Failing grades soar with AI usage, dwindling math skills in Berkeley CS classes

339 comments · 24,354 words

Complete Created: Jun 4, 05:34 AM (00:06:14)

Models: Claude Opus 4.5 (analyze) · Gemini 3 Flash (tag) · Gemini 3 Flash (summarize)

Article URL: https://www.dailycal.org/news/campus/academics/failing-grades-soar-as-professors-see-greater-ai-usage-dwindling-math-skills-in-uc-berkeley/article_16fad0bf-02cb-4b8c-8d88-888ffd9f8608.html

Article Summary

Comment Summary

Topics

• Cognitive Decline from AI (65 comments) (Commenters observe thinking skills deteriorating among LLM users, including PhDs unable to brainstorm or sit quietly thinking. Comparisons to muscle atrophy, concerns about inability to code or write without AI present, and fears about losing the mental discipline required for original thought.)
• Proper AI Use for Learning (49 comments) (Discussion of using LLMs as tutors rather than answer machines, having AI ask clarifying questions, generate practice exams, or critique work rather than produce it. The distinction between AI as learning accelerator versus shortcut that prevents actual understanding.)
• Value of Struggle in Learning (49 comments) (Arguments that struggling with difficult problems creates learning, like exercise creates fitness. Using AI shortcuts compared to bringing a forklift to the gym, and concerns about losing the pain that makes people smarter.)
• University Purpose and Motivation (44 comments) (Debate over whether students attend university for genuine learning versus credentials and job opportunities. Discussion of intrinsic versus extrinsic motivation, vocational alternatives like German Ausbildung, and whether deep interest in subjects is realistic.)
• Standardized Testing Removal (38 comments) (Many commenters point to UC system dropping SAT/ACT requirements as the real cause. Yale and other universities reversing test-optional policies, with standardized tests being the strongest predictor of college success regardless of demographics.)
• Education System Restructuring (37 comments) (Proposals for flipped classrooms, daily quizzes, teaching AI literacy, and rethinking what skills matter. Questions about whether curricula should adapt to AI like they did to calculators, and whether traditional teaching methods are obsolete.)
• Future of Knowledge Work (35 comments) (Uncertainty about what skills will matter in five years, whether learning fundamentals still makes sense, and comparisons to memorizing log tables becoming obsolete. Some see opportunity in others' skill decline.)
• Skill Atrophy from Automation (33 comments) (Parallels drawn to losing navigation skills from GPS, arithmetic from calculators, phone numbers from contacts. Concern that unlike physical muscle, cognitive skills may not return easily, and that reduced friction destroys ability to concentrate.)
• Professional AI Dependency (29 comments) (Observations that even experienced PhDs and senior engineers are becoming reliant on LLMs. Concerns about productivity gains masking declining creativity and originality, with short-term output increases potentially harmful long-term.)
• Grade Inflation and Curves (28 comments) (Disagreement about whether grading should be curved, threshold-based, or standards-based. Berkeley faculty opposing Harvard-style curves, concerns about hiding declining performance through grade inflation, and questions about what grades actually measure.)
• Mathematical Preparation Decline (23 comments) (Reports of declining math skills predating AI, with French universities having to teach calculus to physics students. Discussion of math being poorly taught and whether mathematical fluency is analogous to literacy.)
• Cheating Detection and Consequences (22 comments) (Discussion of how to catch AI-assisted cheating by asking clarifying questions until students reveal gaps. Appeals to honesty, sending cases to conduct centers, and comparisons to pre-AI plagiarism detection methods like MOSS.)
• Social Media Brain Effects (22 comments) (Comments connecting AI cognitive effects to existing concerns about attention economy, short-form content, and smartphone addiction. References to pre-existing inability to focus, constant communication burden, and information overload.)
• Top-Down vs Bottom-Up Teaching (19 comments) (Proposal to invert teaching from basics-up to concepts-down with LLMs available. Counterarguments that intuition requires concrete examples first, and high-level knowledge without fundamentals produces only surface functionality.)
• Addiction and Dopamine (17 comments) (Comparisons of AI use to addiction, smoking, and amphetamines. The quick dopamine hits from instant answers replacing satisfaction from manual problem-solving, and difficulty retraining reward systems.)
• Credential vs Education Distinction (16 comments) (Discussion of degrees becoming worthless if everyone can pass using AI without learning. The gap between students seeking credentials versus education, and how AI exposes those there just for the diploma.)
• Appropriate AI Learning Domains (14 comments) (Observation that AI works best in domains where user is already expert. Using AI for unfamiliar areas produces surface-level understanding without retention, like watching a professor versus doing the work yourself.)
• Teaching Quality at Universities (14 comments) (Criticism of professors reading from slides, lack of teaching training, and research faculty treating teaching as burden. Counterpoint that specific professors mentioned are actually teaching-focused with strong pedagogical research.)
• Societal Nihilism and Motivation (5 comments) (Argument that students cheat because traditional career paths seem obsolete with AI and climate doom. Rational response to broken social contracts about hard work leading to success, home ownership becoming unattainable.)
• Examining AI Detection Reliability (4 comments) (Questions about whether AI detectors trigger false positives, particularly for average students whose writing resembles AI output. Concerns about logs revealing copy-paste behavior and sophisticated notation never taught in class.)

