ȥֻޢ���1 endstream endobj 426 0 obj > stream x�5R9n1 �� } �n��� H��� ��qtR�;ZTvˇ�T��t����7�we�� ɫ�R�H��V!��2��5g�R����qɂ?å� ��������X#�D�2������f[~�i��x��+�-�XR�8���y-������ V�~A�G�}�TX ��Ip��P��v��H 9{,vQ�3(1��E�A�%��9źt���б��،7�:*�[ZYc�>��,2�zX�t��ǃ 4���� 4e�����9���QVx�C� �1�!��� J� > stream x�-QIrA��+�f��c�+����� �@tZ⠌� ,W�呫h��31���~+�mX�@T IcP5�v�z��q� ְ̓ f���)����1g��e��]�=�Aɬ\�%L�a#g?�2.�s=7 �FX�a�nV3�4�H!�����A��F�6r��/�t�U��1�s-m_�M8k��X���]q.'/���6Qi��@H��n�s���xK����{����T`ś�n@6��߼���������_`[�endstream endobj 428 0 obj > stream x�=���0C�� �@��)�|O�j����#K� zc������Æ`����%Tk��@%7ș`���G�zb8\�������f}��B�%h�{�Siܦq�5�)ꜣ����g�4��4��s� �{�S��������w.�endstream endobj 429 0 obj > stream x�5Q9! ���@$�/xO�(���͌�V�\���!��R��!�S>z�.�����j4疴-s����/����f���i��� ��n�c����!� b(,AͩEm�'�5�^�ޗ���S�Ɩ�V8����=O&�t2ԫ5� E" > stream x��YY�ܸ~�_яjd��!��E��:p;9�lp�t7�n�������bQW�3޵c�i�,��u|,��a�7��W����U0%��.8S:�T��7,UE�z���/�Wo�����o�7.�F���?�ݼz������ܽ��r�uƊ4���7�N^o%O�ah�v'�,ɾ��TƓ\�:�[w?���i�y��y�ٳ�{W�l���dZ�?���?nS�X[y���cآ�u�6n��"��s:��_��'M�Cx���%1���q�����/�:(�tZ�6��(t�=����m;��5���n�ӫ�_�I( �����k��w��cm�R� f��\�6]P���*��}�bo�֚np$�\J��dz��g20�Ѫ( ͼ)vR*V���,�m 68 ��yR��ζ4��XЏ?h�4� .��� �]Mqt �i:��\7�� �rHsn(�M�ݸ��a)��?x��,�ͥ���`i0�'3�S�ˁ)�3hg'921қ���3���h�޽y���� ��*w9�ڴ��CL�K�ܟ�9� �c� ���*̹~\�����0�I�j����6� �8˒�Hմ��4 vX_X��� ;�qˣ���6Cد6��]8O�h������z UO�˓���q��� f�������Fݜ,��PJ�IU0�ل�9�ca�� B�*_ {�F9�Ϡ�,�R4���5����:���:����� !�����j�ֆ]>�8i�w\�|c�^� ���۸g}��T�#y��Ҵ�BV�-TC&z�{�3��Or׌����[4ɘ��v�)��łS���K3�SVi��$����n\�T�B��I)�Z�2�� '�vYǼ�-b�,��~�~y���hv�1���4򎂛���_�����ݕBl�w㛚��~��3)�Tu�� s���8BP�Fݐ����&8c�cv�� �AѠ2*�� QZP�ԫ, ��Or ����ZG�H*Pz �yF�Cx��…�{Z� 3$0m�z t���c���u����;yR�Fu始�U��Ybr������E0��wZ�������.��i��N�,�+JĹ�r��2�K��yN�هP��d)�7S�>�*V!%�z��|�c�D�w(ថ�6�Q8S�.��~��"�+e9h���W�W 9���ZY�I5%���cI��4�_PC��h�J=���|շ�`��� ����2��ۗv�� ?s.�7��u���嵥��-���3�/sW�)����J �p` ���m��� /�A��B]a?4z�L�@�'`������ե���UC�I�����3���� �:tF������t�F;LW}��j��'�� ��>%����~��{�� �?۝��I㺏!��1�3U�ta �]��J��UC�Ɲ �#�l�.G�� ���B*���-��p��U_ \�^��f����Cf~���5�W� ל|�%�����Ų}q�}�L�ƫKɳ�V��ț1\1�d2�L��~�hB2�|f4��c1��[���D����wͤ.}�&p�@�?���B���)�r���$�'� > /A2 > /A3 > >> /Font > /Pattern > /ProcSet [ /PDF /Text /ImageB /ImageC /ImageI ] /Shading > /XObject > >> >> stream x�ŘMo7��� O�Ie q4 ����B� �L6��!�b�P�d�Z� 8���~Eа�ܬv�d �^^������K��_֨����s���f�N����u �.����|��� ����R����1yj8�!�mυ�鬫juI�9% �� ���·��A4�P�ޅ^��@�G=XG�Q�6��O �M����Z��Aת S�F1�3���7�#�jƦ�GY��آ��qmK�a��`�%��Ӡy@� �Q��8��A��3f�����, ��fn�U=*��S�/7֣.u�t(����w�2%k����^��o��Ҿ��y��S� �˛����P��,Y|endstream endobj 447 0 obj > stream x�334T0P�f�� �F� )�\@>���eh`f�p�)��Y��P1��������9�e��L��4DjhWW?��endstream endobj 448 0 obj > stream x�=�Kn1 C�>�.��ٞ�(����'%��Ej��LYS�4����p���;l�ff�Z������b�|��F]Y'��f:��Q96M���,��.x�&�[�?�Р�5. 7tW�e)4c���{��2�jL]lR�{ > stream x�=P;�D! �9�/�$�#pF�-f�߮�)PL~�3$��G 1���%������B�n��� �CR �z�t�6�:�3?a7c��E1��t�=&9�� �se�VH'��"��3�)�*{�x,��6['�=� �RR�ɥ���?mʔ :f�,��dM8˻IR��2��v"}� > stream x�=RKn�@��\����yRUݼ�okCR�*��1�0}ʐ��K]q�ɷ^�[ �ܗ�!�]t.�8G� 2�*D�ͪѡ�B���N� �}9��/��װ ��=2A�$�)B�nQ�Aa���P�Y��Q���2jo�� c� �BmH��@D�T���g$��gb`�Ѳ�TD�{�Lj�Ψ��D�>OM�(�L-V�nS_����|t*�4���U�� X�y�9�H���l!�:n���3�2�`K9`���G��Yu����t���pL��~��O�t�Z�u�r�@�MA��F���2>��)z�,���F3�a�����r�4�k"�X"��bD��ls=��L��9���l�֡�33*!�ں�j�@v���p��?3�mendstream endobj 451 0 obj > stream x�MQI�0 ���@!^���C�C�����9 �����X 1�,=��!s7�~�ٻYz������"SQ�R�.bB]�ϡ�=�kY���9,��s���3�c}I2���!N uZ�¸kb��Z� ���v�Bz�M;"� �2;-+�{��n��?�R��������\�endstream endobj 452 0 obj > stream x�E��� �TA �O&�����2|`�� �yF�&^XJc�P�@��r-��p'��T���wi�IU)ܤy�g&��y�c�� c�# > stream x�32�P0P�4�& �f )�\@����B.H ��� ��%���g�� m� D���D����J%��endstream endobj 454 0 obj > stream x�32�P0P�4�& �f )�\�V.L,�іp "����g 'endstream endobj 455 0 obj > stream x�E�Kr D���#��� > stream x�E�;! C{N�#�#>�f2)6�oc�lR��X ����ڂ�肇�%��gc��6n5�u�V��h�R}��i�t�h6s+ �fz���:����r����Cp�_��b�9�������S��6;�������ܬ�~+�UaiYK� ���f@ �S9ګ=? :�endstream endobj 457 0 obj > stream x�5Q�m�0 �50 ~%�� H���;کHӼU^2�\.u�*Ya�Cu��|�f������!�S֖{��yJ��,�p 6��s ���+�Q�D�n���Y@�蚫�Xb �����V�)X���v$� r1Y�SE�)��|,ّZ��rY �n}a�8쳋D�y�L$�d�l�>[���E6v�8�Ѐ����$AS8�>�a�j���('89֯�1� N�KR�W��ϠA��3 ���.�I�wcJ��d)�6 ��S��b��Vo�����fQl������ܠbod�[���7�}�_{dendstream endobj 458 0 obj > stream x�E�� �0D{�`~&�>Q��޿ %n�'�=\ �)n3�C� ,�]q���0 \GP���aվ����R�������Bfendstream endobj 459 0 obj > stream x�36�P0P040�F�@���B�!H��� �Y@�8�&�+�+ � �endstream endobj 460 0 obj > stream x�E�91s��'H\���˵�����G��4�M�zh>�F���k�,J�D��P�c�;(HLHc�H��#�X�]��T*�R�i�¿�Z��D�p����T�O�\��M���1�T�;��I�^E�Z1��D0�� oE&����پ�`��� �8�� L�7�endstream endobj 461 0 obj > stream x�-RKr� �s ]�3�� �s--|FЀ_䟯����i��mӮz�L;���lOs^&_Zg�:�����R���3�s�(�VF�v6Hj\lo��XJЅn\F�#���" > stream x�5RKn1��)�@��O�󪪋��ok�L0`�)/Y�K��%ۥ��[�%�Bޏ��R����-�G��- Q=X�/"����:�pJ��W���I�j��3 P�@ƻb2R�$�siq�r&�����'����=Y��ߓK��MN����I�]I����JD���}�B���<>G1ɀ�6�m�&��%��:���f��U�`��)�W ��l�[�����0��0Sb����be����5�ˆ�Q�]�p��w�h���c��G�� i8����^8�^�9��c��, ch&�z���О��a��Z��T�|�endstream endobj 463 0 obj > stream x�E�K!D�������q2�s�퀝nc=�T����=.k�!�񷩖�EY>TzRV������!ܠ,��I*�!P���-en���6 ���S�����0���l_�["u�?��*/��ȥֻޢ���1 endstream endobj 464 0 obj > stream x�-QIrA��+�f��c�+����� �@tZ⠌� ,W�呫h��31���~+�mX�@T IcP5�v�z��q� ְ̓ f���)����1g��e��]�=�Aɬ\�%L�a#g?�2.�s=7 �FX�a�nV3�4�H!�����A��F�6r��/�t�U��1�s-m_�M8k��X���]q.'/���6Qi��@H��n�s���xK����{����T`ś�n@6��߼���������_`[�endstream endobj 465 0 obj > stream x�350P0PеT�526U05� �LR ���\$��S a��s*s�2��sQ�endstream endobj 466 0 obj > stream x�=���0C�� �@��)�|O�j����#K� zc������Æ`����%Tk��@%7ș`���G�zb8\�������f}��B�%h�{�Siܦq�5�)ꜣ����g�4��4��s� �{�S��������w.�endstream endobj 467 0 obj > stream x�5Q9! ���@$�/xO�(���͌�V�\���!��R��!�S>z�.�����j4疴-s����/����f���i��� ��n�c����!� b(,AͩEm�'�5�^�ޗ���S�Ɩ�V8����=O&�t2ԫ5� E" > stream xڥXKs�6��W�KU  �H�� �>��v�a&r�h���T�B%nަZ�� � p�r�i��TF')WǓ�%��'s!���\���Ϩ}�"�����d|U����4�Ύ#��E����e�3�3I�`8;�����ݻq9*䂩R'JL��I� /��PW� 0�[�5��cH� �@�r+_�#s�(��d��~8��6Ɂ����"�V�@�=�L�p�:�FS!�@���yO>ms�X:��u��tV���l5�1N�Ē(+\[ ��|�Q�:2���W���Z��SD����H;o�O-��?AC� �h>3�ׂ-��� y�����qw��2��T�hN���2�/�;(Զ���0����@���\�Bf���r��W) ��\ " 6 �+04O�9W]�z'� S|U 7&���>��ޢD� �����φ�.$�a�c�Á�Ȁ�oO�03�N��{��e]u�sȄ��7�8�����) 6�}�'o�ݑ���p���~��'P����] �h}��k�0�^C}�s>J�P��l��HnCF>�jE�u����_�.R��#���j*�x�(�a��n��v7)/�F|)�Z��������Si�s8E����A�!�_w �~']s�������a���6��+f2f�2���� �i�|���cأ��p�.2b�z��88�=�.]�ƞ#$ �u���6wƅ �����+]D1�" � !��5�6a�`�#O� OW����;�ƅE�_�&�� qnk:T�أ��>����W� J��jz�m��á�1�7n�����CBЯ(,��ؠfa/�ߕy&�$�X}���/���endstream endobj 479 0 obj > stream xڭɒ�������`���5�$�r))�Y$U�!"��x4����A��Ȓ5���_w�}��v�6�?Q�|��ɷ/m���OU�7on6���a�F��׏��[��������Ά��;����կc5��b%@Y>�Y]oM��J�K�Z5����膟mU���!�Vy�Pv �� ����E�Z��f��;R�9]!5I.��FjU2�Z���~d�*���WǪ�:�������/d��WX����hX��o_�z�i��pX��~& �N)���`��bu���M3�K��L�����C�\q!�ov��\@]���m�*_%���֟�m�E�}Dk�����6l��)��cml� �Z�?��"?t�a�`K�����q�'8�xN��c�|�$���+����D7�3k�k�=A}�)`.a��,��I$�(���c :Il��Z�/����X��d���� �U`���}�!�7Π����IT��8�+(%�UKQ싿R{5�d�`��k�X�$IJ�&x�Z?�v��� %`:���X`+\�(�� %I�� {��e���E_l.U��垇���ƪpOY�IQ�]/Zm��=�}P�B��W�;\lM H�R�UyV3x�d\[�Tp� ��Z���� �\p�xV��h*C����n �������Yo����9�a�+kV���?�U�����n��^�f��P�x e�0�H�zG��A�.N"�t�U�F�� I�;���V��Z�ki0�� }B�اHs/R��+,��r�թ�Ө���);{X^�O;Ya��,bۆ��$#��� D�Me��9�:�K  ��z�ň���}M�%� l�� �^�Hr2�~�� �G�"J^B������.x}/�ΜkѺP��I�(]����7�v�vp�����tO,v� �L9��Qu�6��|�t�s��8Ŏ��� ~�f�0�5�1�Mx.;x���f��6���os���(��O\��)�}d�Qt60�B4�f��(r���By�nx}~�5�+׎ʳ�BVz�j�+������PI� y�8n%`���I� Q��]R�Ȃ@S������];��D;�f$}rP�"'$I?�d�e�� �g$��2�`���� � ��*�^a}�����/ن���N����_ſR��U�����YM���Z�34�! �A66So���?,��{ܮ���Κ�*�c�s������C�N��ˑ�՟��V�?Qi(-�C}�`�\F�H:`�4�"j`X݊���)�i�NF!�~f��Rt�i���*���݆ �~e��W?XK�˼=��C�ؒ��X� �� #�gWq0��F/K�b�w�a�2x^�Q�k�r �c}��'����0�(��i�?攄5N,�r���Bm��x�c[��j0G�Զno�Em̼��f������1���Y�Hw�j-��Ne� > stream xڕYYs�8~ϯ��R�����x�̦lO�V%SS KC�� �8�~��M����هD�h����[l���ǿ��N��d�{n�����i��4�� .GL����cG+Y��g4�hP�4p�T�:O]/�n�:�.�yp��������wO:����"�D�h >�F���8���^�d=��d����A#�Y� ��5Y�� �>��� ��k�������NגªA���o1�%`ɿ� �8�eKc��"�(�h���K�P�x�[-6�5 ~�ߤ�-9�=&쳺a�Xqj�m���WE[�LԬ{�([�8���z]�4�d^����D��l ���6�XB'�A p"���Hl����P��w��@� I�q�� �BG�I��3���G���REB ��6-���GRk%��J�Ϧ^���$9 �T�Rb^Y��I ��m�P�>�u՟�n6G~aj�D��y �O58XC.F|�����LN1Ē�9�\A5,��kp��8�8M��B)�1zצ�Q/҈5���r�!J��� :��z[�y�|��a�+b�Rk����G��EGĜ q�S�y� cP �P��{�mé�;���S�Vfh���D�����O��h�Gn%�� K��I��,gȷ|�z`;�)Qٳ�Ͳ����"��� ��=�}��6�A 洨�Ս�6��v@ �>5zGFހ����u,�(���ܗ��k?��d{��YG�g�kJ�d #����K�MR���j!�����/����5��(;վ���la�-�����n6�1���`214z(�Z+�,$�.aF0#r&BH��m(ޜ:[���҆yF5O�jM�g�p�Ut����Tyje��=a��W�U��8o �[���������@`|W�1�A@�*  j��@#��lo��e��d���wk���]КS �7��Ow��󁆺�Գ� N�%w }x�1'go@W� ~:��l�μ |s7�km�voNON*-]�lN�0����Q�J�i�y�Y=�S7 �88�k��n��v�C%/��-�c/wn�e�\��8ԇJ�G|رW���4/�U��>��4���b#p��t~u�� �YЎ�V|o�(B��s��37�S^}����������`q�����W��a.4���R�-��pw�~~�G�in zN����9�t��y�$"�˳E�p�����e�pDxo��g�D�ŷ(�P�H�.��"��>����U�(U���*�j|�� -^R��Ո������Pt�MQ�*zj c7�)4�R��� > stream x��\ms���_���[��v��)�襅��2|p�%��񦶓6���9Zi�v'30��{v���t��� =��� jxr8he�A ?0ɇ��g�$�Hrt8��䐆ޠ�N�(/x C� ��ሆ^ g0d�(DKQ z٢�B����&����� ���@J��b�R� � �.Nk�0 4�-8� ��g��B6~�R(���@�Ry�}Q���)xE���>����!��� �P�R�+�UJg�xQ����ݶ�8��M�9Jl1�6�Xp �Ї8J�����C#;�cTqL�w�th���ס��hDŽ��C;6�z Qn����Ds ds� �Y$�ƭ�A$�� � D�q "9уH.+r1�p1�~��1��{ ��P{ ��P��]���C�&��X3����'�H����?zS��/�hY��>0�uML'��2O&���@ \��^V���US��w���y��="��:u�Jji��_�Q�`AxFqv6���.�1���Wպ홭PL1���zޫٽoG�����ZM���H�VیOO�Owwa�W��Q��ܩB�u�����kއ�b���0aF*�?�I,�F)����h����h�̆ӱ"ճz �O=�����20�h�u�KO�*�sF8�t�f��1�����7 @$fi�Yn����~z�a����ߎ����14W�F6�xWr�C�cFFae{##����v^PY��NnK-u��ea ?[DYx��(�=�]ņ���-��R�? O }E(��ٯ"��1]DV(fDd�^>�� ��z�a��0LzyR�B/��ׅanҍY �t!���M�3��2���. �؅CZ��!-���>ڇC�p� ������ |V� ���Q�W��a;�L8b ��3���P�:3��!DU(e]��*]����]Z��ua��@�Gօ]~1Yv�%�\ۯ��s���s]��C���W�^��z�]�k�^c���]�+vY��e�b���]�+vY���z�.�>��W��d���� � �T�ehڱ��~�e�O�Y�ʮ��[�S��� Nv��鐕�o�k�߬�#�=��#��p氂x-���\-@���hEU\�4�0�ZD���h��/D�f6yQ`��1+a6IŠoG*��i]dWq��l��,��mϻ"�-����� ���r]) C������cy@��&H�U߉ԙי�Y_�R�����vɠ+X���$�)����;�G�.l#���6-�O���nN���bw:�*�8&��I���I��Y=�)�9��w����ޔ�2"2ty�B���� �i3Nw����� sXfn���&�����p�C��e&�������5Z֧���t�T��k���Di$�3�$ڻ4�!i����g��ϳ3�>ccƢ���,��a�T���1�b&� �dT��ҒY`��UY;��H�jWem�e}���s�~��N�eo�7��Ϙ?zjm�h�Y��j���k�z�O���-:��i�¯6�����ˏ/N��ٝ��7��t���S˸\>X`�i�՗{WQ,y��+�XL�a9HqE��Δ핟y�Ĉ��:#:�È�Ԁk���0�j������)��t����#��FvG��0�;:��`Ew�( #���|r��(� лc��8Ҧ#�a �����.!��kb���b����t>Y�4�4D�H� sϢT7kT]�hF���R"T�DB�tx\��eR �ۼ���[$!E�����\�bF�%���+��(n����X44�iF���g�����������]�����*�/�6��-�2�6W:[���KV~����|������}���}���l�� �t1j�u]m �����Y������l�ݗM�Քm�6�);�mTS�QŔTD1eQL�A��$ԏbl?t���]�.n���������)��))��=c����%���%���>B�G(�e�{�:0:�0&���e^i��M,��!��It. 3���4��m�3�J}� ��۝"Z��Ϩ���!�@ +\�3� ��@����� �u����t�ȔcF=���]����~�Fp$s�W���}1��]F�mT�����lH���|t���:��ղ����?���,�m����ec���z9���e�1��]� ����ګ�W���q:UH�4�� ��F�L�Ȝ�ׇ���%�=���0P�5i�N�$���:͛��p:�SoPj�C� "4�p���)j�Q��z�lf2zM,F3m�v� �Ƙ��g[�9ψ���2���Z��g�黆������9voyBt-_��jJ �7��zB^W����]:+�: �6����?v���ww�y-1�wd����)�endstream endobj 498 0 obj > stream xڝYIw�8��W�4C���@�dnޗę~��33q` �Ц5���_?U(��B���E$����WP4YL��هh�yx��S!', �(g�ۇI�'i�����|�=�~t{��ߧ��^�&�g �]�p�lۆE ������(���O���x�$~�G,�-7" ��Q.�����d��Yʃ��^ר�$ ��CǠ_���C�P�WNo�N��P��4_�W��+Q��V�V��) �����u9�º0��4vO�5է7�xp`!�o,P-�(?�}�X����q� ��W�AE3/��ฺ�����ֿ�]�� ��6-}��.*瞶^�:E3fx0f�����xܙ�x��}`��y�nǬ�� ��ep�\)bAmK�斍�>��n�\�������(67M��O'���?�+q!�$BX��6?h�u��hЌ�4܁W���H��c���B !�'����}�p�ٞ'�L�{B� ������R�tC�'n��h��ye�@c ����]�v���>� �zT�"����a֖I {������ď���l?�Nĩ5 }���K[y9��nC��W���$��!��GЏO$ �:�:�N�_��7���P ����Ų�pn��z�m��>�`%L���.�$��t����  n��N1BUB��GJ�px��;�UC��(�L���q����3TB`�������e�Q$��|(ľ�yU�Q1��,���=E��`H�,u�FX���O*������S������=s8'=\��=�ܫnE����CW���|��3�QC��n��&R8����t�S���.m�X���+�_,}�RP��%-�x�$����>��0���[q�+�k̠^)��)�U�~�&Ipt��}\���+뎿��uM�͢c(�ޑ4�΋~��8��H7W��\��# �3N�����U�ֵZ�STX�Ă��^��侳�|N� v�J #�1���k�|�FFa���9;D�=�8��C��0��~ �k��fV�3����NH��/�C��W5��]}���Ӿb������_��.֪�W�k��ޢ�YYO� ��r��# ΍[�E�~V�f)����݋����;���K���4�0O� _��6�>πk��U4v@�����̓��O'~���΋ߘGz� �ԥ�%@�܎��M��9�]�D�b����+C����R 4h�+�k��"�t��]����w�J *�' �y޹/�Ѐص����Gz�[�vhݺ��CUέG�kR�xʑ�1A�A���Z5~F(���)ͺ�}vֽ�G��@�՘�� � �c��ʖ��Jd��܂{WC'޲�Wo������1����໢��̼�֘*1����n;��k�ֺ�����/���z][��, c�����=����N�@���kR� |��*4��g���r��kq�,�p�lI��~�[�,�endstream endobj 502 0 obj > stream xڝYYw۶~��#unĒ׾Y��D�˧i���`5E�r����;�����M`��`�o o��x�������~~��﹙�����IL�(u3����7�K?�s���b�l��,���YX�߾��T9ot�j: ��9k�3?r��|`��T�7����Պ�?M��1��"�ҝ���w�u9 r���͘�yn�_�OcA��E]��aB�7x��ޙ >om^h2���.���j��4_�e`�o ���߆IW�a_��Œ����^�����{baVςc�~,5&/@�� **s�7w�8��o�R��;ڢ�x!���J?R6�KY}���my�\w������}��鬈���b̐����4�;>��4�I�ޚ��F#��L _����Z�`f���yi� 0����Tsu�|i7u�,gx ���U� ���m�ߏ)���0��/���;�]�{���h>ѓ$�����j�L��k&…��u�iͰ��lW��P����,Mū��s�R^4(A�R��wkJ[��8� 8�?�E'�f}>`  �yQc4w^�e����n�����,s �Bo6G���^��C�Qr^��Q�ð���� 0\Z� �[Qc�K�DQ�R�x+J�}�O �AY��{q5����1R ��� >��Z�}IB�\���l!�J�\ ����փ��m^���=T��B��m�Qh���$e�\Sʚ��爎U�þ6��-&�1��N��W��K�-��aNp�L��x[1��6�,5���0�C�saK���}�����x�Au�D�� �q�,�*�=��r�[���!v%�/(g.'��"��B�2%s��B�H�D�La��NS�i��:ǀx�F �,n�#6W�Ä�b1� BHH��%�3lt�O�σ� E΋�_�� N�'׆ 6�"JL� �m��8ͽ�z$$.e^|T�M�b(����5�"���||�b�@'��٫1��&�W��u��n��1�g�`Q����OGDG�VYC�r�1���c�G�@�1vW�u�����$�"�`�1��/���Ag� �f��!�f�������^f�! $��3��B7��_�E�G��-R����u�s���i��iϗvRv�2���k� "�y� LD����"2)K�Ynl�5 �R�O��n:sL9 �Ig �����?M�M��+ �+}HY�o��� ��萝�ˀq�'O oi�)7 0������h��U���/� c:�G5g��}!����5Srq�ɇ̿�(GNjx��˃|�1��OD�5�/Bʁ���c9qS=��p��4�I`�Z4X`�U���į@^+�? �n�������+��j�>�5�m���̏C{�0�Sт�F�Љ�O�8�5�j��t`�x[q��f��s,A����=�~�|述���!�� ;�R��z5�=!>⹼��Y�L=endstream endobj 506 0 obj > stream xڝY�r����W�VcP��7IC�)ɉ��4�"Pj����E2��Υ� �_�FVւ\^�,z���[|�ś ��/Η닏��k����n���J��t�t��_N��% b����Ӷ-~�޶,���u�?�f� N���D,"�h��ԍ�t܇ �9ғ�=��J��E�\/�ƸX�����q���9������.cw�t�U��Z���V�F�Q� M�@��Ӫ�M ^��{\�W=�h����A�Ę޲����7�N�-,����4u�B�s�}� = �;��^gEݔ��y�b>�V��R���>3���US�5Y7#Q�r#BF'a��_=H�H���N�^ ���p)�W�#�S�=��$�9�G�K�`�@����� ��DƩ�U�����б�� �R��X�%��*��� ������8d�g�F΂�l(���'oF�i��^(lD� �UZ!$t����3+U���Ͱ)X�l�Z���Q�9.3�s�x7(� �N7��\t��2�DG|�B/�?�$ŧ � �� �$F�� e0 �����R�3�'�uS� d� ����>`�SL�f�ҋ�sU���㷆��;��v�#��H���!"�ȹi2JK��(n�5�3����bS6�4/ ^�5�є ������4�@��!b*��9��������/�H��!4*B ��D�a����cp�keX�е�}h5��®i٩yc�8�@�J�|���]�p���� #i��+2�Q6�*P�s�Q����!eAh˗Θ��,�蚍��nL9 �9�($���{���D�:;P��WtD�rf�������n�׮�8���n��I u�k���0v�d��aZgA�_������B�� �'p%&��*�i�1��a���ۘ J��f�%������M� 9�����a���A�p�M3:Pwݘji���c̑������+�I��� /H ����e���7rW���@ � K������ ��)k|LN[�A?F2���?1������F |7LӽQW�H��� �f0�������m�~��7o�;/�1ھ��p����.Qs~�P"����\_�૚ �D,J'tꊝ��{ſr��Қ����hBj~�ȩ��F��� ߗ,z�:I)q@��p,pc���1�)�H=����:`_�N�Gc�����7)q���.���9�����������?����Q=`����ͨ�sy.�d�ҟ@�v��6������]Xy�4�� t�0���'��\��Q���4�xg(�a���4LY�o����%�聚���d��i�\�:}� �����5��:`" v�Ob� x��X ���5���Z�� }JE1����Q�6]h��&��� ����c��� �ax��mC���+��*_��\�X��T��#|���;�'۵%8x�H�#�g�b(Ws�@�mbƋh>>�����c�p�wx'����b�ܠ��'�inh���w�9�!΍�7��;;p g�/~��Ȩgr�!og����r+�����j�5�h-R@44�&B2��7���v����w�)�Xrc;D,w.�(�x����0� ����[~4#�f�+�N�d�^S`�A��0a[b�5G�>.t�QeO��[� ���9��"�l���-��@0��;��\)���厯���k�n����i����y��E�hS��'@]}� ~����WW�t������l�F�K�j��)^�]B�/��r�G����uX�W�਍���Pb�o;������M�7�m�?��G]{(h�-e=Ժ��i���n5K���(irf,p�j�����$ؠ�s* pI�tx�����’�6�����!uԾ���#�/sP�䞡�� Ct_��� !*}B�?�}����Vx�)U*��1vN~�B���tY��^>�C��[�}b��n6�� g�Q2�`q�9k��R���ߧ�-�Ke�(f�.�\q����� �z������� �������v�G@U���C#f�m����L H�C��Į,y]q�hu��&�`�uS��R�� `��?A=��#�����xdI���;z��$/��$�S�����v�Y��".�WE(�����R��'��!�3�p1l���#5��L�5�su�{p������y/ �Mڗa^��������b��bJU� �*�:�߲���'�s~�������٢4�Nff,���]T��K�m���[}�]ߴ����z��XC�[���*��莽�q�W kVb*�� iQRb0}%��ߟ���xufǎK-h=���5����]�$�{�`?��)]>���v Mendstream endobj 510 0 obj > stream xڭW[s۶~ϯ�#9�b ��%oN�8����&m�>�$,bL�:�ĵ}w��,�L�g��v�X`��v.��pq�* Jc�٢�4�%.����{%�G��͵1���8�Nu����v�$ ���)i��(�i~[�$A�䰩�Lv���fס}���&V�_G (��H��O���t�wR�? �Z @8�^bD������*r ��J������w����|�m Q�j��9�Amϳk 0@MέP���j�����0����O\����y�8?�� Vœ9������齒 �#�0[X���U� .o�z[�p�xqk��`S�Q���Z��q@���\���y�[����#4'������`n���=�-l�P�/�v�� ��-�g�9|$LL» ��P� �SR�{0�� �$�u[��u9�}�5VOf�h�7�)���5�tQ���I|E�+5_���#`�]D� ����� ����f�:���{��B�����̬eߒ��Qu-�5��Ĕ�sɊ�tz���]^9�%4 �����;�Զ����a��T kI�>㶮�$���0��tԫ@H��'�� h�6N�^w�D�Dᅆ�uF�w�M�}�l� W����Z�����o,~4gJp� �1K�푕�ا�� � ��J��:%�z���PV)���-��g�Xv�]�'��� ��f1�t�0�e�~�e�"����,��� %���-�=S:ۼ� �xs�endstream endobj 529 0 obj > stream xڍtXl�6]ҡ�������c�� ���H��ҥH�"  ���� !JJ}�������q|߱��v_�y�}�7;��!��������A@)������@!"vv#� �������!�R��PFB��� u�F��� aHL $.��@ɿ��@�sh  �P/"ve��?�䌺���_�����Pt�"a0� F9C�o*B�nCE��#��3 �!%(���+v�@ ����0�3��E�@�F�ݡF b9ü�r"Q�`$pcp�A�p��o� ��0���z@������,�+ݟ�_�`���`������NG����%��C��p�_@���&�����o�[���� ���A� P � B���'"����P'����o�Pǿ�7����,�7���>��g}�0������W,hfdjn`��g�9��~4���_HR ����ŅA�L���i�?�5����_��,��}����B��̥���.��o�[E���/����C�4���e�w������������an�7��F��@q#�CM�iW��v�o� |#E��ۿ �R��A�`(��_���n�Kkn08T�����A@��nq�yA�nh�����?K��!�_B��H�?��5ߜDhЍ"�~�� �#P7!����$ѯ;� o)��22�)�^z���b7cv ��0ʁVZO@;�.-Y(gIt��V�� {K�b&�ԉc2���b��W��T��w��x=��J��^��t�52�\�x�:1��@?�j���j���"[�gh�T5���Ӑӯ�X��+��l�>���?}E jܺz����A������+N"��F�(�Z%7!f���r���lõ�'M=��$cz�k6GFXJ���� o�Ʒ�jI� 6$�%���N���B�W���� v�E�i��N�mv�#x�j���^u�M�k���t�p��'� �m.�������M"�S\�%J�tC��QGJ�ϸN�aK%�������z����]Oeql�� F��8�]�H���&5N�G���Qx�K�pwF@���eHB8� �hvwI. c�$d� ��������xY�������¶�A�\״޳�v �8�1��a̝!�`��|��.�0�8��$��λ��}t����,(>>�a�ح�Ĵ���}��W���� ���f�r�kEhP�3����{u�b��cK��!g�A!��uf�{�QG��6�\6s)ږ����^o9�Z�ou� ��@���B�)O Uu���|U�-��-�`��:s\��=��W�:)9��8��Х*6��I�.�9���潿>�*w��L�92y*픢�9�g)h�P`p��7���Q���� �>�U���'�%����e�LM��os\�+Y�Bf�Z�9�t�b��k���]�m��>�]&ب��ɣ�|yghi��S!��G�$܈&��֜O~h{|n;�V ^ �>�Y'N�eUZ���/O�~����O�R _��;�*3�SY��U������s��Eu������,�ȋ�E,�e�o�YBBQ�nq�C�ͪ�HX���(�E�O?�S�����Xj��RDU� ˹����ۖ$�I�Hh�5�ʪ@���l��6]��^�r����tf_ygaR���c�hC�����z`�-�������vZ��Ϫ���4� �Z�A���J�� ]��3�i úc��[�qd���x,��Q{��#m��v��6iAŋ`^�_���CH)o��}��*�l�l�5��DlF72:iv���f���g>۬�a��g�_v�(��MF�C"߱�I�N�S9���^���j�_+�[ujb�.P6��O�d"G��hc�輥�������S�V� t�����������Qab��S-6��U��~/p��jFnt���Q�rǷ�qW֢���Tˈ:�)��I����l"�"v�~*}�o1R���Y&�v��TtFT�3k�o �GYʊe�i�.����"1������=�ŵ�R�ں��1^����������3 j녾��H!�F�GL��+��غ>��ؽ�] �!/��$�o�F�1�Fߝ��� �Bc�����8B$ ��������oZ ��.�a�0� �"V�i���o1���m�8�gr�2 +H��{w���O�����L�W��! ��Ɯ���9�f������)��R�a�yn����W����5l�y}sln���u��'�3����_dP�Y�03LpO��G* 7v�W���(f\l��mU��~'�6�U�s��~�ai�䬒`�~�E�=\ۃ2*�9�����׊�����Z�UGOas�/��V����E����XlwϦ�?�vu7QlV>��6���N�D��_XL��>E {�ʎ�0�%�:?��������c��G�o(��ù %#>V8���F�D�Y�d)�C|I3hrH�\d���O��O�'�nP4��Syl|P c����Z��5�_�I�"D�Ifr��veVi�͙ŖU�5Sy��sX�)�6�jQ~�� e�oYw,� ��;�P�H���U�w���ζ���_B� �jc��73�;�!\�]2���G�C��X�8�Y�|PT O l���>\aVGX�S��>��1l�M�$�v6n�r��i|���:X�6�]��֑��K�@{�~�{��\h�\����#Y�����-Ʒ�p �;[�ߛ1$�xdx�W�X��������iu&G�GA� -|��g��bǿ6�i /��S�ZI�\��'����Hn�VC����p^% ڒd�/���q"�XtW�9�j���2�_N ���|� 5p�5�.Q�\��|�%�#�[r�g�3kXEF� ��:�|I˾M@���:$�r����Ϻ|�4NUe�[l�}ڶU�NS��Tɬ�g����lj��6�g�2R�AA�q�s���"ܽκ�R����5���\Mo�D�������Ywq˷*���ȧKM)n��J:�c�y9��5�߂���q�q����t���"*s�o��dm�?!�GsA���Eq�2ҭ�~�s��`�I�|n�uf �K�E�-�����Ot��s��Ć�ƫԪص��ⓔG���x�� ��?�v-�I�J�4��T�����R=��S$a]7객����I�눜�?��{3-��\����*��+�,�.*���� �6�Cc�� a#�mR�����4+�nD��u�3���'�ޘI����D�7�#��ڵG�x��8��i��VF��Z�G����'���Ū��#*�u��^�lFx* 2`�|lE����� �p$�F�ijHn�1�Å*c��l+k�R/��y-WD]q�nU���_-�Q�_��ģ���Cx�O _|B��;��Y�k�5 l`��=�]ʁ'֌��`ъx�\zc{+�f��N �"z[yg\���hU��������� �Z�9���j-�+ k �X�� �do^��m+�D#��} �H���Ώ�;�4e�!O )����(Sm���2�>�����^��V��7�&.��vX��kh �=�я#:N��xU6�"S�|�E=}�-���˗�r�!0� :{w�n�bw�y�곁�Q=�:S��(�Џ(���ưϹ���54���HX��� n�-wc��]b� �AÍ}�����ޛP oҬ7����W�^$�����.  �*?�m�z$�^� ~%�zo?'��?��mF���� �s�%'��[i)�|�XOt�� �kߎEX��#Ձ�)O�rjq~������X�n��p!ϝ�#��2��AK��.�W�bS ��#GPD�k�`A���5?�n�ˋAi�O�oa�A��w� ���KzbO[C$옯?�@�1�s߽>+X��i��5�0��>�ob���(]K�Q� �׏�|���f�t�H��\��L`�S�i8#tQ��Zz{�G1�I�.�cf��I�9�˫^L^ժ�}���i �V7�H�+ g��|v|ko�%�0�\o�3���ħ$KQb�C��!�����������E�/�@N[�S ^yd�Ո��L�.��ȥ�7��ɉ�Aw5M�$���떝G���/8B}N�*u^��~#`�k|���8�\� r����Eu$�?�i8-)����|c�7Ѥc���f���7.$|�c�ڒs�Ҿ �h�;�В,UV���� �P���q �p�V �� �S_.\�}-�����R�8�>v��+� ��>�-�� ���n�����d�5�$��G{�w*n�!|d�+�)�7Z�"�&ի�dvt��4�3���\c�a�4�����my�Y����i���ή����i����V���q�ߜ�� >Kduq�50|r�"ת�r�y����-��tI=�(��B�D� X�sP�x٠�� ��-:u�O1�$��s٤Qq > stream xڍ�T�]-Lq�R��� ��^�K�` $�k��Ŋ��"E ���-(Z��W�����_�ޕ��g��3sΞ������� � �q�spsE�����@�� �`22��N��혌� 7�"�/����h���?�!.w'7/�[@�[P���!n"9K� @��q�0e!Po7��=�q����,naaA�?��� 7��� @�nr~\��� ��������Y���pqyzzrZ:�8!nv,�O0�� ��� *�������#�Q�֒�f����8ed ^_^?� �>>��w�����W���-�;��z7��=��_���d�G���(���~|����!�2��������H����?���������Q����1P� ~�����7���?"~�/��ڀ��P7���  h~��oӟH�Q^�A��.��#S�?H����p�= ��� jo����e��c�p=���� �Z��B�_�o=�}���l*WYҮ4N��NLi�Sz�-Rlc�}���K�2nE;�6�y�h�Z�$�/a�}�`�+�T�*��U�f�&�~���+O �1��'��Wm�73���@�x^�F�%�F�ò�oZ8����t�>c�c,�aߩ+Q�$�dH�ˋ�Ψ���j�6�Vbt���w_x��%ô�Z�G�ָ�U��昛�c�uf}�������_�׺�ڐ3�Ŀڧә�hm��u,D@�W��G��EW������ 1��L���� gi�؝6o�\;+�o[!I��0�z8�}{B=eKS �ޟ�h��5N��p�^R|l���y{~ �t=��vg�x0:�p����s^; �q\�.s�����J��՛��Q��S:�#2ĭ���k��{�� 0�D���b�c��v�X�w��J?����/�&��v5�s;�� �+�E���j�Q����貏{k���izlm�tH9i0D����$ʰ'�i /dB �'��sL��Ι,o��1(��|���v]��ֺ���@R����>~i��`�V�1��C��2:�Wa����f�VIY'>���X��|�2���D(�K���7��l� �O� �xF���ۜ�(�./��y��cS]��^�,���.�[��-�}@V���Ж��R�� �h�h43O��)�ֹ n?Ҿ%��h�lM��Y� ��l�h����R���F���V���1�� o�̺llK&��NH�;������?ٵ��E��ۄ�%m,uށJ� ���EB3��EQ�.���?��8�#�!饸�g�� ��Hz ��y�B=s����C#�4dkvt�/f�dz8 ��������?xY�� �r���vcN��N�h[�WGƢ����ԒE�1m��T�Z:��������8k��]��&���# ��U 4n W@�t��@��p����Yj���6A�X8aQ���@ӎc!m�7�Ni�C�k��JM�n�S|��kj�:L�A�7�;�{,T�>U_�n���@��'h�l3��o��V� c��/^��w�GwN%�sǸ���~���Σ�������oP�}���Ȉ����WW���;�#1�������٩��U�N4B~���$�9K��̳s� ���5u&���6���/Կ���T%����k����o�u�+�+��H�$ ���l�?Q�Q~}�ve0���|��� _�������l��� ��� �vb2z9���f� �}3�x���6)����Ψ�x�Y���}���v0���t���ٕ��gb:���eD�˘��� ĺK��q�,�Cɵ�n`�)ME�_L����!�l;���vV��z��_Yǰ�ָ��1E�(e�d ��7Yl���a�V��HZ�p�y� �n�r��9sD- �.��g�j��E6O���O|��`\=9��Z� 6�̇�>d��h8� �/}�h�� ��>�E�S�cLg����� Ӯ���p@�Z �Ejp m�DJ-�R7�T�u���]�����RO4�C~�|; ��P�"Ž�R���֯p�Nk��L�Ɵ�~�+�x��m�:g���qu�~/����D���B���O�v������o5 �g��P���mּz���ŇB�F��&��t�z�q:�+��e�,���㔼Bμ�0ƨKǜM�v��n��v\CJx�E� ��k�ۅp�a����5��Ѣ� P~)�U��'484I`�°)�q�K.��ZI�h����j'ƺ�%E�xODԒ�}m�~����EU_7��R�|=6/*/��9��7I"oM����W�'��s/�ʐ���Cm�㕤��~/ �I��2z�9%�����aD���v��͈���Wd��a{o��{Q1�������N+[W1f��u�ܺ�Jij�����;���n����|΁Y��o�r/d5�TMZ�Ow$��h�L]����;�/�'I����`� _�B�|%�C4.�8B .�E��N��o���O�F-*������#�/��e�h`�r���Ij��4#�R=(�C�3b����mƹU�u&6\��d=0}$�obz�K��e���6�B*#M3�Hܫ��c%f�K��΀�C2�[z픦�v�����uo3 ��Ѿo�� i�K��'I�+7�s��Ob>;�J&����D��2�ed`�ÎIcX�N�It;yǶ�g��\���/�^b�'7Wo�7����%�|�����6�$UK���|���M�F�8�%�)�#���� N�3���N�,-�4%ie&e ���g�ͦ��8�L�y=w�����wT=@t��¦)�$U ���]�8�9�Z�m��%�$� �e�0�+�K�[ �oX^��#iT'߁*� t�鑇"ODs�ߥ��K����A)��W���c���� ���B�Va����B �ꆠj-�Q��ւ�P�IH���p^��W���R6��y���dV���{j+ȧv��ǟ+���;����+��̽^;>���yY:1YB�% ���,!�a x���t>f��h�:x�o�#�P������x���%��w�b��j!�n�mU���iJ ��v{L�� �nz�)��zRTU�q;�&�}T�!������F��p3 �᧱�!ԔזM�be�y�2 ��8�\�׏5���Y�p�b�M\������s�}�P�+ܛOU��> ���r?�\��P����b�D���q��`t��o-���m� )&r|�f�-���G-af�e$���y`O$�^����B�t�Z$�L�lw�U����c) ��Ct9/"�҂���M��ď�໡��iSU8�*�? [�X����AOfk�� ��;��yّ��_>��vm�C��L�G�m�m���y�m�ٶÜ�➙���`��W.Ky����k���+6N���л��A��2%���:��� J�U����#�n��'�7r}U���K���j�+ƽ$�kby4y��V4~�'����4�qv�/v~rz�/�Ї�45���F�*5��`����x����ļ(GפB� U���ŧzT��b�{��47���𺵩̬��C>�*O�Ȧ����i�Zyލ̃]U�,�$(��D��ڸ� �j���ƪ�b�z����Z����3qCt� ��U����7ƪ1�o��1��u�?1�I���T|1�SfLa�d|�go��K��EIG6\��S�Q���*#0:+Vm�z?�{��-��'���G����׍j���"f�'��+'�74��=�g���i��?�ԟ(�[�.��yҿ ��(�=�8=�!L��A�M�"����F �Ù�AE+�n���`yL������QҎ��h�I)R�����o�6݋�J�?��:�3wGX)w�Y���Y��é�Dvw)�Y�'���ARRG��BJT�*� ���v���Z�NI�����^i���� %�$oxA��峍k� P�xۓ�@���v5�M�cu���zM[���'o�q1>X ��y��+�8�.��[�D�]�k�7���V��Ԇ\>n��щ�}������6�&���Vk��evb�tϝغ��Z��y[ �o�����_�����8JK���tV��,���8vF�g�߲���u�����+�s JQ�6|�fA%�殌F �����U�sb�4�J��_VK�T7�#��^-��ܽ0� H�;AM��Sn m֏0��ڒ00�*����L�b��E�gS�gOy�@�/�/��1``�2!�@n+�a=KFvPL�h�s�bPL��g�CM)�:��N �"��G�1¡��^�$�?x_>��� ���~2���X�r�����:�����ˢw�7���ʥASz�R�� G � 4]�7 4P��6��V@i`K3�~�"�Ђ�:GA䊖����*��|#�ш՗�d��d�^ �������>d@��᭱ƃ��pzʀ� u�n���I�N\w~���1�/��Ru����\��L]�N����M.r。3�YO�R����8֏U���9V���A�ꓥ�� }�Y@��SWc��"?�ya�E��ڏ��W����f*���`l�!�wQ�pf�9g?�� 3��~�Tk�U@���{�ɁcK5�j|�+��ԌثZ��D���y�O�C+��g ����hj�w�>ƃ�%�I�le�s��7T�h�?m�v��&�0Vg�����7����K���g~n��y6�b�F��{�8.l�ul�ܩ��~�lش~�Rd̉�����2����t[H,t��s�L�u�5�rE�y+D Z��4��X�Б/�)0���:E�yS[۝|i��a��>���řS!��V�@�x��I�ڷ.[S��*��Tn� �_�殡�&�:�\���&�|�z �M�H>�h����R��}�7X\IJ�˽�_J��M7�d�QI��V��ǩ^7O > stream xڍwTZ�.��H)#��CKwwI 043� -Hwww#���H�t JK(H������9߽kݻf������g���k �35Mvq �TC��8��Ieey~��ra��k� �����u��8L���P��&A��)�a{��� �. P�?�pgA������P�à,zI������5�n��,L������t�����(C��P���!�M�� ��L��H�� '�����w�af�� �P��j�E�q��aƁEв�A��k�-�ng(��`oc�!�2\`Pg���My%��#�'X�O௳�8@�[��_�l`��!��pG��f����Te�8��H6f�+b����C\!6����ߝC2������!̝m����/����ܝ�4�B���!X����q����矛����`^K���/.���0'���_!w&��mVP$ �����@�Pwsk�_� '��7 8�f��o ����pZ���6�;q��(@^'�7�sg�;���w��?��C��Cͱ�g��B��o[�k����7> �吏�q�,4�D�HO���kfe.(���w�d�m�U$�γ?�y��R�y�S�lYQ���ܞޛJ�:����mB��M��,q�AQ����x(jaգI����퇹7�'�2| A-*m5��JD:�V5�����6��$�9�1k��a)*�4��mX�n�x�}����ݔ�~ ��?肑��B�R �G_\���kfʤ}i����ϐ�f�mj��;�n.t�0������}�N�"��Sx��$�m 8��~ū�r�aJ�FQ� ���E�n_�5�3�%J3re�&8�W��S�:���Z�q9U��� ��,�8SI�B�W�Jx�Oϣ)��J�'Ԉ���-(���VŨ��}sP�i��� Ҙ1��t,��i������`��"*�AY[a�P�=?t=��fJ|���O�5oO��祦4q�����ď��;ֱ��]؎=�0C/�&V��Q�)6c�gk�KY� ��^���:z��+�g�d�2�^OmiqxA�}��u1]g�!� ptnrh�c:+��p�~32_^�"����*z�Ȑ�Y�=��|{;SD �9��]L�s��8��j� ښ�%�`�I���n �N7H���'nIEța� ��D��p������'���wWj�uWfb٨�(���kG���u�4���3�C�d2���ntx�)/�#B�67w�>�Y�{�=^�6]N����GS��M��-P!�g�!��B�t� yÞ�a`/���m�Au�SG���9��n� �B ���qL�x.�i2�޻h W(kۮ�mAx��)Y�E��p���n��H��r�h�)p���lS vJ�/%���v���gZ��`��Ӆjo��N��/�~�y�Rۼ*����]s2`���)r��^%�p���X?�KfS�{���������u9ޝ�^dF��3�$U ��K�i���]�'&uZ)�{�sĆ�w�~��`��������~�|�l���jws`+�����,��E�>�Kƀ��� ��c}�"i �u��a. ���v�’XiƐD)��7�x�Lb��%ɀ>�x=k����J�W�̝}(�/$��5�I��g��,>�V1&�D�Ɲ�i,w��̺�f�x8D���`X~p���A�~D"zl&�1�Gn��EC�픝���I��\�t�10��?Vs�~)y��y��w����Tf�ƳГ&a&� ��ި��z��܆Y��+1���!��\�;O��m�(>a�Y����ImJD~4��1���;��3 � ��l�V�-�}5��>n���j��`E�����i�ŃC�L �ɲ�a��L���(sMɃF�ľ�,�Qn�8S�CV�C I�ɑ�L��Nz}T塛N� NQv]l���ޯ(���g�y�O��Z��=k���d:�;�z;��q�_Q����f�'V6�'3C�L�(9���v��SC?g�����1 �b�(E ڋO �p������۱���t�X_�� ��L�eV��"��;֙։�/lod��c�aGp�����]h�}�g�:�J�r)*��߁ �ٓs���{kh�=�Oļ��J *]D���uM�=Y�Ǽ���!c�;��wiZ��ƨ@'� [��O��FK��iV����X��>0F.|��J[���q�]f� �Y�+aMMZ9� ��>�����L?����(Y����.@���d8��Y5o':IՍ�,*Nv�Y�⇧��v��Ǵ!A�XW��+���˦�s7gT����K���. �7k��C����#ip�O -Ƶ77cGqce��ǖb���T��S�T��Һ�+��4���k�ܱ(�� ڎ�,��ӣ��8*��=/G{�o_��跼˸�Kj����EmP�My���Jϐ��{O��3 �T�|�D���(��|~p� 1-�e���݅�d�g屵����V��v�� ��$���_N$��q�B��N�Ϭ��Z'��_�Fy� ����~��\�HB%� �_�[Ƶ����o?t�3 p�cN�jK:��D2l�Ir\1�t �uD ���Z�l���p�=o#�kW �Ffd^B��V�Ւ$"+�DW�m�V}��jr[��^��r"�� ���f"��Y���t��8w��^i� ����n�&{� ���qB�q�\�ܐ� gM�7#�XQ��� ��#����tʼ� ���Y�|1��P ���_"٩;�56��#��R���ؽ��a��48:���v����������O����� �{q�|Cq�Y�~T����߁+Ͳ��� ��B����%���" �͵[��P�Qq�R�.�s�C��-�����c�䅏�Xt*�3 ��Ӳ��nx�m��D���}E���S����������g��^h��0P`T=��x5n���� �V�b�׌� 9���қZ����;�ߟ�� ��R���J�Շ=�%���W��ё���C?��ύ�WLT�+ �2��n]3�Q&M���.X!��>�S� e �D����ZkhH���F�b����L�%�_z@�U R$�F+-�����J �h!ņ�����i'�7u����LD�7�����j�2k���S�ܼoذC T����I�;�>��ǎ�@��{vM���n9G4P:��]��=ڻ� ѐ%��9��v���u�\.AE� �I>���{��^��G)��W�Uu�j��j��j�8U��� ��=��"�/6�9ۆ��,�a��3ci�e�L>��{;�F�pN>�����x,�lo+7�qd�w�i*��?�O5:'�o�̒��G�Ɍ_��g�&�M�f�(��Õ |�:-Շ+ ��h��qȰvX5���P�4R����$�rP�Y.�.��P�&M"�ydXqY�X��]�)� ���%� H~YO�����i�!��_ ���^+*���s�e:^��w4�-H6�K�%�s�X׵z�[Ȗ��� �&�/_�W�k�~̐�v���*��m��7,�w� ���R������⎕�������G{�ʃ�A���~�LC��œ��5���q��(�lS�J;�>MQ�&���9AV�GR��F�rEZ=`�']�O� ub �*O���Ue��D���Ho�up�v���[֩"�n����K��$������lb�����z����w�G�B�F�?{�>�L�}����'���߫撎�սLfqI�S���Z�]��U��e��b^����㳒�:|g�F5�*O���!��0~b�%/��c�r��p�LW�V� �m�W���L����м��g���,Y��ra�{�<>�v�S�/�L�ejG����ehs� �NS�������(,>�2����C�s�s�q�rQ�/ �*�:St3 E{�J{[�0��մ����Þ��/+����'4�X_ �02�_𐐞V9um�Xc3=�oz��:���e��Jn+�|@�#8[��J�Kn��-�-Mk �u�_�X�'��BO�K�3��𘿩*�4��n�{Yf���Q�ǵ��:��={'��2��m��?�~�[$�7����F�J�����)�r#Rf���,~�j[qMo���J��I�j��*��x������+5��{�ޫp0���R�n����8ΦMu$��3`OT�\E�~&��-UǬ������ſ�R��|��:�\*�+�)�G�!Sl�:C�n��I�灠���PY���� �7��7���Ư�������v҄����L3x��v�B���x�3v�CQd�F�_���0-� d��@�)+iѶ@k��t�o\�\���崔LU�S�X{ļY��~��{]���i�� 8oJ�����2����T�D��.^�� 5�%a賎��WN�R9��2��[��?�v��K� �7��W�!H�����tJ��D]n Ǭ���x������A���ǟ]���Z� ��6ܤ��L��_c�∧�9�_��7LǓ}�`u|yr���f�F�H���A_O��� s:~�ϫ,qE�5�G43ו ��J1qڄE��8#u���! ��H4xt��W�z ��H.�AW��M ���-���7E�g�� ����2�'�X*۝E�ED�k��u�k�D�$lD�~�9���D��r6�N�V�)B�����?�*\Xh����in;��u�':16�O9;������3��:W�mN��(]u�5c��d+ʷ5��2 G[=���~�Nb����􆽀$��.3� ��ɮ0�BkQߊ�"��ڔ{�5�9����� 3{����xc���s�Qnu׼6�QHm����b�N��.�:�h��8R-�ϟ`q��pɀC$�D��6��I��&��X������ ȩ� ��G8�����g�endstream endobj 535 0 obj > stream xڍt4�[�.�G�.���ޢ��E��1��B��(Q����EQ������_���?߽kݻf����y���~�w���+:��A���H~���4PY�$���@�0�_0�� �������W�2F�`*`�M�����D� qi����PXHH�?�ni� ���j��!�ve7w?��y���@.{n HJJ��w:P���ڃ�@]0��z��=4v��B�~�*�%�D�K ����]=�Nr�|@(�h� �!�_��z`W�fv��3��n���# ����=o2���fs���P����'����A��������;lo�����A�N@G( �W�@�"��`�ï@0���&� ���v7�O�)�7���i���#= �� ���������P��_7��B� ������;��ghu!P/���j"�7��w��}�PO5�/����w���?��)�A�7O�g�Ӛ��݌�������F��]�������p{7�_#&,&#`?�M�o,1`�f ��E ��!oR�7􂀎n������*� � �� ��F�� ޼_��B@A��G�8"�AA��7,rü Ug���)� �8(d߭s�&%)ߤ�sɟ8�(ـ�}�sz�b���I�v�^���������=���a�1��È�q�k�O/�� �ETm3s�f��>ɝM�4���TG���˨TJ��u�-���w��j�����w�Z����I� 옕�6$���,o����W�u)ED�����߾U���p3;�+�Q�NX0���}�l���S?]]1�֛���v���&J�,�5�$�&�Η�Z����Ci��sN�vho�FN � ��P�s�r�KyM�/$^�J�;j�4�� b��4��.?���Y �Qq$�\������s�]E���i��w�{]�ϖ�z/5:�W�,��_��t��G\�p��oZ8�Xu����j�G����N۱ M���9�}����M�Kt h%�b|��~ �l�|x0�œ�p����h�-��N! �o��EY�9.��_� &8�E�n���6M��3���?�XQ#�P�3~���Vz��Mit�4��9-˛E�l�&V-5�����N���2&����W4T#*�*�;���*�)u�QJR V�b�?WO4�X��Z➂��ӷ���N�P�r�Z J��՞�X;���~��R�#�,�%�R�=;jä��ؤ�SrĜ���sC��b��7\ڎ͋�t�FJ֌FJ`اM�69cYҕb�E�s�;S���x e$�wB�/] ��;&rY�fmSw6U8�B�Pˣ����p5��Hz��e�!]# � �p���4R�[��6|�f��KHE=~"xZ� y�Yjk��)'�g���ԟe�m�� lk+E��C�[��+�kr��O��ɏ��� q� �5��8��� �t ��3^jSi蝀��d��H�λmq4�����u���c�����z�i��� R��0��||��`^��{P �۟l�SC�vrn�Fyb=���H�E(�N����b�'���6�xəe(�e'�{�6�Y�ٸW�bG�=�^��J����տ��dD�͆���>|W��(ǽ�����a(,� K̭��r-w@M��O����Zk�^ϰU�@c���s�����¸�.͚�5!��������� ��棸aM��uj*5ΉS�Lnq[� �3E�}�.�c�ժR (c��8!� �7T�� �4~�|Gn_]Y,��&Ҩ�d���G`�f�q�V���@b:J*�IhŜ�6},�S�����V���)M� �j+րq*�]�I����^�j��X�QGs�=ܪI1�$�?M 3�}����(b�Py K�;��-�����q���$,�l�t#�Sx��R�H\b- 4����:_>8�� � �G�TS�J�:�F\��EvG�� ����G-?8Ԍ���ʴ��@�T�p@m�5I��$� ��4�9Cq qKU�2X~E׫��Ƨ��IpP����lRR&��� ��*��YY���� ��S������� �����Ť�F�~��Ge����3?8��l4�,[ە�w��NqU�������sS�>MY����cA��[��>�A��Qt�T�c+�wb��[}���^�fj����i�Ӎ͘����+��9#����s��zUw��4 $Zi*��-�ͻ�+�z���#�&����jD��,�5�ɲ@_�H,P��í/��&N��0�bB��xt�%�9o�b���B�EH��UI�8����kp��z�R� �A����cs*�-�2%*/~� ×����é�Ҥ%�A}\�C�EX�K�lm�Ћ.