UC Berkeley professor Richard Borcherds shifts to completion-based grading due to AI

Richard Borcherds, a renowned mathematics professor and 1998 Fields Medalist at UC Berkeley, has transitioned from grading homework for correctness to awarding credit based solely on completion. This decision follows the widespread use of artificial intelligence by students to generate accurate mathematical solutions, which Borcherds argues makes traditional correctness-based grading “pointless.”

Borcherds noted that grading for accuracy creates a significant incentive for students to use AI rather than engaging in independent problem-solving, which ultimately penalizes students who choose to work honestly. He also stated that paying graders to verify answers that can be easily produced by AI is a waste of time and financial resources.

The professor highlighted the increasing difficulty in distinguishing between human and AI-generated work, noting that while AI previously made characteristic errors, it has improved to the point where such mistakes are rare. Borcherds expressed concern regarding the future of the mathematics profession, suggesting that if AI continues to advance at its current rate, it may soon surpass human capabilities in the field.