According to a report by the New York Post on the 26th, University of California, Berkeley professor and Fields Medal winner Richard Borcherds announced that he will no longer assign graded homework to students — because AI abuse has made such homework essentially meaningless.
He admitted that nowadays it is nearly impossible to tell whether the homework students submit was written by themselves or generated by AI. So he has moved the focus of assessment to in-person exams where electronic devices are not allowed, recognizing only the real ability shown in that setting.
Borcherds was very blunt with the media: online exams are now a pure waste of time, as some students answer directly with AI. Although he cannot catch every student who uses AI, he has clearly seen the average scores of online exams pushed up substantially.
The honest are punished instead
An even bigger problem lies in the grading logic. He pointed out that grading based on whether answers are right or wrong effectively encourages students to rely on AI to complete homework at home rather than doing it themselves, while also punishing those who honestly solve problems on their own; and spending time and money on grading AI-generated answers is itself a waste.
The university has not failed to try to find solutions. Berkeley previously considered introducing AI detection software, but ultimately concluded that such tools were not reliable enough, so it delegated authority over AI use policies to professors, only requiring them to state the rules clearly in their course syllabi.
As for Borcherds himself, his attitude is rather relaxed: if the worst impact of AI is merely making it harder for teachers to grade homework, then he is "perfectly at ease." When a scholar of Nobel caliber has to retreat to pen-and-paper exams to preserve fairness, it is clear that no one has yet been able to easily handle this difficult problem AI has thrown at education.