���-� ��=������K>�5���8D]in�=M=y5�El�I��@�o[�)Q��ew���҈�v�,��M���,���'4rT0�3>�Iڅ�� k)�/�ǐ��2���f��y�g� �+�P�桄:0-'���d�������1}&-h�) �'�!3� �Rl���>�,+��X>{9�ǨNU�+�%O�ۅ�G�� ][��{/C:!�NQϡ�u�.�����1e2�,尢���t���K,Qp�s�j�gמQ̗�{O8a�n;� � > stream xڍtT�k�/%) ))Cw ��)]*� 3���)Jw�4�)HIJK��HH)]�z�s��޻ֽk�����v 4�᫴�B��oI��P�4u���K2��峮kx��Ê�}T���F�R3*z�Y#ѥ�9W4���֒���&�^"�F*QCb�}�0^��!.���^?5(+����I�U�ס�)��#{�C�%��K�O�e�LiT_�4Vݡ2�C\^ �����7�i�?��c��496����������c��P���Q5��T.>>~�Tm�������1�h�J��� � ����7��R���� ��Ё�O�� ݟWBԸ����n�D�&�"f��8�.aW�=h����7Q�*�d,�%c[�֝x=a��x. ��RS���Z����TP����a���|c�ϯ�ĥ] �^T���K�䒎�Ϩ�|@ub K�A}Ӟ�8rzMG�4�g��o�=�ި f�\P��yZvF� �w��*8i�RE���r{۔� �O~�ٛ��Fo���1�Fh$ya[���}������r��K_�m/�� ���o/ tIzN���D�42�DYU�K+��Iͥ�#7���V�W �caQ��!}�Q�[/|��𺈣����˔ ���m֗�0��?�NG� �������)گJ�m3��9zUK@.�e|�`W�D#ʼe���r \s�i�1e�6-�ő"4�Z�}أ^�{�q������ �6s��w�7>�pqR�Y/���&�v�^��� ;!�6��F ���4_Fz���o���_t��o�{&��s���$1��oc\���c�����r�Q\�ꉛ_b���6#��j��dtg���ްX�:I��d�~�0�MK���3��{pŸ�,�}`?}2Q`E�Η{z�_��.>�h�fʄ����|kE��%�PF���� 3?a��䣌a^W��4�5��=���BcW���V����xL�� �G^l/����C�ּh_mHtVɭ������S]�Ї�D���,o�S��=W.� 4�� w�F�Q�G�O UcgU��v��/�ηwIѭ��]%}ߕN��$�I�)�Jm4��N�� � uy-${�Ά��*���sz8���a�:�C��c���*�r�K���:�~���W�a�-�.j�h���|Ў$�G��9^�%zh�Tm�mC6�dËtu ����L����}")���^S#w��E�"l��J��J�ħw�R�'��ZA��O��/Y�G�@�T�q��-� �~g�o��������0��BޗY��mi����S@Y��?��G/�f!�Q"w�)�\�G��/�ħ�T�$�j��y�$LmG$�Y9_ P���u(⫏4o��`'b ��$��Y��ֳ�^���y��2|(سUzkw���M��P���>����.*E�"e[�}K�W&��u��p�z��tUF���k��7\���|"��X�#ʨ�L����:h 6� Dž���P����=&'� 5k|utm�29��ܮ��DjR��߫�I��Vckr�E:6� ��r��K��Y����ۂX,5�EUj���[%��| ���^q�;)/��W��vi��>)�D���5?� &���[�� Q{Cx��B��M�x���v&d��^!��\����)�@�P��݀j�uNxyܐ�o�a"1��J%-�Hm�����~og}�n���P�b�=�����W������ �Ѩ��[�ȟ,j����~����>��m �rJ�:�d���Gx���.W~�E�3.�*ΧkjEq���+Bk�e�]X+ �1�� ѡDx˘f�m�ˎ�&Iƣ����Jݖ2��+���������ۢ�[�}y��s����Ķ�e�:ėz��5��\�{��&>�!&�ۅ1Z"�u��l^��͇l�����>+�������r�_D�9A[(��U�� Xcg��G�5}�F�ȯJ���MT%�y���10�ށ���u���6�Nh| ބD-�y���7>馝d�_��3��P%o @i}Q�Q�(�r��rwAaG�\�^}���� 䆊VZ��mS���_E1������,�(Aݿ#�+8�h�!% ��I_v���Wy�b�yb%��3�c�\QF��Ĝ\��.:,�z��l�A��hB��(�|B� ?���u�n�� ?ȬԾQ�Q�r)e��I��h�7P�ȗ#�j�΋�'�%m1���˭vC ����C#Dc�s�1�#�2���������.w���������դ��k$Ni��N5vR_?��Kf�r� �b�0?�1�P�x "�0\ũ�|�vR��x�$u[��}$̈��wHM;C���h3라̨F_�����(���0N��w� �؊r���s������K�U���wq{ X���>��FTD�6gP�Nl4y�2��Q;=(ejH��:����zeG�\�,$�B��;��>�|�.�2K�~��?w�My��Y��C�,_{�����8U�� �('� �ij�Aʦ���#���/�&�࢞��=��Jm����t�"���/���GW����&��u��;� ż��/�y�v7��T����p�~)��pO�5{���` ����aB�e�W?h9��t�2!ѹ?��G������:��=�]���T��A>��l�1t �j.�i���\��|G�:���"^r�ޫ"���D�u�ݓ~� j�t�9�rF�xT6 -� �H�l �$�ęlWI�fYE=T/�e�6�� ���A}�Wm�ƀ2�պ�NA����x-Ew�nc�%H�����f*0��G�bU���Ő�� /���/*-�{��J �3�3l� ��7�S�zyOG��Lmc��҃�GB�5A�s3��ޣ?X����O�;.̪��d�z�e�?�8`v�G)g����r�"Իοm_{�\d��6�����7�����¾o���=����3&��A��7�4��m�3�����*\|N�}�? ���k�T]�+2�"L�*Q�'�g�b�9�dP�>��٥�۱j�d�z� O�F>�u�p�����m��Z�E�ɞ�����R���/��|�p���lTSJ]��Sދ<>�&>ޱxx!�/��]���Ah{.Nb����`ނ�/4w��A50��+�m���2�(?�N܁������wvd��_r���k+"�K�������Q@(7L��%ң�M����͔\�h��V)�T�������F�B���jV��x�'_\3T�.Ý��#׾i/�L Pͷ�B��6ä��b;� Q��w��j����E�c���ָ���O*5f��s�X��FJ��o3��_>�Y'�p. �zw,)�zE�~� ��"pLes˶� Fu���z���k�!��t�^��~��e����Mڥ��p�N#O�lk1�o���5}>�PXFQ6��� :Y�����>�=)�Gl ��8f�|s�� G��1��y_ݗ�g?����;]�� .��˚2�F>�-OpH�T��䝭v�����7�#�!D�u�R����oٴ�O�R�8�Fv�O~3Vendstream endobj 539 0 obj > stream xڍxT�۶5"]:�R H��H�ޥ��AHB�:H�"�� H�H��w�(HQ:" �����{���1�#��\s�����3X�O���)��h>A~�@AS�DA�@��>������\\�(��0\`�h �h��5QH�#7@��B@���D��@��h���0W"6��� ������#������s���!�H��5����F�P8 ������hg ~k'W~���4/�����\a.�0(�W�-k'؟������p׿ z([��� �p �qqCBa.Lv���@��������gs���� ���W 8�5�rr�Fz��v[8�V��G{�y�H�/�5���v��#�m0�ߥ[��t֘��� q�;�]�]�_= � ��f%$T��C�]�~էw�A0��%��p�(���+[8j�� ������ ������������a����� ��� �J��� �m� cz��qF9l1m���0�����; �vq������+"AAAl`vp$ѿ�c`��_k����=f@���_�=Y`E!^���>be]��� ϟ��e��Gy|���|�"@����@TT���8`k��:��W i���U.f��.���8� ����P�����9P�� ���������_Q�W��wE�n�o;�_���n�Gx�a`���L�& 3 �����]M����V5�5f�vE� >�>� ��*�=aP0 ��K5��� G��(W����� 3dG�-⊑�o 3C�̫������MHX`��b�E�9k�J�#��J(���H�����E��:XL\�k�~~c� �ts� ��h�3&�ߘ� @� ���_�?j����`����׿o�!Z�GA$C^��]T��{�}�[]눌3�F��}�c����2�D�1���H*x�$ljޗ���b�Ӽ;+p%E5��Կ!ė;�����E��M��FnP|��a${l�cƑh4*�ڪ����e*�p��3�У��ew�a?3af]C����P�����L\c?�ǜ˷E�~��0�I~��M�@*�8O��J�"�����#�쑄AZ���G��X���g %�q,�Na�+7��|��h����i)��~:��V�����Y9f�����&@r�)!Ҟ�:�U�� �u6��� ���ša��v,�ԜA��h`/5D�? Y~��{��yF���E�Bl��+y0��y��h��a2���� �vS��a�m�>/��W��2�{�����7m�=�G(�CF���8�������F5�Y Cn%t��[T�Wlv쌿nN6��CE�TV|a�%�D��O�V_c +M��D"T.`����񰄱r�\d��TP�T�X����� E��X�[:=�p-�l\��,�s:L�iq k����Tq>:�exO�R� �h�r~� HY�;����/9Tg��ӽ- �'�A��x&V�Nš(�Lb�F��JG�BVBͷ*�KP��>D�w ���IO_�Oٝ���X�v��P8�ZY��������0�%� �D�p�~��q��Q[����2�y@B�;�{0J��ʝO�س�ܤ�On�-1��_6�/lϘ�Z�,߲6�G N���X�sG.��oT���L���N��PY������DL ��̤� ����h�a}�gs�V�s� �j~���O����pis|0�Ӻ$��t[f�������HL�'�����cU���3Z+��� >`�q>'쎣?ov9o�,�5M���t_`w$�����WM�r��Sz�Ht�=̚�8� �(�i㭢 ��� {_���]�2&�|�@�����Y�g�,�P{�����(Gy��Nz���7BX���/1~ዝ�� ��;��V�^���i��Mt�.>����O�n/b�Gej�̶��;=i��J���$���2��+��� '����h��c>�Z| �!�5=�i�D��Y Z�x(� *�Ƙ��3�$8���Q˨ޤ�@y�ᶼ_mʎc�r/�����܄��9[��z�)�+z�l����y�_�V�c uEH�� *���%�{��y����0�%)�Cbf_/�L��{+Pq�H��^�y� �[+���kn�������d[� ��݉�N��ݪ��l6����R\� R��7����cu �]�Ň���H��&����norM�gZo�w� �����!�͑�7��}[*87Dљ�]6�� ���DԾY�]h��Lj�nG#:j�U�sͰ�=]�5�C���u/a���B��,ǥ�L��\�űp�瓽��X�� ?��z e������'[b����П^��{� �?���>)�S��?����ՒqC����~�Š�@�iK|Zn��� U zݼ`�O��B�y�:^���܏�7K����x_i6(�ٹJ����k�J��g��8Ry+�b���x0w����׎-��� � F[��Խ� �� � ԭ�pB��%���ju��@c��e��EӓQ9�%.B���ĭ�9��$����e��U�G{��w,���f�G�_�k�a�r{� ��7+�>�7aÐ=�)���'�@��i_�����3=�ھ�^�m� �/EL1~���y��x�y��|ʄ�)��D3��'|N��O]` ���P.��j楠u��Z���Y �yr�ƄS�eQ�{�Y�Иc���u��p���C���|a�@�NG�}飔@=\���#p�v-�����pm˞�u������j�`���b۔{YEު�c�"{ ���E�m:LY@v�U��S�-�2��9�g��V?��������b�d}-u��C��,�]�,�En*�K!ݝc��b���Se�K�)ȖT��ZK�rqn����p��LAG� ueD'�p��+���8s�Tp���Ȳ��A��U3��|�YPKS����6��ɀ;{C����G�V���k��>��'�Ƶ�e�������EEI�����`�V��0���i��(� ��3�����Ш��s���4Q㋋F 1q�����-:wy��R'� �pa;B�Ը,=��;%TSޒ��A'-���h�CH��� >�i�J:]:䕴c�Wq�>d,��"��yhP���ֲ� ����D.M�~���K���8J-Σ$Zs���G����8�vj~!N{k� yv���(�;�3�0 �����+��Y ȖY�ڝ����D�o�y�"��樓�z�u��jW����X��>�����l"�&��Q�m�|6�x���ta��;^p؜9]����Tﱦ�2�0P��8���G!#qi���F��E_�)�P��%ס4��ח�����a� �%��� �V�v����������k�NEu&m|��p�LN;Y �B����]��{т��v�,��@�儃�ڙ��U B7܉��/?�7�\��߶��y�,��k"a�&X��8'� Օ���*�'��e$�y�)}_��%���g���%��zU%!>)G��mDK�잱-G��%᳟� ���8��n���X=c؟X���9��}����A߁u��������L��5���Ө�\�=)D�[�SG��f{��^�N?�����d�z۵�ukv&��RD��љ�dsM��ל��[���Hg�bGM�FY /�[��99�v g�A�'J3i��I%ȭ��S� �Ol3��v.��+��5�I���@{���o�B��ǡ�F� K��nR* �]�݁�>ډ� B`��JO'�g�s�� y���!z���*r)-O�F^WA�;i�1:��^#`vA�c�(�^Xjl?�?�aW?����)dC��t�kWKuT�I�Q}+�0��H ��S"5�O����p��K+��ߣ�ۼ���Z;���C���uFe��G~�q+�� .�u_׬�^�\����������jnQ7C��tZ����Q��uw:u��� > stream xڍuT��6-N� et�N��a��m�� �� �)� J(!) "%-���HH��t�� ��p��E��D��DD�b""����r@5�'�uPH8���r�A#��2�z�B�����҂�Áʮp4 A�!XG�+�"�4AAp��?R��;b�nr��^^^ �+�B;(� �XG�1G{�a�_�B\�&������ ��AÁ8�  GbpH ��h� ���?`�?A�߻��D���o��D��`�ru� }H�=�4��a���@� q��p�O�b����P6Bp�E#ܰ��kD�_ip������\]�H,�?5�]����:#Q^H��{f�k�������V� ����9��@IY)) Y ���: �Jo�����e�M��r�ㆀ ���� b������� * �!�X����';� ��s�-���Ep�������G/ ������ k���R�3�}**(o����PHVR(**# ����3�!� ���j#�Q@�?���_{�%�_q����. �Z8��?$����D����;����_Y�_$��4 �+n�nD�þ�q�+�>͊��@d ��0 �n�O�.qQs�c =(*kK�^m�~��cw�������� �?a�}�}���t e����O��H�����0�����|�P��wߡ��ka�wо��[��������� G���(>���\(Y��\3{D�Q��ЮL]�ɫB�s�O��Q�Aa��0�w!,_���#�$+��������$�.y�Ma�r���o�����U���g�9\c���V� �`�+@4�, ���R%L/�%��S��=����;�C�:�9����n��ɚ�?\c��s���!�l�wR�֡�1��;��>r���\Ѻuv��V�>L�M��dln�� U��n�d7�l���Ԇ ��yh�!�y"�?� ��mn7:�x�_$��V��]�$�}��� (� %侦0PHFH�HC�T���ƢΦ���Oݿ��Q��$�|u |�"�#��Y�=����{бK/?W��Ĉ@�(�G�˒@�A������@:U�&R7-EN�К}#àNqV�3:7�G��-9�,�ˬ�+��T;�YU�D@��Nڒ4(���i����!�;]Dz���ʢ��޻C�|`��![�C}�C�&Z�aZ�����.�1�pf7t��\s��:$_�3U���J�O�l��~�,�ƍN���0�����N@�c҇y�eSO؊W�^��闒H�}��S�Ft���͔�%� �y�#�c �6*�y��O�D���j"t�j�xĚ �#�.*�-L�Z�yU+(K�?��- �� �Yl�@IW^2���}��ˢ��W`B-�M�ڇ@��N���@�r=Ƭk{��ܻ�uq+�V���Cd����K�n��uw��NK�UԀ� ��Tg��j�M�h^mΈ��s��5c5�� F�]蟡���w��+���m+���)U���({���;_�>!:#\\�O\�F�p+�SG:�C��S ��ͻ��1�6�� V&g�͌3r-� �R��? EG(:I6;�� �x ��^�zV�O�Y�0+��������n�Z5F��!T:%�gҐ�f�0)ro��(��)�ؾ � �n9���q]���(�5��~v��#��ƙ6� ��� ��|�k�`��rŖ�|��#_�7?fҥt��3���w��A4},��G��#͕�5�R��g;�����wIB�uy�Y�����;�y�:I��)] �()��me��2��P��ˎ��E���޽��J�jr?���J������)���T|{9��J�4�b����_�OW���M7�� 7�\\�Ӹ^4^aи����cC����avM �r�(3���)�_\�hCG���O3%�iX�3p N�=-Un޻=��S�g�3�\���t_!��K�9#�����Q��B�+������:Q����=u�Y���a_+5׺���7�������(t�y��i�z� xu��A�̭��I���(ܵ@Ƶ�Յ���)d�o��,p�kp�R+�j>�69d6-/I�C�Q/ E�l�a�0�I�kQ2x�I�~J��Iek���Vw���t�RZ ��:iͩ�䥄e��A! k[E����5�C'R�G)�.^~�%��4�:ߜ�P�'p�}�n�r����h2J�A�� ����m�Ƙ��S\�t�Q1�{�Y\�AȭޝV����"�$l���zC �"5�-|���,/9y2��_VF��mǎ�����$w �p������b�g}�ӷ��"J�G�"��CQL/�Z�Ą�Xc,$lC�>�M\O��T��DtT���UfzH�K�|��>`��}�W�X -�����˺A(�jo�}R�2�B��[���D���q��.F$��� ��RQ} -׷)��^��P���>7c�P�\M�yo�>ԇ�~G���UAH!��ӵLJ�F�{$�(�-�]��r�əoZw�d�:s�ܣ�ڕtUч�L��$ʖIAy���n ]���� !��m�W�( �z�B�� ހ{M����֙��Υ���;��g�x-��r?I����6�T]d�p��g-��K�,� �z)���>�ׅ�t�� �$W#"Z���7t�@��6����0���������T�9�~L���أ�G3ɱ[$߰m{��>sJ��|h���,k�����7p�=>����e�v��؇8�w�� ���������Fw�� ��5Vb ��(|9H0N�C�l��G�゚8��N=Bv}*��G �����ܻZ�;&�v��0,bp��1~�B ����E]��|�l֏d,�8���H�a��9���&���aT��\M�^���z�d���\�#.�l;����MI0��=��¯u��Oй;\�C��vK��\�B��_gI9����� �Nj�~��Ě� [S�/��[�Y���{ڔu E��eAkQ���d�����/�H�������~������!מ�ʗP�_�t} �N���l�۾sA��Bd8��v�N�i���/�R�fv�_u��w���&�����N�3��y�=����@��;˓�����^ܐ�%�ԑ���~r�[ ��Emv3TC�*|,ɺ�9�a���� ]r��k��"����g��gW�2�}��3cy�z�(e�l玔��~7��t}�� �)q��+��i�1R�w0mٞk��d����^j����Ӫm�n�ň[�Dz��* k&�S�}iB�F|]�����}�ȖjNc� ��6'}A���5~~��ұj�wʲ�R��5ˡM���� > stream x��VTS��*M)�ϥ���^C�^CB�! I�������P���*E���" ����ެ73k��Jr��η�>���%"���;R�#K�ʀ����C I[!�>X8��DDl0d,�2��"�$ ��W� 'S�p2%��X# � ��&��&/ȁeU����S ���֞p"� ��'1hO���%eXI��S@ቀ �0 "�?�;|> p�e~�nG\���.a8�0�1���~� ��w�'��p>$J!���T'�3��'��቗�7�E�IH��$dO$`��L0 $ - ��?��I&�@ �ZOD�P�t�G> t!�onf#mb��gf�'C�'���� �`I?�����*JKj���RGz�`|�X�!k  &`��!���yjyX�>�)�G��*`Y@�����dH(� ���S硃����d��A�(���@�P6$ K��ǽ)�L��N`0|��� F��� �G�� �,t�M��$u��Y��x HZNP��TTe�_�F��c�d��FC ���Ç�s�?� ��[�� O����]���� V�KsRde�Ǟ����o�9�e��� ��"����7���1��R�;� },��,��9��C�9��^l�O����!��g��yJ �pJ��wC���#=,0d�'��c):��C�c18���9�)�e��_�(=@\�!I$@�%$��?��!��ST�D" �����+� ƗrHQn(+�(A�񧠂*��~I�@dO"�(������0‡H����B��{�"�D0��#ԣ.VF=�y9�'=��wu�wp�ld�a\���{D����U0�q��WQ����(�1nh� ���,mv�ңg} ��a�{��)� �ףm��-�qLZ|���V?ָ����4Dݸ!A��Yn�Z��^�0�M�;�ײB%s{��;���{Ѥ6/��P����#�R@��wԂN��ZC&^ _z��^��W��� ,��@�����ǹ����\�3�m�.���訳�g����n�Tbܪ����D�Ω�|R7A�����NnM0�*���'n&��� J�L����J�WJ�ȇ�&��s��P��+��G�^�p 2�X�e�K_s���E���~[8Jm0+v��+�,2���3 �J=����=�v���~������{���P������[���-X?{�g��U�j�b�>�� �M�����ܴ�*i�/��J^�kC��s�V�~c�����4���L�WE�d����=�M��e��} {��cF��65�-�m� ����V$��w�m��[%�(k����AN��J:p7��/�[�z�Xm�V�]S����lK��b 4�F;sdt�06��Wwi(WOI�Ӿ��e�A�KV�n�Q/�N"oj�ε`�i�TWKg�O�֧�1� ���9?�o��Y'�Y���3�����Z�oAL똬�����P!����G/[��%� B�i�%�^N ��gb,n՛/�$md���`nT�V�~n����3I�>_E�����wY]�';" >�_?C �@, n��>�����k�(֤1X��R�yv+2^")ƽu�[f��k6 2k�*߃��l,��g�o���6-�������Z�"5��������ߚ=�d����yRI2-��7�5v 8�m���J�l.G������J�wѐ���e�Y�Y.4͵n�;�g��Zq.�[ 2���}\�;o�kuN\mR�����n���=���K��u��觢VG"\b���oMu?y�H 9p�1�xB-97?��PU���9�8��wk�*h� "��0g�XD�~V��>�Y� 8ܾ��ɴ���%�9�e�/ɩ�4����������Td�S羕�4MCg����6�# �4#7�w���7B��g��xqTT�dg�]V�2~��[u�Z.W��(�::����������&��!�o*�҄��;�њUyA��v�쪲����c=�Y���c���$>����4ut~�&]�et/������D� �+������hje��#�*^Hҝb!�޳ѩ�[J8e :)�bE�l7fK �v����p�l�����7���9?'g���,9�%M����=+��oO4Q��J��1�l�M��`q�((���uU��[tv�|���I�̽��Ao�X@3��R���[�� �K�"�;u� k!��A���{��7/�O�~M����H u������J �������+������f�^9�R�h�Rn����� ߌ���T��}h($ k��R�=�����aDJѪ��B�d �Dp��3�Z]�I����$�rG}xĸ_� ��x>.�D�XU�Ǜݐz��F߹�����^zyb#�*�� �i�/;��7a� c�\��X�4�]E�r�F�\hk��=b�v���P�-����p��x� %\���Ɣ�2�zUWq�w�ii �AyR�KD�����q�w�.��Yn��� ��~���d����M�L�Շ��Z��551���z!f���m�j��cB�T�3y:b�������Z�X��ia|� L�2������X;�e]�T;������'K�{2�����C3+�����~������ �lJK�39 Վ֕H &�L���L��ߍ������ϸ�KXC�[S�w�"�J��!lP�i4�� {�&yt3y�=���6DR.>W`�m��βH�e�XW��4 W1V+�^��յ�~���ע� $������4ʺ���hD ;L�ȍ.�tܮ�yg��rc����('�y��ϱ��+kn�_`8��ym=�`�����s%����p�،�"����RA��=BJ��GB8� ��|�{���ˇ/���J�J5��떪�-����lJךmw�٤��ƒ��4Q�b�Ά��J����'q&G�Cw�43�䑹hf`��� ���N �関�V�ޞ�5���R��� y���I���Ԡꡪ� z�^AY�XeeQdz�FV��.[�� tZ��nM���i1�� � �[b�v�Att`�m WW�L�%oy������k���]`�bʾ15�f�*�G-ԭ���w����&��\E���n���@zZ� �8�k���s �xIoovX��r%�.��S�y���� �"n�Q�A���6C X�V*���ʙ��4 ]r5"�  �ї35ڛ {��O8�|Nd`y�&��軵�;jˡ(���꤮�����i�v �3@O�{;���]�����Bg���lLh§�)�������Im.���0�onendstream endobj 545 0 obj > stream x�}�y8���%K Y����Yu,�T��eb� �1���XK�r4�X�}�D%�!&"ʮ�$4%KBF��T�s�s��w�� "��AL ��K�� ���� ����kk @a`��@�C������C���3`pc�����#��  ����8&,�W.K"sDM����� ����� ��Q�X@�gc�ɣ�q���W���PM,��� �C[��X�3�#��Q�'PHa؟aWK��X'"��@ ��+��C�d2`�gʑH��9���;��N��n�q��y(2K�ƿ�����0������VV�����9�I�������"�P�!C �8�S,�s[0�@�p�!a���H�>&8F}�$c��7�`�_��0fB0v���&�����!� �6Upb�Mȑ�oB�v���ho� 刅lBLڄ��ɛ�S~�?muB������w�0�';SH� �C �\•�pTO±����B�wn��3�`P}j�i���Bt��y?~�����n�X*-02HD�9x�l#��R����S�?ƛ���.Rԇ����l{C>���;3����=��՞cwS����^��ϣpٴM >�t�.6�c0D�xF„�����;} �*W�Y�ks�[��|� ?y,�8�p���݆��g�\��H�{("`��3a��d���zaz� �QT���?����܂O5���)q7ċׅ�� �̥w��ÿ$=�3�s���C#A�e2yn�� �LV "�Yq��n`~C��R>!� >�{ژ��)��B�>~�J�x�;)���)�J��DE?'��2i��� &e���z m���&��>��ՄK*���j��)}�~�/W�c���L�g�OW�A>N!��I#|�'�q+3��ML�Q���M��ݩ����K.Ǜ�pg̞�� ��*� �=0 _���F _̅����“����^WyX��:(v�������o�ڻ�t��S���S� =;>7j�ng�n�!W��i�O�`]R܁��$��! > stream x�}�u\�}��%�A@Dr �n隠��`Fl�����T )%$F�ҥ(5$��뺯 ���x��g��s^���.�!�a UC����B2��' "��RAB�h~����E��E���`4�dp@��0#]7��O���@DRRL�LD�٠�P;�L����p[@X�o;����;�������JBp'/j��ssr�;C ^� yY nL����q�M���z]B��/zƝ� E���� 0 �~�����������!�p�� h��!�`4�i.$ �{nB~��{����2�Ӈ_��p_��8.NJD���lܐH�v�k)�6�?l ��D(�jC6�a#�P������~����D��Hq��� �TW�h* 6����W�U�n&V�R�U��quZ�,r�+Eő�v�k{�QK�u�Oa�#HK��Sv�3�1�UD��S�2B�t-_1�6��r����E�����P�ġrh��w��{��b�_����7F���KZ������k�$( �5�� ��D���GY��x����'qq����L��8�� #�Es��uG���ɋ�F�M}�"ޤ�P�� O��=�qʝbu�Йy��|VP��Ů�Ω�-�H�Ϧ��e��oS�DD��� ��phv�_�j��G�X����RU����r��O*v����x&=��^�1(��*!���u6Ɣ��u����l�]�C# y�}m S>�uܟ� �AV��U�{����(��k���}]1c�,º �}*-��)� �����׌��fklI{T�Z��$q��c�n�u�,""�0I����G��E؈o��� /߄�C��,5�c�jK?la��q�5œ xWW��:�i��U�wSs�e� �YS���3I�g(�����\�f���dK��WOx �b8,~*om����s%I'��|��;� N -*��1�b?��o;���6�؆�#�zVOU�7��݈�6=-|�+޴%�ԏ�~���߼�ٲ2�~���?��-2�ֿx;��h&�r�{,&�K$Y�����2/�H�%_/�9-�+ۜ���L��}�W=~uA{ڱ!eі�g���c;�����!�;��M ;������"׏�����?p���S|����."�=7H���믳{?����XHZxnɵ�C�^ab���,���F2�A�����]ùQ*9�Ӥ�j�(�bR�����*� ��M�AhS��N��f”4�gNU vh���Ͳ���I�=]�.Q��c��hb�՟W��jd���f��B'bO�wT�m�[1�uf����y��^b�O�s��/�%�6�#H�k��W%WB��b0��)��2j3���@I�\��T�����W� cR�ė�M���O��U�XS�*�9%�O��Ԯ:��o���d�O���!���>�Foְ� �N�:83�I)4%�K����R4�=�{�T}�.�=�F���FȱUu����~�p�c�}���U Q~�Y�&u[�F�}L�ԚJb ՖQ����k3� =����Cx�̪�$�j���S�C6�j��N�w6ē�v��o�-j�/MS(l��V�X9�Mf�� T�E3?JRm��ȯ��� fO�죣 �H��]�$�g� ���%8G�d�����T�JY��� ��~C4�Ѻ���l�ѫ�L�Gl�LA�MK!���mt��7�6�bN[��fЯg�LS՞AVʩ��j����A.�%�a�L>�|��x-ka eC���� �3���7�=C �������]B҂LVۙ�?�f;�V��;n��:kRg�z/�e���h�`GGOf,����(/�5uN���:{c�-V?�~�B��G�0��R�U��F���'���1*a�+��%�U>� ��snH\3�0��->b��l�)����~��1�E��X�L� �P e)�Y/�%�[��-�U�0��8o]������} -T���V� o��0����xR���F6.͎R�o0q�H��V�V��q�s X�%�a����`,���4E+~��4ū*d�N�r\p?�4��[g�����&ӈġ_���̎um��r'H?G,�!^��zf!Ux+0{P�J��Me���{[��˄f˦�cCO��?�}���!�lssC�LW���ߜ]��#B� :�y���0K��`�p��SO�Ÿ�3Sg���li��S؞t�5�����Tz�q0 h8���F.' ���!?n��u�ø�ן����~�f3�eܪ�@������ pw+A��8��,|8Z�ϔ�A���L_��²���F�D�Cd�9&Rn�1i�0�Qb=G�j�_��{�|�aO��u��J�3����u� 8x�j[4�ա�}�J����p�f�x�N�a-�'V CW'�aDeT����d6�!�oM��3�Q]���o%]}X�I����;�$����V>�(�u��Q�6a�R�����^ iαU�����F'�ђ�č�4m4� � �t�Q��� ��%���+v��|�����߁�:�u���ŵ@J�pL�ð=}���Q/��w�J8�6�?⚐�.��q�8^6�� �x��)q3�:��dx��f��� ��y�|�efO��x5��huR:����)�$��U���&��wY$z�B|�W�����'��Вx2�c���{5|�K����T�|�u�.�Œz� �F_��B�Q�yx~����J�I���.�2�FC�Y�N��`�5k��ݓa1�� �@5�� ��m��#��{��d ���zl����>�9hDW�#��o2k����p �.� a�\����~���XV�_��Ӛ������I`�s�ב!��m�}.�.� ����'�n�Oӥ�2Dui�i�[} �� J� g��|�� �hY�F}�8�E˱��f�{�d�V�l��tf�{���"��;���>��"m���"��2�#��-��&�J���`��~�[Uڬ���M@`B �3k*^U�+�a�W�o{�Z���z��h|fl� v���_� ʃ.o��a�q3$���4�|͜��cX�8�ٔ�0k!�2AJJ��3����LJ��Fw���ػ�Y���\q�c��ݦM���3wV 'n-΃e�8(�m�ձ�ט��-�Lm2?��Л��ĸ1V�ř��[aJ4 ��oť�����A���N{�O-A��L��0-{��Mg�Cy�əg�8��u4�n���;]��t �{�m���%e��GY�o\ �z-5�v����v4�5�Snh� >�gS�����܊9O���W�f�]���C7>�6;�>��&��t����S����y��8}�IP��s��d^��Y���Y�.��h���G2Jq�=q"�_�g| �e�V�s��Z�ۧ����QS��2�ȷ��ڂ+w�BI�����i��Ҳ8�*��˞gR˳���z��B����8�i���_�b�jn��~��v���a���/1 ���T��� #R�(��F����;�IJ�Qy͡\u���H�vq�x\)D��W���3l�\/_��%g �� ��1�':7���4Kx����n|%nLd1=}]t +��CX����w��R[,�~Wl:�ue�&���a(٨-��ML.�y�y�!�;XJ����2�}e�o���A�E�����d'O;-���q�ؐ;A�O5 �X��ڞ��֧T�-]ݡ]s�~��I��y�ُG{��N�:1?�:4o�։ng[����Y��+�ny�@JKA+���t���M�|�6L�lԋ��;�n�d�Ns��q��N�)�:�s1�ZVfF� �c%q�e��^/��x��敵�\ﳃH� G��T��ɮ��~mp%��ؚ��j � {���t�ioBj*�;B£��, 2%�Y>^�T�)�!����΂}�j��hU\.1�x����[ߋבMz��N��9����H��("�k�U�"�c��dұu�7���O��{��/�U���vfƩ�F��_?ތ�r���+H�J�������ǗS����u�22xwdS��VWe�d�ѓ��2w�髂to�L R*��� �u�f��^�%l��LL��*cyf� څtj��0\��(��G�?��6v���k�?i���魥���Lg| ��6����C{i��HP>ͭ�_�7ӅJL1�U�Ó #�O�BA�HL {ލ폧&3z;="^i�Y �#�܊�{!Pt>r"�zN٦���9��z.���Hbf�oo��5٦u��Z��e�Ȯ� 9�.9�9 w��M!ћ���U^�*P���ۢgZ]�퉏��ko�endstream endobj 549 0 obj > stream x�l�tfM�.�v��vǶm��mvl�IǶ�t��m�c����9�޳�k���f���� ����SA�������R #J6��E�7�8�:�z��~#L ]��V��{MdG�b��P;� �����-�z�I����L��H��ce�q8���:%�J±u�����ui�A��\�����4�d��]n��\�����1A01��D^�� �{�qa��åU`���x�9���wn��N�5��/��$ޜ��X�M�+:�Dxn�������?����ׂ�� ,��R�� g�;�x��� ux �g�%Ej��4�/�+��Bu {�Y����{�) ��}��oD��R;]�Y�� p�7?;�ܝ���y-�q��te�:���a?����x������6��N?E���� �ˍt1���d���h�{�|k[���F�W����M}Ȓ8f���J�]k�*'����n{}��>H������Y��B��,,Z�Нc��֙�r ;ﻓ� ��[�*pJ��kX ��J��:�8���R{?�Jg��B����-2Y,v�l�l��v�{|����uG-�b��۳'S�������p�������-�����n�̹�+9әOb駚�Lj� ���K/�I&;�U;p.Y4c��L�(VF|�����so�GK٣��k�9�#`� J��$H�-��.ͥ�� j4���t9�AkQ"�~;��B6���� %H��f3�����Mi�K*`P"�����34��� �Q�L�fl1� $��uZ����D�������fL�a%E;�s��l��f��Y���/����7W�h3+ ��w������v1��a~u�E�g���ƪ�;Ul���)��,Q��:�.!�#�د�^�n�t�7���a}�&#�'��!O�n$x������X��rR%��}�JP�����L浨S".�c�%�ԚT�o��y����ڈN����)I ��R�-`�� �߄�#°�6Tv�g��O��!{�_�o�����w��ىM_|V֎����U M�W�q���x� ~��ӉPn ͦiۮS;=SM�zA�7A�^�ἦxh/��� NV�p��� �Uu�xn� ��D���M��EL)A �b͌"����O�+bM��=��XeON��7D9����(���� 䑈džf��U�*f�-v|a���k��KL��6q�s��z��h��������v�����A�@��X�:F����ˌWV���=�ug���-$����+�ֆ��p����lXw����̸�i�|m}���@f�Y$\��� (�w�t鍴��M]�珇 �՜��\��O%̭4Q��O+FmG�`�q���jʼn����j�Ͽz�\?C��Jk{���]��UY"��].���h �u�Rxm�#h� (����Dz!w���������w|Knu����V�F�`�w�:��u��\���S� n�T$[V�eb|NbJR��7�zcH��?���b_��©�-�RJ3#N_CC���W����DA~��\���GY�u4� R��=Q����kyE����~��9z���8�ܘ�n�[�����iW�j��r����d�7I��U�ڷ��3 �p�Q�hF31+���' ��nG4~��#���J���i�R����5A2��K!��m�ى^�]�,��u���Epdc�Hғ@�Ǥz��"Cl�,$�{���)�hR�ό��O��d�';i � �ӥ �SMǟ�G&��!?��9{� F�~Ş��m��|7+�n��tm^�f-Q���Gtڶ1]X���M�Ilw��"� ̡@����)�5H���R�2SO�t ��"f3��!��GK5��m��f�� 6�������P5���[m��W �%/�����D���; ���h��wCT�9 :O���m]����H7��NL7�Q�Z�&b�_=���-� �)��?j&e���_�� V£��+" r��4���cBW���\"ݜf��t��y���#R�:Kp����M����rR�S�Ͱ� ����+d�w�JU��V�Њ���b�5�icR�թM�A���ߧ�Z.��3T����]c��G �+n�˷ ���E�oM��ہ�D�� �B�~����`��#�/���t�����y/R � ܳL����|�r "��4�����:����#'L^ͩQ��N����Cvo8� �u># ?�I ��Dn�� �r��k[�+��A ֟x������, &�~i��ݓ�KGZ2|�� ᾶ�/!f����.^�CҠsE{�����\� .��H�$��/�T[C��/��:g,�Syk��''�/����m6��1���[PƸ&ӥ5x:4/�kI�gz�������'��� '�E#��r$:�]ZV�I ���w`��b�d��xs������0��� �ĸ�m����E�Z�>O*�X�� >[�x�6�w��4��.��d=w ��e�=3]�V����C�R��X\12��nJDO5�>w�����Ecj��8��՛�g1��O �Pڛ��h��z�^�};�w���._�ڼp� �>��&�T(�j�� ��x��b����;�� ��K�o���;�s|~?f��$�摇�J ��J�Ӏ5מ��jN� o�u=�|K6F�4��/�7{� `��+��� ���`:�3ï#���|��G�~f0��0�VF���,s=T�)%�l���=�(������0����Y��=ވ�!?I`@���E��믋5��:�=!��ӹ�A����h�6���ie��i��fx˹4B�U7e�� #��荣*�;%��Q��F�R�/I���6�1*w�B�΃U�� X�ד7zD�,��������ª:6 ��W�5;q�W��k%�ŵo�� ��� A�Up_t�T� *x�֌��~ @���?!ij��r���B�Mh��KS��m* A[�p�K��q`��(/&�y����ⱹ~*��yMd6q!�R�D>+��:D�� ���1&Wo��=r��Z��0��x�uР�0�4�P���SP�? 1)|��U(��n� ����^8[�Q]I��c�:�P���8C�����[`����4�O9�0֐��է�ا4����� �TP�`"1sw��1�$��wڷ�^�oب�U��AAgcZ>��ݨ�����qރfQ�I�@��(�� �B��z&��/ V���-ݿr����K�kY��}!$��Ls������"!�Eo ���\`T�&�x��e@�z���?QS���}�N���?�B�jX5G����^y-���m纖}�����h%�?��.�j3!�rE�]ei�eF���H �V��p9k9.�f��K�N4�J��#r��P^v��]A5��o���Ka�9F�^��>�(�/�w���'�����N��泞p�37V�����v�$R���ܘ0�}��KM���Rc�&�Fy}�6tp�3!�'31��s"�m�� `�۶����$4���Ec6��b��j�!-��dP�q�0�1����L�/�|���}t�E��~� ��Y�$m���/�?�4�y�_F������fҕ��v�S������ҍ|1q��Jw�݋�׷gqG�j��Ϙ�ʗ�&��@���U=�WI�ܫ߆̓t=wj��n��+�>� $��$�[��j���A2�bX0�r�kL�6�P�� ��]��sg���7����1� Yb��M�ҞQ *�#�,��[�.P����r�6e�37L>��*���|��QX!�릙-h��6m�O����Si� �ҋ�) ��o��������N��򁲯c�zmR)]�t � =�/�&Hvͩ�C�㥒{t%�$��8�0� V��{�3E\]��Q�S(��"���[��U�F�>��i/j�0���.�y4��; �V�*���(��jJ@�J�%k cf&�TJ�" 2R/mt P����Ԋ�c�����yJ�L逦�b���L�z�(��R��0\EGīAH8O$){�%��^�=�W�!;\��K|�!�~��N��]WLJ���[ϻM�GI����2�������{݆?�6��/{a� ����>������HA����o�� �S��1(f��r)/n� �����!�ܐ��\���� N�K��C$ίvd7�ʲ� �����a��|Ìl8��Kr/�A�;�R���a���S�� �&:���ʳ�n�I� =>�J��2��Kfe R#�I���^�"��/�:��:�i\��l�v�(�X1Z`��7�au���o�Ç��?D_�Gp�R�K%��~ϝ��'t!3s��0=�������Z��o�uن&3.(U�\M�t�� Ψ��zVn���x�#cnU�'|�O�� � @�ӤA�4hM���D���`Y=�q,G�o����f�XS����/�#��D���[Flg��R�`�J_d��PedK7��I+��WI��f~eh~�������>~b��9� �x}1RL��m.��ֿ�v:rp����c ��{Y����A,������㳧sL�\�q�-�x p�P�)]�H��(e_|�XE���������9�pv���54.#���g�o����0 ����t�l���^����������*�rJ{��W�&��M��W��w?X?��i�u�UK�3]� )Q���RGK�&��u�4 �Zb0�S ��% �Xh�y��5�~LT��~�D&��ȴ�������{�7yfHy�L�� �l9��(�>UH��y����sD�^=d�D�U-^�e�a�H�Z/Z�=C(��[�I4֡�_W��o�H1CK�Y�45��(�4į�8rEqɭ�Kj��Y��C���J�.�� ���LM̴�O,N��j�������c����v�O�-�n.S�RL�/ �u/�@m(vN��=��O�8f+� NРU���d���s'�NOm(���uD��j ��j}��a��0/׋H��f�m˟���oᒸD���}8c��'7��O��%�G�k,�ﳌa��p�&$��������+Ց���*|$۶a�t�x������솎�84���L��� �3��*5��)��i��lDAX�Tm� �ߧ����>�⫎�,}��vɞ��Y�a�7J��D��gYvEhTQL��\� �`=��%P�A��r��L�t8� sX6rEVؓ���+�P[[y �^ Ëʮ$�T���@��s�f�ld!3;^��1Ucx��4��hu,8�b��8�T��^�|h�{sK*E>���.N's���>0 T ��V��0�l���VoY`Ht��6��} �`� M,�M��g��K�M�b��^�����Īi���J� �z����3�����:��r��R���%�|�fc�$����K�w"����G�p/ev�"ٷ����QP='��X�AWU5�"X�aYI- �Mȉ�vW��)V)����(w��D�0�̕��&.��` _���r[y�˭�"∔ �z��Tcr�i���x'uo�{�Ξ�El����~ ��O/�a|U����zE۫ �;5������]a��mhG/�4!fq�_�㎃�f��Wv �h �K�f�)rUq��!�"�1����=m���:��:٘X������\5_�"����-+�T� ��0�t��G�� B`��uUi7iQ�2���i%p�{kŝE:�q�T�ʡ��BR[�7�L��ep�—��%n�)�!J �h� ������:�Tk�C�(��:ZZE��?��'�-#����U�פ�w���]���3YKJ����-�I���#"�i�H-�������§E�l\�=y��l�в��G�ʦ��nM��l��Ĺ������ �o�R�������4����w�#�N 3�� ���=����T.��`�=��#�K�;B��AzEwUlè#u��Q�A�5-c7�&��Ʒ&� ��g�#N�8f~��^Y���-= ����B8�;B%ڕ.|_��᪑�(����,����]?���5��U/§0�cP�`ܫi%0��2������[��ـb�Fr��WR4�S�3)㙘���(��:�w���x :���U�P��K����u :��O��=���IB��W��=��$1��?�c e�F�#"R|��h�cŅ8��;���h���Q��X ����:1�©u�u�GL� ��oBb����p�+��G#�����2�'��!U��u�2�(r�B�#CS�Zk��_�g>�A�L]lmv�!�ڠ��x����Ħ�,EDe����Gf-7� $��gU ��M%�����+���{b���os���B��9,^��j?�\�r������ƞ���в��Em�^�>��y���˭��r97�Vt�������W�WQT6�Z�6�y��͞#։PO���մN�j7�e� �+�h檠'g�Q�S�ں,��N1 ��)]�Vx���_c���JT(%C'w2�v�eΨ]'�3Տ�G�R�U0��J��d�$���D��\*���"9B|آ��&� ����\��#/4Z_L��N��g�,�7 M���-�Q �ۅ��}U��A�q�d��$���W�U�qo-bR;�|����ɰRj�? '�h��R�O�s̶�D������.�h%�F���U��,��϶_4���$����V��S�\�c*���D��:4�VLq�u�1q�sؿb��5 qM� �����MU#�]*$]V?)x�V���=��� p��+rp�Y���|oax"�q��LJN�%�5����\#8E����@L�����k��+�ٖ� �ޙݖI*b�����1C�e w�5Y��թ�k�qc���U/��]Y�A�B9nR�Κ�S������ÔP�$]�QO�6�ü��M�������=��2K��G�ĵME�֊giһ�{6�>A`-�����e����|+����%�9q]�}��u�t��)��F���4���ط�����\�5��F�۽Lk������'��8�Y�T�ȧ�A\���{ڛ(��ϷC�qy�oپ���v@�!f��)G�N��㞢AŴ)+;��G+^��C�Ri�J��n�ڦi��cޣ� �af�ڒW5�/�u�@6"5X��Vp��_�'��d/-7���nT�lg�~Q���`�v�N��3�%=׎�����bb+���؅[�>w�ÿ�v��ݩn.f�MB�b�*4��u�a�y�$6���lp�w4�J�GC.�;� �Z3���RZ�N�����z�&�M�?��=dP�ե����^���1 ��E��]Ր�7i�} x7�Ƿ����>�+��"s��9mU�ǎƏ4�[��^���*���D�Ug`�C˲�V�c%O�GT8�Df�}���)������>���2��um]�#���U�`.��v���o�����͘y#sR���>��Q�ܾ���/%mZ~_vI8ր�������s�� v��E u���L��ށz�l�eÅᱴ)��lQ��9�=���j� ����=7KL����V�?�k�f�"S� �[�f[��p�D�LE����#"��I&ծ�ѿ�(4���L�(HC2>�Q��"l��ܼ�a{�3�I ��d�AAt��S�[�����*U����@��tg�K�"MP���f�����V�M>L6���z�(k�����c�X&��95�*N}c��i� k-`��H������dvW�5�X��$"`)��1���F��rJMB���4�$V+�ACֶ���vt&�쇰�x7$���ѦN߫�h�X��O������}�zgz�׏��; K���� �NY�GۓBilo݁�ّW-F޵���#ҷ�#�{ '&�a��m�U�U������S������d�mM-&��rX�m�῟$4���>h�p��H��������_}�M�'e���}=�D�Z� ��x+ČA����]M��ҨP~ �߬*�L`q.���gZp�|��R�� �Z�=1}k�f�)���t���Kj�9��wZRCkP">� ���{����a���3O�S��)>y�^׉�Fܭ=�2�4�3����|q�b[������*�5�&V� ?����U��+Ҽ���P��t/g n�2�K�ZΪ�����~�� ES�l�_�4�(���\[�]i�h�|B�� NtCi ���K������~���kף���W�Bݱ��_��Q��`Qك�5���l���z�� gL�w2����7�z�9-}�.��e�9�& "^gɆ���1!���ݭ�T�A��?��`u��3V���{g���}vq�l� )v����X�qi���,���m۶m۶m۶m۶m۶���������UI%q��3b ��) ���� w\p_�����Ћ� (_�#�i����g]ANc��o� �|r%��^}|;�>^�M�>�J_R/�wP���)�Q"V1����S��C~���Y���&s���u��� �/{�74z �%2?2:,�����1"�B����{9�PU`6��ؚ��g�q��� �u$�M ��o�rt?��9@���+�� c�D+W�@Z)g��U~���•���b]~��2ԀUub�{_(��ݓ�F as���r^�T���S�Y%h�k���U������Z�}B�Q�C��G9*�+ n��*WC��V�ە�KG@E-���P����@r_!��'bἂ�(�lb��sr�F����Pŕ �q�|]V����(B�Lc�����iAE`�:���*�3��ᾴ���o��1a��zF�ܺ\�li�8�ɦ��06a�H`�$?-,� ��$�E��8�t�u���o �P����#�'����K�Qo'Q�*��h�oN[�-���JgJ5���"L��=�D^�4�P8��6�:dĵ����@A�k9��}�X�sX�At�c�S͖d�����jKp�67�%����`��lIt���swg�k�$E�"�x�*&�F��5[آ�SѦ�(����6��� ӿ��p�K���{^|�v!��������e��F�ӱ���#7�u^>��T�*%O�UZ�V�ac��.8�����#��;MCU�ka��W�"�ov��I{�T���� Ǯ�,�i7�g�.[�"w�mQ:s��'C(LF�Q�aTq�}���Ә�;�#B��\�ޮ�B t��$��B>��u ��6�k��YXFW�����o)k-�H�V89���]Z>�t���:��`�˞�JW��9it��0*=�Ǖ(�t2i����EO�n|υG*@9ےw�N�4� ��`3]D[�п�v�j��9qҪy�U��j` ��aF�3׿�큢��;�WR@��1a����s�l3���λ4�bA��n�S��d�y��l�4PF�z���kpR��h�ӡ�;W���P��ȼSݫ\,���b�2�w>@��b�5��d�ˀ��CR�P�tg=r���h����#�;���W��y�~6�e�`��3*�{G�����k�12��󴼪$�pba���i�1�@���݄dA��Ԟ'JF�?���\e�_�I�Z͹C2 i���jf����L� ~�����!F��Ql�l��!�2�%d�_tm3��t�%`�}Q���Bq{�A��oo��85 D�ݓJ�d���Q�D� ��& ��[�zk2�� r����=�(������,��a �gv�\���'>՛����>겊���? ��_y�ąT)�L��yޭ '-3�R n�ZX~!y���� �,�f�W��[��ohuW�$3�4��K�i&u ����}�L�����g�-_�C���į��1^�#.��Y�1S:�e������dP�KߺY��^��ak� {��|�5��Y�B:[OM�(~Ub���޲q8��}�3�vjҡ۪Q}��LӍ�d7�b�“�AZ�:\�� ���J�w�dI Č Jt���W������%^�p L�n����3qfB c޸@8�ӑn+>����ȡk>U���=h��s���1����#B��O��+Xj���p�}�G>�� h�ZGs�C(Ě�����XԵ�� K����Vjm�~U��m�rnw#����K��g���� j� �oqP��R_ (`~��Ź^��"�6h�⇈N��Q���eX-LJnRj��JW��:�`��Hz1�@�c"+3M;r_uȾ�z���>��q�$˨���q�r��=gT�lƅ�ɓ�� �>~�XAg��V8ZbG�o��%�ЎL>�;�n�݁F����5 kS� վ�t#� �՝�*�!���`g*2>SR;�xݞs����ς>"W�Q�;���m�ِ.�gF ΓRJu����d��/���S�H��N���x�0�XO��3�̊�aҙ2�P��F��L����F��?�w�Y/�'�BR`6B�0�ch�. wSB�R+��|�m��\{���d�I��^�geߥ�wA@ v��h5{���d0�9��%�?X��� ��|�%��+H �L����E��tE���5�ݟ��EK ONU)��kŷ!��� {fʿ'&]��yƕ�c��n��p�2������bi����O[;9r�� �;� � |%r���` ��6A�PD�u�w�|(pL�'�rg�>/����f�6 ��X"���4,�§�$t�겠G$�4 ~��8�{ S]���/�ijNhJ �� {���_I@=­��ȅ�H��Qz���!fv���8 ���j Y��7�C_�� ���ͷ�S�˒��j�ASK K�,md%�D��g� Db����B�m�b~�� ���0rq{�:�����?�8�!�3b�ڲ,T��b8��*�׷u��/F����wX5����| ;�ք3�7x��������х�(�7kR�����o)08J�- �PPU�i��-iíkj�6�kP�*���3���?��WO�3���7�������P�=�2���2u 54E��f�_#!}T�� ��uJ�5ׯ�i��;�k �M��a��{�>R\R����o������ n�[fV�S�v������|��������B�&�����I��䨤��>GU��� �u�H�o��É��͌� ��Z����nc kF]8V�� ͡�>���J�f��;Ce��B�]�E��x��2�0�vw�*��Ω�%ia ��j���:���)�P~zNB�_�Pl��`��mGE�dD���F� ����߷f�k����P[X ( :��)��s��@Na�ut|��^�&]Z XM"�)�²�4~8+~�'Vw����D(zg���>�8)�cGk�&�ѕPU�!�"\Y\�D�=�b\r�~[/���9�Ax)�~����N�T���d��%Bl�C�M����XQtL�Ή��q�� ��i�"���G����k )��dMD_��U1�яG,����Ua�CS��.uM�!6c��f.�u�׎.��*Y5{ez��&�0�������0u?a�ޞ�E��/�5�?����q�S���>������x�57�| Һ1xV^��^�!��,n&N_r/t��tW �C����~S��EWj��U�ժ �p�>��f�WZ'�;�q�su�Đ���,3����A�K�Qŝ�Lu��sA���+�>�n�:c���뀯����s�Y!�F�c��̣#��-B���f�n)�GhH��A��އ��U~B?,w����(p���5P�V ��җg���������eDf�Z~���:��%d���%ީ�xnGE�/�$��l%�EaL��Z.um�we� g�t@/�6%؀�%�(;�Ħ�'o���4�Bv*C��]���L"�v�����+h�e�����q�Q���|���=ޣfE+�Ϟ��H�!]�;�M�z[546�{o�z�U6^�xga���/��KH�i��W� ��o?��g�H\���N��~ 1�d���K����ž�yd��Mp��*����g� NYt���=�� ?��ݎt����S~�����ɭ������ ]��9�gЮ�)RO�U���n!Q�ݫ��K�U����=�?ټ��rh��\|UG���^���!pZI�p��}�Ķ/ sRˮ1��g$@zs��W-����H��F,�� B2�0�R�j �eV̼�@�dը��^?�*��F������+��j,�dҁW���+�M;&-0=bH�f���b�o��53��A�׃�Q2̪�9 0Q�|�i�pb� d4lx\� ��N��0q����������}��y�h�Nz� X���h�|޷ک��>��w�A#��1��#�ȵNs�$��a^�[J|h�!��F�+�F?�����_�m �na�!�GxR f�?WI�2I�4���^z���U�g4��L�7�� \':�^��"��J��G`��6\�Z�áヰ �z@�@I��|X}� �?��D�h�� 2=�$����O���=ͱDIr�Lis2)i�B^ �4Ka��l��ETf"�XR�"τ/"� �xW�uI�m�L{��������]jF*��|�,n4A�����K�K�R���'��t�)�5��.6�r�u�|�\F��k�2�՘(�F���~*�\�T��3�T��W��9���;����� F��0T�[��!�ŀw?*F��2�lj�:�{6{(@��_]�� ���n�����T��@#�72Hr��T�2�|��~)��(�Yr�����&|�V]��qB���=A�JZ����x/��c�:�V��.�fc�dl_x�ʟ]��Q��4T f$�~�TQ"V֬Q� Z�V��_nА�mX�#18�i�=}�lPyC��c��x��HGً�LʙXt����st>H�i_��Li�P]�V;s���g��ư�#�cP���O�*����*�g���=�_>�/ ���h1p�ݸ@���r�1�Y�Z�l�{#c!�5n;xf�����k�\��:�(���췕�}ӱ.����j�@w!�U�L�y�D ����-�^��c]�7�����3���{=H,>�.�m �M�C��t�a��$�n�,�X� J}��|�`㚧2�W�A���+o��h�˦��oa�$+z\��T#em�m-R�t�5Ir�L"���s:�\�*�%����2QJ��kAd�^p�- �>�$[�Ⱥ�0��zv[E�dP�g|�䷜���d��[��^I 1�BX.4���{5 !�7��=w�$�����PI�]ۚ�%��;H�%�Gz�CHFQi��#�(>+����Ah {�U���% ժ?R�>-;�o�#��: ]��K����q\P���ȫ�s��3i�24_� �LmYʌpW������ ��Ճ�� ��2 �/^��C�u^$.������'&e�S��iY-r���";�y�x�\�:�} \��3TN����Rt�� �g⡽�U�@�e׃�s��Pkڈ��� �����PD�(z;��w�x;﬷�ǂ�v���P���m�ɕ�^Z�ʖ��[��碖�*� D`|��`�ݒ���! {� �ln!-���h���s�H�����J2V�'ja�M]�:������s���yhLc��3:�}�r�/X��}V4�����4��$O�!aA>����\u�w�3ă������*ܧ��ѓ_��Y���m2l�[.�|�?�)$�L������e����H��H*/+�u� X�˼����j.���Gq{ީ�]RS\E}+�+��)i�UW%AŁ�_��sepR$��$��T=�R�`V�x����)�WO�'Һϙ���,ĥYu��niIbI cH���yH@���l��i�q�����d(����t-;��8��ʱ����i�Kc���I����L�;��BbT�p�0�}=��A��t�F��h��U_�a��u U"ǀ) "Wq���w�������ߒ�2��������͊���4�����bh/���C�R~��?�����W 2f��Q +8/%R�ش�TdX�L'����Ʌ"$4��F����4�&��/��IY��z8�M#ᖋh��j����� E!�s&�g� �~d���ryu�=� qU�������>�,�^�l����%���K�$]��06�l��VV'�ԹΧ:���%��� ��i�a��4�� W������z���ź� ��c�J�'ݺ�Ny���Uw�#��0B�8L��\sS��k� �S��Y�~A/�;�Ng�j�!�;�鬰qF�Q��V��r �=\z)��>- ��x��u2:dz&o�k�����LPĀFTL����%&R˱����c� Y�R1/A�Q���(�3ZT���&��IT���e�l��>�Ը[}?j����{ �k�U��[��YSm�'v�*�����[%�h�O#�d�7�;��s5V{9������5����g�5���T(>ʗ�zs�����A*Fq�ͭ��ľ�@rO���Ƶ׆;S��i71��U1{�a�e�$� J���8���!�S@�>]��u |��� �w ���,�D ���2Ȗ���7H�Ee˭~�c`� �} T�O�,�01a�d�1l�Y��\�1F�Db6�{���{ڟ�N���,�D� $f=��*� ��e���;� ���gB�VyϏV�93w&,��f;"�$�-�f���@�]��ю��V�p厖�3r[�㋣�ew5��� |W����[V��C�x�Dw>�錄 `��231K~�ݱhb ��߭vE�gpS.�����t�����)�|0�w$^>lk)��z�*In�G7��@#":K�4|R��������@̵3��c��cb�A����A^�-Ń�jŊ���g+��cv\�d2c�Y"G[(���R���WCy��� a�w_�֧(g�pـ\�Ө�l-�d��yj(X�&��&iju�no]���� ?f��>o��ڨ/�G�V �i�::�e�7�(�z��b�Ȍ�����؃ v��cw���1#�=�W�Ծ�[Cn4��V���g���Bh���q;��t{����[�坯�eD��c.����/n�>���b���t�i��s�W�3���r�����Qy���JT�`�m��w��m����-�"���H9���oؙ���RuB\�;��pw�0F�s��@7\�ߚ�Ġ�tw��հ�@�#_O! �HX�H���-^wY�`���XIv�0�W�١x�^ތ�n�(A�\Y�g�� `C ���� ����g���x�Q�����iY$�3ml�0�E�r��_ٝTC���Im�2�zV G����C��1-��\��BUF�&����'Ry�a���J�;�`�4WF�h%�C��Ё�� ��T&���^8��$��eER"�9�)�{%��K�wm-Z��'�ﱻD+��k�1��J��-斌ў�5�{:P>�l�i U�{.3�Tv^�� 25��GFfm��(S�|�3~�B&�kN'���/K�����`��a]�g:j"-��|>�b��{U��p\�>�Δr W���� @F��ֲY?��G@v��߭Yl~��"v]A�E@���'D�Sn�ˊ��f\T=�[���H����e��� �fH�ܞ�>i��H��B�8�����L�Iu%�F� �e��Lal��^��Q*4 j ��C)�������>2>e��.V�C�P������t�����GkE�� ��I�(�ע�/��(��;k-���.