AI use leads UC Berkeley math professor to see 'no point' in grading homework
- ID
- 39415
- Status
- summarized
- Published
- 26 Oct 2026, 12:28 AM
- Fetched
- 26 Oct 2026, 10:16 AM
- Provider
- Hacker News
- Category
- dev-community
- Original URL
- https://www.dailycal.org/news/campus/academics/ai-use-leads-uc-berkeley-math-professor-to-see-no-point-in-grading-homework/article_bdf1eae2-c08b-491f-b9f1-81a7cea177de.html
- Source URL
- https://hnrss.org/best
Summary
- Score
- 7.0
- Created
- 26 Oct 2026, 10:17 AM
- Tags
- Audience
- developersvibe_codersai_ml_learnersstartup_founders
What happened
UC Berkeley math professor and Fields medalist Richard Borcherds told the campus paper he no longer has homework graded for correctness, because so many students used AI to get top scores that paying graders to check answers felt pointless; he now grades on completion, which he calls "not a very good system." The article, published Oct 23, 2026, ties his remarks to an OpenAI announcement that AI agents apparently solved the Navier-Stokes existence and smoothness problem. The Hacker News thread drew 200 points and 292 comments.
Why it matters
If a homework set graded on correctness stops separating students, the same is true of take-home coding challenges and written AI/ML exercises used in hiring or course assessment. Anyone who screens builders with an at-home task should decide now whether to move to live explanation, oral defense, or process artifacts (commit history, design notes, debugging walkthroughs) rather than scored output. Note the Navier-Stokes claim is reported secondhand from an OpenAI announcement, so treat it as an unverified claim to check, not a settled result.
Discussion angle
If an agent can produce a passing take-home, what replaces it in your hiring or course assessment — a live walkthrough of the submitted code, or a task built around debugging someone else's broken system? Ask the room which format they'd trust after this professor's experience.