University of California, Berkeley professor Richard Borcherds has officially stopped grading homework for his mathematics students. The decision stems from the rapid advancement of artificial intelligence, which has made it nearly impossible for educators to differentiate between human-penned submissions and those generated by automated tools. Borcherds, a recipient of the Fields Medal whose current research focuses on quantum field theory, noted a sharp surge in average scores on unproctored assessments as students increasingly turned to algorithms for answers.
Prior to this policy shift, the university evaluated several automated detection programs but ultimately deemed them unreliable. Consequently, Berkeley left the formulation of specific AI policies up to individual instructors within course syllabi. According to Borcherds, traditional homework grading incentivizes dishonesty because automated tools reliably provide correct answers. Continuing to grade these submissions, he argues, unfairly penalizes honest students while wasting both time and university resources.
Moving forward, assessments in his classes will rely strictly on in-person, traditional exams completed without the assistance of electronic devices. While other educators continue to grapple with the integration of language models in academia, Borcherds' traditional approach highlights the growing friction between legacy evaluation methods and generative AI capabilities in higher education.
UC Berkeley Professor Stops Grading Homework as AI Makes Cheating Undetectable
Comments