�5+E]a5���zI����� $ZѮ���@PC�Ɓ��6���ui+l(�P5�mj�~ ����:��>3" ����I�)��� ��rU��x��z9�t �4��o� �e�"�R�j��h���'Ѕ'&��5*΋�"�߈���X^҃Q˹�/4����\ķ�_���ܙj��}&sͱ&�|%�{ �ڠn��WC-|��0M|H��6����i/�����}�׆Xo�h�i��O�>PD�ki�y�m �} ���grM�e����_*Ƥ�B��򔞢]��{��0,T��s�6g��_$��R26*�`G�@>�/���Wx��N�� ��fW7 _b�&��%�5/���.�F�G�ŷ�'��l.����~��O>ʟ��,�J|3�@����,1���e�s��v�CX}O�����e �pǃ�wj�z���I����u ]����DZ/I� 'e�^�cC�h1�G�80��an#| CY�Ϻ�I��L�(�b�;e�0�^؇�:��9�H� �:瘼ʖi��w��٠� ����B��e�����E������ ��~�xC�5Z8�L��[�,� ~ ����Q��!��$b��e�����B€:�Z�NʅiO���"�zn�U���B��Aq>��!\h]�~���D7jq^ʛ� j����S 0�O_8)�*���)��#Y���!G^�v�(r"�=QS͠�ʍ����13��Go^{�}6n����i��H�����C�����4�8ITe۰*�'���ۜ���#��Ubi�q6��y�Z����®6G��m�Hճ5/z����� ��%QKn�T-&�[7����ӊ�*&Y�MEO�3����D�1�X���s�R���d�Z�bSq�jT��K7��P���%�S��8ui�u��ϝ��2�@)Ks �0���様b � &ç?^����$�*���.H\�ز���i�[k���_42��M"���:�)��g.�*� J�����6�{WF�9@�2̥ès�-�jV�8�G@q�le������,��ᓖ�&:�]����!7����={�`a�q�2��{�{/q&d��wX�/��z[ėddSV�A�Y�������sG-��$��gE n��lz0�o�Ǔ�r����JSŹ�u��V��@���p��7! �%8�#0Z'h�k%{�O� �̲��o 0�X,8{�]�y�tص�������e$VĢy��L�t�W[ ��n}֦���-�NC�Ƨ�W���������e�kNM4��AAC��.+��)�h_B����NS֗��@>��z^*O�R�8:.���m+���QH�G�2�oh]?-����'�^ƹ��:������U�K1�� Ie@I_������HH7*� �jJ�������Z�d��/��v?���^'u2�U�;q7���O�q[�g�� �kL��u\B�6=�� ���57Jc���{( �\$���_�ldGQ�� �6_;$$��ɸ>�8q J Ao�擬D챯�`��8p�0� _1�[�X ����s~��Z��J�y�,�k�{��ŧ$�O�1�y�Ђ\��&�A*?� �a�����}ظʉ��/ �m�I�kz>���~�����HS5q��cZ�eN�P�v��}�0����i�m\TD��i�m ���s��U�~�a�����%������EE�i ��n��jk�l��9c�5�H���]��ǕK&�!C��G����Z�s��BZt��g���P m9/+�Hݿ嶵{L����͒T"$�#1�ΔfV��Y�g:\e��~�S��ץ�ư0F��ГE���v]Q���q�I�I`d)Wz-���X���(��-�.��m�W�f����@u�K>�������B%���*����~�~&�!���[1���%:ظJ�J �ݦ��nN� ��\.�DN�&�D�wM���e�`��L�� i��N�\��_$ӡY�Թ���jZ|F�m�?y� S����t8��P���V��x߮Q��m��p (u����q^��0]��.��a�q��& �� ���E� ���1�l�wO,x�OMݹH,x������`1O�f�K�xQ p�/�. �X�_G�)���j߶t�B�Rk�Z~m%�Q4a� ش�t������� �^��'�1�=l�R ��T�"�ۂ~gȓ ��ϭF��4�/�e�^����M5���9�I���f�g�x���;�,���1pe��$��"�j�� �Ф)�� ������SV/�1�D�d��b��pO@�#����g=Fo�����Y�S������;�O� �S'����[��սn��r�w�a� A��l��q>��9��m6��ù"�t�_\�a���p�)��t8�9� H_,~���/Y�F`-�s�W!-�{,�?cW�-h�Arـ�S���Z�$�����P���-����������_��R����F�� �&�y�ª@�X��h�D�{� ��͊���*=����>CO�7�u56����&~�������A���Bf�cI���d���f%|�8�7��b"�Y:��af*&p�:��H�4zv��f���H�NQ������&����k�@mu�PC=�ޥ�hL�jF��ʋ ��w���h�m�9g,��N�׹�R�s��F�jd�][�i�XJ�X���&c��d�[_��>.z)��3�zw�l�-;,�k�/Dˮ�D��9���'vm���3s�g �&��z� c�2x��5�fA���RI��9M RT�|�)�t����&�q�Gö2i.1H7]&O;��n �D�[�� �->�����/�1=Y'DU�Ur���9T��kob�'{+�o�d0Rg���۫�:�>c�-�.4�~.Z�����m�tE ��T�#�R)�u��±����>LR���A�)rE�X&1&گ�� �FȮ�Ԃz���҃H�o�I�����8�b��ar����ke�h�U��#O�eR� {�9��0,-J����n���.�s�0voذ���7���j[Ϯ8�M�*0�1r�ق�mM'��D��z���FJ$����*�;���m��Ǵ��@��,�� ��̲���x������g��Q�/�90�y��VIYT�Om0ݴ�,���tG������/7�Ԉ��/Gy=^s����!��{Q=� ��}w֛������UO)75���ͭ�k=��^ \/@3� ,%�&(a[�'+֥�{�lv��Þ���!˙`�%�Q݆�Y��+pD 0�n����:��_10)�b �f�x�� �����%�>޿tKE�JCƾ����˩,��D9�A�(" )�a�,7'pٟ4+��#�ގ�@�KJ�ԾF[.��A�����`2KCy����� ���I��v' c�Qd��:ME���|�QWI�pM��/�H��tgK���^*�B;I���V��-� 3�����\W�����7�zg�p�9K5��t���s`���M?��~����R0*�2^� ���f���N�`�U�z[.�!�f̖�����q`��# ���k/������D9J��� ���{������|����K��#��? ���1��Հ��}7!�����z�=w��&huQD�mf�'����'��O�q`�%�2��F�b�NwtH�hMzd�0�5�?z�K�[��t��W�r�KH8o8��N��~�1��+36�f�� 㠹��w��)��ۧ�z:��b�])�V76�X�zTU$�Y��m �2Cẜ����� ` L��jX��W��PC���1�/u�f�%��aٰ�4�3�R�]+h�辩!b������c�J����L���<>+��ZWסIy�k>q��qMהA�FU�Z� ȅ �Zt��!��0���'���Cz��$� �7�Co��:��f�N�����.R��n��oCi2�� �д#&*����"6J�Ge2�:��}(0� �F�l(gp�k����N�p�r���i�> �ۿ�ɜNN�� 9������4W�@��%�����{Co�T�Mǿ$\�w�*ҡ�Q�_�7DN�� 5b������G.a���8����f���?rY2I��F��#B�q����%�Џvֹ^]��"��,Ϸ����v7ikA�0��K� &O�_�ïV����� �8�i���6k"�*ߨd��}+ i�\�4D�t@(8�7E�����y ��$J��e��I���H7�y�Se�cn�f�z��} ����];9��3��ct�E,*X?V[�e��"��"� {���-h�7��ǩ_�7��1���8�EA�� P#�1��IF�Sva�=v�S�Cm����R|5�D�y�v_L2�]���L���v���� z��}r%��9��Px��q��կ0�3��>�+���� 8������y��D�l�� C���,�`'�����c`��J���f$v-H-E�j�w�M!}�ʃ}��v#���5��ƪ�(� zE����%�:��ԟE�@A�@-Y��t qO �*w��ʹ&����qI����GPt�٬��� �>�B�n��x6wa�/6j����þ��z�1�IQ���Nt�($�V�lVR;��h���� ׉)ǣ�R�j��p>)~ݰЂ�� nz� �D���EzV�4R$#vR�w73�lh�zƩ���[zV�������X�H{�+�>f���>J�g?`aX���{�&A�>�dMs'|��P�-l~�����;@�kՙ�A��3zW��`d��D����h��Pt�mi�����mS�jR�(p���u��o4�T뺱Fӟڅ�htv q�B� �4���+y̒�`�x�FAi���|�\z LNc�3M�*��t��ܓ���V �ބ��f�H��b_�0� �_��kI����:ⅈ�� �2��,�TZ]pc��#$���f#z��f��0�Yuj��B���`>���0��p�).r`���I�g�'�0 ُ���@ET��a��m�# ּ_�~��)��G: ���Hs��,��ɒ68�0-�W����PuQD~�Ԁ[U�4��C��P�V�|i� ���ϛ��˾BG�"�G}S ���\EV!2 N��w��F�wP4s�ʾ r�nP �����i�fz��~�A7����7E�_ ��=�}�����I�}�*h�)H���c! > stream x�t�T][�%�Kp��A����w��Á��{pwww �A�;w ���������c���Z�jΪY��}#SVc5��@.��L,|u���'�lb ��3�fTZ�ڙ�lL,,_x��f.S�����QJ&�@������ Ҁe� s\��bQ�{NR�aÚ3��a��1�yr�\,H���}� ��,��������௴t��I����ν]�u ΰ��1c�94$�B���d(k!�f�%7��㵱�uk���!�\ ՙ1&�r")�}c�p5 $��f�}��Qh� ��P=4'��'�Mvl��ƕ�L3e��v�`����gO�f��z �X��!��tl)�\t���(��\��� ���]�a�0�ͬ�گd'p����@�^31��uKu�����`,[���'L �+�"�fПUϜ�r��d-��/�oU��& ׅfp9���D�#~Z������Q+��I�� ;U�K��m��������2��cG����%N��ݔqBi90@r�{�=}K�1`JI�:�j��o~XX��hꉗ�?p�;���V?��t����ЕD�)!��(TD�'��>?�ąӧڏF�văQ�Sη@��Ot7���c){ßUs��:@�(� euC,�Wi�I����V�r�+үo!:I��x,����Ϗ��|/����砹�m���BHn����\hT�y[ֿ�87�_�)�*4��j{�p��y&��M�j� �}3�IA���M+�|Q�b��9nH|���¸�1��B�2���nǣ�VN5��E��[f��� �J�񛏵��J�q.( ~9��'��í�V�`cH8m���Zg����궏4�ld����� `�b��� �@,+>F�K�Ne�����P'��u5E��-�1(bU4�0��N�]��L��q*���^���n�� �/�� �����K���� ��F8�l���㿡�~#�D ��b ������f|���,@v\L_�( ��3n�� O��m�&�����c�ڊ��9|i! P&��ɝ�-~��!�X�ٝ]�Q���R�Zs��a�y��7R�H��JG�����*"]d����=�H{ҌG������c�!���4u��;����m��n��{th�0�#%�[-�ii7ᚷj�^������%V{"z#�ig�?�!�%�U\�䁳@�]�L`X>W����K���{���@e-�������Q�y^�V&�}�r/Z3��mul��i_'�#��]�nj�kL L.��(*��!��a���F}ag�C;O�+�w�&g��E�DFV���xӾ�=&B��s&9�xQˢ�{p;�7�x��AO�zB&52�X�P��e�'�˧���hO����H�H(��E���\.��Ee ���]��;;�ZY�^� HE�_Q�Em����V����wl����2Q4΍��(�qяY���� ���+�pop�h Rto���������� ޏ$���EC}���=)@>����*�� ����'�h�L�VCMx?7~�� �KL$��c�G%��� �ؒ �$^�?w��)0ʖ�y/� ������( ��|n���\�5g�U���C3eM����,e�ii�����T��M���.2�z�E�Sul�_5�p������Ž��K�I���q�������g-�˽�S������Q!J)�2�݆�&}�m_x�tg�pT%��]���~ �;s�ù��˖.�3���R��^ o b��Ԛ �r�E-�_�=X�L�s���W��v�)}6�֒�!�(�K@�>�+� ��u�3��� ����D��O�e�@w�c�_�6�u�(�-�3��-�m��p����c2����h�"�7y6х0�53�yg�m+7{ �[�rZ9�����=u �gHn胟����Kڟ"�^�Y�5œ[L�����%]�������p����1UU�X�{P�O^�����������}b�C� |���Ҁ�� _�����N���{��Y�'�=���V�E|�j�EҪE��R�n�󯂂Сe���-gBw����S�����2C�]?�X��ÿ��Y���� ���ep���h�R�8��A0kS���ax�k���rJ��w�{�b_5���� u'��F�^p���A4��+tll�tC�dbe7f���w[_��*QM�,�rh��%|X�d�+_��,���GL �H��Ô5� �BN�����W �v�(XL��5��\�?�� �zݮ�_S �.)�]�ni�aK��/�SH�_�Io��:�d�ʉ��^Q�����ȭ�%F��?��N��y�U��&�o��ƒ�m�'��䞤��{&��x�:�����JBe|r����k>k�|�;4@�~�Z��5m9��\�% ��^N'�A��!!k����?�ϲ�EM�v�'�o���t��v�#��{j۟Slc]s�=*��N�ح��ԧc]� �0evR�D>N4����K�%]9��".��WN���E�������- ��+kHU�}h���F�)�q��XÝ�=M��&z�M�C�`���f`t���+(�FF�8���Ub����ou�~x۲���z�>p[Jsb/�7�:���J� -��hg��Œo��`�2��K������_���;�J��&T�\q_1���%�;�ѵ" ���2/��ʋX��%�Na����'�{9����!qB�vL$X������s���{����)h�����7>�ᐮjX!^�/T�zM�g�%��8���r�O�Mtu�8�G��a 4,H����,U�K_�`����!Td2�� mg6��Ƃk?m�z��*p�QA%�ȝ ��2T� �:rlN!}�L+UZ�� u�4��JVxhf B��\h+��c���WY�O��`�Zɵ&+�j$��mfbӜ�� T��h�ǭ���4/� �͇w�~y��X���Ѯ��mt{�Re�Qi!FR�� n�ݻN=�xn������Ps�z��eo�=\�_׎����asF�V`�>���w� ��*#��c#����%�S�W�M��^r�3ǩ�?� �Ťxd(G(��/���" Z3����� |�\@�H�VF�L�I2���9�b���tX���%�;Z4*�����L�ӡ,t��o�6��}R",J��k32�^�f� g�x��=cj(�X�d�i�Kv;?W�d����BOnb�Y �l�4 7�p��?��VOhp�5��j.m'�U�F��e�1�+�v�m�S�u���{�׮z����r ۹ɡ͔9�Õ�䳚��>���x�_�b�!^1>�_�ؚ8��1 �ԉ.�O�O� e�KjL3 �� �HΡje3+�M�9�0M^�|u*��#h"șgQ��~��!q�eX���3�sT����=��-��޵>>�����E���op�H� ȏn�h.؈����A�Z���:��{��N����8���8Z���[��.���Q�0Y�������9X �Ñ�J2�` �������y�G{3G�zm�s����O�ƕ�U�������+�j�@>��aw�D��O�(�$p�.Ԯ��i�Q�c�����z�=�L(�SR>��鼋"��W�Ƨo�~=~EW���G}4��9� )��A`���'�s=(�'����$���Q0�&�?D'��tk��K#� �3���b�g�_�Y`�芰�Nv�E���^%VC�X��ص���e�$ ��N�րx��Hhь|��Z�ر:�v������K�ľא�1��`d2��|�g]��AZ� U+��Yq\��j��$)���)�3t3@;�8�E/:��wb� ��a��d[��g=��%~t���CjǛ�����j2��~P�r+1�U��- j�mҌwxW�;���Nˑ*E�[+*J�s�Y%��zɐĐ�6�\v�Y�� �H���Ҋɉ83h�&c�h͉}�V��a�ܐZ]�J�r�+�ই��H$]��%9���-���=�6^�d�Ie:O��)X击��=z-Tpb2�J��ӜQ��d:�+�)�$�,���R]eN(Vr����h�Ic{�l2��o��@�@T������vB4g�UO`Z 9FiVm.󅎿����b�oJ�z�Z�� ,oڅL�L�G��)mC�$�P�^ �m��m#o�U�Ǐ����8�ac�Q���^䋾�R�`�G�x�.��;��7?f4�����ܗ��}.I�*YTb�\nqt��b��Ծn�9��l�{,�ʎ@V��D+��4�?�3��C� �!�ƃ��� ;a+����F��j�����Gd^d8;A'P5�E��mض;C͞z��J_�.p�˫��̙�� X����S0��!���\.j?\�'��g1�P\�Y/�����B��9���c�]Xho���[�b��,��痦�X����d���ޫ�WE5���'�mC�����˶�>���c�x')���&�ϻgQ�B��%Jz����9Ƹ�+���LT2�?��)����{r��Ѭo��P�x�/��y�߭�r9� 58��2Y���Q�'�:N��u9.l��R>�J��I�'����� �V{���d�Q�͍��>w�(��Z���z�r�+7�C�y41Ԗ�O6SJ1��6���n~����⣀��N��O�(�}�߫M�t�$m��P�?_���J�a�6�����q��z$�'�U��5���\�3�{!ՙ�e3�Ȫ���Ė@͇e�W���j�M����(s���Ʋ�e\�D�( z��� �:4fZtj���Y��B���F�� ٴܨ1���]6��H�e'jΉ��Y��V�IE��}���P��in!��S������$��ƔE�=�!Q���t �p���-�� ,y O�x��&A��II�,����Qeʧ�u�a4Lhv�d�D�;5��&ϔ�͢��_��2,xH�GON.B��u��p���]re�j� J��X��mҀ��%�O૴"����*-�0�x�Ԇ�� � ~��N䄴�'���'q�N�]a�q�:J�'�A6D�U �\��j%�a*}�&K�w�vꔤ��K��]0,�����T�K��"D'� ��u�rҨX6�-!�i��5#��F���69D�������̓ �r���U���w����]�]�S�wϨ-������5�pRq�9{�CȮ��Ӱ���d�tM��s.&��X���qK 2�gIĤ�f�_� j��:�I�7X@��)�PWZ��R�Ua��/_ٔ�f����\~6� ���4���yX�4�s���O\ޚ� �2BM��ɵ�b��R���ʤ0��l5�2ٶ�#� tmXp�O�f�A�ݲ5.�G �L� Bu��4�qn;ARhQ�_�����Cz�b �M�uU��2�����m�mRPa��V�U9�Za�7݃ }nu��:"�Rᚈ9���BR����x�˄h)����n�5٪��n�u���=|u�����hjP���Jj(hysͲl��%��W�F��0�9�M#��R�n��xa^�5|��_������dEٓ�ATϟ˄�5� ����:ht���LD�g �hx.(�XW��!>�퇬���?-֖-$��ǻ���J�J��ӂ��">���v;˴��喖yr]`BF�{"V�4�D�h�c#�CZ.�J�^K/�Z�� �||�������H�S7�_��� ]��E���s�E��Q��U�t��2i} IbS[؏�����ئ�!6��;іnz=snF�9���L��F��*�L��'�}�ʾH�8��]E�x���R �C��4���B�M�7��OLI�L��@�D;��5U/��hz�!JO/~Y� �:��+��� ^�����~�阱��p�JW�P�}����Z/^F��2���y�ß���d�1�w[��M~����Q[��CA������d]H���{�^ )��1�঴l���~�xκ�FK�~( ��E���U���?PU�j���.�e�E��I+��� N��G �V�������y�l�� Wx3(�O�LA��o���#�F�(`Ο�c ZA;�" �����}dЉ�÷�ة�㫂�L]�����@x ��UXc#ʐ�w"P�߿����m'�+��G�L��n�j ���?���3����A��8����Fp�`���[h� +��֪vb��w~Jy�zWT���I��P����&7a! �j� y�ڈ4�����)[\��I|�6�I���>ou�)}�'Vq�H��p����E�u��Q���U��wv�r��"�)�"��%���_��df�Zr�������3��n/�!nrhw�?����=)����a�+ImŒÄF!k�:$���[Mهu&��dq*�Ճ7|����ƻ�ʜ�_5���~љ��*�h���o�ĩ�x�J�X���6O����6 ���r5��Ã���;B'X|��̒f��W��Ym�9��8�C��y��:}�A9����:�_rB����ͽ�/X}�F�a�.�C���*؋�C;,�b���� ��X$�4Sy/�|v̮}Ex;� �"([[7��nN�?ⴍ%�yF⮱ ,Ţ�}�Fd�Xn�����Nt A��8�IqWL}ʎ�[6���0��ﺎ(�B�~��A!��$h���ۦ&�j�I#�"Ǫȹ�SO� �;���oO�P2��seK�7_�Ձ�c�/�+�@T������XJO&����J�\��F�� �jMB\�ڕ@�i'�?�1��\�Ф:��Ʒ�Z��& ��� �>|G&�A�^��o�����$��2����1��2����Ol���o�힗RÝ��N����?�l0m�ܺt|Ij+5�A�}F�oʑI⇭g�=���Į�I�^���"22N�+�}�gW0d��/�P> d �S����oNk�������0����Qz�q�~�4���������:���m���f����=�SB��G3niv�|]wwm�Eu�:��:�]�����c�jtb(b�^ KP�#[����kK�,�*( ^�������G؍uEn�i�D[��3g" x�yk�fpW����w� f4>�'÷{�6W(\�L��|�y��/%�Y���,�ֆ�O�����)�*�ϣ��>��mvV�X�}� ��!ߺs� �"��ט_? �ծ1��Z�Â\��F1 ��ϣvx�O����X��=⾙kj�*x����K�$ �Ж�ՈG sBrֽI`-[SXw���}4Jޢ�f3)���ev� ��M�H?�_�7�8%/�]��p>C ���إ�R�hf�سSp�͇B휅���GY�@w�AK �q�#������hn8�+d���ms�R���#�/�b#���Z�#��j"$9^Cfwז��Ӎ�1Ȏ��۴B�O:'� g8�E��}�J����2���;�$�񋥵&p�0�;=�j~�*���g�G`�c;9U�����C��V=��;̍���^�~��d�dG�yI>�~�í�kF-n>a��o%���%���Y��9R�2UR����vG"��*���*�!�A��C��;���+!�L�����Qr��7fŚ�#�����t��g�9�4��T� ���q�Sꋆo�%����[�Ҥ�f�X�w��� m@Oh)�fZS)m�$D�!���C[xmx�yP�TEPDR1�ґ�+'r�=v�t������)�D���:��N.�����D�\�΀����Nō�o����J �S����)�M��Wщ,�����"{��L6�jm?����`E����֣Y�*?��h��:NJ�`����5#o������q>m��`:Ճ6l AA�2�A��S��"�G��y[o��a2�����-NC�/lt���{D��{�\ &�ژ�w&�a�#�C�NV����F���d�"�Pv����S{��= ?������ y�J��Qǃ����%�H-���4��+��=^$���4L��j���'��b�rӡ�ƒ]�����a&/k+V$�r�ecl�ꉊ����ϴ!j�K�z��P�������+������w�:���������s� ��O�d� $/?4����6�bf��`�����ӼwM�v�Ա�> }g�o%P����0�xѫ�$e"4[^G3�Vo�dy�J�jH�/��1�m����`P��W-��T~�)�?nf JY��(C��V�IV#� s#�`�d�?�]�r{����|�����MN��]�����ԗOb����� �/�D�/�w(eZ>,g�j�F)C�N@��*��Эņ�"IF#j��3�|��יִr��T�i$=��0���5 ��u�L�XO5�V�y�#1�Y��%n���r�`%1ݗ��l�g��aޚ��}�#;�f���@�U�כ4��[w�� �:�U1 Z�C���ǀ�5'����~�TI��0�H�N,�-�욊u'';����B/�L�mX��تQ�D�9u�+���W�����O�Td�L�pxsU����vk�9qCA�� �W쮭��S(�AC�gh�*�o�+���¥l=�� �j�W��0�,�`�BeH�}"��)�T��v�c�����Y5毐��r�8F)x�4�u0c�&����m���� ��D�ݽ�V���墷�� ��o���� "��U��yQ���,���q��CX+��=6`kʌd�>��7���\�۽iй��Z%[�%3� �[���lj��-?��Mz��^�}��Z }**80�n��w0ȉ N����f�Ez��ҿ�Q�% ��?\j� ���^g�?'VRٯ��xz�p��L����0�Bu Aܢ^ 6���K����x�}��4Zj���A$����r�'��0o.�ޘ{��(���$�ͰI��e���q@��������ȉ�nQ���Z���r|X,û}��B�F_��/)�gMwp"� ыh6.�L����� n�dШ88y�ڄ?���%���K�z���x��H޺OQ���7�f�vD�T����J��-��>#�� �}�h_�����̗�Jqf�f�ċ��0��S�v�虯ͳz�k�w���0؛r�G�kD��~u hK�36�Fh������tmn���7U�z�x��� u/5��(�#��1����r��v���D�)*���{���޺�X��"t���3ucl����=�:dC�ɏ#���7TQ��B��i�����|����;_��� n��M_l�r�k�ʄ�:�P�b�4T�s���or��~>Q>�W J��ޗ��2�,��/����鏚��X��C,�q^w��cC�\]��Y!Y3I�!^E��He�b����Z/��s�x�v$S=�H'`�]rt��"ZKAFW�]��H ��Q�Z���d�� �m�o&*�`�k%,J���d�=f�cG{�~q�w1�ʥMN� 4��`�#�� ��D,����'���|\)���-���?*�]����R:Vr����l� ����5��W+'P�趞�^�m�m��.�w����L���l�!���i���KSn/�i����o%AL���^|X�P�?Q�x�T4-M��=���e�;�Q0O���2��z���Ϯ��br�$I(� ���� &O�9���RI~j%��*�C� ��p�{;Z����̲��ʨ l'���>�QkyI�B* �mxlN�n��nڬ=J�8���IH�M �IÕ��mg/��{���9z�ꏢsu�Bll9 1����+)��oɆ�S���lS��s�t��l±��p0$O�K4����OO���X��D8�U�L����_n˰x�2�۞� ��?n$��v�@D%4�?��8�-�W�� Q��,B�yH����,���h��#���?Z֐�P��.0�Ƕ� x��ڂ1���]�l4���z9�i��*��?3�;h�>>:e�Q��o��z?�i!D ����b��"��ߥ�K������6�L�ŏO�'h ��c�U(a-�q%��P��6�壖e�5r1�s� Nbu�2W1��˅�¨J��Q��a�a͔j�2s�O�� #�]3�v�&/���#٘ޔ?X�������� >g��~���#y�q��'�w︫�i�0}�s��#�P'��*/�$i�H�ΫrU�j�=�%�>V �J����@�zd)�~���AEfI%���N�æ��W%z_,�xn�@�!�� :��p��A��K�����d�U"��{N�/�J%5�.ƢnY?�MH �����ְ��b���xW�c6V!���+�: �Ų����Y�հS�j�.��Bto���M���2*�ɀNi�"���e�DO��[c8� �0�0�o�`�zB�>eި�|����\4�t�F��(4- ��y��c�1� [L�|pk � �~�,(d����Ƕ�k~!j�M�7�����a�`�)��3CU ���Zq�n���Z�M� q��8�fb�=k�p��FY�3��zkMŔ��ilZ�������k�j�Lt,&�k�2 �O�% ìf�|M~�j9�����$� ��5X�h�J�F�E)a;!��=mX�� ��v�ᇷ�~�w�a�n=�42��RػIZW��8�,6�ckG;_��}v��R���վ3��#w�9F�M�P�i4�ES��M:l�q���K��E�v���5�Ē�� 6h��Jn�ώ)iL>�Ȝ�c�P����|����Q����N��@8�(Ow_�9����} ��V����h6q3e]|Z��I�O�aG��GUEѨ���5ʟ������CQ��TT�θR�c�-b)�f�f�$�����I�,��,����!�oP�l$��Y��-ݧ��1�;X9Â7EiaY�ڨ�9�n�$���[����P�_�:���˪����� �qy(o�[��P���~}�pB]�)j0��>%a�Je=�y��&M�ːWm;0���&%CV�)��r�L����OMF�(��ߛ�{�.o���0W����B3�n�z�b��o����� t��'�@����ɼ�}�B� )`?F�I�x�,��kѯ(�)����@B�%���M�K�s|�8=S�n�Q�0>9M$�� %�"�����ŕ��8���Q[>ӌ�Kh���r��PÊ?޽�xm����6hэ�mv���M�[�g*Ԃ�M��f)xZ�©�!/b��i4]����ƚ��g'\�s�s�dΖ�F>�" 6I���o)R�������S���Zu#1����?�z����Қp�9D�?�vy�,h��%���D���^7�P����=���zZAJ����w��A�{����]�W�t?N@iYD�E7�P�|� �#�&��$[3U`{�e T�hD!U] �K-y��;L��; ��ʜ5���?����� �{��k��ז��Z����%�BO������_/�%a� `Up[���q�����$nyk�۔�ϱY������zИ {S�)1�́�0�����E- Kd���qQG�uW�G&T� ��0Kd��(�4��Fp�8��w����r�&������7���ӷl�5���z.U>ffj��+�j���X(�"����PL8b�7BEY���8�{���a ?�Xf�k��7�Ų�:O�E�ڭ07O��-�}��G�{�rLg,!�f1�Fq�ABF$�E�o?�H�H=�|Cj��:�Ľ����񅣗��(#�L�RҦ��:��eg��^��z�f``��L��Z���L���X��}Ls³�A��ߦX�|{;���=x����4���9P����9i�22:� �)��ez��*����*�$�D:�S2����`.i&E�=�?C20 �'R�����e�l�ͷ�\G��XեYy���D�"���g&w�e�����T�=���e��!'�'�D�L|چ�������T?�&����Gю �"�`4͔E>K�HFܫ���J��>�X�t�洉I��������*u���&�w��F���0�#W�:�ʢpVɯr�f����W���͘P#��m����{�Q�(-�Is���E0�\Mf���_ �����N��V�x|��A�%0�&6�ҋga�;Dϥ+KME��L�I�T��7ߐ�M^�c�� ���jP�V�>�y�G�љ)Zz� "��1 X`�)��l/�^A{ņ��T��!�;�ѐ�~ΩZ0nY�eĜ�7ُ��z2�ܖٞ�I@4�Kz|vS/��0ɽD����IW��j����\B0A�T'��Gu� Q41����L������q'�2$JE���S���"� �0�?��]�k�p��-XW�a�ہN�$�6v�P=֖ �崼2����6RS�9�mN��*�;���9��P��ֈ�-9�g��y�TE({U���V�����0XU/M^�kl�Ɍ��\f��e&1`�8$��d�=شx�s�-M7\�hf�/(�����C���Q�G����(GX#�� �t`����>�Zfᯣ�ه^�Ï~$ja?8Y��*�v!�x*�%�L��#�ks��Ѯ���t�I�+�A�h�2mض�ID 'd���~��A�]��p%[�q1� }����ĸ�x5�e߼�o�zbO�j!R�߹�&�O�� j�kМg��}����Gh�f#�IFq%g2��«� ���+F�vr���vmN�M��/�좸r�4�Q�|m�a)��&u� p�v�x. �B�;Be���a�8�o��BK䝬�e��K��)�!@޲�*t�B�#> ����Ky^�+�}n�7?���k�Q7_c�l2и��>oȬ�#of�9�� ���"H����7 ��ė,����0b,�˩�X� Ŧ�`�p�/ ���KP#5�ETfHE���X�V�ܯ�)t)��i;� 5l[����7����2�~@[]W�V������w veY;�bZe;]I�����,�^�_��`9 ӓ�� ݥM�ֿ������ p��2m~y�7ŧ��͡YB~?��AV���b���좽\�ץ���?�ca6a��܍]�A��A��7��B��R�St�G_�5���0q��0ʕp�4S�&Wej�;�� �w��_?N��B�O�C|� Z=�aՒ$W�R��ch��;?�2�r����P �w�d��6V[^ ؍w�>Lwڇ�4�R��Y��>������RV����R����NAH�oĭ�W�w�vzù�Jv�&g�8G���;�vaڟk���ʯ�*�/3���Ȇc�� ��*��$�h*#� �.� �C�b�6��t�����.B��wuB>h��/f�� �2s�T���:�]���O���{( e*�F��j�7/�ܔ�����Լ�? �V'fߴ���57!�;�l�u��7�I��I�E�g��L?݌uLbC�5�b�V E�8���MOB��^,�ܳ��K��ˮv�v�Zޕ�.�[mXЭ�%4��y^���4.�=��j��˒%x��jEw� �� �N>o&��)�^�IE�"�iV�ʇl��;�o#��s���; o���!�Tƴ`-�U�#E\3�����\6L��K�������f�@!���S����6HOZf���7�:R�P�қ362p�b&�m���E�U�bn��6�`g =@��v�(���4��p��Al/)3p� )�Jz̧=��&ZD��g{������'�6m�>#�7b������f�x)I)�U�؅=o؋ �P�Z����\�"����n��#� B�}�a��y�6ag�LEy�ʱ�.PXT@pu��G�)���8�癙 �naV�P�S��z��YQ6>UL�2��'3X41� #�U�&lc·��A| ��/c⒗�4� 9_4Go_s��+��:X�[Rl #')2b�E����rb� !���Q�>9ӭ�$P��c �_Wiw�Q�f�U_�\�hPՄ�{B�LD�������~?�)Ѷ,s4b-�lMB�i��d;�%����3"7���ol���'�ZA��s���Z����(/���g���hJ7]�o��v��}h� ��qs�7�����8��� ���Nko�E`��9&�]�ӆ3�a�9�?�t(F=���ڶ�ꍫ==q�e��&�Fkt@ o�W*�B�&P��T8MX��Ӫ�l�厹�1?.FR禤�V'E= b��\��˺#�uטɎ)�{�q~'A��&N�>� �RH%�Gp�m" 3/ ���A�I'�>A���8.�DZV�X8�Y^�� 2hFh�@�����!�g�Z_}m�&�u�!ϴ���W~7 ����iA���E ��t�j+ς4k�|Y!�dK7wu2^Q?�� ���(8� �-�y� t�r `C�6%�ư�|��g:�ؙ�ٹ73�'j�WR���1E䨸-��M�t �mP���ʶT�om^��Z�� �wq��ǎ�V�>��l�Y���R�3A��qb�"�H��_i\����Cu�����dⵘx��愨�g��/�E�հ��t� �=��I5���$�W�P���_d��k+$�i]� v/���ϫm���T�#z��4K+��%R�ߒ��g�@ ����N!����̐���>�'�L�з���Ha[��n~x���3(3{Y�m>���'�5�e, v�O�?ZA�&C��8hR��kF5c��*��I�O����������6ޢ���}I�3���$r��e���K�]�=Ǥ��� ��Tí�C�o;`ՙ�.r-�:���+��+@����ܳ�Ͻu�� &�2�Z�qk�����B�����",��p���y_b��w��J:Z���|{mk�L�Y����m ��;�fU�d�߽t?���?�q�VR����^X���"�%�������\����G�y��6#�1�] +��� TB����� ��t ��d~:��u��\D���oa��q��)����|���#��ظ��:%���AІ��r���تB|V:+擒��b� 1��b� I_"Uh�yCn! ���0���L� ���ٗ�h� �2�}������f5������ �� cFX�j���(��(�jt#f���$?� J ��R��6� _��5��!d �z&�o��KS��&���H�[��T*ñ�뮎��(^� $� ?�ֿlk��y�H���x���3`���W��� �K^�&hmR�\Xt� D2$>f��{�]sP�1|��-n�R0]�����\�N}��7� �}��% �+M��|Y��}*�Yf���px����g�����R��+$��'?�_ ������ †s���ƚ�7q���=�[(�p�c���|�k���3�Vd~�1�]pt�}z�_���VD� dl�^b�� &p�q����������3�Y���'*Q�s�8g�~!�=4�@f�U��fm�X��=�tA'�f~�B��K�$'D�F��M"-L�t�C @��4�F���秴\�#��ȼf �d����b~��j��:�$��1�����k(��$��K��� Ʈ��l��ˁt�ʚ�x�̥���o�|ω��k���s��_�,$ixDit&�j脤�����>l�ť��Wĝ�{�>�V�� �I�#�E@� %&i;��MG�_�,�pS��,���!��~�I� �TbٌEF��f�(�na?�&rf?�ј%)Bz�B�,Q���=(�$Tqx�)T �D��s�f( G� 2܎ng ��i`��F�R�,�����|j[�D�8!�\�"����Y�+��:��݆"ux�hO^�PQ���E؈:p��[���9�"���/�'ٓ g��TL�N�U&҄I� ��撳�H��i�Y��N���nW��OP"�N�R���1�w]l$R��x �������I`�}�T���竺����,����2 .#y�|������abd�|��u�5���esq��R�|yV�b!���4 � �Pb��ʑ�AࢷD�W4x���n�x�ӡ��--�Iᮎ��Ók"� �f�c2�0J~ȟ��m����"���;�]��y6&� *�����p��hX ��X�zO#��%��) ��M�|K=F��2���S���56)㡲J��ZZw2��t���Ѩn�^,���q�d'Q}��N����4��\�,�� �r���Q�;]�(���90`rjrnB^Ѩ}.ll�+��!VZ;|TƌJ\q�4J�?Z�Ӛ�8ƌ� �}'�L-�̓I�J�`t^]=��e�������v@�s����/�7�K����Qp�H,x*AԎ���K�Lo�h ��Px��9j��e ���BAF��?=��(�������W0 �l�Mx�[M�(�5�[og%"�vqh�eQ�g �cz�� w+JA�����l���D}�E��ez�b�B^��ij��q?s�z�v�>����F �=c%cQ���WC�h��y��'&�Z��?>F7�r�g8ry��l�v�.Y'��7��`���X�|4 ����!iܳ���N��U�X4�6�ZXu*��^}�� ���^S-W� �q,m4���o�v�YH�\��p���� OBO�we�]C�c�˻m���/a�;d����bي��Q�X���H"߇����:Iyl�uF��ψz��&CY������B��Nﴇ��8� ���j��W���J� ��T@�����q�&cO�ՍA��D���*�� �%�����֦��&2 GZ�(5���c����X����Op���['{�jkÖm���x�. WI�g�#��jt�x3������[൘;���}�c��%=� =�H�����^�=ݽ&QD�H�\3~�fA6(n��J@��v���x���Di���M�KA��ጀUBꃬ�F&����%־�H�:�.)�J�&�g��� �!� �� �k�~#�kz�%Z�+1R�ѭ~�Յ �8�u����K[k=���Գ�{!���qzR� �_���.ķ����� 5�'{HL}�>�"/•>�=���,2:5��?��G H���"z�> a�9ʹ����a����i�;�|E�YSZ������)���\�H@���fx����h$>�W�;���0���x�>sP�Kژթ=%�]�ԓ�8���6�>�~�s�Q���p�dK_'$l�eG��z���9�C�M��A5�z���kj��e m���w��1ҹ����:������X��� D�DF:�g����4Ɍ��.@�5��L[�%��zUј�~ |��`*Hl�����H+r�G?�T������l���e���+zL�*�TA0gYLB@׵�6AV��L�C�m.?�E1i���3� >J}-5�O�a�v �\>��T��������%Q_��L��#�1f�U�)m)���(���ːQ3��!G�������J;�*��l�T�Γ�ዀ�����~ͼ;?�u��T��eP�$L�b��kk1�a}�C|_: K�j��g����(�ZP���$k��ZH�%.�w�H�ʙ�+}�?a��E=b��)e��p��Xk�|�> �"p� k� ��Q�\�?0��Mb�n�a÷H8Q4���p� ;r����hR�xu�?�:�d�I8�U�yD�~���ZzG�2YW6���G���%Tۡͯ9DР٣ +V�ګ2�g p�:]��N8��9Pg=�� .�ߔW���(�ƣ�7�&Hsi�;���2����[��soLjrmG��ac`�.�G:�������|���[�A��z��a � �����}߇� b��3�$�⃗Y��JO��>�}#���tR Ȉ�e�~� �~;�G0����Ma�%Ŭ�`����g�����_�?�� >B�$[�M^�������Lb�d��k�IT�o#� ��;��;�gw���� ��B��+{��� �9��)�z�(r�Z҉ma�����Y�[�=4�� �e�� =��v�� *�9)^��Ցa�!�� S �S���κ4��A?� �c���}%n�^���+��,C����4L� �8�'���W�eƓ���!5�UF$G���Z�Vˍ^Y����� b��HRU��Z��[R��J82�����5�5�}\w����;~ ޺����J�r')i^z��Tg`�b9�J��% l/��X�[��c��+6��3�yIql��t���'G�#� ���+$��L�dȥ/��:Rq�������X�m���z� },|.&�&Q���I|��c�؈�)��`�������@���H~����.4�t ��/ZYH[EE�џ�7���>�3�8��5�Y�g��m銊i���A�������SV|���i�t ��W��){��P4�F�!PqxI]�5�h X��Q���u�I�췲YPfJ� r�z��G �g�&֪�Mv�8F�l֤���ᚋ�%K�k�� P���,)�����jL ����Io5�����zm|x ���!�U��y-�����v�dG�Ai�߁Zq�������y[[� "�P8�Ugo���}[g��$�"�)��:y�K��jڀ�\�c��؂;���� v��7��\iWC:�~D�R�k�)�����ڽ�j*��5!*'){ڛV�����\�\�j��bioaq��L��g� ,����O���Õ�@%w�_��@?��`[�i``"e���0�}"Y��J��vJ3��K�ӽ�c��Q&�m'�E��;NQ ��>3y��=J�KH����'�S����@�� &S�nNث����x+g.����4�D�+�ikK�l�T�q+��h54.�E)�����Q2��~R��^�l�17ݻm�i�|��/]���+r�d�� :�F��j�w���m��}��L�,l�bd�Lhu5J|)�ѡ��kA ��Cg#��V�v~HZ6�m5�.�i�E�gw"e� ��σu�O��tj������+�����+k�ڣfs�d~�{ IS�;��$��MQ- > stream x�t�T]ݲ%����A�www �p�����=xpwwwwww��������{��Wɬ��jm�F��B'dbk��q�c�g��5$����f@++C :)'C+�1������ `2v�@6 ��� ��Q9M5)���h�og)S[��0q���8.@G�� ��_I�@C[+��m�@�@�������� ����N�Ϟ `j��c�?i@7'��?�F�az��������%`hc�����ۺ�U�:T� �L��������������Q���jr��:X��V@CG� ����p2$~���) ����#@Gp�NNv� N�f��f �]�������N\A^�NVJDL^E�����_��� AV���{�Lltť�L9���!� �bn�`�����/#+[[˿�i� 3g�߁���Z�������������ݿC���ߺ��A6�Te����`mk���f�;[Y���T��^���� +��+�?]Ձ��E� 4s�2t�O3�Q�4Q9�L ��2����y�l̬��?㿕?�i���h�r�gmѱ���o{�-m�����2�N���)fclk�1�8��=C��_�/��!��I��o3�������d9C'�@����f���w����Sqr����L������¶nO:Vf3����������?���@��]k�o��wf�@7�1�ʢ�1O�EjF�S5N�ศFk�g�W���[��{٩����'��]J�����k8 ���{n�9���̤�&"�]��DI����-�r& \�q� ��!O5I?�`}ZL�v Ė�����B��3�4�w& R���h�Ԃ�B[� �Z�Gq �(RؓK��2�$y5�LK��T���CQ+���{��J�xy�t(zl�Y�����U8�707��}S=8�E��tq�Fs9�e� �f:u!�۩�� �+�����5f�9���'~T�~��"�� o��^��X�~~�ڔ� f����qq\��r�L*ol���D�w�����a��E�N���6I����S�Sn�Q������I���5Q�N�����Z�O��C���5k6�����=����*عle �7�r?%%��m������!;Τ ф +�U�q�f)I��/�>���6��4}$ ����Ҋk�Ÿqy �0��Ӕn�n=+�KO��m�[�ΏvP5l��9�^d*�s�� n����Vl�ۨ��W*Yࢎ����)�� 5#�=K`��. %vD�H�u���o�o9�Q�����l~5�ys'C����1b0�g��MBC��C�Cd+R]IX�>. ��O ����)�����&؅�em�O�� �|3��v����uH�Z�� ]=�T���qyA��̏�8��ʜ��)�OȺ�I��Rx�b:A��ɳ�ޣ������;U9!�O��+�����2 � ��2��[��r�'��"{kŒ�� S���;�eGM����ۍC'v�M�2oQ�v�@�=?l���o )�8�S �%k���� ��ոo��?�h -ɥ��:�g��:�}i,�u��9�0��0�z�[O��h$�i���X��4ǯ�����|�"v��m�`�`�?`T���z�S�;lW�kN�*϶(NK:��H��]�#�/� "f�jmb���W�[r�!58D{��ޤ�x��zE���5��C�K� v"��F��T�e���I� J"M�ts�^�Px\���I�z�7�dū���Η��Ł.��l�[�CҲ0�z�&�)f������Z�.w^�B�^^Gwa�W����m�����_B� ɾXZs|�\rʀ��3K� :��Fa�.a�|�\���a��nߩ�, �0�ύ�m�mP����0��_�:t}\�����Rd�� @��Y��V��j��i�\rS����l�����;�d�뒣1yR�o��F��rt�d�*�x �$��xX���� ��� �W���#�|����sxxĨ�x��PP�˰"&�� �B�g�\oje�4�S�Q ���v�'��a�nl���%p� y+!��� �mEͬ)fZ��@K��|�����t��.��"�4Cxuh�;�~����y} ��|Ü]�ua�ⷦ�꩛�z |_��8=�)V�i�C]H�GI�]��3��9�W�o�O�P���eH�V���\� �h=;�����Z������� �&Q]"_�qP��B��@di;�t��^FW��Ԍ.(�0�Z�4�B�[������ɕy��⻷�� 6>_�f����޺�����E��=�h:�q�_!T�/P��S�����k�]�!!�a�k ;���-���B�_�/NV�־�(b����m �Rv���z��tT��@��GU���ͼ�C2��&� ���`�̴� ��@�����m�x�̴�򵊢�f&X�ǟND�9D��^YD�y��I5I�R�hj��Ƽ1:�_oy�e�@~?[���1k����\@p@�?kj�,��N:O/?��6yk0��p�țO\���4 C��3T�RS��Έ�&�r:[��ƄC����Ոt�� �� p����O�sW�/ ���d]�a>��A}���,��m饁$�T�ʌ�,(C�ޖ����D�芮^R��=,az�.���.�^����J���)�p���A�����+�v�j� s�+��b|�]���>�a,�M� �J���X�M� f�6��\��oRd��[Wa�`]��(sr�*��kN6��X���m�Xz6�R����ɛTTN��KM�+Tz�K c$�,������u>س5��T�7�˪��B���XƄ3�4+h2U+�ê!�P)���� ݚenK�ό�9��)Ow����h� ʨ���$�b�I,t ���(��S�Z�I�UH_���)����o����VR0�7;�\v������&��M�H��N!p�L�M�7\���k}��D{�F�$l�hȇ��@��!�T֕�޺B(x�Do'����T���Y�:m�:�v�lw�}��v:�0��!z�nB>A����E|�]/�m] 2���(M�Py�ר zŦ^��d q�k]��4���&��n7r�W�W��� �si��9�a@ C����?�;��dul��e]+�{�2��A����\L��YK�P��/?^�>���'�h��5�8�6xb�CBo��h��%XK�����.�ܫɺ��a.����?��B����+'� �H�j��h�t�[BfLA�u���w!�sS3���݂W�S7���.����:v�oYP? 3j��ܦl3�.��7 ԽR�����a�h��+f���j7Q�C��]�Bl��1��� �����_�׋��� �#u�{�����=��F��j�v9{*6;���Y��Y8�sUZ�L��{����S���ݨ9��SV�-���F�4d��.��r����T��i����]!NC��I�4��j�LcCԙt�}i�q���el��3S�W�]Rߥ��F����]w�B�6O�O4!3�v"� �K���zS;i���Ճ���ν�* :�z�����+� I�"�TcW��sX�e44��� ֭�����ͣ% �ud#�◺��Nxv�Ŧ�*����I��/- f��I��+�����FB�[�u�%�h��3����oD��]Y�@�n�S- �l\ �����=�o��B�� ��e/"I K|B�س6G�Mb8l���V��е7!�/5\��o���Rl�3���r�����Q��T>8m�9~o-�S�� ;9�����sf�Y e4�-I�mRI��2���#i|��V݅5�(~V��C�����n8�-lnYv+�-^�ӿ�3�u}�T�.��[}�12u���Z�RD���5�S�4�£��H�+�g�XRƟ��,���s݄ن ߖ��ň�#x�f�K��扶S]:� �,*v��5���k݈H��Ub-���^^�-�!���)����6B�̇g�&T�I��2�K��F�~r��"Z01�on�zǬLK��r����� SdO�OqGM{���� ��+b�4�u�N-1L;d��=^���U�ܿ�20P���j��)ȔXbH�ل7dɛx��ԅi�QB{��n��$܆�BE�RR6 W!&'Y�)L�Y� ��yB��V�X=��p�5A�����Q/m�):� v4�S!��h� %ҳA�Sݠ � q�?�"�6�R���!"��{������QA��'�: U�'�H���v��_�D�Uc����G���,w�-��dI�]����#�a�/���}�N��m�K 95z����]e@)�^.��j��)��(^�A5 ��Ht�*6����]F!���_|;�d�2Z�)��M9�(�,g{nݧT i�-Vھ� _,����+���t��X�F����U�Y� %��u�W�[l��B��9XK�)WJd+���z�Ks^c0�m�/�����^���)�|�ĥ��S�D^lYl�V���m�78�ȫ�h�^���`��5�-چ�=�Xݢ!���B)�P^�|� ��U�L��j�'�;]Pix� >�$���D(�tU��ǘ�da*&��1��"�E��6;Y�z�;g�e�Ur���R�L~[������{+-�c�`���U&��C��~�����}8B�"���@�m���z��,�O���􁹩�N�IھÇ ���L�m�,)���|��O������ ����}�0�b�g,;�u�A��D�} \���g�����} �2&�D]H�X��Ev����� ����j���QF��+�ў~�Q��24֕�h4���5'OV� �O�N�Ϻr�Zn ׻��ӛo�H^�-�OQ k��f��.��.d�{��o ��\�r�ؽ�G}XV�:Û��^� �E3� L��i�@“˜��X��:7��sANJ�e��{� �"� ��M.�N�I2�R���j7B�P�w�ތ�Ya��B?)�j�� ������ ��dtT��l��s&V&�0�1���(�#Ǎ=���9z��J����ls�"t?_]fv�b��8��߆�.��/t�\�Y�f���y~U_�oҧ^�m3����r࠻��_���7`�*98xe�C���}tݏ��� -4��wJ� �#6 s���`H?��:��-��� �����5�*F,���������� |��3�p�����$R^�$��B�4 ]�uj�����8$�7]�-T}�>Oh �{��`���z��h�g ����,��S��?q�������d�.0�PjJ봡�CoF���l���O��Q��.����Š��{NoJ!]�����H?�м�).Of匞�L�� �v������(�_/x���� >VI��K��K��47�)Eg����|����l\�Fh�2��Z��W�F���F�"u|I��jn��LP�j � �*�es����D�V�����o+�0�Q㝩��Sr9Y(!���O��#���k��T�I,�r��H�d� T##^��: ��tp��ɛ�nF�d�����ZR���S����L�Ϳ(��kD�$w��� �Q^��{s�6ȏ�P�kJYƔ�_��I�-?��������$��BaV�PX&1����x�W ۸��ȷ�w�!�v?�k��p��I7ώ8��_G�J4/����4�#Z�?䛸bT�x�� ��rb2��� ���V�jbBu��������-��� �0@J�we��}yʇ�sc��Sa�湥�n��`B���W��޷#FC��o���[���#R\4e\�ݜ�|�f2ζhb;��v�gZ�m��f�8�,�6���9U�Qa��|_�|5�얿#�9j�^1F��ﲰ��`�B'T`�c貥IF�'�&-:��qE@ �1V�����B>���V�^偠l�v��b������!����G�ʩ��}�¿�a��p��X�2껡Eb�:ď�v�� ,��#p���=]�.~;�wr�2��^Q�=���#����%����F3U�$�����\N�.+��L�������`/��2�����5�� L�9�s/B6������~��;���5>o�㋨�W������g����v�=�$�����D��_ s�i%1d"�8&Y���� m����B��MOQ�?'kݓ�� �*�½!��kT�s�-!���%�ٗ��{�T7xx�pr �:B=�� }���K�Ѧ��YX�LJ�n��e�|�L�!d�n ['^wrL��VBj�����D��T7���/;)N0�����E�5�/7�C�;�CPD��.�N=-*6�_o��s�a�!�q]e������+�;�>tM���ۗ��PKF�no:�S�H��^�!xD���R�;�{PV죊[��ޯ����X(�C��i�p�_S4���ɉ�f�SY'{����M ��#k�è�2)���/e��C%��=���e2`�˶�_̈�FM������5/2\g�� �� k�>��%�O��ET�_=��%R"�� $a$97��JS�&a��#�����M� ��p�H�/�6{�ޠ�����VYB�W��2����7�V�k�%� m��3�;)lEŁmЀ����H��y=�t�@�����W��@�D.��oҳ�t�I�jD�7���'p�9��Ԟ�`��Ɠ[��+c�c�����d�s�쪓����P� ���y���T���T�_�o�}�LXy�~E�|}�ܹ�l�Ѡة��O' �;a�����rGè"�mn�cT���I,6t��� 溱渾l�~��Y��2����k�O��9 V�g�1�H�U���U�3�؊�S8GzHO R �g_������� o(��ޝ%���-W��%���?*����*4p�77�Vg�vRFS�.t>!�����M��K�.$NzNV�k����eimx�R(�{&4�K�.3�f6�� �� �[���yj(��u���[.���=�}4��_Rwݮ�~P���tA�ƕR���� d'4g���].��F~"1�� ��f�-|�VnA&���ɴ��*BǑ���p�m�:d���D� ��+0=5���h��bduҫ�L�޻��G����q{j���. ��!� ��V�� )��ߑiYm��B*̓�3�� ����:C�'/�"�rhq�-U+]\���T[PI��J�����$�-l;�+�1͊�����x���/�8-}��I����,��~��6NEכ��?�- D��hzP���ݚ�/� p��b;�ܩ^�Hi��T7\}z�g��A`��NA �#t�xm�,(6�a/f��{� �r�q1cI�,���dfZJZj��t�rҼ)�x[�$�z�q���bin�g�RD(�bu��T0�N¹� 4u�^�A�O�?����mtӿ���Ɩ��;f�&���$��M�i�:b�p�9������?�F����غLP�헛�녺.Ғ� }�>̪��1��奸�����A:d�,i %�� �+\�$^}�JO`���JQh�Jn�O�(Y��. �UҢ�#���� ��M�����%-ň��(����:����-Q��o �� �ࣞ\�4��"�n�$�޾�6YK��W|��20�l �{L0����2t@�&,�n��fB#)b=���)P�M�fWl b�| 6�('�_۫�h��k��g�A����J��v^r�J��f�IQ=� �q�7]q��$� ?��`O�f�7����~�Q �k�'� ,:R�|3�@��7d`��� �� ]��$�'�S�c�fZ�T�`����j�,�C]�lg1���W6$���k�- s� ���WO�J�R- �Ǜ�gu���7,_��R�Q)�}e�ws����S#"n�ptHd�O��:q4.��� 8�g�x`f�m��['���A�Y�>���%W��Ο��c�qά������/^���U3CU����8`�Z�D�+�(��d��%QTe�jdgr�G?�X�����j@>���ޜm�`���j��� �G��� -��"y�����ž���A+UٟЉB���b�� �5|$��.F��VU1�ˏ�D�й����-�Nh%�[��y�H ���K���Q�E���9}/N���j�_sCBu�$�ݐQiy�c`��s�.W���k�Bk�-/Xd��고@x��s�Ӓ�WS���C����ms`;M�����I#WDi�U\�MC�C8ڒ�f���Ѻ������1���q��b��0F���� * g�����R�Z��B�|O�؝$N�Ogwq���;,j �� �I6U�� ��@��/�� 2x�����Nxzl��rt�����X�3\��E|z_��M/aj���:�,�ʚ�?�Eâ�o�֗�M!`�-$׊�(#�{[���,��_�y�|��f���;�m��-lU$����uh�ʜd��>�H��ޕ��P鈒68%�V]9� ��5Ϟn��:��gûw��% �7Rǩ��7����|F.;�tFH�Qk�rT�������-s� �U,���zM�fo^� ���X��jJ[ ��5�+Y/�|E���(N�^M�?��3W�K�NHXN\'_��|��i��b�OMCRb{y6i����A;|�Հ*�|�U�Z\&�Q+� �A7V��%�:\�Ԃ?���\���-�d��?��Zh��f ����lt��I���;�������H�c�j��������(�I���yZ��u���V������B�qH�Q�v����2-��BȌ=�yj��Tċy�۹i|�'�^�p���7�%��[���bd� ��L�| �[�㘿��� N��ى�yV�1��Ǭ��;��(6ܼ�Z�`��������+��� �� ��ܼ( �%��z����7�p��KR1P#�St�-� t�"�6n�eG�($�DSQg���������}N & q혔��_�+���y˦�}z��L�M֣�%��; s�#���W��_�l�}�f�&8i �r����� ���[:�Ҳa]��߉��V"�e��QOuD�8�~�B_���R?��}W�RcU��S9L�$��7*6��ŧM�SV�䌪"���%Gc��U���/"�^�/S!> ���H���2]N�R��f�y�^`HIr�*�0!����������9��y���@�l�?��3�~M�u2Ѧ�ޣK�@�C�P�K}H����A��x�P�U�&�o���]ڙI��@��M0�4 :��=���� ��$�V��5UZ(Q��z9J�|/�����%��m`4��PdqF/���fR#^��,+ J�(�3[n�N\��ƩE_׉[5�&o#Ț ��˩ }Ƿ+L��Ծ��� �z�� �\�(��$m�P���rmBW�����h��- �d�� � (�C���j� \�����M,�T>2��GU#��W5``%5��ېSMH�g���]2�M��^���!���A8p��2 b��^�m�[���iԤ�`��q�lP�A�0��B���kg��#�b3A@^��V���D��.��g�k�&���M��� ? ���9���:�E����X��T��C[�~��T�.s����ʿ�u���v I�Pc���%��Y�)����E'���.J�(f�(��/� s:�T԰uzՏ�D/F�`��Ln��L�3O:��~UPGi���ܧ���eB/:��"jx8����,��9���a��{�Ɗ��n"���d���;���pq].��JΚ6����x0q ��1�Dw�I�e$�_R�JTiY!w]��6=�znv����CUmX�h�H��J#��i��p_]S�����m�D�ԇ2h���z�����������1o�v�?�8` Bƃ �����LP9%ƁR��p�, ���O%�$:���CFZ$o����^�/�58?)O��@#�>&��3x�j�nS����?-mL����8󱂤�Fb�Җ����&���b�z t�� ü9[��u/�G�q����D�K� ����!�k2�u�\��� 1�^~�G��� 8U�|v/.°C�`���4pkV3�XŌ����i� xD�h��q#�/����X�恡m(L���J���?q|�����X����R��6_ݣ��5�U������(�q���5�#�Nհ��q3�Z�q������w��[�����控��^eMIT2Er����&d�'X5RS��啙��`���˃��އhb������Υf�2�����5߬����� ��h��w魖a�z�؂A�Ogm��۶�{V`�p/����€Hf�j�"�a�@�� �O� y� ���oq�U��J���im/ ^����0��ik�����6�g,֧��uzc�x�����uG;���|�����ؑl��w� ��l{�b���i�V�DU l4�ŷ�5�0�I�ɭ�g"�\���υvO�� Q��c��dN��t��^��=�ޱ7�`��oA*�O �ӡ3)X�0��D�O,�y#!S��4g����Y!��]uy*�U��t�x� ����i�靏z�E�f�����"���X@e�  {��������ܦ^dw�zV�"�&�8 ��T�Qu�)ɟu$�?�_���2�8tOzC���.�ظ����5 ���/�G�r����f1���V��*�w��M� e�VK7z~7RzA����nKy)���=����X2����]#�>�/��n:�d�#G�wf�f���E!�7���R稖)�'h 2P��:nSGt ��[�+`d6�N#�ɭ1"A8Ɩ�Fw���m�X�d;�U޲�gPH��2�{�5s��EI�P��?.J�`c��G�^̬��t�Dh}}#�T#����ĉ��t��s�AwJ'� Eywh���I��]�w� �a��������/!Fnէނ�A`AҀ�M������ڼ=�w䎸_W7���-]Mּ�pC�Mo�� e��x�y�"�q��������;�X��4rj�y��vA������=�$P\�RQ��=!J�4�c*N��!�6�k ��ju ua��U ���w������u����3�Y�v��Y��~�#��l�횞�����]���Wݳ5�3�H�|a�ᠰ�k� D� �Z�K�ׂ ��e*���*w�:�# "��T  �oک�Lmy&ɓH b��DL~��I�� �,Z�1 �qg8x��O񦿄�A��|�.n��R�8��2)�4���� �~���`|e��\ ���4��%h�N��� �.G��� U���οl`k��e�nů$B�a����H���R/3p� �m�l����_瑆`Z��4�ܞ0�����(!KV���eSr ���4�F���'1�m�ߨx���� 8��� 1p����4��ǪLG����o4�(��n5�&����&��LDM��}醗]R��(A�b.1�� ����:�%+>�N��B֨�"��ۿ*�-W{u9�V����ҁ����x�K���>v��j�Pq둠2q�%뙩�>�d`� ; u��_��`'2��� �-�4��ʽ�����5�p��oJ�I���*78�4Y� �E�Y;p�e�iC|�n���,R�a*����SI���B� �J�&���*� Ux��D�)s�С1��H ;}�q%�1G���WPNw��6��;���i��T��n��X�p�mV�ֺ� ���n���T‰И�� ��f^T�}��n���)d�Q�tc�\��B�H@�[OF�T�@M�g�u2|�L%��-˚���@�[Y ��p`}2ؼ���C��� }����"���5�:.