Overview
- This week Richard Borcherds, a 1998 Fields Medalist and UC Berkeley professor, said he no longer grades homework for correctness and now gives credit for completion because student work is often produced with AI.
- Borcherds told the Daily Californian that he now bases most course grades on in-person, device-free exams and calls unmonitored online tests a “waste of time” because he cannot tell which students used AI.
- UC Berkeley’s Academic Senate asks faculty to state AI expectations on syllabi but leaves specific rules to individual instructors, and campus testing of detection products such as Turnitin and Rumi found them inadequate for a campus-wide solution.
- Other California professors are responding in different ways: some are switching homework to completion grading and paper exams, while others are building supervised AI tools or keeping AI as a permitted learning aid with strict in-class testing.
- The shift raises wider questions about training and career paths in fields like mathematics, since rapid AI advances — including recent high-profile algorithmic breakthroughs reported by AI companies — could change what skills universities must teach and how employers judge student competence.