Overview
- Richard Borcherds, a Berkeley math professor and Fields Medalist, said he no longer grades homework for correctness and now awards credit mainly for turning assignments in.
- Borcherds told reporters that widespread student use of AI makes it hard to tell human work from machine output, so he relies on supervised, device-free, in-person exams to assess mastery.
- UC Berkeley’s Academic Senate asks faculty to state course expectations about AI in syllabi but declined to adopt campus-wide AI-detection software after testing tools such as Turnitin and Rumi and finding them unreliable.
- Borcherds warned that continued rapid AI progress could change the work of mathematicians and said even asking ChatGPT did not produce clear career guidance for students.
- Experts and campus officials urge alternatives to strict policing, recommending AI literacy, classroom-controlled assessments, and collaborative policies because detection tools and punitive approaches risk eroding trust and may not solve the problem.