\��I�9��o@������⊞��j7���:�NG��v�p˟$uȏ��gsҡ�cSp����B�wkK�\�Xgru�˯��~���i7 �t|���o ��O��Z��J��^^鮛�eu;���Ed~(ԉ� ���-@N��� ������ϴ*I��z��-��[��V�3Y_z���f���?&�����P6:��p�nJ`w��m?ͽ ��|��H�}�5~�5i���Z���4+��Gu����5���3�$4F�W���v&��������$��A��t[�o"��,xp�}b���e� �8��f+b���rP�~r'\����4u �-� �-1�W�/Є������5k[un�[RV6�2 �x ��f���ۤI��ߜv�Tl�_�R�����>���.G�ܳ,�ɾ�@Z��ο���خ�#ƾ�˝���� bu�����Q2�Ӭ�|��Y`K\39iK����c�Z��x�ɰ��ÿ&`��o�LL� ѩ� �I�xC���n����C�/+��L膆� s s�|��B��8@�� ���$G�$��)I����F? FIx`-�gV2G�Y�X�=��-��T�Ƀ��#2��L B3j>�Hlt�m/�&�L1�lr�ȸE �9�C��:����0�^ҿr;b%�L|@>�S��c��P�B9�ixǓ>��@(�(C'��jT���AMg�P�p��x?�k�P��UeҶɥ��|l�[1#��Ct�q; n}��g势R>. �P�pS��`:6Z�{���K�� ��*��t�ϑ���K)� ���p�,C$WVH�r��i���Õ�����}��� ���vRq.L��J[ͭ��Ժ�OYToB��q;�G��1r+B+�w�phS荑����9��U��E܍S�х#��'��a'e`p5�RSk�M�.oG��l�9�Xn�����5�V�z�{b$n�؊�e�Y�*`d������#�ߺ�@F\I���[ �Mfq݆�E�J#GZ��#x��GU��T�{k�H���Z�zij�� NI��G����K����w�[ �ݭ3�/N�[�_7��+�m#Qr\�� �6�ƌ7Eo󙬐P��,������k�S�c���z�^E>ߍ����2���N]�+h� Kė|� �']�(�3�3� F���9[d�ZՐ�x�нtoQ�,��k*"+�vvثĀ)>z�E�����:��i3�Ց31�0N�w?��[@ҋN7Ф� ��@��)���e�e+ȁ�E�x��¤�a��q���4��t�gI����$��{1�v4,��h����Y� �]$ @�� 8 ����vtO��w�[�2�#C-�����J�z�ww��0Lb�v��|�M�p!�z��Zp������ZX��=���:�z�L� ���qc�^*��u��� ���az�^{>��oD�E�~�5yO��9'5Χ�Ơ=ga?�� ȩ>N�\��&B+d����Ű��b�׻�Q2�Ӭ�|��Y`K\21:l��՞�iѨ+�L��&5�1|�,�Թ�}S^�4�}�ۏ���}c���t�� u�4k"T��oNKp�;v��X��j�0��K��#�l-��r[މ�$��5�[-o`�j���.m��!���P 83y��&58�� ��ly�]�~*��3p�&��9C���iӨ�����������\�Q��W��iֱ�+�?j ^�������&��e�H��9�Y���j�v���rvCK_������֕����­G�̴��4TN���&�l��d���h��+��z���=���K�j'��!ء�NBa`����Q�)c�/��Z�٭>�������2y ���*!FZ�? 0wv� w8��XUm���k����N�.�`f0`����k떽]�)P���ʡf#S�tSS:NV�|VFSt�!�?C捧�Q}�?/��lF$f�����dO����cY}�Kl��v�r�Ѵ��_��y��wj��Dˁ9ʢ��E��&]ƲCIIé�P&�)�SY�]ku��mk��r��5Yp���}&�) 0��*c GShp�t�O2r�����6��́s�k��j�+�� ��L� p����������5��th1��JX.���h ���3=��y��P�?L� i��������ŧ���`(.�LU�Tp��+��D��� �ɭL�E�`�{�2�^I�1����y�� �5Y���1/�i5���ӏ�t���"��L��P���4��6�.Ї ��7��D�ʬ1&4ݱ��|�r÷F��T��y�ڞ(�|�?�-���λ]��͢���^��o�~�~"8�>�ʛ/0,�UI���k���εTC��M4�'�w�A|���?�Îx��qdZ���H1!�ւ.���ٕ� )���Q������H��I�����0����:����l=K�D3v>B�$L��w>LR�&Tߺ�d!lj8LWԇ=���0�p2 �5�a�� {4 �E%��Lt|��P; ifX7�wn\�+j*y��|y֨J6���vw;Z�����J0N�9:��uiY���T�W�^�甍=RO� /�v6�(�2�?����̹��\.y��DQ��q�0 L�]���j=�yT �[��J�]�J�d ���V��t�l.e���^��`���Q�o�4X��a"~#u�������I3!���Jvi��� �nw����4�D_����S�/*��J)Jp@AE,����}�%��$G��e��?���r���$?2����tZM�W�q/������ ])��if�]���񬊂=t j�+��mN��07���X��!*���/�N�s��&�kc����3����'. ��pR�i�qB��R\6�2+�O,MlE͐��6�[[����� \�@ �@n_X�`�*6Q? �$���Ǖ���^����n �'���b�ހs{�ګ��@���V�)�k��:M_>L'����ЗZ�o��.{�r��?�x��6P��q��e˴����'�%����O�[l�́�p��E��h��� �a���~wbk�YO���q��f�d�8�D�����v��ȇ�Hї1 Ѝ$�hhEi�`{Z�M_�u��G��{5*�.H��Z�#/*�^x��Rd~�ힸ�yU�C�Y��_��ˮ��ʰP�f_FP���־*ߪ�Y]�ʅ���˜��Y���X��˜i!+s3V��q>�Շ����A�Z�%���qG�z�T4˨4�9�z�H{��-['���~��fϹr�&x^��F ��g���؃r#8S� �S1�[G� �^����u�Iq�?��~�L??�z��0IZ~ �@ƒ-l#Zܓ j�S�Dz�-_V�[^�*�~���;ڊz�=]̭����7��RJ~�7(0Xǁejַ6 *>Y�=�K]����3ܾ�C@�*!����I���d��.Gg��kF�� �x��3p�hv�V(�fw�Y�D���� �c�N=E-Ƅ�qekh���{�;[��s��jh�kɾxj1Va��O v��Rqԇt�"F��H2JE�A혋��w4-�@���A\��eg.-G�_7&��R���Vp��A!�� ��Y8���,��3��FP����"��3R���珿֚�����h���A_c�� ��-��9��b�F1�c YO�GG���S��y���i�����n[b�+�'�}:�W�=S �dPN� �z��oa ��B�D��1����t}72 �-�Ɛi���vQ�1@�'Cg[�|��u@�iԵ�"X/A꾩��� _E�������_t��;�D�,�ɗ[d���O��n���䤓���W��#?�+����b,��IJ�od/�� '=\���Z�8�����:j�&Z��d�L��FS�G�z���udK�߁����Џ�É,b5�nbL���;�g�NZc$+��RN"J���#:{�j�np�#��R��0�*E��K0|:��ř���@���:5b�$�l�[�*����ͥN��3Cp�������;1��ˮ3A�=��6��/�Ԟk07-�l5s>U�pr�@[��.��"ү����&�LU��sh D�_?q�`�LH��9����L�9/�����m��|Y����W����������h��_�sS }��iİ� �t�7 潦����? %� ϟ s�/����-u�D��]�cY2�E�'���0|��q|�2™�XQ8�˲ k3��^�t�� �}��Jy�� %I���S�?���%�QE.� ;���3dYos�U~��a�̎������ޢ�f��� �bs��^XD ��`�L�ߦ M+Tz�!�r� BLڵ��4����&6C�w��~X�~�����n���[��ł:�L)חh�4Uhh�(Q�endstream endobj 555 0 obj > stream x�m�Mo�0��� �!Rz���|U��A�a۪�V�&�$E �6�~=H�UA�g���ɯ�����~����uo$����ƛLD-� t� �@������Z�c��N�t�=Y�N�k�`T=��R�o �æ��eC���ڕ��(>��Պ� ��A���i�Z���sn[�6u�c�^0Xa�h�\je?��0b���prOY�����[�A����KS|�dۙ����oF�)��MZ��}�4W@�{Y��mG;�� sb�gE�jq�(����po�$��}Ido����n-p!J ��m����-��O��[��L�endstream endobj 556 0 obj > stream x�mTMo�0��Wx���$� !� �8l[�j�WHL7IP�����V=M��̼� s��u;�U����ٛ=w����������]yil; ������MS >�_��P��{=��s�@�d�����k��x�;`�VY�`�s4�Ja�Qܡn������.�Uu9\��Y6�>��� Dӗ��}�~�������r:-d0�V��W����k,��8���y�Lһ�ʮ��Ӯ�����ђ�[�*�m��Lr��?���q��� �5F8@���=�@��)&�� �8�����Rx u�D�\j2H�����V���0CzL]� �b�ct�I �g$`htы0��\F��0���s���� jd &j�ؐ��u�������1�V�������ljOu$՟qW�S/%1{\�xB!���K(hH��TЖ枃�J�ρϯv����=k��2��U�Kς_:~�$�������/��� ~�E�+7��ˢ/ �l���(/}�� -+ZXuko��ԝ�E?Z�����K�q��endstream endobj 557 0 obj > stream x�mTMo�0��Wx���$� !� �8l[�j�WHL7IP�����V=M��̼� s��u;�U����ٛ=w����������]yil; ������MS�'�K}�v��}�����t�ƾ`�R���\w���s��*��p��Wl:*;�m�_Ű=�EB��.��=��]�����6���E�%�������‡h��W���v�E�;�^N��� ƣՊU�� �ٟw�eӟ�Q�?O�Iz^U�U�|ڕ�ߵ6Zr�bˢXE����I�S����:.��t�r�A�&�T�H>��4�"��P�X ��H� BM��@5�*08WfH�� �AX v�.2I�##�� �.z�Әˈ0�Q�a8�tc�pN��0�A����2 @݆s�>^�l>^w�o�_��j�4Rrt���sľ�� ��x��[���%QLu��Q.ݢ�T� ��܂��P��Kߗp������#}߂p��M�����AM����3�7CB�2>��*�R�{�@��8񩎤�3�� }c�$f� O�#�z  )�� ���spW)��9���N��{=�g-�_Z�� ~�Y�K��/��t��:�������/�~e��}Y�%៍-t����:�UEk �n�m�����Gkp\�x{)���ނ�endstream endobj 558 0 obj > stream x�mUMo�0��Wx���vHU�d�C�mU��^!1�H������#x��?��g����x]OT�m�$|��͜�s_�I�ss ���:L; g~��8�� ����)ok� �A�8 �$`I\���3`�Af�� �tw�8��y*�s��ύ }n���FE>7*��Q�ύR>7����G]�;~��� O:�yұϓN|����I/|���y���Ig>O:�y҅ϓ.}�2�� ��L� > stream x�mUMo�0��Wx���vHU�d�C�۪T��Bb�� A!����Gp��?��g����xYOT�m�$|��՜�s_�I�ss ���:L;�2�6�8{zb/}W����U�j����Wm�?�fd}O��i=���7�gR��d{n�C�N8�oͰ��of�-��%��6����'&9�P�u�`�L/"�t��kں�(a[� �duS �����$�x�q�a�� M���N�����{��}m��}g���ى��x��` �tw�8��y*�s��ύ }n���FE>7*��Q�ύR>7����G]�;~��� O:�yұϓN|����I/|���y���Ig>O:�y҅ϓ.}�2�� ��L� > stream x�mUMo�:��W����5?$R. �d9�M���� �eC�����k�m�C��p�;;�w�~>�|��3�E�_�?O]�5߶����w�]O�c�c]=~?�}�O�yh���9%?��۹�׬��B|Ɯ�>��)�;�v�w%g����4�3������>\ ��6���� �EJ����7��8� 1��{�~���`W(-��;]���%=����x����e_,�b�+-O�;q�\�L}���U��I--=���B����K������E1�p�[���! Mߊyu�>�.N��5K)Wb�٬�8��i�[�_��uʕM��zQ����)V���(T���x��ޢ��jy���!�����Z��2�P="�Zd�0\ÃG�R\���).�2*�Шa!�U��,����H�`+j��.�5Nα@����VK-�x�%���3�%�A�YӀz�Κ�>kP#�����5m0W�o��þ�j�������.��Z��T$X/�)n)�#W��o(�o�RZ� $K�p4��Z�-b�\�1�ܰJ� �P"G�X�Q��i/8�k�^��Zq�:Zs���9�d���B� �)�sL�-�7��x���J�����`�a�ɽ)f��$���1�� dъc�CZC� > stream x�mUMo�:��W����5?$R. �d9�M���� �eC�����k�m�C��p�;;�w�~>�|��3�E�_�?O]�5߶����w�]O�c�c]=~?�}�O�yh���9%?��۹�׬��B|Ɯ�>��)�;�v��z�|�N�8��}No)�e�0�&h�?q:��P_�� X�}��a�c1���+��a� � j�Ң���~�]�ߏ��{_��r)����4��������i_�������px�`!d�Z�>���i���]��� rEqI��b>�,�彐A�$ G#�e�"&��c� �D�`%r�E�*��s����(�Ǩ�5�ث�CI��*�����=ǔ��^p�k������+ ��ܛbV�LbX+���@�8�:��1��3Jp3 �o���� U��^g�{�����_e�{���]��������*�?���`��C��B�h���gi�یt��V�;�۳ѝ�)�(�Z�K�7b��A;��E�^�]|���s��Qendstream endobj 562 0 obj > stream x�mU�n�0��+�C������W`�$ �M�E��Ĥlɐe���rv�IS�`c��ݝI�շ���m���E�Z=��p7+�o���U54��ε������84OnR�����t�ɛ�ٟ[wa}M*�k�P�G]?�_����M�lw��S��4��ݴ��/����AEI?�x��N�[�������E��_Ľt};��������IV���H~z;M��_�`�T�G�y��7�x���֍]���?K�[O��q� C�`�R�{���?���_���y~;:�ڰ�fh��mܸ�_]��z��u� \���g"Nٽ\�k�յ� u����A� )`JbD>`���2���$`�TY'`�`�9&�D��kx,+0*N��X��k0j�9�F�)u���\�c��3�����"/��!\,�1��8"\�� �Es�N[��sS�����9���Ӻ��)9^W�5j��s7�� �^F��p�} �L��.?��t��5=��?� ��/�_�"�矲>p�3������5�N� c�����-��~�`�& ͟%�Q?K'���!/�/ya�%+CsV2�GYŘ���&�{�Ռ�3g�1��??g����?����'���̞�����?!>�L�9g�)�k�:#�e��NC�l1��1��c`�8�3zY����׊��ӊ���1���ߊ��i�g�n�g�f�g�/>C�����ѷ��/�g� �z �r�g�^�������i@X��R�������?��o�RM�N �#��Sq�>T�#�p�aޫj9��������o���/�q��)����R:Gdяn�����}�nߕendstream endobj 563 0 obj > stream x�mU�n�0��+�C������W`�$ ȡM�E��D�bɐe���rv�NS�`c��ݝI�շ��m���E�Z=��p7+�o��U54���ε�=����84�nR��}u�wӍ'���۩ug�פ½v�}����5�̟i�mO����3 �K7�y�W���ԧ����n cv+>�7+>S}���� ������� ~!>�_���S�ϔ+>c�B|&��L��O�r�`�B��,�������&+�j�wRP���{���x��H��^U˩D��g\%���9���辡��|׻˕tȢ�e���:� �ݫendstream endobj 564 0 obj > stream xڅ�Ok�H�����@�q�d� �Вܐ�N�I��H�����,�f>���*'��9$ ��Ch�Ы�A�ޠh�w?���p��g�u(l���X�-$�Z\��~���D&Q�����w�� ͏������Vj�-|x��W��)Z ���z=y�е�����i��i;��h��Z��[G~���I��yw�n�W��/�i��V�&��iP�Pȹ��Bb�Q�:�P(���P��(h����*LG�SZ�t�j�h���cCKh*l�p��a�C�k��ע�3����=��u���$�B8 ֆt����:ir�t zI�ct��:gm� �-�K����^m�Һ�s��p��n�5f��{Q7��&]�G��\S01 ��?��ߊ���;�i�����2phQ7� 3���q�o� > stream x��[Ys�F~��cR)c�k˵U:�ĉ�����$�)R!��ί߯{E��]�/[����>��9�� Y� eSa�*� �Յq�k �Ba��GS�Z� ���+��_[�z���b�G�P��)1j��"i](��u�d��D�Q�X ����|~3���a9�^ %K`D����|��$g\�y\|�0�,N ����1e(|=���6�iBi�v:�Lഭt^���@'U���q5�!�m�S@w� ��~���Ru؝L�s�� B*f����\����� �|�\3�|:���B )ī��8[����e�G���̗�6��NՌ3;7=�·�������`�V�1v��S�@4j&o)��n�Y����J��v:p��)��>m�S�LH����l�v=�Jg��a�-�ܮ��t��Q�����C�i�D�n����j��*�RZ� �W����E��^�}v�*�f��l2٭t�ŒF��e���,�}@�o�Ê S��•~ݕ�[]�vt��ѕn͕v+�w����V:8������V'�k��j+�Eְv:L]i;٦$ԥ�Δ>lWC��Anw�r��,�?@���ȍߊܴ#rӎ�M;&��c�,�"�����]J��q���c�:i�����ΨҜ�>`��J��XD� �Ħ� ��,�bKJ언7mwJu�O����}kf�ޒe�Wl����{��k� �ʒ�'�i�kA��l���;zfi�{��H~*�Z!��>ru$�OX�D:?0e��u�J$y�e�Me"�O}e��g-��lS�kR0W�0t�Zc.k�\�1 ��m�|���o��"^�{�f��Iʛt�lA-���Z .����6�2 y4,EJ����X��"�0��pt1����3�|Lz-%�hI �'e�V��6P��W0�x�����P�#� ;�ZI@@�������J�9 $�i���@�@�,� �k�+���Ƥ��EI��f�y(�57X5�m=�>Gj@I�Xn�e �+�E�D��uQh���]j.���#y�eq��{{�x��k���h���T�fi��˜ě��@�]�P��a��]^^f-i��� ����d�F�ƢDO�MG���?�Gw,��J�( �pԆ�~����n�C��aV������a^��b�a��'�,M6u�� �׀�,�t/aM���]���Tl@��z���饠��ބ&���"!������j������.V�e6�7�z�4J7�7� �0B �HF��e+���O�ѷ)�sS�;����ҿ���u@���W����9��aq�`X�Wc�nI)�`��f[pL�X�B�Mj[4���U��=:�unF�7��l0��'�|��{�O޾���雟��>z�$�����f�C>>{�>@v��G3��������Q��jxy5/pC�ЃXF���y4�/G=|9���@��x[��F��UJgE?�i�|8���bv{}ݧ�������,��|\~�z.�Y��'�/@���7���u���X��m��ii��a[��6:�Q�q(~��W��n���c5�����f6M�J �'17�t89s�y�F�c�7OO^�f �{L�OM���F4�^�+�93��_D��=~v�����t�{1����[�[ا��W#����?�q%n�yT]�smJ}���:�=Y�������w�?"�7���F� ����oT:�{1��f�Y��iU���j��4������W���w�S*pz�c��RW(�u𥼶I�Qu���z8��׷���f�E܎��V�L�����_Y�M.�}��"٥��f��gs$��'$���^���񳧏~�i|�yv{3{pZ]ގ�� ~��� =�FZ�)��]HK�6�1ѵ�����𯪄0 X���˘͵�'����^i�����_����� �F������[�j�qEc����a�Ҷ�t��܏b �E��{)���\Uc$ߑ��YFLג��+>zv��Ϥ�&e����\KWO���������&�#�H�`�},����w�L ?�m��&9���O�SFKވKolZL�]�`5�� h��Wx�x�aڍ�7�,G߉����?�������+ѿ�Yg���bF(iJ�9���� ��8�>Us�G]���3�A�:T�Cبv~^r f����9�X��O^]�adtL���ւ6�O·մ� g[PDF�M�N�����hnr'�[��F� Y�Yu=�:��b����k >~z����5�fDmX2kC)���%��۶R�^I�A�w}:�-6�hm������ V�l�h��eii�̿&���9�����޸ca�h > stream xڍ�Oo�@���ˡR9����?EZ۱�� qM�M��ؑ�H�OϾ�I !��'of����7o>݇�]���;���ua�aw nn걽�0t�s����{�i�{7��j[o�~~��ۡ=\:wu��T��~x��u�྆?N�p;�}·}8�p�����4Z���[�í�E�ˢh�7��qx��;��/l���X�,$�Z\���C7Ir��=��og����ѓC���yv�����J->����L�����sS? kC�\BcqU)tL��$��1:�z �����f�9��_pom�Һ�s�Sq����5f��{Q7��$��G��\c01 ,�?��ߊ���;)i�����p�Q7�3��Ӱ�n��DO# > stream xڍ�Oo�F���ۃ��h����%�|h�F��L�\)PP��w���n�"������웟(�����g����/�u�|���?Gww��]O~\>{������'�u��G����}��C0?��������M��w �Q�O�G����/��~���!^L �V�hc�l*�� �� i($���2��q� ���&�� ��B�p48��AG���[Z8v�������[�:��� hq- � +�u��0�I����Ӊ�FxL���B����^�g>�����^]��.��w�퍌�/.�&2�Kendstream endobj 568 0 obj > stream xڅTMo�@��+��H����"�ˇd�M�ڪz�X�H6 ������{v"UUr0�f� ��|y�L��}q��^��nh�}�&�]���dmy>�f|t�r������ܸQݦ�l�����My a_Ko� ?6�s�a��u?�~,� t�̠��X���H�GX1�cR��cě��)x�zpL`�`��k�ԡ3����\�`�y!~B��!�т��EF���,�(y�����g�7�l9�( ��䖠7��Zp(u��x!��[�1��,�!f�y�z"Y%�*��p���i/(�X�o1���cɖu��9r+��W�_66��,�}O{�O�e�_u�������?>a�'� �/��tendstream endobj 569 0 obj > stream xڍU�n�@���C���z,�ei#�Цj���ʁM�d[j���73N���z���x3�恗�7��W��*y��g�������~����=�p��}�����4���?��궾��� ��ù���E����Q�����tX�Ӳ��χ��:��0�����4j��!���\���FQ�/~^�qx��;�u vCW�G�Dk�֗M����żz�*��U׷'�E��1������䏷��m6j|��i~� �F뻹�s? �h�U���F~���f.y|�hwA�����F�ba*)���$11�� 8m����L�9u�G���ar(*���G�55��P�h M��M.iдAIS1����=���=����~��Q0���.o����� 8!\#}C�1:e��;��ks`� i.�47�lK�x�l��m͘��]5)�m�g��S�-� }�����?����gT��3�^�����?#=���`�9�/�gB��3�Z�i��vq�3vt�s ,9�Fr�,'9�q�3�:�y:ə4�3vw�3�;�>��ݝ�ܜ�L�%g�w�3|��3斒3��� })9�O)9S���Kə��3�-�k �9�]J����?��R�5�L=��_R���g�ȡf>M���!�󵺑ӃN �x%���y��YN� :�q��}�L�*��;���į�&�'X�endstream endobj 594 0 obj > endobj 566 0 obj > stream x��Xے�}�W���Xč�\���:�/�ݲw�R�-i`Q�‹�3_�n�"4��%3�}9}�I���Lʌ(���$#,�'�"�$,���ɸ�����yFrw�9�4#��WpU2� ��א��I��p��D����[���&lߌCm4w��Ba��ٹ#�c>ڡ���a��ݔ �;s�M7��3��lʋ�)�#��csK1YҤ��w����l��̒�(�y3\�d���[:w����ܟ=:�,A�xdy�ҥĊ�%�h����H3���8>��J~t"�Ή?���� ��D%-0�(6�8% {�V�? e7�d�!�>�Ƕm����� �R�Rz)e ��R�A��T���[��{{�����+ �2'�Ӝ���P670g~���w���í���-���m_�;�� 3���]]����?�;�A�T DNH��>���q�P���[�P��m���nme����L�֨�w�1]98'J\��ﺋ4AU{_6`��GS ���yfS�� @�ͩl*F�p,AF���p��5:�{�5��� �La4�1���e �0�X���n �D�m}���QŸ�{c��=�.4�T~kG8#`��Y�a�������vw_?�q�#4Lc��z��B�H���λ�sD:��ei�4���G��l��ln�X��o��c��~�֡s�ĩ�a�E�ض�L��g3n��|�8n�n+�v~�Pmڦ��Ε��>z035��چh�p�ں�=�6���Eb�}��9�=,� q�iD�# �{�Q {�E5L�1�v|(���Е���ĕ��mi],�68�@*f����p��q��)���� 6���Ijҙ�����ƾ�i糴e����u�=v0��f�Xa0�`�-�(.q�f:�e��v�x�2���XP�����#�K�&�f��+*��G�X��o�a� zPI������`��g,1� �4��b��b!�S�� 1 �x ���)��p�ń.|� [� ,†*��"B� )�{����9� �|=E$>��H�}���r^C��aE�{��W�M�, �)V��9:^^����+���^����V����7��7��U����ӷ��y�n��}o�>��9��+}&���b���endstream endobj 595 0 obj ] /Index [ 0 596 ] /Info 594 0 R /Length 1542 /Root 593 0 R /Size 596 /W [ 1 3 1 ] >> stream x�%�glWU������̲G m� �e)Ph���B'��Ԗ�[�6j�c�&j4*jQq�AP�@P����ĸ�����7� �@F& ����ݺ�#Oe�h��-�ρ���@VY`m�5\����h����z� I�p?� n�!n潷����,��� ],zi�.y�؜%B�[V���[[�K�m�x%�����kc�>a�WbW��V���_%��%�n�֊�ܱ�y��o⭯�;׈�w��A�r�/��$>�G�ᥰS��Gsı+�q�w�q��h�9T��c��U�F�=-��z �d ���;��(ȥ-�,1�/�n�4���?��p���A��)�H��2-��zuj�8���zX�Ig����+ğ�ߟ��9��A�W��{������oq�q�.Rfj�,�(�ƈ��rVc���_���l�I+tIQ'1�ILy@��7��4�6-����E)/�9>e E�r�*�m�r�����]���j�&Ԫ��=�^Q�`� e=P�e=P�e=P|�D�7���_\�L����?����,u2P@�3�0�W�'J�"��Aw�x�O� h(�"�jq� ���7��k*����aOr��I��]�{D��Tl��g�g����bc�x^��^�ҵ����M����"�����qb��ж�mߋW�۵����^�';v��rX�P�0�l�^��*�6��&�A�4;�g`B6�C���M�U)��N��.�_G։��@׎i������;��*�_��r�V|������̾���XB+��6���|�`_�f'7�ža�R:9c`$��������m�C�ɖ(���R �Z���m� Z�"B4.�l�s~���+��P %0��t���f�o� P�PK`6�1vx����h��+y4��u=��x�F�X�ս-����UYRj�GՖ4�*�j,iR�G��T��Q�%'��Q�%?�ãE���$�G��z�HӖ�H�O �#�yJ�D~O1f������QMT(s�҅�y�:�����詐a�Cd�`C! �!�������v8�����V�N4����endstream endobj 1 0 obj > endobj 3 0 obj > endobj 4 0 obj > endobj 5 0 obj (��Introduction) endobj 6 0 obj > endobj 7 0 obj > endobj 8 0 obj > endobj 9 0 obj (��Data) endobj 10 0 obj > endobj 11 0 obj > endobj 12 0 obj > endobj 13 0 obj (��Methods and Findings) endobj 14 0 obj > endobj 15 0 obj > endobj 16 0 obj > endobj 17 0 obj (��RQ1: How has the percentage of scientific and scholastic research that engages with AI changed over time?) endobj 18 0 obj > endobj 19 0 obj > endobj 20 0 obj > endobj 21 0 obj (��Methods) endobj 22 0 obj > endobj 23 0 obj > endobj 24 0 obj > endobj 25 0 obj (��Findings) endobj 26 0 obj > endobj 27 0 obj > endobj 28 0 obj > endobj 29 0 obj (��RQ2: To what extent has AI engagement diffused within individual fields?) endobj 30 0 obj > endobj 31 0 obj > endobj 32 0 obj > endobj 33 0 obj (��Methods) endobj 34 0 obj > endobj 35 0 obj > endobj 36 0 obj > endobj 37 0 obj (��Findings) endobj 38 0 obj > endobj 39 0 obj > endobj 40 0 obj > endobj 41 0 obj (��RQ3: How is changing ubiquity of AI engagement associated with changes in the semantic character of fields?) endobj 42 0 obj > endobj 43 0 obj > endobj 44 0 obj > endobj 45 0 obj (��Methods) endobj 46 0 obj > endobj 47 0 obj > endobj 48 0 obj > endobj 49 0 obj (��Findings) endobj 50 0 obj > endobj 51 0 obj > endobj 52 0 obj > endobj 53 0 obj (��Discussion and Concluding Remarks) endobj 54 0 obj > endobj 55 0 obj > endobj 56 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 228.426 192.963 268.223 203.964 ] >> endobj 57 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 272.716 191.07 315.328 203.964 ] >> endobj 58 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 319.82 192.963 363.141 203.964 ] >> endobj 59 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 367.634 192.963 409.608 202.832 ] >> endobj 60 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 74.636 127.223 116.943 138.224 ] >> endobj 61 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 351.58 110.902 393.897 121.789 ] >> endobj 62 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 398.016 110.787 447.61 121.789 ] >> endobj 63 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 451.728 108.895 484.004 120.656 ] >> endobj 64 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 488.123 110.967 517.912 120.656 ] >> endobj 65 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 162.898 755.807 199.772 768.701 ] >> endobj 66 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 305.24 757.7 360.91 768.701 ] >> endobj 67 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 408.452 757.7 453.344 768.701 ] >> endobj 68 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 99.236 741.265 144.771 752.266 ] >> endobj 69 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 414.63 741.265 448.805 751.134 ] >> endobj 70 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 452.495 741.265 485.644 751.134 ] >> endobj 71 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 489.334 738.423 517.912 751.134 ] >> endobj 72 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 139.796 724.83 188.866 735.831 ] >> endobj 73 0 obj > /ProcSet [ /PDF /Text ] >> endobj 74 0 obj > stream xڥ�vܸ��)���1 n�$�e{��l�2>�:�I��l��Ŗ�>U(�"exFϹD�P���^��X�_���J&A�2�*�� �ʘRq�F��" ���&��ԛ��y�7Z�m�8,��ks{�&�FJ�H��V$,�ܥ �K9��m�V{G�ˉ��C3l�Q�C*��amd�]����*�n��J���]W=J�rV��� jq��T��Vr�%1���8N�]���H�W* ��2�k�`�R�dI�.������di�W����&N������G'J[��I�3��8���(�1z�T�{G�{^���q�X��?l.�f�O���;xYA���+�e)J���x���3����1�9gi����!r ��I:��"Y�9�]��Xt����6�Pa'�1��i�x��� _ y�o�F�_���yQOص� ŀ���9��=b*�׺�'����01�(��}���]^6|f��Z�A���O���ݍ��� P�T�8�p��TFΔ�~�rxR��Ť�u*u�\ �!E��c��G��'�A�5f�ƒB AC!�̳'W鞮�ua��*F��Jk�0M��7͞Nǝ�8��{����ʵA7�A�۩g�Šk�iS�(({S"� �����r����*p'�s&`3�”F���1IB}_Mc�%�taL ���l��]� A�!Z�љ�B��R�#�6IZ��a���T������6hrr�-��E�?;V`��x���v�ME90����� �2������ �nк�CVm]Q0blE����p�9�f�J����'Q]�LP����dJ���Ep�=�qSR>ava���ca\����i?����u�I9�C��'�dqx"|2�$�$�cԬ#NX� ��k=A$lng�#R�q���5���O=rz���l��D�ToF�֕!�4� 6�H��KG*0�2�T��-wR ��4��5�C���|��� �9�#� ��� .��%���:$��u ��d�$��v5�"�r|�����c5#5D����TӴ������$~�B��{ը�i2#�ԈO47T"�"?d0 �2W-:��r�����X�����بď�V ��dl; ��މ@/�j՝�|��D��[�Ǿ��t�1_��x���d+ ��¦�C�~$\�{:��h{ �\�&��y��� �!�{����u|@:u��*�zqGY��Mԛ��0�L�l��\� �x�'� @4bi������! �̳��:Q�a*H���~�U�,�xLS��X�X&� �e>cy�(��2�,�N�mG�YQ����B� � ���=;x~�ep��_�� > endobj 76 0 obj > endobj 77 0 obj > endobj 78 0 obj > endobj 79 0 obj > endobj 80 0 obj > endobj 81 0 obj > endobj 82 0 obj > endobj 83 0 obj > endobj 84 0 obj > endobj 85 0 obj > endobj 86 0 obj > endobj 87 0 obj > endobj 88 0 obj > endobj 89 0 obj > endobj 90 0 obj > endobj 91 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 476.184 642.654 517.912 653.655 ] >> endobj 92 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 431.109 560.478 473.415 571.48 ] >> endobj 93 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 476.352 560.478 521.545 570.347 ] >> endobj 94 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 71.004 544.043 115.302 553.912 ] >> endobj 95 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 118.427 544.043 146.623 554.245 ] >> endobj 96 0 obj > /ProcSet [ /PDF /Text ] >> endobj 97 0 obj > endobj 99 0 obj > endobj 100 0 obj > endobj 101 0 obj > endobj 102 0 obj > endobj 103 0 obj > endobj 104 0 obj > endobj 105 0 obj > endobj 106 0 obj > endobj 107 0 obj > endobj 108 0 obj > endobj 109 0 obj > endobj 110 0 obj > endobj 111 0 obj > endobj 112 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 74.636 541.921 119.517 552.923 ] >> endobj 113 0 obj > /Border [ 0 0 1 ] /C [ 1 0 0 ] /H /I /Rect [ 397.872 489.774 404.228 502.818 ] >> endobj 114 0 obj > /Border [ 0 0 1 ] /C [ 0 1 1 ] /H /I /Rect [ 258.724 90.845 491.874 103.796 ] >> endobj 115 0 obj > /Border [ 0 0 1 ] /C [ 1 0 0 ] /H /I /Rect [ 301.523 341.858 307.879 354.901 ] >> endobj 116 0 obj > /Border [ 0 0 1 ] /C [ 0 1 1 ] /H /I /Rect [ 295.01 79.886 452.345 92.837 ] >> endobj 117 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 475.758 295.394 520.639 306.396 ] >> endobj 118 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 316.652 211.326 358.495 223.088 ] >> endobj 119 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 171.168 708.395 211.62 719.396 ] >> endobj 120 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 410.632 708.395 444.381 718.264 ] >> endobj 121 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 95.507 675.524 135.959 686.526 ] >> endobj 122 0 obj > /Border [ 0 0 1 ] /C [ 1 0 0 ] /H /I /Rect [ 515.188 656.247 521.545 669.291 ] >> endobj 123 0 obj > /Border [ 0 0 1 ] /C [ 0 1 1 ] /H /I /Rect [ 276.02 77.796 461.598 90.748 ] >> endobj 124 0 obj > /ProcSet [ /PDF /Text ] >> endobj 125 0 obj > endobj 127 0 obj > endobj 128 0 obj > endobj 129 0 obj > endobj 130 0 obj > endobj 131 0 obj > endobj 132 0 obj > endobj 133 0 obj > endobj 134 0 obj > endobj 135 0 obj > /Border [ 0 0 1 ] /C [ 1 0 0 ] /H /I /Rect [ 402.523 459.026 426.334 472.069 ] >> endobj 136 0 obj > /Border [ 0 0 1 ] /C [ 1 0 0 ] /H /I /Rect [ 118.593 426.156 126.04 439.199 ] >> endobj 137 0 obj > /ProcSet [ /PDF /Text ] >> endobj 138 0 obj > endobj 140 0 obj > endobj 141 0 obj > endobj 142 0 obj > endobj 143 0 obj > endobj 144 0 obj > endobj 145 0 obj > endobj 146 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 329.311 626.219 359.099 636.088 ] >> endobj 147 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 362.841 626.219 392.794 636.088 ] >> endobj 148 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 350.966 576.914 375.998 586.782 ] >> endobj 149 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 380.047 576.914 409.301 586.782 ] >> endobj 150 0 obj > /Border [ 0 0 1 ] /C [ 1 0 0 ] /H /I /Rect [ 448.304 409.72 454.66 422.764 ] >> endobj 151 0 obj > /Border [ 0 0 1 ] /C [ 0 1 1 ] /H /I /Rect [ 225.315 101.534 472.089 114.486 ] >> endobj 152 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 262.637 396.127 309.771 405.996 ] >> endobj 153 0 obj > /Border [ 0 0 1 ] /C [ 1 0 0 ] /H /I /Rect [ 222.813 343.98 230.26 357.023 ] >> endobj 154 0 obj > /Border [ 0 0 1 ] /C [ 1 0 0 ] /H /I /Rect [ 338.705 294.675 345.062 307.718 ] >> endobj 156 0 obj > /Border [ 0 0 1 ] /C [ 1 0 0 ] /H /I /Rect [ 192.238 138.891 199.685 151.934 ] >> endobj 157 0 obj > /ProcSet [ /PDF /Text ] >> endobj 158 0 obj > endobj 160 0 obj > endobj 161 0 obj > endobj 162 0 obj > endobj 163 0 obj > endobj 164 0 obj > endobj 165 0 obj > endobj 166 0 obj > endobj 167 0 obj > endobj 168 0 obj > endobj 169 0 obj > endobj 170 0 obj > endobj 171 0 obj > endobj 172 0 obj > endobj 173 0 obj > endobj 175 0 obj > /Border [ 0 0 1 ] /C [ 1 0 0 ] /H /I /Rect [ 178.291 480.336 185.738 493.38 ] >> endobj 177 0 obj > /ProcSet [ /PDF /Text ] /XObject > >> endobj 178 0 obj > endobj 180 0 obj > endobj 181 0 obj > endobj 182 0 obj > endobj 183 0 obj > /FirstChar 0 /FontBBox [ -1021 -463 1794 1233 ] /FontDescriptor 184 0 R /FontMatrix [ 0.001 0 0 0.001 0 0 ] /LastChar 255 /Name /BMQQDV+DejaVuSans /Widths 186 0 R >> endobj 184 0 obj > endobj 185 0 obj > endobj 186 0 obj [ 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 318 401 460 838 636 950 780 275 390 390 500 838 318 361 318 337 636 636 636 636 636 636 636 636 636 636 337 337 838 838 838 531 1000 684 686 698 770 632 575 775 752 295 295 656 557 863 748 787 603 787 695 635 611 732 684 989 685 611 685 390 337 390 838 500 500 613 635 550 635 615 352 635 634 278 278 579 278 974 634 612 635 635 411 521 392 634 592 818 592 592 525 636 337 636 838 600 636 600 318 352 518 1000 500 500 500 1342 635 400 1070 600 685 600 600 318 318 518 518 590 500 1000 500 1000 521 400 1023 600 525 611 318 401 636 636 636 636 337 500 500 1000 471 612 838 361 1000 500 500 838 401 401 500 636 636 318 500 401 471 612 969 969 969 531 684 684 684 684 684 684 974 698 632 632 632 632 295 295 295 295 775 748 787 787 787 787 787 838 787 732 732 732 732 611 605 630 613 613 613 613 613 613 982 550 615 615 615 615 278 278 278 278 612 634 612 612 612 612 612 838 612 634 634 634 634 592 635 592 ] endobj 217 0 obj > endobj 218 0 obj > /FirstChar 0 /FontBBox [ -1021 -463 1794 1233 ] /FontDescriptor 219 0 R /FontMatrix [ 0.001 0 0 0.001 0 0 ] /LastChar 255 /Name /BMQQDV+DejaVuSans /Widths 221 0 R >> endobj 219 0 obj > endobj 220 0 obj > endobj 221 0 obj [ 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 318 401 460 838 636 950 780 275 390 390 500 838 318 361 318 337 636 636 636 636 636 636 636 636 636 636 337 337 838 838 838 531 1000 684 686 698 770 632 575 775 752 295 295 656 557 863 748 787 603 787 695 635 611 732 684 989 685 611 685 390 337 390 838 500 500 613 635 550 635 615 352 635 634 278 278 579 278 974 634 612 635 635 411 521 392 634 592 818 592 592 525 636 337 636 838 600 636 600 318 352 518 1000 500 500 500 1342 635 400 1070 600 685 600 600 318 318 518 518 590 500 1000 500 1000 521 400 1023 600 525 611 318 401 636 636 636 636 337 500 500 1000 471 612 838 361 1000 500 500 838 401 401 500 636 636 318 500 401 471 612 969 969 969 531 684 684 684 684 684 684 974 698 632 632 632 632 295 295 295 295 775 748 787 787 787 787 787 838 787 732 732 732 732 611 605 630 613 613 613 613 613 613 982 550 615 615 615 615 278 278 278 278 612 634 612 612 612 612 612 838 612 634 634 634 634 592 635 592 ] endobj 266 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 74.636 412.972 105.145 422.841 ] >> endobj 267 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 118.725 396.717 163.132 406.406 ] >> endobj 268 0 obj > /Border [ 0 0 1 ] /C [ 1 0 0 ] /H /I /Rect [ 342.626 730.749 350.073 743.792 ] >> endobj 270 0 obj > /ProcSet [ /PDF /Text ] >> endobj 271 0 obj > endobj 273 0 obj > endobj 274 0 obj > endobj 275 0 obj > endobj 276 0 obj > endobj 277 0 obj > endobj 278 0 obj > endobj 279 0 obj > endobj 280 0 obj > /Border [ 0 0 1 ] /C [ 1 0 0 ] /H /I /Rect [ 396.09 495.563 403.537 508.606 ] >> endobj 282 0 obj > /ProcSet [ /PDF /Text ] /XObject > >> endobj 283 0 obj > endobj 285 0 obj > endobj 286 0 obj > endobj 287 0 obj > endobj 288 0 obj > endobj 289 0 obj > /FirstChar 0 /FontBBox [ -1021 -463 1794 1233 ] /FontDescriptor 291 0 R /FontMatrix [ 0.001 0 0 0.001 0 0 ] /LastChar 255 /Name /BMQQDV+DejaVuSans /Widths 293 0 R >> endobj 291 0 obj > endobj 292 0 obj > endobj 293 0 obj [ 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 318 401 460 838 636 950 780 275 390 390 500 838 318 361 318 337 636 636 636 636 636 636 636 636 636 636 337 337 838 838 838 531 1000 684 686 698 770 632 575 775 752 295 295 656 557 863 748 787 603 787 695 635 611 732 684 989 685 611 685 390 337 390 838 500 500 613 635 550 635 615 352 635 634 278 278 579 278 974 634 612 635 635 411 521 392 634 592 818 592 592 525 636 337 636 838 600 636 600 318 352 518 1000 500 500 500 1342 635 400 1070 600 685 600 600 318 318 518 518 590 500 1000 500 1000 521 400 1023 600 525 611 318 401 636 636 636 636 337 500 500 1000 471 612 838 361 1000 500 500 838 401 401 500 636 636 318 500 401 471 612 969 969 969 531 684 684 684 684 684 684 974 698 632 632 632 632 295 295 295 295 775 748 787 787 787 787 787 838 787 732 732 732 732 611 605 630 613 613 613 613 613 613 982 550 615 615 615 615 278 278 278 278 612 634 612 612 612 612 612 838 612 634 634 634 634 592 635 592 ] endobj 316 0 obj > endobj 317 0 obj > /FirstChar 0 /FontBBox [ -1021 -463 1794 1233 ] /FontDescriptor 318 0 R /FontMatrix [ 0.001 0 0 0.001 0 0 ] /LastChar 255 /Name /BMQQDV+DejaVuSans /Widths 320 0 R >> endobj 318 0 obj > endobj 319 0 obj > endobj 320 0 obj [ 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 318 401 460 838 636 950 780 275 390 390 500 838 318 361 318 337 636 636 636 636 636 636 636 636 636 636 337 337 838 838 838 531 1000 684 686 698 770 632 575 775 752 295 295 656 557 863 748 787 603 787 695 635 611 732 684 989 685 611 685 390 337 390 838 500 500 613 635 550 635 615 352 635 634 278 278 579 278 974 634 612 635 635 411 521 392 634 592 818 592 592 525 636 337 636 838 600 636 600 318 352 518 1000 500 500 500 1342 635 400 1070 600 685 600 600 318 318 518 518 590 500 1000 500 1000 521 400 1023 600 525 611 318 401 636 636 636 636 337 500 500 1000 471 612 838 361 1000 500 500 838 401 401 500 636 636 318 500 401 471 612 969 969 969 531 684 684 684 684 684 684 974 698 632 632 632 632 295 295 295 295 775 748 787 787 787 787 787 838 787 732 732 732 732 611 605 630 613 613 613 613 613 613 982 550 615 615 615 615 278 278 278 278 612 634 612 612 612 612 612 838 612 634 634 634 634 592 635 592 ] endobj 367 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 282.588 511.099 316.762 520.968 ] >> endobj 368 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 320.609 511.279 361.809 520.968 ] >> endobj 369 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 365.655 511.099 409.724 520.968 ] >> endobj 370 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 413.57 511.279 441.625 520.968 ] >> endobj 371 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 445.471 511.279 486.256 520.968 ] >> endobj 372 0 obj > /Border [ 0 0 1 ] /C [ 1 0 0 ] /H /I /Rect [ 417.794 376.776 425.242 389.82 ] >> endobj 373 0 obj > /ProcSet [ /PDF /Text ] >> endobj 374 0 obj > endobj 376 0 obj > endobj 377 0 obj > endobj 378 0 obj > endobj 379 0 obj > endobj 380 0 obj > endobj 382 0 obj > /Border [ 0 0 1 ] /C [ 1 0 0 ] /H /I /Rect [ 221.258 394.224 228.705 407.268 ] >> endobj 383 0 obj > /Border [ 0 0 1 ] /C [ 1 0 0 ] /H /I /Rect [ 283.738 394.224 291.185 407.268 ] >> endobj 384 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 461.588 282.02 486.883 291.889 ] >> endobj 385 0 obj > /Border [ 0 0 1 ] /C [ 0 1 0 ] /H /I /Rect [ 489.901 282.2 517.912 291.889 === The Rapid Adoption of Generative AI | St. Louis Fed (https://www.stlouisfed.org/on-the-economy/2024/sep/rapid-adoption-generative-ai) === Skip to Main Search Explore Our Apps FRED Tools and resources to find and use economic data worldwide. FRASER U.S. Financial, economic, and banking history. ALFRED Vintages of economic data from specific dates in history. CASSIDI View banking market concentrations and perform HHI analysis. Search Toggle navigation About About the St. Louis Fed Contact Us Careers Economy Museum News Releases Our Organization Vision, Mission & Values Our History Annual Reports The Federal Reserve System Tutorial: The Fed Explained FOMC Voting Rotation Fed FAQs Our Branch Offices Little Rock Branch Louisville Branch Memphis Branch Our Leadership Board of Directors Executive Leadership About