Raw Files

Execution Log

[2026-06-04T12:34:51.642Z] Starting step: fetch_pages (attempt 1)
[2026-06-04T12:34:51.664Z] Fetching HN page: https://news.ycombinator.com/item?id=48392004
[2026-06-04T12:34:51.833Z] Fetched HN page: 558001 bytes
[2026-06-04T12:34:52.017Z] Extracted title: Failing grades soar with AI usage, dwindling math skills in Berkeley CS classes
[2026-06-04T12:34:52.035Z] Extracted linked URL: https://www.dailycal.org/news/campus/academics/failing-grades-soar-as-professors-see-greater-ai-usage-dwindling-math-skills-in-uc-berkeley/article_16fad0bf-02cb-4b8c-8d88-888ffd9f8608.html
[2026-06-04T12:34:52.054Z] Fetching linked article: https://www.dailycal.org/news/campus/academics/failing-grades-soar-as-professors-see-greater-ai-usage-dwindling-math-skills-in-uc-berkeley/article_16fad0bf-02cb-4b8c-8d88-888ffd9f8608.html
[2026-06-04T12:34:52.246Z] Warning: Failed to fetch linked article: Failed to fetch https://www.dailycal.org/news/campus/academics/failing-grades-soar-as-professors-see-greater-ai-usage-dwindling-math-skills-in-uc-berkeley/article_16fad0bf-02cb-4b8c-8d88-888ffd9f8608.html: 429 Too Many Requests
[2026-06-04T12:34:52.282Z] Completed step: fetch_pages in 624ms
[2026-06-04T12:34:57.392Z] Starting step: extract_text (attempt 1)
[2026-06-04T12:34:57.533Z] Extracted HN text: 157968 chars
[2026-06-04T12:34:57.741Z] Extracted 339 comments
[2026-06-04T12:34:57.831Z] Comment word count: 24354
[2026-06-04T12:34:57.883Z] Completed step: extract_text in 475ms
[2026-06-04T12:34:58.321Z] Starting step: analyze_content (attempt 1)
[2026-06-04T12:34:58.357Z] Calling claude-opus-4-5-20251101 (article: 0 chars, 339 comments)
[2026-06-04T12:35:34.026Z] Analysis complete: 20 topics, 32626 input tokens, 1305 output tokens
[2026-06-04T12:35:34.057Z] Completed step: analyze_content in 35714ms
[2026-06-04T12:35:34.473Z] Starting step: tag_comments (attempt 1)
[2026-06-04T12:35:34.508Z] Tagging 339 comments with 20 topics (batch size: 50)
[2026-06-04T12:35:34.525Z] Processing batch 1/7 (50 comments)
[2026-06-04T12:35:59.011Z] Batch 1 complete: 96 tags assigned
[2026-06-04T12:35:59.028Z] Processing batch 2/7 (50 comments)
[2026-06-04T12:36:31.558Z] Batch 2 complete: 95 tags assigned
[2026-06-04T12:36:31.566Z] Processing batch 3/7 (50 comments)
[2026-06-04T12:37:06.060Z] Batch 3 complete: 94 tags assigned
[2026-06-04T12:37:06.084Z] Processing batch 4/7 (50 comments)
[2026-06-04T12:37:29.406Z] Batch 4 complete: 74 tags assigned
[2026-06-04T12:37:29.421Z] Processing batch 5/7 (50 comments)
[2026-06-04T12:37:54.106Z] Batch 5 complete: 86 tags assigned
[2026-06-04T12:37:54.127Z] Processing batch 6/7 (50 comments)
[2026-06-04T12:38:17.336Z] Batch 6 complete: 79 tags assigned
[2026-06-04T12:38:17.354Z] Processing batch 7/7 (39 comments)
[2026-06-04T12:38:38.663Z] Batch 7 complete: 74 tags assigned
[2026-06-04T12:38:38.679Z] Tagging complete: 598 total tags, 49021 input tokens, 8535 output tokens
[2026-06-04T12:38:38.696Z] Completed step: tag_comments in 184207ms
[2026-06-04T12:38:38.832Z] Starting step: summarize_topics (attempt 1)
[2026-06-04T12:38:38.854Z] Summarizing 20 topics
[2026-06-04T12:38:38.939Z] Summarizing topic 1/20: "Cognitive Decline from AI # Commenters observe thinking skills deteriorating among LLM users, including PhDs unable to brainstorm or sit quietly thinking. Comparisons to muscle atrophy, concerns about inability to code or write without AI present, and fears about losing the mental discipline required for original thought." (65 comments)
[2026-06-04T12:38:46.112Z] Topic 1 summarized (6595 in, 147 out)
[2026-06-04T12:38:46.146Z] Summarizing topic 2/20: "Proper AI Use for Learning # Discussion of using LLMs as tutors rather than answer machines, having AI ask clarifying questions, generate practice exams, or critique work rather than produce it. The distinction between AI as learning accelerator versus shortcut that prevents actual understanding." (49 comments)
