Mathematician borcherds abandons graded homework amid AI cheating concerns

On October 28, 2026, a professor at the University of California, Berkeley, announced that he would no longer assign graded homework due to widespread AI cheating. Richard Borcherds, a renowned mathematician and Fields Medalist, stated that it is now nearly impossible to distinguish between student work completed independently and that generated by AI. He emphasized that online exams have become ineffective as students can directly use AI to answer questions, leading to inflated average scores. Borcherds argued that grading based on correctness incentivizes students to use AI for homework, while punishing those who complete tasks honestly. He also criticized the time and cost of grading AI-generated answers. The University of California, Berkeley, previously considered AI detection software but found it unreliable, instead delegating AI policy decisions to professors, who must clearly state rules in course syllabi. Borcherds expressed concern that the worst-case scenario is merely increased difficulty for teachers, which he finds acceptable.