President & CEO Alberto Musalem Publications Publications & Podcasts Search All Publications List: Publications at a Glance Educational Explainers Open Vault Blog Everyday econ and the Fed explained Page One Economics Personal finance and econ basics from education experts Economic Analysis On the Economy Blog Insights from our economists and other experts FRED Blog Data storytelling using interactive charts Review Scholarly research on monetary policy, macroeconomics, and more Access to Credit and Financial Services Podcasts Teach Economics Timely Topics Hear fresh perspectives in research Popular Topics Inflation Trade Labor Markets & Unemployment Subscribe to our In Focus newsletter Research Economic Research FRED Economic Data Our Economists Our Research Associates Careers in Research View Our Research On the Economy Blog Insights from our economists FRED Blog Data storytelling using interactive charts Review Scholarly research on monetary policy, macroeconomics, and more Working Papers Preliminary, cutting-edge research Fed in Print Research from all Federal Reserve Banks Local Economy Population, Employment & Housing Data Eighth District Beige Book Local industry trends and economic conditions Fed & U.S. Economic History Federal Reserve History FRASER Digital Library U.S. financial, economic, and banking documents Communities Working with Communities Request a Speaker Attend an Event Regional Engagement Advisory Councils Dialogue with the Fed Popular lecture series with Q&A Your Fed, Your Voice Listening and serving as an active partner in communities big and small Community Development Subscribe to Our Bridges Newsletter About Community Development Community Development Publications and Reports Community Development Staff Featured Resources Disconnected Young Adults in Our Region Bank On National Data Hub Investing in Rural Prosperity FedCommunities.org Education Economic Education About Our Education Team Meet our team of education specialists Educator Advisory Boards Student Board of Directors Native Economic and Financial Education Empowerment (NEFEE) Education for Everyone Page One Economics® Economics and personal finance explained by educational experts Tutorial: The Fed Explained How the Federal Reserve supports a strong and resilient economy FederalReserveEducation.org Teaching Resources For economics and personal finance K-12 Economic Education Events Professional development and other happenings for teachers Glossary Explore FRE.org: Resources for teachers Banking Supervising Financial Institutions Consumer Resources Federal Reserve Consumer Help File a Complaint Banker Resources Appeals Process Ask the Fed® Board of Director Training (PDF) Community Banking Connections Community Bank Research Conference Discount Window Supervising Financial Institutions Contacts Consumer Affairs & CRA Supervision Current State Member Banks (PDF) Holding Company Supervision Safety and Soundness Supervision & Regulation Payment & Financial Services FedNow Financial Services Reserves & Reporting Master Account Coordination Financial and Regulatory Reporting Reserves Administration Structure Reporting Applications Become a Member Bank Application Filings and Notices Explore Our Apps FRED Tools and resources to find and use economic data worldwide FRASER U.S. Financial, economic, and banking history ALFRED Vintages of economic data from specific dates in history CASSIDI View banking market concentrations and perform HHI analysis Search Home > On the Economy Blog > The Rapid Adoption of Generative AI September 23, 2024 By Alexander Bick , Adam Blandin , David Deming SHARE THIS PAGE: Link Copied Generative artificial intelligence (AI) has rapidly emerged as a potentially transformative workplace technology. The large language model (LLM) ChatGPT debuted in November 2022, and by March 2024 the most common generative AI tools were being accessed by hundreds of millions of users each month. While several studies have found that generative AI can result in sizeable productivity gains for workers, firms’ adoption of the technology in production has been reported to be fairly low. See, for example, Aakash Kalyani and Marie Hogan’s April 2024 On the Economy blog post, “ AI and Productivity Growth: Evidence from Historical Developments in Other Technologies .” The overall effect of generative AI on the economic landscape hinges on how many people adopt the new technology and how intensively they use it. In this blog post, we present results from the first nationally representative U.S. survey of generative AI adoption at work and at home. Our data come from the Real-Time Population Survey (RPS), a nationwide survey that asks the same core questions and follows the same timing and structure as the Current Population Survey (CPS), the monthly labor force survey conducted by the U.S. Census Bureau for the Bureau of Labor Statistics. We have used the RPS previously to study the rise in work from home since the onset of the COVID-19 pandemic. See, for example, the June 2024 On the Economy blog post “ The Impact of Work from Home on Interstate Migration in the U.S. ” How Prevalent Is Generative AI Adoption? As shown in the figure below, the RPS revealed that in August 2024 almost 40% of the U.S. population ages 18 to 64 used generative AI to some degree, and almost 1 in 3 respondents said they used it daily or at least once but not every day during the week prior to the survey. Usage at home was somewhat more prevalent than at work (32.6% versus 28.1%, respectively), but daily usage was less frequent at home than at work (6.4% versus 10.6%, respectively). Use of Generative AI at Work and at Home, August 2024 SOURCES: Real-Time Population Survey and authors’ calculations. NOTES: The figure shows the share of RPS respondents who used generative AI for work, outside of work and overall (either for work or outside of work). Intensity of use is broken down into every day in the week before the survey, at least one day in the week before the survey but not every day, and not used in the week before the survey. Data are from the August 2024 wave of the RPS and for respondents ages 18 to 64. The “for work” sample includes only employed individuals (N=3216); the other samples include all respondents (N=4682). In our working paper on this topic , we show that generative AI use is more common among individuals who are male, younger and more educated and who work in computer, math and management occupations. However, perhaps the most important finding is that generative AI use is widespread across gender, age, education, industries and occupations. How Does Adoption of Generative AI Compare with That of Other Technologies? The next figure compares the speed of adoption of generative AI with that of two other transformative technologies: personal computers (PCs) and the internet. We plot the adoption rate by year since the first mass market version of each technology. The first mass market computer was the IBM PC, which was released in August 1981 and sold more than a million units. We date mass market availability of the internet to April 1995, when the National Science Foundation (NSF) decommissioned NSFNET and allowed the internet to carry commercial traffic. The first generative AI model to eventually sell at least one million subscriptions (ChatGPT) was released in November 2022, roughly two years before the date of our survey data. Adoption Rate of Generative AI at Work and Home versus the Rate for Other Technologies * SOURCES: Real-Time Population Survey, Current Population Survey, International Telecommunication Union and authors’ calculations. NOTES: The figure shows usage rates at work for three technologies: generative AI, PCs and the internet. The horizontal axis represents the number of years since the introduction of the first mass market product for each technology. AI usage data are from the August 2024 wave of the RPS. PC usage data are from the 1984-2003 Computer and Internet Use Supplement of the CPS. We plot two estimates of internet use: one from the 2001-09 Computer and Internet Use Supplement of the CPS and one with 1995-2021 data from the International Telecommunication Union (ITU). The samples from the RPS and CPS include all individuals ages 18 to 64. The RPS sample size is 4,682. The sample from the ITU includes individuals of all ages. The figure’s data are available for download . The black dot in the figure above repeats the 39.4% adoption rate for generative AI reported in the first figure. Generative AI has been adopted at a faster pace than PCs or the internet. Adoption of the PC three years after its mass introduction was only at 20%, about the same value as for adoption of the internet after two years. For What Tasks Do People Use Generative AI? The RPS also asked respondents about how they used generative AI. Respondents who indicated that they had used generative AI in the last week were presented with the list of 10 tasks (plus an “other” category) shown in the following figure. They were then asked to select any of the tasks for which they used generative AI at work during that period. (In our working paper, we show similar results for generative AI usage at home.) Types of Tasks for Which U.S. Workers Are Using Generative AI, August 2024 SOURCES: Real-Time Population Survey and authors’ calculations. NOTES: The figure shows the share of AI users employing it for specific tasks at work. Data are from the August 2024 wave of the RPS and for respondents ages 18 to 64. The sample includes only employed individuals (N=3216). The figure’s data are available for download . In a nutshell, respondents used generative AI for a wide range of tasks, with usage rates at or exceeding 25% for all 10 defined tasks. The survey also asks respondents to rank the tasks for which they used generative AI in order of how helpful the technology was in completing the task. Remarkably, the rankings were fairly evenly distributed overall, with eight of the 10 defined tasks being ranked among their top two selections by at least 10% of respondents. How Much Could Generative AI Increase Labor Productivity? Finally, we examined how intensely respondents employed generative AI on days that they reported using it. This allowed us to estimate a lower and upper bound for the share of total work hours that involve generative AI for the U.S. economy as a whole. We estimated that between 0.5% and 3.5% of all work hours in the U.S. are currently assisted by generative AI. Combining these estimates with a median increase of 25% in task productivity from the adoption of generative AI, an increase consistent with that observed in several studies, See, for example, the 2023 study “ Experimental Evidence on the Productivity Effects of Generative Artificial Intelligence ” by Shakked Noy and Whitney Zhang; the 2023 study “ Generative AI at Work ” by Erik Brynjolfsson, Danielle Li and Lindsey R. Raymond; the 2023 study “ Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality ” by Fabrizio Dell’Acqua et al.; the 2024 study “ The Effects of Generative AI on High-Skilled Work: Evidence from Three Field Experiments with Software Developers ” by Zheyuan (Kevin) Cui, Mert Demirer, Sonia Jaffe, Leon Musolff, Sida Peng and Tobias Salz; and the 2023 study “ The Impact of AI on Developer Productivity: Evidence from GitHub Copilot ” by Sida Peng, Eirini Kalliamvakou, Peter Cihon and Mert Demirer. we estimated that generative AI could plausibly grow labor productivity by between 0.1% and 0.9% at current levels of usage. Is Generative AI a General-Purpose Technology? Generative AI has rapidly emerged as an important new technology, yet its impact on the U.S. economy depends critically on the speed and intensity of its adoption. We show that two years after the mass introduction of generative AI, its adoption rate already exceeds that of the PC and internet at comparable points in time. The wide variety of tasks for which people use generative AI, combined with the widespread adoption of generative AI across many demographics and occupations, suggest that it is a general-purpose technology. * Editor’s Note: The second figure title was updated Oct. 10, 2024, to include “and home.” Notes See, for example, Aakash Kalyani and Marie Hogan’s April 2024 On the Economy blog post, “ AI and Productivity Growth: Evidence from Historical Developments in Other Technologies .” See, for example, the June 2024 On the Economy blog post “ The Impact of Work from Home on Interstate Migration in the U.S. ” See, for example, the 2023 study “ Experimental Evidence on the Productivity Effects of Generative Artificial Intelligence ” by Shakked Noy and Whitney Zhang; the 2023 study “ Generative AI at Work ” by Erik Brynjolfsson, Danielle Li and Lindsey R. Raymond; the 2023 study “ Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality ” by Fabrizio Dell’Acqua et al.; the 2024 study “ The Effects of Generative AI on High-Skilled Work: Evidence from Three Field Experiments with Software Developers ” by Zheyuan (Kevin) Cui, Mert Demirer, Sonia Jaffe, Leon Musolff, Sida Peng and Tobias Salz; and the 2023 study “ The Impact of AI on Developer Productivity: Evidence from GitHub Copilot ” by Sida Peng, Eirini Kalliamvakou, Peter Cihon and Mert Demirer. ABOUT THE AUTHORS Alexander Bick Alexander Bick is an economist and senior economic policy advisor at the Federal Reserve Bank of St. Louis. He joined the St. Louis Fed in 2022. Read more about the author and his research . Alexander Bick Alexander Bick is an economist and senior economic policy advisor at the Federal Reserve Bank of St. Louis. He joined the St. Louis Fed in 2022. Read more about the author and his research . Adam Blandin Adam Blandin is an assistant professor of economics at Vanderbilt University. Adam Blandin Adam Blandin is an assistant professor of economics at Vanderbilt University. David Deming David Deming is the Isabelle and Scott Black Professor of Political Economy at Harvard Kennedy School. David Deming David Deming is the Isabelle and Scott Black Professor of Political Economy at Harvard Kennedy School. Related Topics Labor Markets and Unemployment Output Artificial Intelligence Cite This Article × Citation Alexander Bick, Adam Blandin and David Deming, ldquo The Rapid Adoption of Generative AI ,rdquo St. Louis Fed On the Economy , Sept. 23, 2024. Copy Close Subscribe to On the Economy This blog offers commentary, analysis and data from our economists and experts. Views expressed are not necessarily those of the St. Louis Fed or Federal Reserve System. Email Us Media questions All other blog-related questions FOLLOW US SIGN UP FOR EMAIL ALERTS Receive updates in your inbox as soon as new content is published. Sign Up QUICK LINKS About Us Legal Information Contact Us Privacy Policy Careers Doing Business with the Fed Events FRB Services & FedNow Visiting the St. Louis Fed Economy Museum === Notes for Early Indicators (https://www.notion.so/23b1a29f178980e7bb3af34c093d0b45) === David Langer at Lionheart suggested that they might be in a position to help us measure some things. https://x.com/StringChaos/status/1928476388274716707 https://x.com/Simeon_Cps/status/1900218546904293863 Relevant indicators hinted at here: https://www.theinformation.com/articles/openai-says-business-will-burn-115-billion-2029 Example of finding useful data sets: https://mikelovesrobots.substack.com/p/wheres-the-shovelware-why-ai-coding Plan For this sort of project, serious work requires multiple rounds of deep thought. This can’t be done in a compressed period of time, so doing this collaboratively requires an extended multi-turn conversation. I have not had success in getting most people to engage with this in a group email or Slack. The only way I’ve found success is me doing 1:1 engagement with each participant, over email and occasional calls, with a lot of nagging. This is annoying but does seem to work. I’ve witnessed one counterexample (Helen’s CSET workshop) and heard about at least one other (something Sayash worked on); these were both collaborative projects centered around a multi-hour (or multi-day) synchronous group discussion. However, I’m not sure how many new ideas were generated at CSET, as opposed to just aggregating people’s existing thoughts. The CSET workshop did generate some new ideas but also there was a lot of aggregation. Maybe we should just embrace this: perhaps our niche is projects that require lots of multi-turn 1:1 interaction. Phase 2 of this specific project is more about aggregating ideas than deep analysis, so perhaps getting a group of people together on a Zoom to brainstorm together could work. Brainstorming and ideation was, I think, the focus of the two success stories I referred to. We could try to get a bunch of people on a call without being too fussy about exactly which people (though it would be really good to have someone from METR, someone from Epoch, and perhaps also the Forecasting Research Institute). This could be a meeting to put a cap on phase 1 (without necessarily completely locking it down, I’ll continue to get input from a few people I’m only connecting with now) and doing the initial brainstorming for phase 2. I can seed the conversation with a few ideas shared in advance. Note that our measurable indicators don’t need to directly answer a crux, we’re just looking for data that will be helpful in future discussions of those cruxes. → share thoughts with Taren, Ajeya, Helen, ? Ask Helen how far she’s gotten in turning the discussion into a paper and whether she wishes she could do more rounds with at least some folks. Ajeya: Might be productive to do this in two rounds. Do the fuzzy thing, then get someone who’s interested in doing the legwork to do the first round of coming up with concrete experiments – Ajeya might help with this – and take that back to the participants to rate them. Group lists the fuzzy question. Individual – Ajeya or Ryan Greenblatt – proposes a list of concrete experiments Go back to the group to rate the proposals. Realtime, concentrated bursts of time are so much more productive that it’s better to ask someone to come to an event that sounds cool & fun, spend the first hour reading the doc. Or can get people to pre-read more reliably if you get them to agree to come to an event, that’s what she did for the loss-of-control workshop. And keep the writeup short. Siméon (@Simeon_Cps) posted at 5:31 PM on Fri, Aug 15, 2025:I've entertained that theory for a few years and have been confused since then why people expected confidently so much GDP growth. Basically prices if goods should crash so fast that the question of "how do you count inflation" will become the first order parameter of whether and(https://x.com/Simeon_Cps/status/1956514153528742168?t=qJkEFlWoQAD6z6g4HMhNAw&s=03) Notes with Taren, 8/8/25 Post four brainstorming time slots, let people sign up; probably don’t want >4 people/session (though could do breakout rooms) Taren, and probably many people, will do better in a brainstorming session with people with different expertise Could let each group decide which two cruxes they’ll talk about; if we wind up with a gap, do something about it at the end. Or maybe assign cruxes to time slots. Doing an in-person session could be fun; might be more trouble than it’s worth, but might not be trouble at Constellation, ask them to help organize & recruit people, e.g. two 4-person groups Try to do one in the south bay? Could include Sam. Ask Ajeya how she’d like to participate Talk to Helen Toner, how to make sure what I’m doing is complementary Other people / groups to include DC – government does a lot of data gathering – propose content for legislation – Abi could help? Taren could do something in DC Oliver Stephenson (FAS) Elham, or the guy who worked for her? FAI IFP CHT (CBT?) FAS? Some Horizon fellows who are placed to draft a bill? Plan a third stage where we produce a piece of draft legislation? Discuss with Abi Discuss with Victoria next Friday Taren discuss with some people in DC Should discuss with Helen Toner Next steps: First step should be some Zooms, invite everyone. We won’t exhaust potential participants from Constellation. Taren to talk to Abi about whether to make this the theme for the dinner they’re organizing; if not, Taren will convene some other small meeting while she’s in DC. Pre-brainstorm: me, Taren, maybe Ajeya, maybe Abi? Chris Painter? Ryan Greenblatt? Josh Rosenberg? Someone with economics expertise (a grad student from David Otter’s lab)? More focus on identifying kinds of measurements (quantitative, qualitative, horizontal, vertical, etc.) to seed later conversations (+ as pre-read) Participants in Stage 2 Jaime Sevilla (Epoch) Content Notes See Untitled Review the Cruxes list in the early writeup Incorporate this idea from my discussion with Nick Allardice: Our economy and decision-making processes are so fragmented and messy. Trying to answer my crux questions at a societal level is unhelpfully generalizing. More tractable & beneficial is to pick some sectors, industries, types of problems, and come up with ways of measuring change in those, as leading indicators for what might happen in other sectors and industries. E.g. software engineering may be one of the more tractable problem spaces for AI; he’d be very interested in tracking diffusion here: hiring practices, how much autonomy is being granted, what level of productivity is it unlocking. If CS is a fast example, find a few slow examples, get super specific & granular about measuring diffusion. Get an idea of the uneven distribution. Carey: I don't know where this fits but I think the question of "what are the most common failure modes that prevent current models from excelling at practical tasks?" would be a relevant crux or root cause for your cruxes. Anna Makanju: It’s hard to break down someone’s usage of chatbots into productivity into other uses. In the last year it’s flipped from predominantly productivity usage into companionship. If you measure usage, need to try to disentangle in the nature of that usage – apply a classifier to their chat history, or focus on measuring enterprise accounts. Could work with universities and government agencies who will have enterprise accounts and might be more willing to share data for a study. Also see notes from my 8/11 conversation with Anna Why Focus on Early Indicators? If you want to make predictions about a future that is similar to the present, you might be able to simply extrapolate from past values of the variable you need to predict. For instance, Moore’s Law was an observation about trends in transistor counts, and for many decades it provided excellent forecasts of future transistor counts [FOOTNOTE: Though this may be a story about self-fulfilling prophecies as much as about the tendency of important variables to follow predictable trends.]. If you need to make predictions about a future that looks quite different from the present, you can’t get by with simple extrapolation. You need a model of how the future is going to unfold, and you need data to calibrate your model. For instance, if you want to predict the potential of fusion power, you can’t extrapolate the graph of historical electricity generation; for fusion, that graph is flatlined at 0. But if you understand the path that current fusion efforts are following, you can extrapolate metrics like “triple product” and “Q” [FOOTNOTE: I got help from Claude Opus 4 on this; the answer matches my vague recollection well enough that I’m not bothering to fact-check it: The most critical metrics for tracking progress toward practical fusion power are the fusion energy gain factor (Q), which measures the ratio of fusion power output to heating power input and must exceed 10-20 for commercial viability; the triple product (density × temperature × confinement time), which needs to reach approximately 10²¹ keV·s/m³ to achieve ignition conditions; and the reactor's availability factor or duty cycle, measuring the percentage of time the reactor can operate continuously, as commercial plants will need to run reliably for months at a time rather than just achieving brief fusion pulses.] to get an idea of how close we are to a functioning generator. Verify that an early application of the steam engine was to pump water out of coal mines. Make a reference to this being a sort of recursive self-improvement. Observe that if you had wished to measure the uptake of steam engines for pumping water out of coal mines, you could have looked at inputs to the process such as the amount of coal being consumed in steam engines outputs such as the amount of water being pumped or impact such as lowering the water level within the mines. When will “superhuman coders” and “superhuman AI researchers”, as defined in AI 2027, emerge? How is the task horizon identified in Measuring AI Ability to Complete Long Tasks progressing? What are we learning about time horizons for higher reliability levels (I presume reliability much higher than 80% will be necessary)? How fundamental is the gap between performance on benchmarks and real-world tasks? Is it growing or shrinking? [QUESTION: Does this cover “capability-reliability gap” described in AI as Normal Technology, or do we need to expand the description?] [this might better belong under the question regarding advances in domains other than coding and AI research.] Are any skills under coding or AI research emerging as long poles (more difficult to automate), and if so, are there feasible ways of compensating (e.g. by relying more on other skills)? What does the plot of AI research effort vs. superhuman performance look like? How does this vary according to the nature of the cognitive task? In particular, for tasks such as “research taste” that are critical to accelerating AI R&D? Basically redundant with previous cruxes, but perhaps worth listing as something that could be independently measured: across the broad range of squishy things that people do every day, how rapidly will the set of tasks for which AI provides significant uplift grow, and what are the contours of that set (what separates tasks experiencing uplift from tasks which are not)? Can AIs think their way around the compute bottleneck? If the R&D labor input races ahead of compute and data, how much progress in capabilities will that yield? To what extent does this depend on {quantity, speed, quality/intelligence} of the AI workers? Does this apply to all compute-intensive aspects of AI R&D? Is Ege’s forecast for NVIDIA revenue bearing out? Is his model for relating NVIDIA revenue to real-world impact of AI valid? Could rapid algorithmic improvements (driven by a software explosion) decouple impact from NVIDIA revenue? Could adoption lags result in revenue lagging capabilities? Possibly other questions as to whether various current trends continue – I’m not sure whether there are any other cruxes lurking here. For instance, is general progress in LLM capabilities speeding up or slowing down? Are breakthroughs emerging that accelerate the curve? Is RL for reasoning tasks hitting a ceiling? Are any of Thane’s bear-case predictions bearing out? Etc. As the automation frontier advances into increasingly long-horizon, messy, judgement-laden tasks, will the speed and cost advantages of AI (vs. humans) erode, to the point where they aren’t significantly faster or cheaper than humans for advanced tasks (and a few years of optimization doesn’t fix the problem)? Resources from the CSET conference Helen’s working doc My slides Ryan’s slides and notes It's important to detect when the most senior skills start to become automated. This could indicate a tipping point both for progress at the big Labs and or the ability for a breakout at a rogue actor who doesn't have access to senior talent. Perhaps we can look at the percentage of impactful ideas that come from unassisted AI. Perhaps we can look at the ratio of of major ideas, paradigm changing ideas to other inputs. Look for additional domains in which to measure sample efficient learning. In other domains, domains look at the ratio of spending on real world data collection versus in silico data generation. ARC Prize (@arcprize) posted at 10:21 AM on Tue, Sep 09, 2025:ARC Prize Foundation @ MITWe're hosting an evening with top researchers to explore measuring sample efficient in humans and machinesJoin us to hear from Francois Chollet along with a world class panel: Josh Tenenbaum, Samuel Gershman, Laura Schulz, Jacob Andreas https://t.co/dq7NJyXkNk(https://x.com/arcprize/status/1965465501079142814?t=L7y6E9f9cwxgjwWD1FMZJA&s=03) https://epochai.substack.com/p/after-the-chatgpt-moment-measuring Check in with Divya / CIP to see whether their global pulse surveys have questions relevant to the cruxes. Perhaps we can draw from this / make suggestions for it. They’re running “global pulse” surveys, to understand what people want from the future but also to understand how much AI has diffused into people’s lives. Every two months, started in March. Questions around trust, diffusion, how much are you relying on AI for medical or emotional advice, are you using it in the workplace, etc. https://globaldialogues.ai/cadence/march-2025, “In three years, what questions will we wish we had been tracking?” Maybe could be interesting to co-author something at some point. Cheryl Wu (@cherylwoooo) posted at 6:25 PM on Sun, Jun 01, 2025:Are we at the cusp of recursive self-improvement to ASI? This tends to be the core force behind short timelines such as AI-2027. We set up an economic model of AI research to understand whether this story is plausible. (1/6)(https://x.com/cherylwoooo/status/1929348520370417704?t=f9E9Yty2m27EQ-Z_NbbJ3Q&s=03) From Does AI Progress Have a Speed Limit?, a measurement: Ajeya: I'm kind of interested in getting a sneak peek at the future by creating an agent that can do some task, but too slowly and expensively to be commercially viable. I'm curious if your view would change if a small engineering team could create an agent with the reliability needed for something like shopping or planning a wedding, but it's not commercially viable because it's expensive and takes too long on individual actions, needing to triple-check everything. Arvind: That would be super convincing. I don't think cost barriers will remain significant for long. Another: Ajeya: Here's one proposal for a concrete measurement — we probably wouldn't actually get this, but let's say we magically had deep transparency into AI companies and how they're using their systems internally. We're observing their internal uplift RCTs on productivity improvements for research engineers, sales reps, everyone. We're seeing logs and surveys about how AI systems are being used. And we start seeing AI systems rapidly being given deference in really broad domains, reaching team lead level, handling procurement decisions, moving around significant money. If we had that crystal ball into the AI companies and saw this level of adoption, would that change your view on how suddenly the impacts might hit the rest of the world? Another: Ajeya: …do you have particular experiments that would be informative about whether transfer can go pretty far, or whether you can avoid extensive real-world learning? Arvind: The most convincing set of experiments would involve developing any real-world capability purely (or mostly) in a lab — whether self-driving or wedding planning or drafting an effective legal complaint by talking to the client. From Deric Cheng (Convergence Analysis / Windfall Trust) Early indicators: he’s friends with the Metaculus folks. They’re working on indicators for AI diffusion and disempowerment. Don’t share: Metaculus Diffusion Index – they’ll publish in a few weeks. https://metr.substack.com/p/2025-07-14-how-does-time-horizon-vary-across-domains When monitoring progress in capabilities, need to watch for the possibility that capabilities are advancing on some fronts while remaining stalled on some critical attribute such as reliability, hallucinations, or adversarial robustness Nick Allardice has a prior that labor market disruption is not going to be meaningfully different from other times in history… but he’s highly uncertain. Evidence that might push him to believe that meaningfully different levels and pace of labor market disruption would [?]