AI use leads UC Berkeley math professor to see 'no point' in grading homework

Richard Borcherds, a UC Berkeley professor and Fields medalist, has stopped grading his students' homework due to concerns over artificial intelligence (AI) usage. After observing AI being used by many students to achieve top scores consistently, he felt there was "no point" in paying graders to check for correctness, as AI use has become unavoidable.

Borcherds discussed this in an interview following OpenAI's revelation that AI agents solved one of mathematics' most challenging unsolved problems, the Navier-Stokes existence and smoothness problem. He stated that if AI continues to advance at its current pace, mathematics will soon be a field where AI outperforms human mathematicians.

Currently, Borcherds grades homework based on completion, acknowledging that it is "not a very good system." He expressed uncertainty about advising students to pursue mathematics as a career, as it seemed five years ago that it was one of the safest careers from the AI boom. Borcherds attributed the lack of change in the academic system to "inertia" and hinted that the math department might remain relatively unchanged for a couple of years despite AI developments.

He drew parallels to the obsolescence of physical paper journals and horse-drawn carriages due to technological advancements. The Berkeley Division of the Academic Senate urges faculty to include statements about course expectations regarding AI tool usage in their syllabi. However, direct policies on AI use can vary significantly depending on the course level and focus area.

The campus has tried several AI detection tools but found them inadequate for standardization. Instead, faculty have the discretion to determine AI policies for their classes, and there is a growing database of sample AI use statements available on the university's AI Hub for reference.