[2026-06-04T12:38:51.610Z] Topic 2 summarized (6597 in, 139 out)
[2026-06-04T12:38:51.642Z] Summarizing topic 3/20: "Standardized Testing Removal # Many commenters point to UC system dropping SAT/ACT requirements as the real cause. Yale and other universities reversing test-optional policies, with standardized tests being the strongest predictor of college success regardless of demographics." (38 comments)
[2026-06-04T12:38:57.670Z] Topic 3 summarized (2678 in, 150 out)
[2026-06-04T12:38:57.697Z] Summarizing topic 4/20: "University Purpose and Motivation # Debate over whether students attend university for genuine learning versus credentials and job opportunities. Discussion of intrinsic versus extrinsic motivation, vocational alternatives like German Ausbildung, and whether deep interest in subjects is realistic." (44 comments)
[2026-06-04T12:39:04.614Z] Topic 4 summarized (5593 in, 145 out)
[2026-06-04T12:39:04.641Z] Summarizing topic 5/20: "Grade Inflation and Curves # Disagreement about whether grading should be curved, threshold-based, or standards-based. Berkeley faculty opposing Harvard-style curves, concerns about hiding declining performance through grade inflation, and questions about what grades actually measure." (28 comments)
[2026-06-04T12:39:11.285Z] Topic 5 summarized (3095 in, 162 out)
[2026-06-04T12:39:11.327Z] Summarizing topic 6/20: "Cheating Detection and Consequences # Discussion of how to catch AI-assisted cheating by asking clarifying questions until students reveal gaps. Appeals to honesty, sending cases to conduct centers, and comparisons to pre-AI plagiarism detection methods like MOSS." (22 comments)
[2026-06-04T12:39:18.235Z] Topic 6 summarized (2890 in, 130 out)
[2026-06-04T12:39:18.260Z] Summarizing topic 7/20: "Skill Atrophy from Automation # Parallels drawn to losing navigation skills from GPS, arithmetic from calculators, phone numbers from contacts. Concern that unlike physical muscle, cognitive skills may not return easily, and that reduced friction destroys ability to concentrate." (33 comments)
[2026-06-04T12:39:24.691Z] Topic 7 summarized (3744 in, 146 out)
[2026-06-04T12:39:24.719Z] Summarizing topic 8/20: "Education System Restructuring # Proposals for flipped classrooms, daily quizzes, teaching AI literacy, and rethinking what skills matter. Questions about whether curricula should adapt to AI like they did to calculators, and whether traditional teaching methods are obsolete." (37 comments)
[2026-06-04T12:39:35.602Z] Topic 8 summarized (3979 in, 143 out)
[2026-06-04T12:39:35.627Z] Summarizing topic 9/20: "Professional AI Dependency # Observations that even experienced PhDs and senior engineers are becoming reliant on LLMs. Concerns about productivity gains masking declining creativity and originality, with short-term output increases potentially harmful long-term." (29 comments)
[2026-06-04T12:39:42.579Z] Topic 9 summarized (3587 in, 162 out)
[2026-06-04T12:39:42.605Z] Summarizing topic 10/20: "Value of Struggle in Learning # Arguments that struggling with difficult problems creates learning, like exercise creates fitness. Using AI shortcuts compared to bringing a forklift to the gym, and concerns about losing the pain that makes people smarter." (49 comments)
[2026-06-04T12:39:52.959Z] Topic 10 summarized (5777 in, 153 out)
[2026-06-04T12:39:52.985Z] Summarizing topic 11/20: "Social Media Brain Effects # Comments connecting AI cognitive effects to existing concerns about attention economy, short-form content, and smartphone addiction. References to pre-existing inability to focus, constant communication burden, and information overload." (22 comments)
