One signal

Update What changed

Richard Borcherds says he now bases grades mainly on in-person exams without electronic devices, while online exams have become “a waste of time” because some students use AI to answer questions. He says he cannot tell which students use AI, but has noticed substantially higher average scores on online exams.

The account also adds that UC Berkeley explored AI-detection tools including Turnitin and Rumi, but did not adopt them as a standard campus-wide solution because they were considered unreliable. Instructors set out their own course rules for AI use.

Borcherds says he does not know what to advise aspiring mathematicians about choosing the field as AI advances. Those details come from the Yahoo-published account, which attributes the remarks to The California Post and the Daily Californian.

Sources and evidence

• UC Berkeley professor refuses to grade his students’ homework after noticing worrying trend: ‘No point’ - Yahoo: Borcherds says he now relies mainly on in-person exams without electronic devices, regards online exams as ineffective because some students use AI, and cannot identify which students do so. The account says Berkeley explored but did not standardise AI-detection tools because they were considered unreliable.

Independent WittyWires Watcher; not an official account or feed.

Update What changed

Richard Borcherds told the Daily Californian that, if AI keeps improving at its current pace, it could outperform human mathematicians at every task within a few years, Hindustan Times reports. The Fields Medal-winning mathematician said he had no clear advice for students considering the subject.

Borcherds compared the uncertainty with earlier technological shifts, but said he saw this as different because mathematics had seemed relatively insulated from disruption only five years ago. He also said he could not reliably identify which students use AI, despite noticing substantially higher average marks on online exams.

He said he now relies mainly on in-person exams without electronic devices, calling online exams “a waste of time” when students can use AI to answer questions.

Sources and evidence

• UC Berkeley professor stops grading students' homework over AI use, calls online exams 'a waste of time' | Hindustan Times - Hindustan Times: Hindustan Times reports that Richard Borcherds warned AI could outperform human mathematicians at every task within a few years if progress continues at its current pace, and said he did not know what advice to give students considering mathematics.

Independent WittyWires Watcher; not an official account or feed.

Update What changed

Richard Borcherds says his new approach to grading is “not a very good system”, but that he currently has no alternative. Hoodline reports that the Berkeley mathematician now gives credit for completing homework rather than marking whether answers are correct.

The report says Berkeley’s Academic Senate asks instructors to state their expectations for AI use in course syllabuses, while leaving the rules to individual faculty. That makes Borcherds’s approach a decision for his own classes, not a university-wide policy.

Hoodline also says Berkeley’s Research, Teaching, and Learning Commons has warned that AI-detection tools provide no conclusive proof, and that their role in academic-integrity investigations remains unclear. The distinction matters: the account presents detection as no settled answer, not a replacement for deciding what assessment should measure.

Sources and evidence

• UC Berkeley Professor Stops Grading Homework Over AI - Hoodline: Hoodline reports that Richard Borcherds considers his completion-based homework grading an imperfect stopgap; Berkeley leaves course AI rules to instructors, and its Research, Teaching, and Learning Commons says AI-detection tools provide no conclusive proof.

Independent WittyWires Watcher; not an official account or feed.