> If LLMs were around when I was a student, I would've also used them to "speed up" my homework assignments then proceed to fail all my tests. You go to a university because you are deeply interested in understanding the subject that you study. Doing the homework and the tests are just the "goalposts" to check for yourself whether you made progress on this. So, as long as you are not under time pressure (which you in some degree courses unluckily are), there is simply no need to "speed up" any homework assignments. If, on the other hand, LLMs help you with making much faster progress in understanding the subject that you study (which is only loosely correlated to homework and tests), I guess it's fine to use them. Just always keep in mind that very often the pain of attempting to understand the topic on your own often makes you smarter - something that you will miss when you take an "LLM shortcut".

Uni grading Brownian-walks around edu trends, but misses the point that improving one's (and humanity's) lot depends on a tiny loop: - doing - failing - discovery>learning - remembering With learning predicated on both failing and remembering it's unfortunate uni scores on 100% successful doing but doesn't teach failing well, and scores for remembering but not for learning well.

> If, on the other hand, LLMs help you with making much faster progress in understanding the subject that you study My experience is that they uncomfortably do both. You can "understand" something conceptually quicker -- like you have a new brain-muscle-thing that lets you cut through the hard difficult tedious corners to get to the meat of the matter. But then you also can become reliant on it, and have difficulty doing the mechanistic rote work of working through it yourself. Like the really big powerful calculator that it is, really.

> If LLMs were around when I was a student, I would've also used them to "speed up" my homework assignments then proceed to fail all my tests. I agree - I would have been toast. I wonder if the teachers/colleges need to change the way they teach and assess. Let the students use the AI tools they like (perhaps guide them how they can use them professionally), but test regularly and early on the skills/knowledge they're meant to be gaining offline and in person. Oh and don't give Fs for cheating - suspend them. I read a few years ago about a teacher (I think highschool) who put his lectures on YouTube for students to view in their own time and then used the in class hours for interaction, questions, tests. EDIT: Claude beat my Googling: This was 2 chemistry high school teachers in 2007 - The Flipped Classroom https://fltmag.com/the-flipped-classroom/

I’m not saying only teach interesting things, I’m saying teach things in an interesting and engaging way so that kids don’t feel the need to cheat their way through it to just get it done. College is a different dynamic from a middle/high school classroom, but I don’t remember 95% of the material from my college engineering classes anyway, it’s the problem solving and information finding that I’ve retained and have helped me do the things I do. I remember the stuff from the classes that taught me the material in an engaging way though.

True. People don't do it though, because keeping skills sharp and using them takes effort, and we have a predisposition to be as efficient as possible with how we spend our effort; if there's an easier way to do it in our awareness, we will naturally gravitate towards that. LLMs are often a universal crutch or swiss-army-knife that significantly take away workload for many abstract tasks, so all kinds of atrophy in abstract thinking is to be expected. However, when looking at muscle, once you have it you don't need to use it as much in order to maintain it. I wonder if the same is true for skills; in that case, some kind of regiment where you still use the skill you delegate once a week or so could maybe help with avoiding this loss of skill for most part.

“ However, when looking at muscle, once you have it you don't need to use it as much in order to maintain it” No.. this depends on how much muscle you have. The appropriate comparison is mass and density of knowledge/understanding vs muscle. There’s not a chance in hell you will retain mass and dense muscle without pushing the body hard. Just in the same way you will not retain very deep understanding of things unless a) you’ve been reciting it for over 10 yrs b) you go back and push the understanding continuously for it to remain as part of your being

Building muscle is much harder than maintaining muscle. And if you went 3 years without exercising, you'll be able to get your muscles back much quicker than had you never had the muscle before. It's pretty comparable to skills. You don't need to practice as hard to maintain a skill than you do to build it. And if you let the skill atrophy, it's much easier to recover the skill compared to building it from scratch.

I'm dumb as a rock and I don't have a PhD, but since ~1 year ago I started forcing myself to do small bits of coding and math manually. I'm not noticing a "cognitive decline" per se, but I do see I'm a lot "lazier", even stuff that used to be routine when I started coding now feel heavy.

> but I do see I'm a lot "lazier", even stuff that used to be routine when I started coding now feel heavy. Not getting that quick dopamine hit the LLMs give you.. Some say you can re-train your system to get back the dopamine hits you used to get from other things, like the enjoyment of the "old fashioned" manual coding and math. Getting there is hard work. And YMMV.

>even stuff that used to be routine when I started coding now feel heavy. The same weight feeling heavier is a sign that your muscles are weaker :) There's many areas in life were we look back a few decades and think "people use to do it that awkwardly?" And yet results were better. I think the process of removing friction have just served to destroy our ability to concentrate and tolerate difficulty.