. The burden of proof is on this time being meaningfully different from the past. If AI gets good at something, we’ll focus on something else. He hasn’t seen enough evidence to shift his priors. Leading indicators currently reinforce his prior: unemployment is low. It’s harder to get a job as a junior developer, but not impossible, and mostly seems to be due to other factors. Even if capabilities advance, diffusion challenges will leave room for human workers. Our institutions aren’t going to turn everything over to AI. Offcuts qualitative rubrics [MPH is a good indicator because every M takes about the same number of H; cities passed per hour would break down in the Great Plains; MPH breaks down when you enter a city center] If we only measure high-level, downstream attributes such as practical utility, we won’t have any way of anticipating these twists and turns. As I noted in the introduction to the cruxes writeup, by the time Facebook started to show up as a major contributor to overall Internet usage, it was already well along its journey to global dominance. === Tweet by Naman Jain (https://x.com/StringChaos/status/1928476388274716707) === Tweet by Naman Jain: Can SWE-Agents aid in High-Performance Software development? ⚡️🤔 Introducing GSO: A Challenging Code Optimization Benchmark 🔍 Unlike simple bug fixes, this combines algorithmic reasoning with systems programming 📊 Results: Current agents struggle with <5% success rate! https://t.co/ZaphmxrI1t pic.twitter.com/RoifIdWHxH === Tweet by Siméon (https://x.com/Simeon_Cps/status/1900218546904293863) === Tweet by Siméon: If an LLM scores 40% on Cybench, what does that tell you about its ability to help in a real-world actor at cyber offence? 1. We just released a first paper proposing a method to get an answer to that question. 2. Cybench is particularly suited to real-world because it has a… https://t.co/FWBXX7cOgD pic.twitter.com/s4Fk8RW0e8 === Where's the Shovelware? Why AI Coding Claims Don't Add Up (https://mikelovesrobots.substack.com/p/wheres-the-shovelware-why-ai-coding) === Subscribe Sign in Where's the Shovelware? Why AI Coding Claims Don't Add Up Mike Judge Sep 03, 2025 1,006 171 165 Share I’m furious. I’m really angry. I’m angry in a knocking down sandcastles and punching Daniel LaRusso in the face and talking smack about him to his girl kind of way. I’m not an angry person generally, but I can’t stand what’s happening to my industry. I know software development. I’ve been doing it for 25 years, maybe even 28 years 1 if you count market research tabulation on amber monochrome screens. Yes, I’m old. I’m a middle-aged programming nerd. My entire life and personal identity are wrapped up in this programming thing for better or worse. I thrive off the dopamine hits from shipping cool things. I was an early adopter of AI coding and a fan until maybe two months ago, when I read the METR study and suddenly got serious doubts. In that study, the authors discovered that developers were unreliable narrators of their own productivity. They thought AI was making them 20% faster, but it was actually making them 19% slower. This shocked me because I had just told someone the week before that I thought AI was only making me about 25% faster, and I was bummed it wasn’t a higher number. I was only off by 5% from the developer’s own incorrect estimates. This was unsettling. It was impossible not to question if I too were an unreliable narrator of my own experience. Was I hoodwinked by the screens of code flying by and had no way of quantifying whether all that reading and reviewing of code actually took more time in the first place than just doing the thing myself? So, I started testing my own productivity using a modified methodology from that study. I’d take a task and I’d estimate how long it would take to code if I were doing it by hand, and then I’d flip a coin, heads I’d use AI, and tails I’d just do it myself. Then I’d record when I started and when I ended. That would give me the delta, and I could use the delta to build AI vs no AI charts, and I’d see some trends. I ran that for six weeks, recording all that data, and do you know what I discovered? I discovered that the data isn’t statistically significant at any meaningful level. That I would need to record new datapoints for another four months just to prove if AI was speeding me up or slowing me down at all. It’s too neck-and-neck. That lack of differentiation between the groups is really interesting though. Yes, it’s a limited sample and could be chance, but also so far AI appears to slow me down by a median of 21%, exactly in line with the METR study. I can say definitively that I’m not seeing any massive increase in speed (i.e., 2x) using AI coding tools. If I were, the results would be statistically significant and the study would be over. That’s really disappointing. I wish the AI coding dream were true. I wish I could make every dumb coding idea I ever had a reality. I wish I could make a fretboard learning app on Monday, a Korean trainer on Wednesday, and a video game on Saturday. I’d release them all. I’d drown the world in a flood of shovelware like the world had never seen. Well, I would — if it worked. It turns out, though, and I’ve collected a lot of data on this, it doesn’t just not work for me, it doesn’t work for anyone , and I’m going to prove that. But first, let’s examine how extreme and widespread these productivity claims are. Cursor’s tagline is “Built to make you extraordinarily productive.” Claude Code’s is “Build Better Software Faster.” GitHub Copilot’s is “Delegate like a boss.” Google says their LLMs make their developers 25% faster. OpenAI makes their own bombastic claims about their coding efficiencies and studies 2 . And my fellow developers themselves are no better, with 14% claiming they’re seeing a 10x increase in output due to AI. 3 “Delegate like a boss” – Github Copilot These claims wouldn't matter if the topic weren't so deadly serious. Tech leaders everywhere are buying into the FOMO, convinced their competitors are getting massive gains they're missing out on. This drives them to rebrand as AI-First companies 4 , justify layoffs with newfound productivity narratives, and lowball developer salaries under the assumption that AI has fundamentally changed the value equation. And yet, despite the most widespread adoption one could imagine 5 , these tools don’t work . My argument: If so many developers are so extraordinarily productive using these tools, where is the flood of shovelware? We should be seeing apps of all shapes and sizes, video games, new websites, mobile apps, software-as-a-service apps — we should be drowning in choice. We should be in the middle of an indie software revolution. We should be seeing 10,000 Tetris clones on Steam. Consider this: with all you know about AI-assisted coding and its wide adoption, if I showed you charts and graphs of new software releases across the world, what shape of that graph would you expect? Surely you’d be seeing an exponential growth up-and-to-the-right as adoption took hold and people started producing more? Now, I’ve spent a lot of money and weeks putting the data for this article together, processing tens of terabytes of data in some cases. So I hope you appreciate how utterly uninspiring and flat these charts are across every major sector of software development. from Statista from Statista from Domain Name Industry Brief by Verisign From SteamDB I spent $70 on BigQuery processing to make this chart. Via GH Archive The most interesting thing about these charts is what they’re not showing. They’re not showing a sudden spike or hockey-stick line of growth. They’re flat at best. There’s no shovelware surge. There’s no sudden indie boom occurring post-2022/2023. You could not tell looking at these charts when AI-assisted coding became widely adopted. The core premise is flawed. Nobody is shipping more than before. The impact on human lives is incredible. People are being fired because they’re not adopting these tools fast enough 6 . People are sitting in jobs they don’t like because they’re afraid if they go somewhere else it’ll be worse. People are spending all this time trying to get good at prompting and feeling bad because they’re failing. This whole thing is bullshit. So if you're a developer feeling pressured to adopt these tools — by your manager, your peers, or the general industry hysteria — trust your gut. If these tools feel clunky, if they're slowing you down, if you're confused how other people can be so productive, you're not broken. The data backs up what you're experiencing. You're not falling behind by sticking with what you know works. If you’re feeling brave, show your manager these charts and ask them what they think about it. If you take away anything from this it should be that (A) developers aren't shipping anything more than they were before (that’s the only metric that matters), and (B) if someone — whether it's your CEO, your tech lead, or some Reddit dork — claims they're now a 10xer because of AI, that’s almost assuredly untrue, demand they show receipts or shut the fuck up. Now, I know the internet. I know what many of you chumps are going to say before you even say it, so let’s just get into it: “Well, if you just learned how to prompt properly, then you would be a 10x engineer like me.” Look at the data. There are no new 10xers. If there were — if the 14% of self-proclaimed AI 10xers were actually 10xers — that would more than double the worldwide output of new software. That didn’t happen. And as for you, personally, show me the 30 apps you created this year. I’m not entertaining this without receipts. “Well, it’s a new technology and so much is invested, and it takes time…” Yes, billions of dollars have been invested in these tools. Billions of dollars will continue to be invested in these tools. The problem is that they’re being sold and decisions are being made about them — which affect real people’s lives — as if they work today. Don’t parrot that nonsense to me that it’s a work in progress. It’s September 2025, and we’ve had these tools for years now, and they still suck. Someday, maybe they won’t suck, but we'd better see objective proof of them having an impact on actually shipping things on the large. “Well, maybe it kind of sucks now, but if you don’t adopt it early, you’ll be left behind.” There are no indicators that prompting is hard to learn. Github Copilot themselves say that initially, users only accept 29% of prompted coding suggestions (which itself is a wild claim to inefficiency, why would you publicize that?), but with six months of experience, users naturally get better at prompting and that grows to a whopping 34% acceptance rate. Apparently, 6 months of experience only makes you 5% better at prompting. “Well, maybe quality is going up and things aren’t necessarily shipping faster…” That doesn’t make any sense. We all know that the industry has taken a step back in terms of code quality by at least a decade. Hardly anyone tests anymore. The last time I heard the phrase “continuous improvement” or “test-driven development” was before COVID. You know as well as I do that if there’s a tool that can make people 10x coders, we’d be drowning in shovelware. “Well, it’s all website-driven, and people don’t really care about domain names these days; it’s all subdomains on sites like Vercel.” Shut up. People love their ego domains. “Well, .ai domain names are up 47% this year…” Yeah, that’s cause all the startups pivoted to AI. It’s the only way to get money out of investor FOMO. Has the overall amount of domain names gone up at an unprecedented rate, though? No, it hasn’t. Look at the new domains chart. “Well, if you were a real engineer, you’d know that most of software development is not writing code.” That’s only true when you’re in a large corporation. When you’re by yourself, when you’re the stakeholder as well as the developer, you’re not in meetings. You're telling me that people aren’t shipping anything solo anymore? That people aren’t shipping new GitHub projects that scratch a personal itch? How does software creation not involve code? Thanks for reading! Subscribe for free to receive new posts and support my work. Subscribe 1 28 years of experience is approximately 55,000 hours. I thought that was a fun metric. 2 Sam Altman says programmers are currently "10 times more productive" with AI coding. And that the world wants"100 times maybe a thousand times more software." (I agree there.) Altman also predicts programmers will "make three times as much" in the future. From https://www.finalroundai.com/blog/sam-altman-says-world-wants-1000x-more-software 3 78% of developers using AI report productivity gains, with 17% of those (13-14% of total developers surveyed) claiming a “10×” increase in output from AI - http://qodo.ai/reports/state-of-ai-code-quality 4 Funny to me how none of these “AI First” coding shops reportedly provide any training on how to become a 10xer with AI coding. “Experiment and figure it out yourself” is the common advice. Meanwhile, the official prompting guides are apparently not worth paying attention to because they don’t work. You see the dissonance, right? 5 60% of developers report using AI coding tools daily, and 82% use AI coding tools at least weekly. - http://qodo.ai/reports/state-of-ai-code-quality 6 The Coinbase CEO fired engineers last week who would not use Cursor or Copilot - https://www.finalroundai.com/blog/coinbase-ceo-fired-engineers-for-not-using-ai-tools 1,006 171 165 Share Discussion about this post Comments Restacks Mike H Sep 3 Liked by Mike Judge Fucking sing it from the mountain brother. Expand full comment Reply Share Khải Đơn Sep 4 Liked by Mike Judge I am not working anything in coding and engineering. I am a writer. The most claims I heard about AI are from IT engineers who told me that I didn't know how AI has evolved and sped up the productivity in tech and all other fields, including writing . As a writer, with no evidence backed from my side, I couldn't say anything to counter these claims because I had no data. A lot of writing work I had and lost, and then got again, came from the adoption of AI. When AI first became a trend, all my clients stopped hiring me because AI could write all what I could write. A year and a half later, they came back and asked if I could freelancing in editing the things AI wrote. I asked for higher price because the quality of the AI writing was not good enough. And then they decided to go back to normal, hiring me for the same work before AI because fixing what AI generated for them is more costly. I still can't say anything to counter the 10 times faster or 20% faster number that those companies put out into the world. Even the novelty of using AI to write fiction or mundane job of writing has become boring. It always come down to the blame: "your prompt is not good enough." Thank you for the article and your hard work to put this out. Expand full comment Reply Share 3 replies 169 more comments... Top Latest Discussions No posts Ready for more? Subscribe © 2026 Mike Judge · Privacy ∙ Terms ∙ Collection notice Start your Substack Get the app Substack is the home for great culture === Tweet by Siméon (https://x.com/Simeon_Cps/status/1956514153528742168?t=qJkEFlWoQAD6z6g4HMhNAw&s=03) === Tweet by Siméon: I've entertained that theory for a few years and have been confused since then why people expected confidently so much GDP growth. Basically prices if goods should crash so fast that the question of "how do you count inflation" will become the first order parameter of whether and… https://t.co/RS1LXRFIRE === Will there be a bubble pop? / Economics of AI (https://www.notion.so/26b4c916374a807eac83c7b05859c488) === If we’re in a bubble, it’s like the dot-com bubble, not the tulip bulb bubble. Ask yourself whether the Internet turned out to be a big deal. Also, under any circumstances, many current startups will fail spectacularly, and this could easily include some very big bets (cf. Webvan). OpenAI’s International Conundrum — The Information @Rittenhouse Research: Biggest takeaway from this (very good) interview was Altman's outline how OAI's financial model could eventually work. - OAI enterprise is growing faster than consumer - OAI enterprise growth is constrained by lack of compute capacity (e.g. enterprises are coming to OAI asking… https://t.co/RtmIYTst6x https://x.com/i/status/2002038608501301346 https://www.noahpinion.blog/p/the-ai-bust-scenario-that-no-one https://x.com/Oaktree/status/1998430811159409007 https://ceodinner.substack.com/p/the-ai-wildfire-is-coming-its-going https://www.theinformation.com/articles/openai-projected-least-220-million-people-will-pay-chatgpt-2030 https://www.understandingai.org/p/six-reasons-to-think-theres-an-ai From https://thezvi.substack.com/p/ai-142-common-ground: Chris Bryant: The head of Alphabet Inc.’s AI and infrastructure team, Amin Vahdat, has said that its seven- and eight-year-old custom chips, known as TPUs, have “100% utilization.” theinformation.com/articles/anthropic-projects-cost-advantage-openai Meltem Demirors (@Melt_Dem) posted at 8:56 AM on Sun, Nov 09, 2025:reflecting on the recent slew of “it’s an AI capex bubble” w my partner @kellyjgreer - feels like much ado about nothingthere’s some confounding narratives at play:- the magnitude of the capex numbers is enormous and unprecedented - largest infra buildout since WWII- only(https://x.com/Melt_Dem/status/1987565089696694666?t=c2xIrwJRU6pk0uzxpmmHBw&s=03) https://www.theinformation.com/articles/anthropic-projects-70-billion-revenue-17-billion-cash-flow-2028 https://epochai.substack.com/p/introducing-the-frontier-data-centers https://nikolajurkovic.substack.com/p/are-we-in-an-ai-bubble?t=Hr_MPzFeYKIGTDbvWW-dDA&s=03 tae kim (@firstadopter) posted at 11:40 AM on Sat, Oct 25, 2025:Why AI is Underhyped and Isn't a Bubble YetHere's why AI isn't a bubble today. I've distilled the data points and ideas from my previous columns and coverage. If you prefer facts and evidence-based reality over vibes and conflated narratives, enjoy!-Big Tech valuations are(https://x.com/firstadopter/status/1982155280423977243?t=Q-ORfSjvzF7yiM-Gcasb6g&s=03) https://peterwildeford.substack.com/p/ai-is-probably-not-a-bubble https://www.dwarkesh.com/p/thoughts-on-the-ai-buildout Review the stateof.ai report, slides 99-104, 111-112, 131 Share draft with Siméon, he commented on Noah Smith’s post and argued that a pop is plausible https://x.com/ajeya_cotra/status/1977881746797687223 https://x.com/snewmanpv/status/1977931996249928011 https://www.understandingai.org/p/16-charts-that-explain-the-ai-boom https://pluralistic.net/2025/10/16/post-ai-ai/#productive-residue https://epochai.substack.com/p/openai-is-projecting-unprecedented https://www.dwarkesh.com/p/thoughts-on-the-ai-buildout https://x.com/dwarkesh_sp/status/1981074799758921843 https://www.noahpinion.blog/p/should-we-worry-about-ais-circular → the circular deals don’t determine whether this is over-investment, but they may increase volatility / exacerbate any collapse From https://thezvi.substack.com/p/ai-139-the-overreach-machines: Gunjan Banerji: Goldman: “We don’t think the AI investment boom is too big. At just under 1% of GDP, the level of spending remains well below the 2-5% peaks of past general purpose technology buildouts so far.” [see subsequent image] Epoch AI (@EpochAIResearch) posted at 2:19 PM on Fri, Oct 10, 2025:New data insight: How does OpenAI allocate its compute?OpenAI spent ~$7 billion on compute last year. Most of this went to R&D, meaning all research, experiments, and training.Only a minority of this R&D compute went to the final training runs of released models. https://t.co/Jq9768bSB1(https://x.com/EpochAIResearch/status/1976714284349767990?t=UU51tJ2sBPX7dGTHMWK9rw&s=03) Corry Wang (@corry_wang) posted at 6:17 PM on Fri, Jan 01, 2021:1/ Lessons From The Tech Bubble:Last year, I spent my winter holiday reading hundreds of pages of equity research from the 1999/2000 era, to try to understand what it was like investing during the bubbleA few people recently asked me for my takeaways. Here they are - https://t.co/41nTJdrFR1(https://x.com/corry_wang/status/1345192541545766915?t=YtZvbF0fgkf-zBeVWX2wxA&s=03) Benjamin Todd (@ben_j_todd) posted at 6:29 PM on Mon, Oct 20, 2025:The FT is my favourite newspaper, but this seems to be bad reporting.The article "What GPU pricing can tell us about how the AI bubble will pop" points out a rack of 8 A100 chips need to generate $4/hour to cover the capital cost of the chips over 5yr.It then points out that https://t.co/9SEbY5rnLW(https://x.com/ben_j_todd/status/1980446425499726094?t=AJEb5wjbwagwVHE55lNGWw&s=03) https://www.theinformation.com/articles/nvidia-broadcom-amd-face-new-risks-openai-deals https://thezvi.substack.com/p/bubble-bubble-toil-and-trouble https://www.theinformation.com/articles/oracle-assures-investors-ai-cloud-margins-struggles-profit-older-nvidia-chips https://www.theinformation.com/articles/salesforce-ceo-shifts-ai-strategy-openai-threat-looms https://www.derekthompson.org/p/why-ai-is-not-a-bubble https://www.derekthompson.org/p/this-is-how-the-ai-bubble-will-pop Could quote from the first section of “Money Stuff: OpenAI Has a Business Plan” for a funny intro to the question. It really is the greatest business plan in the history of capitalism: “We will create God and then ask it for money.” On a pure science-fiction suspension-of-disbelief basis, this business plan is perfect and should not need any updating until they finish building the superintelligent AI. Paying one billion dollars for a 0.2% stake in whatever God comes up with is a good trade. But in the six years since announcing this perfect business plan, Sam Altman has learned[2] that it will cost at least a few trillion dollars to build the super-AI, and it turns out that the supply of science-fiction-suspension-of-disbelief capital is really quite large but not trillions of dollars. David Shapiro ⏩ (@DaveShapi) posted at 5:38 AM on Fri, Oct 17, 2025: Some people still think that AI is a "bubble" so here's my updated take. They point at a few facts like: The "circularity" of the market The fact that OpenAI and other startups are not profitable It "feels" like the Dot Com revolution ("where there's hype, there's a https://t.co/3WTUQxaTjP (https://x.com/DaveShapi/status/1979165009495171523?t=VNbxocWJqnzeiR6iBkZ2zQ&s=03) https://www.ft.com/content/a169703c-c4df-46d6-a2d3-4184c74bbaf7 [me, on Signal] My two cents: The amounts being invested are eye-watering, but to my understanding, not at all unprecedented – the 1800s railroad build-out was much larger (as a % of GDP), the 1990s telecom build-out was at least as large (again, as % of GDP). "If we stop, the economy will collapse" is a terrible reason to keep investing. If these investments are going to pay off, then there's no economic reason to stop (there may of course be other reasons, relating to safety, disruption, etc.). If these investments are not going to pay off, then inflating the bubble further just delays + worsens the inevitable. None of this is binary. Current investments might yield a 500% profit, or 20% profit, or break even, or lose a bit of money, or wind up being sold off for pennies on the dollar. The economy might collapse, or suffer a brief recession, or a brief slowdown in growth, or do fine. (Granted, it is tricky to gently deflate a bubble, so the distribution of potential economic scenarios may be somewhat bimodal, but we've recently seen that soft landings are possible.) None of this is predetermined; the choices that governments and businesses make, and the skill with which they execute, will play an important role. [Igor] On railroads: I would encourage you to do the math on the capex, opex, lifetime of data centers vs railroads and the implied ROIC required for things to be rational. The essays linked here https://futurism.com/future-society/ai-data-centers-finances might serve as inspiration. Rest I more or less agree with, albeit probably from a more ai critical lens I do think the ai bubble popping will be this generations 2008 Epoch AI (@EpochAIResearch) posted at 9:23 AM on Wed, Oct 15, 2025: One way bubbles pop: a technology doesn’t deliver value as quickly as investors bet it will. In light of that, it’s notable that OpenAI is projecting historically unprecedented revenue growth — from $10B to $100B — over the next three years. 🧵 https://t.co/WCgLhoLb8B (https://x.com/EpochAIResearch/status/1978496866624176507?t=UHaEDOor8F84JrM3JTUPoQ&s=03) Ajeya Cotra (@ajeya_cotra) posted at 4:38 PM on Mon, Oct 13, 2025:I'm interested in more detailed work on how much profit AI companies will capture and whether there will be a bust in the next 2y.Appreciated the emphasis that "market crash" != "low-impact technology" https://t.co/A8SuW9IGB4(https://x.com/ajeya_cotra/status/1977881746797687223?t=9gVZ8D__Q8-HEPjeZ2P_0Q&s=03) Derek Thompson (@DKThomp) posted at 8:24 AM on Wed, Oct 15, 2025:New newsletter: WHY AI IS NOT A BUBBLEEverybody is calling AI a bubble now, even the folks building it.I hate when conventional wisdoms get too conventional.So I read and listened to the smartest cases for why AI is NOT a bubble.My summary —>https://t.co/9gjWvh2WZL(https://x.com/DKThomp/status/1978482238229561816?t=dvmpuBl2JhgiJ3AhZKANJA&s=03) https://www.theinformation.com/briefings/anthropic-said-target-26-billion-annualized-revenue-2026 https://www.transformernews.ai/p/what-happens-when-the-ai-bubble-bursts-crash https://www.slowboring.com/p/the-ai-boom-is-propping-up-the-whole https://x.com/ecommerceshares/status/1978392637682876551?s=46 https://www.linkedin.com/feed/update/urn:li:activity:7384323912786141184/ https://x.com/deanwball/status/1978458776832266669 https://davekarpf.substack.com/p/its-giving-enron https://www.wheresyoured.at/ai-is-a-money-trap/ https://www.theinformation.com/articles/openai-working-softbanks-arm-broadcom-ai-chip-effort Derek Thompson (@DKThomp) posted at 7:06 AM on Tue, Oct 14, 2025:What's the best source -- from an article, bank, analyst note, etc -- for the level of "AI revenue" that hyperscalers are seeing right now?(Yes, I know this can be fudged in a million ways; eg, Meta can always claim that X percent of its ad revenue is thanks to AI.)