[2026-06-04T12:39:59.989Z] Topic 11 summarized (2450 in, 133 out)
[2026-06-04T12:40:00.027Z] Summarizing topic 12/20: "Credential vs Education Distinction # Discussion of degrees becoming worthless if everyone can pass using AI without learning. The gap between students seeking credentials versus education, and how AI exposes those there just for the diploma." (16 comments)
[2026-06-04T12:40:06.742Z] Topic 12 summarized (2292 in, 112 out)
[2026-06-04T12:40:06.787Z] Summarizing topic 13/20: "Future of Knowledge Work # Uncertainty about what skills will matter in five years, whether learning fundamentals still makes sense, and comparisons to memorizing log tables becoming obsolete. Some see opportunity in others' skill decline." (35 comments)
[2026-06-04T12:40:14.134Z] Topic 13 summarized (3364 in, 126 out)
[2026-06-04T12:40:14.162Z] Summarizing topic 14/20: "Mathematical Preparation Decline # Reports of declining math skills predating AI, with French universities having to teach calculus to physics students. Discussion of math being poorly taught and whether mathematical fluency is analogous to literacy." (23 comments)
[2026-06-04T12:40:22.619Z] Topic 14 summarized (2480 in, 135 out)
[2026-06-04T12:40:22.645Z] Summarizing topic 15/20: "Appropriate AI Learning Domains # Observation that AI works best in domains where user is already expert. Using AI for unfamiliar areas produces surface-level understanding without retention, like watching a professor versus doing the work yourself." (14 comments)
[2026-06-04T12:40:29.959Z] Topic 15 summarized (2673 in, 149 out)
[2026-06-04T12:40:29.987Z] Summarizing topic 16/20: "Addiction and Dopamine # Comparisons of AI use to addiction, smoking, and amphetamines. The quick dopamine hits from instant answers replacing satisfaction from manual problem-solving, and difficulty retraining reward systems." (17 comments)
[2026-06-04T12:40:38.980Z] Topic 16 summarized (1847 in, 122 out)
[2026-06-04T12:40:39.008Z] Summarizing topic 17/20: "Teaching Quality at Universities # Criticism of professors reading from slides, lack of teaching training, and research faculty treating teaching as burden. Counterpoint that specific professors mentioned are actually teaching-focused with strong pedagogical research." (14 comments)
[2026-06-04T12:40:46.187Z] Topic 17 summarized (2751 in, 140 out)
[2026-06-04T12:40:46.294Z] Summarizing topic 18/20: "Top-Down vs Bottom-Up Teaching # Proposal to invert teaching from basics-up to concepts-down with LLMs available. Counterarguments that intuition requires concrete examples first, and high-level knowledge without fundamentals produces only surface functionality." (19 comments)
[2026-06-04T12:40:54.013Z] Topic 18 summarized (2199 in, 154 out)
[2026-06-04T12:40:54.038Z] Summarizing topic 19/20: "Examining AI Detection Reliability # Questions about whether AI detectors trigger false positives, particularly for average students whose writing resembles AI output. Concerns about logs revealing copy-paste behavior and sophisticated notation never taught in class." (4 comments)
[2026-06-04T12:40:58.500Z] Topic 19 summarized (447 in, 124 out)
[2026-06-04T12:40:58.524Z] Summarizing topic 20/20: "Societal Nihilism and Motivation # Argument that students cheat because traditional career paths seem obsolete with AI and climate doom. Rational response to broken social contracts about hard work leading to success, home ownership becoming unattainable." (5 comments)
[2026-06-04T12:41:03.466Z] Topic 20 summarized (705 in, 147 out)
[2026-06-04T12:41:03.484Z] Summarization complete: 20 topics, 65743 input tokens, 2819 output tokens
[2026-06-04T12:41:03.503Z] Completed step: summarize_topics in 144654ms
[2026-06-04T12:41:03.537Z] Job completed successfully

LLM Invocations (Total: $0.2872)

← Back to all jobs