I do a similar version of this, where if I notice a mistake in generated code, I fix it manually (or at least attempt to) instead of telling Claude to fix it.

This is the right balance for me as well. I use an agent to generate a first-pass attempt, and then (deadlines willing), I manually read every line at least once so I understand what the code actually does. Then I manually fix the inevitable slop that is mixed in with the good stuff, and only once the code is up to my personal standards do I send it. This probably reduces my “AI performance boost” to 30-50% instead of the huge gains reported by others. But I retain the ability to reason about the codebase and use AI much more precisely when I’m trying to troubleshoot production outages or subtle bugs — something I notice the rest of my team struggles with, since adopting “agentic workflows” everywhere. I think actively working to retain some cognitive flexibility and “muscle memory” around coding tasks is going to be rather advantageous in the long run.

When LLMs and ChatGPT first came out, it struck me as obvious and dangerous to a deep thinker or a knowledge worker the answering capacity. So, from my initial use I did not ask them questions, I have always "done my own work" and then asked the LLMs to criticize that work. This has been an exponential ladder of learning, and my cognitive growth is personally noticeable. I'm not hesitating to scribble out calculus and work it out, as I need for my work, where in the past I'd have found some other way because I felt uncomfortable with my tip-of-my-tongue calc skills. Don't ask AI, do your own work and ask for criticism, and them improve your own work yourself. This creates a learning ladder that you will climb.

i use claude a lot and i find that it is best applied in domains in which i am already a master. I tried applying it to domain's im unfamiliar with and i found that i produced stuff but as time went on i understood what i produced less and i almost felt like i do after binge watching a netflix show, 2 weeks later i barely remember any of the details. I wonder how much you need to "do" to learn and remember. LLM's give you a shortcut to doing and so you probably aren't learning either. It's like when you watch a professor write a proof and it makes sense while listening to the professor but at home you have difficulty deriving it. LLM's give me the same sort of feeling. I think the way forward is still going to be doing things manually to learn and using LLM's once you've mastered an area and people who don't understand this fact are going to slowly descend down a hill and forget how to depend on their own thinking.

LLMS didn’t invent cheating just made it easier. When you cheat you’re the one who cheats yourself because the point of an education is to learn, not complete the assignments and get high marks on tests alone. No one benefits and no one other than you is materially hurt by cheating, but you are absolutely the one who is hurt. There’s no way to learn than to force the brain into adaptation which it is resistant to do through challenge and stress, just like your muscles. Similarly you can’t play e sports and get into physical condition any more than you can use LLMs to do your homework and learn. It’s going to be a hard adjustment for a lot of people to recognize that letting the machine think for you is as healthy as smoking brain cigarettes. The smart student uses the LLM as a proctor or provide challenges and feedback on attempts rather than an easy button. They make great tools for learning if they’re used as an adversarial or editorial tool. The future belongs to those who work to use the tools in ways that make themselves more efficacious, not those who use efficacious tools so they don’t have to work.

>The smart student uses the LLM as a proctor or provide challenges and feedback on attempts rather than an easy button. Yeah, this is how we used wolframalpha for Math as students. Whatever we had to do, we did it ourself as a group of three. Afterwards we checked with Wolframaplha to see if we were correct. If there were any difference between us, we went line by line to find where the error appeared. It was helpful, because we did it ourself, but because the work was graded, we had the security, that it is not a total failure.

My son just finished his first year in college, and had no trouble getting decent grades without using AI while many of the kids around him were using it. At least in his humanities track, class participation is a lot of his grade, and he said the "AI kids" tended to suck at participation because they hadn't actually thought about the material, and couldn't dynamically work with it in class. He also said their AI assisted writing that he'd read was dull and unoriginal, and all sounded the same, which he thought likely helped his essays stand out. His English composition teacher said he was "probably too advanced for this class" when he told her he didn't use AI to write his essays, which made him roll his eyes, as he has clinically diagnosed dysgraphia (learning disability in writing).

You can ask LLMs about high-level techniques, and their answers will usually be good enough. What you can't get from LLMs is the taste and judgment, which you can only obtain by having a strong CS base and coding manually for years. High-level techniques were never a problem. You could Google tens of articles on this topic. They are useless too, it's like learning how to drive a racing bicycle from reading a book. Sure, you will know a lot about nuances, but you will fail miserably when it comes to a real race.

No one seriously expects a food critic to be able to cook a Michelin-starred meal. The job of that kind of critic is to be insightful and entertaining, and it's very different to the taste required to create top quality food, which is a combination of solid technical skill and creative flair. Taste in coding is a combination of insight, experience, native talent, technical skill, and flair. Tasteful coding produces clever but straightforward minimal elegant solutions that an average developer can't imagine but can adapt and maintain.