(https://x.com/DKThomp/status/1978100071896944905?t=e48XnPEWU-DmRQwNBr6VBQ&s=03) https://www.theinformation.com/articles/microsoft-let-openai-play-field https://www.theinformation.com/articles/race-rent-nvidia-chips-cloud-intensifies?rc=jsaoww https://epochai.substack.com/p/the-epoch-ai-brief-september-2025 https://www.bloomberg.com/news/features/2025-10-07/openai-s-nvidia-amd-deals-boost-1-trillion-ai-boom-with-circular-deals https://futurism.com/future-society/ai-data-centers-finances https://paulkrugman.substack.com/p/technology-bubbles-causes-and-consequences https://www.noahpinion.blog/p/americas-future-could-hinge-on-whether https://www.exponentialview.co/p/is-ai-a-bubble Andrew Curran (@AndrewCurran_) posted at 10:02 AM on Tue, Oct 07, 2025:Jamie Dimon told Bloomberg that JPMorgan has now reached AI equilibrium. JPM spends $2 billion a year on developing artificial intelligence technology, and saves about the same amount annually. Their plan is to gain first-mover advantage by incorporating agentic AI at all levels. https://t.co/nLA1z1zFGC(https://x.com/AndrewCurran_/status/1975607594556563555?t=yHaY5hPnM8LkyG-tR2Rm-w&s=03) https://www.theinformation.com/articles/internal-oracle-data-show-financial-challenge-renting-nvidia-chips?utm_campaign=article_email&utm_content=article-15862&utm_medium=email&utm_source=sg&rc=jsaoww https://www.nytimes.com/2025/10/06/technology/openai-amd-chips.html Follow up with Konstantin Pilz from RAND, per email introduction from Matt Chessen after we were matched 1:1 at The Curve but Matt couldn’t attend. Open Philanthropy (from https://www.openphilanthropy.org/research/ai-safety-and-security-need-more-funders/): We’ve supported the Center for Security and Emerging Technology at Georgetown University, which compiles detailed data on AI investment, semiconductors, and governance efforts for use by policymakers, researchers, and journalists. Epoch AI (@EpochAIResearch) posted at 0:04 PM on Tue, Sep 30, 2025:Announcing our new AI Companies Data Hub!We collected key data on frontier AI companies, including revenue run rates, funding, staff, usage rates, and compute spend.This free resource will help you understand the trajectory and economics of AI.Highlights in thread! https://t.co/bk4h5PixbA(https://x.com/EpochAIResearch/status/1973101761964462582?t=4ldNUkTnYFeFUnp6g07usQ&s=03) https://peterwildeford.substack.com/p/openai-nvidia-and-oracle-breaking https://www.theinformation.com/articles/openais-first-half-results-4-3-billion-sales-2-5-billion-cash-burn https://pluralistic.net/2025/09/27/econopocalypse/ Dan Hendrycks (@DanHendrycks) posted at 7:12 PM on Sat, Sep 27, 2025:OpenAI's recent benchmark suggests that AIs are nearing human-level economic capability.But the best indicator of utility is usage. OpenAI also noted that since GPT-4's launch, usage by early adopters has only grown 1.4x.The actual usefulness isn't increasing that sharply. https://t.co/vvxrNUY6KB(https://x.com/DanHendrycks/status/1972122329787593164?t=y53UwIR3qIcecdepfhUq5Q&s=03) https://x.com/lugaricano/status/1965765898540822820 https://pca.st/episode/f3b124e3-e29a-45c7-b373-fa6efef8e676 From https://thezvi.substack.com/p/ai-135-openai-shows-us-the-money: The investment numbers are even more dramatic. AI investment was already responsible for 20-43% of Q2 2025 GDP growth. Heninger’s numbers imply that AI labs (collectively) would be investing $720 billion to $1.2 trillion by 2027 if they remain on trend — that investment alone would generate 2-4% nominal GDP growth. I think it’s unlikely investors will pony up that much capital unless the models surprise significantly to the upside in the next year or two, but even still, 1-2% nominal and 0.5-1% real GDP growth coming from just AI investment in 2026-27 seems entirely plausible. https://www.exponentialview.co/p/is-ai-a-bubble?utm_source=%2Fsearch%2Fai bubble&utm_medium=reader2 Derek Thompson (@DKThomp) posted at 11:49 AM on Tue, Sep 23, 2025:New pod: This is how the AI bubble could burstInvestor and writer @pkedrosky joins to explain- how AI capex is eating the economy- how financial wizardry pays for a zillion data centers without showing up on Big Tech's books- how it could implodehttps://t.co/rnKFwuWDpv(https://x.com/DKThomp/status/1970561295834341438?t=O8dwl-Iltt1uoXKqMHjCsg&s=03) How sustainable are the current economics of the AI buildout? https://x.com/deanwball/status/1970490882047779111?t=obmfFWrMdMAcXEKYciWdug&s=03 https://www.theinformation.com/articles/openai-spend-100-billion-backup-servers-ai-breakthroughs “AI Agenda: OpenAI’s Spending Spree is Reordering the Cloud Market” “Applied AI: Salesforce, Microsoft Find Selling AI to Enterprises Is Easier Said Than Done” “https://www.theinformation.com/articles/openais-350-billion-computing-cost-problem” https://sundaylettersfromsam.substack.com/p/the-amazon-of-thought Look into Ed Zitron’s writings https://www.lesswrong.com/posts/KW3nw5GYfnF9oNyp4/trends-in-economic-inputs-to-ai https://www.theinformation.com/articles/microsoft-hopes-hastened-ai-rollout-price-discounts-can-fuel-office-365-growth From SemiAnalysis, a pointer to what sound like useful resources from them: We’ve forecasted xAI’s CapEx on Core Research, our institutional research service and are now closely tracking the ROIC of AI investments across the hyperscalers and AI labs in our new Tokenomics model. Anecdote from someone who works at a major tech company: The other day I was in a meeting with an internal AI guru/proselytizer/etc, and the guru was showing off the various tools they could use. He also dropped, as an aside, "your team is fairly low utilization, only 20% ... the gold star is HR, they are close to 100%". Turns out the metric is 'what percent of the team uses AI on any given day". I immediately told my team: "go out, buy a second monitor, always have an AI tool running on it. Ask it some sort of question, at least once a day. I don't care at all what you use it for, but what I can never have happen is to be told that I don't get more headcount because my team isn't using AI enough". Another: I have a friend who’s a senior software engineer at Amazon who has said that the mandate and metrics internally work just like that. She told me that all of the seniors she talks to believe that AI is garbage that is going to take their jobs and have no interest in using it. The juniors are happily using AI. Apparently, aside from the mandate, there has been very little training about how to make good use of AI in her org. Cite the thinking machines crazy pitch meeting is an example. We're starting an AI company with all of the best people, but we can't answer any questions and we're looking for 2 billion on a $10 billion valuation as mentioned. Maybe 10 minutes into the Derrick Thompson. What if it's not a bubble podcast? Overall theme is You know bubble may be defined in terms of things like stock price, stock market valuations. I'm going to be less interested in that. A lot of the signals that we have available to look at are levels of investments. You know valuations things like that a lot of that has to do with people's expectations. I'm more interested in the reality of what's actually happening. Expectations are useful to the extent that you know if we if we believe that in the efficient market hypothesis. If we believe that the people who have these expectations that are acting you know based on their revealed expectations to the extent that we, we think they know what they're doing that provides a signal. Now we have a lot of evidence such as that crazy pitch meeting suggesting that you know people are flying pretty blind. That doesn't mean that this is all going to collapse, but it it does suggest that the signal of the efficient market signal you know make have come disconnected here as it generally turns out to have been an impasse bubbles, so it it is perhaps unnecessary. Though not sufficient condition for a bubble that the investors are just piling in based on vibes and momentum rather than hard analysis of some kind of Rich and Rich data on it. And it's clear that that necessary though insufficient condition has obtained here. What I'm interested in is at the you know there's this long valued AI value chain I'm interested in. What's the level of investment that's going in at the front you know spending on chip spending on data center construction. Probably spending on you know labor costs, salary and interesting question. Whether equity labor cost should be included in that? So what's what's the investment going in and what's the actual value coming out at the end user end either consumer or business end user. You know people draw all these diagrams of like overlapping circles and things, but if we just look at what's going in at the front and what's going out at the back then we can eliminate a lot of that complexity because it just factors out of of this analysis. People also talk about the you know another scent of bubble being creative pricing or creative financing mechanisms. My feeling there is that. Again, that's sort of a necessary or at least suggestive but but not sufficient evidence for a bubble. You know, all that really means is that there's an appetite to invest more than can be obtained through conventional methods or or simply that these investments are getting big enough to no longer fit into the historical business model of the entities doing the investment. You know if Google or Oracle or Microsoft or whatever go around and do these creative financings you know I presume that the kinds of financing they're doing are very standard in some industries. It just wasn't standard in their particular industry, but this is perfectly reasonable because they're essentially getting into an industry they weren't in before. So again, that you know the creative financing is it's a nudge. it's it's a bit of evidence that like something funny might be happening, but it doesn't actually prove that this is a bubble and and so again you know what I'm really interested in is you know is the level of investment warranted in the sense that is it going to translate into into returns in a reasonable time frame that gets into the time frame question, which is interesting. You know even let's stipulate that in some long-term future. We will want the amount of gigaflops billed out That's greater than what's being done today. The question is not whether the world is going to want all this computing capacity. It's whether it's going to want it soon enough to make building it today. Be a good investment you know versus waiting 235 10 years and then building it out with you. Know having saved on time, value of money and also you know using more you know more advanced and cost effective chips and and even energy supplies and so forth. Not to mention. Also just avoiding the the rush Factor. You know paying it paying extra to get land and energy and other inputs quickly when you know you could probably get the same inputs for at least a somewhat reduced cost. If you weren't in such a hurry, you know certainly xai is doing things with like trucking in portable, gas, generators and so on that I I presume are not the most cost efficient way to earn money into into kilowatts. So these are the kind of factors I'm interested in exploring and so we need to dig into whatever data we can find about what's actually being infested to build data centers. Just very tricky because you know things get reported from a lot of different angles, so there's potentially a lot of you know double counting and double reporting, but that's what extent can we tease out? What is the actual total investment in AI Data center construction? I'm also someone interested in how that breaks down across the value chain. You know from Nvidia to TSMC and tsmc's own suppliers and and like the various other inputs for power and so forth and then and this will probably the be the first heart very hardest part to investigate is what's the data coming in at the end user end and then how does that break down across application developers like cursor model providers like Open AI You know Data center operator is and and so forth. Note that I'm not going to discuss potential impact on the economy, crowding out other kinds of investment. You know electricity use of data centers, things like that. These are all important questions but they're just out of scope for me. Justpp just prior to halfway through the Derrick Thompson. What if this isn't a bubble episode? He starts talking. Azeem starts talking about the ratio of of capex to revenue. So in the rail railroad construction bubble I believe he said capex with roughly twice revenue in the dot-com bubble. It was forex and currently in the AI build out. It's 6X you know this gets directly to the kind of question I was presently look at. I do wonder you know where he's getting that 6X figure from and whether it's the you know true end user end-to-end kind of figure that I've proposed looking at then a little bit earlier in the episode he discusses depreciation schedule for GPUs. Yeah, I think it'd be interesting to look at. What is the breakdown of of AI Data center CAPEX how much of that is you know building shell power and HVAC networking equipment GPUs and so on. And you know, what is the how quickly do each of those things? Depreciate how long do they physically last? I've seen at least one passing reference suggesting that you know GPUs might actually physically wear out surprisingly quickly. If they're used intensively to how long do they physically last and and how long do they depreciate just in terms of being you know outmoded you know no longer as as cost efficient power efficient. What have you as as as newer hardware? That inbound revenue can be can take multiple forms. It can be you know customers and businesses directly using products that are essentially pure AI like Chet EBT. Even there it gets interesting. Like you know look at like something like cursor or shortwave. You know what freshen of the you know a significant chunk of the revenues from those products is flowing through into revenue for operating AI models, but not you know, unclear exactly what that percentage is. Then you have you know the the big AI providers themselves using AI internally to power their own products or optimize their own products. So that's everything from you. Know. Google selling Gemini services to Facebook using AI to improve AD targeting or to generate content on their beyond their own platforms. There is some AI, advertising and product you know. Purchase affiliate links built into AI products themselves ; I feel like I had one more variation in mind but and now I can't remember what it was. Azeem says that growing into the current valuations would require revenues to double each year for a while and then slow down. I don't quite get that. I would think they would need to more than double you know you. Can't you know if if revenue sour currently won six of capex then doubling each year for a couple of years? Given the type depreciation of you know the the rate at which these assets age I don't see how even that's enough might ask him to review my draft. Then of course there's the question of AGI and you know how much. How much revenue would you know replacing all human labor bring in? I'm not really going to go into that because you know I'm going to assume that that's outside the return on investment window of today's data center. Buildouts you know it just kind of that's kind of just kind of a separate. All bits are are off thing. So you know if the question is is this a bubble or or is that more deadly defined it is? Is the current level investment a good idea? Like you can just have this separate theory if you want that you know. Maybe the investment is is justified because it speeds the path to AGI and AGI will be so valuable. But even that I would book a scans at because most of these chips that are being deployed are not being used to develop AGI. They're being used basically to the purposes that are ultimately about generating revenue along the way to support the further R&d. And so if the revenue on the way isn't going to justify the chips, then you know then that indirect argument about AGI isn't going to justify either Azeem briefly hand waves toward another another benchmark which is you know. Basically look at the total market for for you know digital services. I think you mentioned something like 1 trillion a year, although I'm not actually sure which scope that that was meant to refer to it. Anyway you know if you can imagine that all digital services are going to be are they're going to start? You know needing AI needing to use AI dreaming, competitive or and or AI might enable you know all those markets to grow. Then you could use that as another anchor point estimate revenues will be going. Although I'm not sure that it makes sense to look at that as an anchor point for the next few years. That seems more like a total addressable market kind of a data point [Child Page: AI Financials; potential for an AI Winter / Bubble Bursting] Lauren Wagner (@typewriters) posted at 5:16 AM on Mon, Sep 08, 2025:A few ideas for what can explain this:1. Businesses are adopting AI tools that don’t actually meet their business needsa/they aren’t adequately assessing product value beforehandb/they may be adopting general AI tools when they actually need more specialized systems (like(https://x.com/typewriters/status/1965026517940904151?t=9g4i_5N-k9korSlaWCCj_Q&s=03) https://www.theinformation.com/articles/openai-says-business-will-burn-115-billion-2029 “Applied AI: Census Bureau Says Corporate AI Adoption is Slowing” [Child Page: Does R1 Undermine the Business Case for Aggressive AI R&D?] The concept of a "lead" is funny when it's easy to fast follow. Is there a world where advanced techniques can be kept truly hidden? If anything OpenAI releases can be copied (for much lower cost) within a year, then all they get for their troubles in pushing forward the state of the art is a brief period of exclusive access8. [Ben Thompson of Stratechery: On the positive side, OpenAI and Anthropic and Google are almost certainly using distillation to optimize the models they use for inference for their consumer-facing apps; on the negative side, they are effectively bearing the entire cost of training the leading edge, while everyone else is free-riding on their investment. Indeed, this is probably the core economic factor undergirding the slow divorce of Microsoft and OpenAI. Microsoft is interested in providing inference to its customers, but much less enthused about funding $100 billion data centers to train leading edge models that are likely to be commoditized long before that $100 billion is depreciated. Or from Hacker News: DeepSeek just further reinforces the idea that there is a first-move disadvantage in developing AI models.] Which isn’t even particularly exclusive, as there are at least three frontier labs (OpenAI, Anthropic, and Google DeepMind) who are spending the big bucks to stay relatively close to one another, even if OpenAI seems to be slightly ahead overall and certainly is best at generating headlines. If you believe that AGI is coming within the next few years and will have value measured in the trillions (as the leaders at the frontier labs appear to do), then spending many billions of dollars may make sense even if it only buys you a few months of advantage. But if the pace of progress slackens, or the economic impact takes time to manifest, this logic could change. An issue here is that the collective experience of the tech industry leads everyone to believe that you respond to competition by innovating harder – we’re not used to a situation where it’s consistently 10x cheaper to follow than to lead. (To be clear, it’s not certain that this is the situation in AI either. But we’ve accumulated a fair amount of evidence that it might be.) Eric Gastfriend captures this idea in a meme: (Via Helen Toner, who adds, “it's not a perfect metaphor - the Chinese companies are working super hard and doing real research to keep up”.) === Nick Allardice – Impact Cruxes (https://www.notion.so/2381a29f178980eda6c1d137ccc01aac#2471a29f178980878a29c548b6409c1f) === 8/6/25 This is a project to advance the conversation regarding when, how suddenly, and how strongly AI will have a transformative impact on the world. In other words: when big things will start to happen, how rapid the onset / escalation will be, and how far the escalation will go. This is related to the concepts of “timeline” and “takeoff speed”, but focuses on impact (things actually happening in the world) rather than capabilities. Get his high-level take – review Questions for New Participants. Ask what aspects of the question I might be missing, and how he’d like to participate going forward. I want his views on timelines / impact, in his case especially impact outside of the US. How will adoption play out differently in the global south? The thesis he’s found most persuasive is AI as Normal Technology. He was at a conference called Dialogue, industry leaders engaging with unconventional questions. He should nominate Taren and me. He went in as a fast-timelines person. He went to a bear-case session, lots of tech people, including Jonathan Rosk (founder of Groq). They spent 90 minutes discussing how timelines could be longer than expected. The core argument he came away with is that diffusion really takes a long time. E.g. there are Fortune 500 companies still using fax machines. Our economy and decision-making processes are so fragmented and messy. Trying to answer my crux questions at a societal level is unhelpfully generalizing. More tractable & beneficial is to pick some sectors, industries, types of problems, and come up with ways of measuring change in those, as leading indicators for what might happen in other sectors and industries. E.g. software engineering may be one of the more tractable problem spaces for AI; he’d be very interested in tracking diffusion here: hiring practices, how much autonomy is being granted, what level of productivity is it unlocking. If CS is a fast example, find a few slow examples, get super specific & granular about measuring diffusion. Get an idea of the uneven distribution. In the global south, there are four problems that need to be solved for AI to have a meaningful impact. E.g. in sub-Saharan Africa. In order: Hardware Rapidly accelerating mobile phone adoption, but in poor countries it’s still mostly dumbphones. Connectivity challenges, cost of airtime. Expensive; regulatory barriers. Will be 5-10 years before we start to see meaningful levels of hardware adoption that make the basics of AI accessible at the consumer or enterprise level. Software Language performance Google is convinced that within a couple of years, language will be a solved problem – just needs some fine-tuning. Other leading labs aren’t as confident, perhaps because they don’t have have the training data, that might be 5 years. Content (didn’t get a chance to ask what he meant by this, maybe region-specific knowledge?) Accessibility (will partially be addressed by hardware) Multi-modality is important for non-literate populations Connectivity issues interfere with always-on voice mode UX stuff, e.g. support voice memos. This won’t get much attention for 10 years, everyone is focused on the land grab in the west. Economic opportunity Until basic needs are met, AI isn’t very relevant. 800M people aren’t in a position to meaningfully benefit, because they don’t have the infrastructure and economic opportunities. E.g. medical advice doesn’t help much if you don’t have access to medicine. He’s doing research on all of these at the moment. At the end of the day, this will be just another trickle-down technology. Doesn’t affect the impact on the world at large. What is he most uncertain about? Cruxes that feel important to him: He has a prior that labor market disruption is not going to be meaningfully different from other times in history… but he’s highly uncertain. Evidence that might push him to believe that meaningfully different levels and pace of labor market disruption would [?]. The burden of proof is on this time being meaningfully different from the past. If AI gets good at something, we’ll focus on something else. He hasn’t seen enough evidence to shift his priors. Leading indicators currently reinforce his prior: unemployment is low. It’s harder to get a job as a junior developer, but not impossible, and mostly seems to be due to other factors. Even if capabilities advance, diffusion challenges will leave room for human workers. Our institutions aren’t going to turn everything over to AI. How offense vs. defense net out for various x-risks. Increases in defensive capabilities are often under-appreciated, but he’s not confident and the use cases are all different. Frontiers of autonomy – will agents be able to avoid going down unproductive, token-burning rabbit holes? Feels like a solvable problem, but he’s not sure on what time frame. He thinks our general risk aversion will slow a lot of things down, for better and worse. E.g. we’ll forego applications in mental health or AVs. Things that can have a large impact on society and don’t have meaningful regulatory or corporate bottlenecks / need for human process changes: the intersection is fairly small. === Anna Makanju – Impact Cruxes (https://www.notion.so/2551a29f178980188ce6fd8bb09a566e#2551a29f178980c1914bf155bc912b0f) === 8/11/25 See personal notes from 8/11/25 conversation. Some highlights: She hears from colleagues that AI researchers are able to automate a lot of their coding work. Building a production-grade consumer-facing product is much harder to automate. Her trip to Nigeria was a reminder that no amount of GPUs is going to fix people’s ability to use this in any way. Electricity and connectivity are extremely limited / intermittent. Most people on Earth aren’t eligible for decades to access this. Could try to monitor [inertia effects] by finding highly reticent groups and monitoring adoption rates. GPT-5 didn’t make a big impact because everyday use cases are already approaching saturation. With o3, she saw people in specialized scientific fields seeing a big jump in applicability, there will be large spiky impacts in niche applications. She’s fairly AGI pilled but she doesn’t expect it to automate all work. Partly because of limits to adoption in many parts of the world, and also many jobs are too varied and granular. But yes in finance. Transformational to medicine and health, that’s the one area where she has a lot of optimism – personalized medicine will be highly impactful. In the legal field, it’s difficult to say. It’ll transform access to legal services, but the infrastructure of the courts will be behind, so unclear what that will mean for legal recourse. Impact will be highly sector-specific; some sectors will be very resistant to automation for non-capability-related reasons. We are nowhere near figuring out how to make AI granularly effective in some sectors – hard to make AI capable in very specific sectors. For example, food service: you’d need to be good at forecasting traffic, sorting customer preferences, understanding tariff-related spikes in ingredient costs, 100 different things that go into making a restaurant successful (without even getting into physical tasks). Will AI be useful in running service businesses? Unclear. Education is another interesting example: we have wonderful tools like Khan Academy, studies show good results, but not being brought into schools. Entrenched interests in many fields will slow adoption. They have a whole team that’s engaged in [adoption], she’ll look into whether they have any insights to share. She hasn’t been paying close attention to their work. It’s been clear from a lot of her projects: a large barrier to many people’s usage is language. There are 10,000s of dialects that we’re not on track for any model to speak. Need transcripts + voice recordings, and many of these languages don’t even have a written form. === [2503.14499] Measuring AI Ability to Complete Long Tasks (https://arxiv.org/abs/2503.14499) === Happy Open Access Week from arXiv! YOU make open access possible! Tell us why you support #openaccess and give to arXiv this week to help keep science open for all. Donate! Skip to main content We gratefully acknowledge support from the Simons Foundation, member institutions , and all contributors. Donate > cs > arXiv:2503.14499 Help | Advanced Search All fields Title Author Abstract Comments Journal reference ACM classification MSC classification Report number arXiv identifier DOI ORCID arXiv author ID Help pages Full text Search open search GO open navigation menu quick links Login Help Pages About --> Computer Science > Artificial Intelligence arXiv:2503.14499 (cs) [Submitted on 18 Mar 2025 ( v1 ), last revised 30 Mar 2025 (this version, v2)] Title: Measuring AI Ability to Complete Long Tasks Authors: Thomas Kwa , Ben West , Joel Becker , Amy Deng , Katharyn Garcia , Max Hasin , Sami Jawhar , Megan Kinniment , Nate Rush , Sydney Von Arx , Ryan Bloom , Thomas Broadley , Haoxing Du , Brian Goodrich , Nikola Jurkovic , Luke Harold Miles , Seraphina Nix , Tao Lin , Neev Parikh , David Rein , Lucas Jun Koba Sato , Hjalmar Wijk , Daniel M. Ziegler , Elizabeth Barnes , Lawrence Chan View a PDF of the paper titled Measuring AI Ability to Complete Long Tasks, by Thomas Kwa and 24 other authors View PDF HTML (experimental) Abstract: Despite rapid progress on AI benchmarks, the real-world meaning of benchmark performance remains unclear. To quantify the capabilities of AI systems in terms of human capabilities, we propose a new metric: 50%-task-completion time horizon. This is the time humans typically take to complete tasks that AI models can complete with 50% success rate. We first timed humans with relevant domain expertise on a combination of RE-Bench, HCAST, and 66 novel shorter tasks. On these tasks, current frontier AI models such as Claude 3.7 Sonnet have a 50% time horizon of around 50 minutes. Furthermore, frontier AI time horizon has been doubling approximately every seven months since 2019, though the trend may have accelerated in 2024. The increase in AI models' time horizons seems to be primarily driven by greater reliability and ability to adapt to mistakes, combined with better logical reasoning and tool use capabilities. We discuss the limitations of our results -- including their degree of external validity -- and the implications of increased autonomy for dangerous capabilities. If these results generalize to real-world software tasks, extrapolation of this trend predicts that within 5 years, AI systems will be capable of automating many software tasks that currently take humans a month. Subjects: Artificial Intelligence (cs.AI) ; Machine Learning (cs.LG) Cite as: arXiv:2503.14499 [cs.AI] (or arXiv:2503.14499v2 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2503.14499 Focus to learn more arXiv-issued DOI via DataCite Submission history From: Thomas Kwa [ view email ] [v1] Tue, 18 Mar 2025 17:59:31 UTC (27,258 KB) [v2] Sun, 30 Mar 2025 17:53:28 UTC (27,258 KB) Full-text links: Access Paper: View a PDF of the paper titled Measuring AI Ability to Complete Long Tasks, by Thomas Kwa and 24 other authors View PDF HTML (experimental) TeX Source view license Current browse context: cs.AI < prev | next > new | recent | 2025-03 Change to browse by: cs cs.LG References & Citations NASA ADS Google Scholar Semantic Scholar export BibTeX citation Loading... BibTeX formatted citation × loading... Data provided by: Bookmark Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer ( What is the Explorer? ) Connected Papers Toggle Connected Papers ( What is Connected Papers? ) Litmaps Toggle Litmaps ( What is Litmaps? ) scite.ai Toggle scite Smart Citations ( What are Smart Citations? ) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv ( What is alphaXiv? ) Links to Code Toggle CatalyzeX Code Finder for Papers ( What is CatalyzeX? ) DagsHub Toggle DagsHub ( What is DagsHub? ) GotitPub Toggle Gotit.pub ( What is GotitPub? ) Huggingface Toggle Hugging Face ( What is Huggingface? ) Links to Code Toggle Papers with Code ( What is Papers with Code? ) ScienceCast Toggle ScienceCast ( What is ScienceCast? ) Demos Demos Replicate Toggle Replicate ( What is Replicate? ) Spaces Toggle Hugging Face Spaces ( What is Spaces? ) Spaces Toggle TXYZ.AI ( What is TXYZ.AI? ) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower ( What are Influence Flowers? ) Core recommender toggle CORE Recommender ( What is CORE? ) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs . Which authors of this paper are endorsers? | Disable MathJax ( What is MathJax? ) About Help contact arXiv Click here to contact arXiv Contact subscribe to arXiv mailings Click here to subscribe Subscribe Copyright Privacy Policy Web Accessibility Assistance arXiv Operational Status === If AGI Means Everything People Do... What is it That People Do? (https://amistrongeryet.substack.com/p/where-ai-falls-short) === Second Thoughts Subscribe Sign in If AGI Means Everything People Do... What is it That People Do? And Why Are Today’s "PhD" AIs So Hard To Apply To Everyday Tasks? Steve Newman Feb 27, 2025 38 15 7 Share “Strange how AI may solve the Riemann hypothesis 1 before it can reliably plan me a weekend trip to Boston” – Jack Morris There’s a huge disconnect between AI performance on benchmark tests, and its applicability to real-world tasks. It’s not that the tests are wrong; it’s that they only measure things that are easy to measure. People go around arguing that AIs which can do everything people can do may arrive as soon as next year 2 . And yet, no one in the AI community has bothered to characterize what people actually do! The failure to describe, let alone measure, the breadth of human capabilities undermines all forecasts of AI progress. Our understanding of how the world functions is calibrated against the scale of human capabilities. Any hope of reasoning about the future depends on understanding how models will measure up against capabilities that aren’t on any benchmark, aren’t in any training data, and could turn out to require entirely new approaches to AI. I’ve been consulting with experts from leading labs, universities and companies to begin mapping the territory of human ability. The following writeup, while just a beginning, benefits from an extended discussion which included a senior staff member at a leading lab, an economist at a major research university, an AI agents startup founder, a senior staffer at a benchmarks and evals organization, a VC investing heavily in AI, the head of an economic opportunity nonprofit, and a senior technologist at a big 5 tech company. What Is the Gamut of Work? Most contemplated impacts of AI involve activities that, when done by people, fall under the heading of paid work. Advancing science? Generating misinformation? Answering questions? Conducting a cyberattack? These are all things that people get paid to do. Of course, if and when an AI acquires the skills needed for some job, that doesn’t necessarily mean that we should or would give that job to the AI. People may prefer to talk to a human therapist, or to know that the song they’re listening to came from a human throat. And we may choose to preserve some jobs because people derive meaning from doing that work. But we can still use jobs as an intuition pump to begin enumerating the remarkable range of human capabilities. Studies of AI performance in economically relevant activities center on a few specific jobs and tasks, such as writing computer code. But there is enormous variety in the kinds of things people do. Just to name a few: therapist, tutor, corporate lawyer, advice nurse, social worker, call center worker, salesperson, journalist, graphic designer, product designer, engineer, IT worker, research analyst, research scientist, career coach, manager, and CEO. Of course, a complete list would go on for pages. And I’ve only mentioned paid jobs, with no substantial physical component. What I’ve tried to do here is to sample the range of non-physical jobs, touching on different kinds of work that require different sorts of skills. Even so, I’m sure there are gaping holes, and I’d appreciate help in filling them in. What jobs require capabilities or aptitudes (not just domain knowledge) not touched on here? In any case, simply listing jobs doesn’t tell us anything about AI timelines. The next step is to ask: could current AIs do these jobs unassisted? Where would they struggle? Again, the goal is not to hasten human obsolescence; it’s to shine a light on capabilities that are missing from current AIs. Where Might Current AIs Struggle? Why, exactly, can’t AI plan a weekend trip? Where would it go wrong? And what other tasks are beyond the reach of current systems? Here are some real-world challenges that current and near-future AIs seem likely to struggle with. Building a successful business . Mustafa Suleyman, co-founder of DeepMind and now CEO of Microsoft AI, has proposed a “new Turing test” : turning $100,000 in seed capital into $1,000,000 in profit. Some challenges for AI: Judgement : selecting a good, differentiated idea. Deciding when to stay the course and when to pivot. Setting priorities and allocating resources. Hiring: interviewing candidates and selecting the best fit for a job . Avoiding being cheated / scammed : it might be easy to scam the AI in various ways, especially if it were “outed” as an AI. Managing long-term context : keeping track of everything that has taken place over the course of the business, learning from experience, and juggling activities that take place on different timelines. Planning a child’s summer activities ( proposed by Kathy Pham ). This requires understanding their interests, sorting out camp availability, and coordinating with the schedules of friends, family, and other caregivers. Some ways this might go wrong: Consultation and input management : managing the conversations with everyone involved – what information to ask for, how to ask, when to follow up. When to make assumptions and when to verify. Maintaining coherence over an asynchronous conversation. Detail-orientation and complexity: tracking hundreds of constraints (camp location, start times, end times, ages accepted, vacation schedules, etc.); proposing solutions that optimize for the factors the clients care about most, without making any errors. Writing a high-quality blog post about a specified topic. Insight : synthesizing facts into a crisp, memorable insight that brings the detailed picture into focus for the reader. Engaging writing : identify an anchor – such as a topical policy question, or a story from history – that naturally fits with the ideas being presented. Organize the post around that theme in a coherent, appealing, and natural-sounding way. Support specialist at an enterprise software company . Learning on the job : teaching itself to recognize and resolve common problems, and to navigate the knowledge base and other internal resources. Ability to learn about and remember new features and bugs as they are introduced (and resolved) on an ongoing basis. Getting help: deciding when to escalate, who to get help from, and exactly what help to ask for (making efficient use of the helper’s time). I believe that current AIs would struggle in all of these areas. This leads to the question of why – what capabilities are required, that existing models lack? What Are The Key Missing Capabilities? Progress in AI is often aimed at improving specific capabilities, such as working with more information at one time (increasing the “context window”), or reasoning through a complex task. Here are some capabilities that come up repeatedly when our group talked about where AIs would struggle with real-world tasks. Managing complexity: Holding the details and constraints of a complex project “in its head” when breaking the project into steps, and again while carrying out those steps. For instance: maintaining characters and detailed plot continuity while writing a novel, or writing code that meshes with a large existing codebase. ( Ethan Mollick : “I have spent a lot of time with AI agents (including Devin and Claude Computer Use) and they really do remain too fragile & not "smart" enough to be reliable for complicated tasks.”) This requires recognizing that much information is not relevant for any particular subtask. David Crawshaw : Avoid creating a situation with so much complexity and ambiguity that the LLM gets confused and produces bad results. This is why I have had little success with chat inside my IDE. My workspace is often messy, the repository I am working on is by default too large, it is filled with distractions. One thing humans appear to be much better than LLMs at (as of January 2025) is not getting distracted. Metacognition and dynamic planning : Stitching together different skills, including high- and low-level planning, to accomplish a complex task. Monitoring conditions throughout the course of a project, and exercising good judgment about when to proceed according to plan and when to adjust course. This should reflect both internal conditions (failing to make progress) and external conditions (the situation has changed). Metacognition also includes understanding its own strengths and weaknesses, what it knows and doesn’t know. Accurately evaluating and reporting its confidence in a fact or judgement. Seeking outside information / help where appropriate. Noticing when it is getting stuck, repeatedly making the same error, or external conditions have changed. Learning and memory: “Learning the ropes” of a new job, or assimilating a user’s preferences and circumstances. Incorporate this information (continuous sample-efficient learning) so that the model can act on it in the future as effectively as if it had been in the training data. (As Soren Larson puts it : “Agents saving users time is downstream of context availability not capability” – like a personal assistant in their first day on the job, an AI agent may not be of much use until it can learn your preferences.) Leveraging external information: Making effective use of large unstructured information pools: a user’s email, a corporate intranet, the Internet. Intuiting what information might be available; weighting sources by credibility; ignoring extraneous information. Judgement: Setting priorities and allocating resources. Making good high-level decisions, such as which research avenues to explore. Deciding when and how to push back in a discussion or negotiation; reconciling different stakeholders’ conflicting preferences. Creativity and insight: Coming up with novel insights, compelling explanatory metaphors, and non-trite storytelling. Are These Serious Challenges to AI? That’s a nice list of capabilities. Does it tell us anything useful about the timeline for AGI? In our discussion, there were a lot of questions about how fundamental these gaps are. For instance, would current models really struggle to “synthesize facts into a crisp, memorable insight”? Could this be addressed with a well-designed prompt? Perhaps some external scaffolding would be sufficient: ask the model for 20 separate insights, and then ask it to pick the best one? Maybe we need to wait for the next generation of models, trained on more data. Or perhaps some fundamental architectural advance will be needed. You can ask similar questions about any item on the list. What will it take for models to assimilate new information on the fly? To successfully coordinate activities that take place on different timelines? To know when to ask for help? We didn’t get to dive into every dimension listed above, but there was broad consensus that one of the biggest barriers – one of the last sandbags holding back the tide of AI exceeding human capabilities – will be learning and memory. The expectation was that continuous, sample-efficient learning and sophisticated memory are important requirements for many practical tasks, and that meeting these requirements will require significant architectural advances. Benchmarks like FrontierMath and Humanity’s Last Exam are very difficult for humans, but don’t capture these fuzzier skills. We don’t know of benchmarks that require, for instance, insightful writing, judgement, or continuous learning. Open Philanthropy has begun funding work on benchmarks to measure capability at real-world tasks , but the field is in very early days. We hope to explore these questions further. How can we usefully measure an AI’s ability to ask clarifying questions, or to make judgement calls as it attempts to build a business? How can we identify leading indicators for progress in these areas? We Cannot Predict What We Cannot Describe People love to speculate as to when AIs will outmatch humans at anything we might want them to do (or fear that they might do). I continue to be surprised at how disconnected these conversations are from any discussion of what those things are and what “doing them” entails. The next time someone predicts that we will soon have AI that can do anything “a human can do”, ask them what they think a human can do. You may find that they are only considering capabilities that are easily measured. AIs keep racing to 100% on benchmarks, because they are easiest to train on tasks where success is easy to measure. For a complete picture of AI timelines, we need to shed some light on those capabilities that are not easily measured. Please share your answers to this question: what tasks would pose the greatest challenge for current AIs, and why? Share Subscribe Thanks to Carey Nachenberg, Rachel Weinberg, Taren Stinebrickner-Kauffman, and all the participants in our dinner conversation. 1 A major unsolved problem in mathematics. 2 For instance, Anthropic CEO Dario Amodei recently put it quite plainly : An AI model that can do everything a human can do at the level of a Nobel laureate across many fields... my guess is that we’ll get that in 2026 or 2027. 38 15 7 Share Discussion about this post Comments Restacks Player1 Feb 27, 2025 Liked by Steve Newman As a mathematician, I am annoyed by the common assumption that proving the Riemann hypothesis *doesn't* require managing complexity, metacognition, judgement, learning+memory, and creativity/insight/novel heuristics. Certainly, if a human were to establish a major open conjecture, in the process of doing so they would demonstrate all of these qualities. I think people underestimate the extent to which a research project (in math, or in science) differs from an exam question that is written by humans with a solution in mind. Perhaps AI will be able to answer major open questions through a different, more brute-force method, as in chess. But chess is qualitatively very different from math: to play chess well requires much greater calculational ability than many areas of math. (At the end of the day, chess has no deep structure). Also, prediction timelines for the Riemann Hypothesis or any specific conjecture are absurd. For all we know, we could be the same situation as Fermat in the 1600's, where to prove the equation a^n + b^n = c^n has no solutions you might need to invent modular forms, etale cohomology, deformation theory of Galois representations, and a hundred other abstract concepts that Fermat had no clue about. (Of course, there is likely some alternate proof out there, but is it really much simpler?). It is possible that we could achieve ASI and complete a Dyson sphere before all the Millenium problems are solved-- math can be arbitrarily hard. Expand full comment Reply Share 1 reply by Steve Newman Vaughn Tan Mar 6, 2025 Liked by Steve Newman I've been thinking about this question for a while! When we consider what humans "actually" do, we often look at tasks and their outputs. A different way to consider the question is to understand the subjective valuations put on tasks and their outputs — my belief is that this alternative is superior because it provides clearer discrimination between [essentially human actions] and [actions which humans currently do which could be done better by machines]. I call this act of deciding and assigning subjective value "meaningmaking." A writer choosing this word (and not that word) to achieve the correct tone for a blogpost is engaging in an act of meaningmaking — the choice of word is the result of deciding that one word is subjectively better than another in conveying the chosen tone for the intended audience. These meaningmaking acts are everywhere in daily life, corporate life, and public life. Deciding that this logo (and not that logo) is a better vehicle for corporate identity — meaningmaking. Choosing to hire this person (but not that person) because they are a better culture fit — meaningmaking. Ruling that this way of laying off lots of government employees (and not that way of doing it) is unlawful — meaningmaking. Humans do 4 types of meaningmaking all the time: Type 1: Deciding that something is subjectively good or bad. “Diamonds are beautiful,” or “blood diamonds are morally reprehensible.” Type 2: Deciding that something is subjectively worth doing (or not). “Going to college is worth the tuition,” or “I want to hang out with Bob, but it’s too much trouble to go all the way to East London to meet him.” Type 3: Deciding what the subjective value-orderings and degrees of commensuration of a set of things should be. “Howard Hodgkin is a better painter than Damien Hirst, but Hodgkin is not as good as Vermeer,” or “I’d rather have a bottle of Richard Leroy’s ‘Les Rouliers’ in a mediocre vintage than six bottles of Vieux Telegraphe in a great vintage.” Type 4: Deciding to reject existing decisions about subjective quality/worth/value-ordering/value-commensuration. “I used to think the pizza at this restaurant was excellent, but after eating at Pizza Dada, I now think it is pretty mid,” or “Lots of eminent biologists believe that kin selection theory explains eusociality, but I think they are wrong and that group selection makes more sense.” At the moment, I cannot see a way for an AI system do meaningmaking work. I've quoted a lot from an article i wrote on the problem AI systems (and machines more generally) have with meaningmaking: https://uncertaintymindset.substack.com/p/ai-meaningmaking . It's part of a longer series of essays about how the meaningmaking lens helps us understand what AI can and should be used for (and what it can't do and should not be used for): https://vaughntan.org/meaningmakingai Very much a work in progress so would love comments and suggestions from this community. Expand full comment Reply Share 2 replies by Steve Newman and others 13 more comments... Top Latest Discussions No posts Ready for more? Subscribe © 2026 Steve Newman · Privacy ∙ Terms ∙ Collection notice Start your Substack Get the app Substack is the home for great culture === A Bear Case: My Predictions Regarding AI Progress — LessWrong (https://www.lesswrong.com/posts/oKAFFvaouKKEhbBPm/a-bear-case-my-predictions-regarding-ai-progress) === x LESSWRONG is fundraising! LW Login A Bear Case: My Predictions Regarding AI Progress — LessWrong AI Risk AI Timelines AI World Modeling Curated 2025 Top Fifty: 14 % 377 A Bear Case: My Predictions Regarding AI Progress by Thane Ruthenis 5th Mar 2025 11 min read 163 377 This isn't really a "timeline", as such – I don't know the timings – but this is my current, fairly optimistic take on where we're heading. I'm not fully committed to this model yet: I'm still on the lookout for more agents and inference-time scaling later this year. But Deep Research, Claude 3.7, Claude Code, Grok 3, and GPT-4.5 have turned out largely in line with these expectations [1] , and this is my current baseline prediction. The Current Paradigm: I'm Tucking In to Sleep I expect that none of the currently known avenues of capability advancement are sufficient to get us to AGI [2] . I don't want to say the pretraining will "plateau", as such, I do expect continued progress. But the dimensions along which the progress happens are going to decouple from the intuitive "getting generally smarter" metric, and will face steep diminishing returns. Grok 3 and GPT-4.5 seem to confirm this. Grok 3's main claim to fame was "pretty good: it managed to dethrone Claude Sonnet 3.5.1 for some people!". That was damning with faint praise. GPT-4.5 is subtly better than GPT-4, particularly at writing/EQ. That's likewise a faint-praise damnation: it's not much better. Indeed, it reportedly came out below expectations for OpenAI as well, and they certainly weren't in a rush to release it. (It was intended as a new flashy frontier model, not the delayed, half-embarrassed "here it is I guess, hope you'll find something you like here".) GPT-5 will be even less of an improvement on GPT-4.5 than GPT-4.5 was on GPT-4. The pattern will continue for GPT-5.5 and GPT-6, the ~1000x and 10000x models they may train by 2029 (if they still have the money by then). Subtle quality-of-life improvements and meaningless benchmark jumps, but nothing paradigm-shifting. (Not to be a scaling-law denier. I believe in them, I do! But they measure perplexity , not general intelligence/real-world usefulness, and Goodhart's Law is no-one's ally.) OpenAI seem to expect this, what with them apparently planning to slap the "GPT-5" label on the Frankenstein's monster made out of their current offerings instead of on, well, 100x'd GPT-4. They know they can't cause another hype moment without this kind of trickery. Test-time compute/RL on LLMs: It will not meaningfully generalize beyond domains with easy verification . Some trickery like RLAIF and longer CoTs might provide some benefits, but they would be a fixed-size improvement. It will not cause a hard-takeoff self-improvement loop in "soft" domains. RL will be good enough to turn LLMs into reliable tools for some fixed environments/tasks. They will reliably fall flat on their faces if moved outside those environments/tasks. Scaling CoTs to e. g. millions of tokens or effective-indefinite-size context windows (if that even works) may or may not lead to math being solved. I expect it won't. It may not work at all: the real-world returns on investment may end up linear while the costs of pretraining grow exponentially. I mostly expect FrontierMath to be beaten by EOY 2025 ( it's not that difficult ), but maybe it won't be beaten for years. [3] Even if it "technically" works to speed up conjecture verification, I'm skeptical on this producing paradigm shifts even in "hard" domains. That task is not actually an easily verifiable one. ( If math is solved, though, I don't know how to estimate the consequences, and it might invalidate the rest of my predictions. ) "But the models feel increasingly smarter!": It seems to me that "vibe checks" for how smart a model feels are easily gameable by making it have a better personality. My guess is that it's most of the reason Sonnet 3.5.1 was so beloved. Its personality was made much more appealing , compared to e. g. OpenAI's corporate drones. The recent upgrade to GPT-4o seems to confirm this. They seem to have merely given it a better personality, and people were reporting that it "feels much smarter". Deep Research was this for me, at first. Some of its summaries were just pleasant to read, they felt so information-dense and intelligent! Not like typical AI slop at all! But then it turned out most of it was just AI slop underneath anyway, and now my slop-recognition function has adjusted and the effect is gone. What LLMs are good at: eisegesis-friendly problems and in-distribution problems. Eisegesis is "the process of interpreting text in such a way as to introduce one's own presuppositions, agendas or biases". LLMs feel very smart when you do the work of making them sound smart on your own end: when the interpretation of their output has a free parameter which you can mentally set to some value which makes it sensible/useful to you. This includes e. g. philosophical babbling or brainstorming. You do the work of picking good interpretations/directions to explore, you impute the coherent personality to the LLM. And you inject very few bits of steering by doing so, but those bits are load-bearing . If left to their own devices, LLMs won't pick those obviously correct ideas any more often than chance. See R1's CoTs, where it often does... that . This also covers stuff like Deep Research's outputs . They're great specifically as high-level overviews of a field, when you're not relying on them to be comprehensive or precisely on-target or for any given detail to be correct. It feels like this issue is easy to fix. LLMs already have ~all of the needed pieces, they just need to learn to recognize good ideas! Very few steering-bits to inject! This issue felt easy to fix since GPT-3.5, or perhaps GPT-2 . This issue is not easy to fix. In-distribution problems: One of the core features of the current AIs is the "jagged frontier" of capabilities. This jaggedness is often defended by "ha, as if humans don't have domains in which they're laughably bad/as if humans don't have consistent cognitive errors!". I believe that counterargument is invalid. LLMs are not good in some domains and bad in others. Rather, they are incredibly good at some specific tasks and bad at other tasks. Even if both tasks are in the same domain, even if tasks A and B are very similar, even if any human that can do A will be able to do B. This is consistent with the constant complaints about LLMs and LLM-based agents being unreliable and their competencies being impossible to predict ( example ). That is: It seems the space of LLM competence shouldn't be thought of as some short-description-length connected manifold or slice through the space of problems, whose shape we're simply too ignorant to understand yet. (In which case "LLMs are genuinely intelligent in a way orthogonal to how humans are genuinely intelligent" is valid.) Rather, it seems to be a set of individual points in the problem-space, plus these points' immediate neighbourhoods... Which is to say, the set of problems the solutions to which are present in their training data . [4] The impression that they generalize outside it is based on us having a very poor grasp regarding the solutions to what problems are present in their training data. And yes, there's some generalization. But it's dramatically less than the impressions people have of it. Agency: Genuine agency, by contrast, requires remaining on-target across long inferential distances : even after your task's representation becomes very complex in terms of the templates which you had memorized at the start. LLMs still seem as terrible at this as they'd been in the GPT-3.5 age. Software agents break down once the codebase becomes complex enough, game-playing agents get stuck in loops out of which they break out only by accident, etc. They just have bigger sets of templates now, which lets them fool people for longer and makes them useful for marginally more tasks. But the scaling on that seems pretty bad, and this certainly won't suffice for autonomously crossing the astronomical inferential distances required to usher in the Singularity. "But the benchmarks!" I dunno, I think they're just not measuring what people think they're measuring . See the point about in-distribution problems above, plus the possibility of undetected performance-gaming , plus some subtly but crucially unintentionally-misleading reporting . Case study: Prior to looking at METR's benchmark , I'd expected that it's also (unintentionally!) doing some shenanigans that mean it's not actually measuring LLMs' real-world problem-solving skills. Maybe the problems were secretly in the training data, or there was a selection effect towards simplicity, or the prompts strongly hinted at what the models are supposed to do, or the environment was set up in an unrealistically "clean" way that minimizes room for error and makes solving the problem correctly the path of least resistance (in contrast to messy real-world realities), et cetera. As it turned out, yes, it's that last one: see the "systematic differences from the real world" here . Consider what this means in the light of the previous discussion about inferential distances/complexity-from-messiness. As I'd said, I'm not 100% sure of that model. Further advancements might surprise me, there's an explicit carve-out for ??? consequences if math is solved , etc. But the above is my baseline prediction, at this point, and I expect the probability mass on other models to evaporate by this year's end. Real-World Predictions I dare not make the prediction that the LLM bubble will burst in 2025, or 2026, or in any given year in the near future. The AGI labs have a lot of money nowadays, they're managed by smart people, they have some real products, they're willing to produce propaganda, and they're buying their own propaganda (therefore it will appear authentic). They can keep the hype up for a very long time, if they want. And they do want to. They need it, so as to keep the investments going. Oceans of compute is the only way to collect on the LLM bet they've made, in the worlds where that bet can pay off, so they will