I can't speak for other disciplines, but for math and CS, both with a really heavy focus on abstraction, the final result of learning is to build a nice intuition on top of the abstractions we find useful/expressive. And to build the intuition, the old, usual, and perhaps the only way is to see and practice a lot of concrete examples, after which the motivation of building some abstraction can be understood, and after which the abstraction itself can be fully grasped. e.g. The "group" abstraction requires one see a lot of int, polynomial, modular arithmetic etc. before knowing why we want such a thing. It's unskippable.

> As they pursue the topic they start to get down into the details, although probably never learn to do it fully independently. It's hard to claim one has mastered a subject without independent command of its fundamentals. A less charitable take on this future is that students only learn to hand-wave answers and correspondingly cannot evaluate statements beyond "sounds about right".

The place I've come to with AI for writing is to have an idea for a chapter/article/etc, which I take to AI, and tell it to either ask me a bunch of clarifying questions, or try to blow holes in it/challenge it. I'll keep talking to AI and answering questions/handling challenges until the AI runs out of steam, then I'll ask the AI to write out a condensed outline with all the pertinent details of the conversation. Once I have the condensed outline, I'll re-order stuff, clean it up/tune it up, then do the final writing. This keeps my voice and logical train of thought while avoiding blank page syndrome and some of the organizational mess of condensing notes into an outline manually.

We also had exercises for which the solutions were given, and we didn't reach for them immediately...

I totally agree about school-level homework: it was many years before my pre-frontal cortex developed enough that I could have forced myself to do the work. That said, though, one thing I don't understand about the heavy users of AI in academia and software development is that the thinking and coding is the fun part . And that's the part so many people seem to be so keen to automate away.

Isn't the fun part having the thing work how you wanted it to? Why shouldn't I be keen to automate the process away?

Depends on the person. I find that it's extremely satisfying to figure out a tricky problem on the way to that end result - to struggle with something for a bit, then finally fix it or fully wrap my head around it. So to me, it's a mixture of both. What I want is the end result, but in the past sometimes that came with thinking in the shower about an approach... Or a wild thought while going to bed that makes me jump up and grab my computer. That doesn't happen for me anymore to the same degree.

I've read enough comments* on HN to know that there are different camps. Some people don't really enjoy the process of development and just want results. Meanwhile, telling me to automate away the problem solving aspect of software dev is like saying "you know you can just copy the answers to the crossword from the back of the book?" *speaking of things I should be doing less of...

different strokes for different folks. I'm def. in that end result camp, i get the biggest thrill out of seeing something work. For me, coding agents are awesome because i can bring a lot more to life in much shorter of a time frame. I do enjoy the process and problem solving of coding, it relaxes me. On the other hand, i really really enjoy when an idea i have is on the screen and working.

Eh, I think it's less like a cigarette and more like the car. We're not going back. Americans are famously less healthy the more car dependent they are, and now people walk/run as an explicit task to be healthy. People will start going to a "thinking" gym, or engaging in additional manual mental activities for sport, like we do with chess today.

I wonder who would be working in these thinking gyms? Nice idea. Extra mural studies for the age of agentic ai.

You do realise most people aren’t in shape right? The idea that most people have the discipline to keep themselves mentally in check is false. We already know this! Millions and billions of people who spend hrs a day consuming media on platforms such as instagram.

In my own experience, the only path I truly gain intellectual benefits is the one where I work closely with the LLM, test very narrow hypotheses, and leverage it for learning over producing. Trying 5N paths is useful and sometimes yields interesting insights I’ll retain, but it’s not the rich, challenging, deeply engaging kind of process I find I need in order to develop useful knowledge and skills. So yes it’s an accelerant for people who want stuff from me, but that doesn’t map directly to learning and building skills. I think that mismatching is really important.

It’s creating a daemon and machine spirit filled world of Warhammer 40k. We already scarcely understand how the world works, but LLM use actively degrades cognitive ability that way it is used by a majority of people (The bringing a forklift to gym analogy).

The risk or difference is that tutoring helped people learn which they can use to do the work, whereas with only one or two different words an LLM will do the work (that proves you have learned) for you. A tutor has limits, but an LLM needs to be asked to set limits. And especially younger people are less likely to "punish themselves" like that.

I didn't say I'm immune to those effects, I'm including myself in this as well. (also, I'm not older than my colleagues). Most people definitely can't meditate for 30 minutes, so if you can do this, it's very impressive. Regardless, being able to think about poorly-defined problems and build completely new mental models from nothing is genuinely a really hard and uncomfortable task. If you don't use the skill you'll lose it.

> I'd rather use that brain power for other problems. Except you won't have the underpinnings to even properly think about other problems. Your brain will be mush. > I don't want to have to store a bunch of location or routing data in my head. This is preposterous.

Obviously a balance would be best, but as someone who went to a very grade-inflated school, I do believe that grade inflation gets in the way of education substantially. When you can get through classes with very little effort and understanding and know you will get a sufficient grade, many people will simply not learn the material deeply.

Some of the exams in Berkeley were brutal, but they never felt like trick questions, they did on occasion require a level of mastery of the material which was extreme, but it never felt like someone was just trying to make the questions obtuse for the sake of it.

In terms of material learned maybe, in terms of shaping logical thinking and tackling hard problems there is a huge benefit.

It seems like now’s the time to rethink how we do education. In my personal post academic life, I’ve found LLMs to be an incredible teacher. Almost like the best professor in the world at my fingertips. I use it to generate quizzes on demand to test for my own knowledge gaps. However, if I use it to speedrun over concepts I should be learning, I may achieve my end goal but I wouldn’t actually learn many of the details. I think it requires an approach where you have to continuously audit your own understanding as you work with the concepts. You must slow down until you’ve confirmed this. Only once you know the concepts deeply and have retained them in your own memory can you then go all in with the LLM.

It's too damn tempting to not use. You have a magical machine that, on command, will spit out the answer to your question in 10 seconds, whereas you'd need to spend hours to do the assignment the Good Old Fashioned Way. Even students who aren't just there for the prestigious degree are falling victim to this. When you're up against a deadline - and unless you're very good at time management you're frequently up against a deadline - it's going to be an irresistible lever to pull. In times past, cheating would mean copying an answer off the Internet or off a friend, both of which are easy to detect. More sophisticated cheaters might spend an hour rewriting the solution to make it less obvious they cheated, but at some point the cost of cheating (time + risk of getting caught) starts exceeding the cost of just doing the assignment. AI changes this - you get a customized answer that doesn't show up in a database with no extra work. The thing is, students fail to realize just what using AI robs them of. Struggling with the assignment is the entire point. You don't learn if the assignments are too easy; you need to have some challenge to push your brain to understand the material more deeply and to build those pathways to apply the knowledge in novel ways. You become more efficient and effective over time as that knowledge settles in and you get more proficient - one of the reasons why time-bounded exams still make sense (being fast is also a proxy measure for understanding).

You are wrong. Some would have failed before, but not in the larger numbers. Before when they couldn’t complete an assignment they would try different things, seek a professor, or seek out friends to help explain. You could find answer keys to many assignments online, but that doesn’t feel like learning and wouldn’t even always answer your actual misunderstanding. It wasn’t perfectly tailored to your issue all the time. Now the barrier to an answer is zero. They are basically watching a YouTube video on how to X, seeing step by step instructions feeling like they are doing it, and the moment they swing a real hammer they are whacking themselves in the crotch. It might get better after a few years, but this stuff is just now hitting mainstream for the masses. ChatGPT has only been in mainstream use for about 3 years.

AI should be a formidable booster for learning if used properly. I know that some students it to prepare for competitive tests, sometimes with very good results. I've also been using it a lot recently to brush up on my math and physics knowledge from my graduate years. It has helped me clarify and understand a lot of concepts better. That being said, there is no shortcut, and to be good at anything, one has to put in the work and the hours. However, information has never been as available as it is today.

Grades only matter as much as being able to transfer just to the real world. People can use AI to outsource their learning, but if they use ai to outsource their understanding they just set themselves up to fail even more. From what I’ve seen, how students are using ai (not that they are using ai) is making them less prepared for the real world, which unfortunately is changing faster than ever at the same time to create double impact.

It’s not that they can’t think deeply, these are smart people. It’s that there is no reward for doing so and in fact there is punishment . The punishment is that for all the thinking you do, someone else will arrive at the same result as you in less time, or maybe even a better result. You don’t get rewarded for the effort of thinking, only for the end result. Naturally, even if you are an intelligent individual, you can still be conditioned in this way to take the easy way out, unless you purposely like to suffer. But suffering is only worth it if you know in the end you come out ahead. But now, you do not come out ahead. People will be using AI in the workforce for the rest of your life anyway, might as well just join the trend. It’s like if everyone started taking a magical steroid and growth hormone to build muscle and look great instead of actually working out in a gym and possibly getting worse results anyway.

Its not teaching. These people cant pass a the class. They never went through the friction needed to learn

I think it is important to learn how to implement it because it gives the student an opportunity to learn precisely because it's been done countless times and debated over to death. There are many analyses and if one doesn't click, maybe another one will. A student can learn how to analyze the algorithms and try out different implementations to assess differences in performance. Of course, if a student just breezes through it then I would agree. That would make no sense.

Some people need Jesus, but y'all need Kant ;) "Enlightenment is man's emergence from his self-imposed nonage. Nonage is the inability to use one's own understanding without another's guidance." https://www.columbia.edu/acis/ets/CCREAD/etscc/kant.html