The Pulse
Richard Borcherds Stops Grading Math Homework for Correctness
UC Berkeley mathematics professor Richard Borcherds now gives homework credit for completion rather than correctness, saying AI use has made grading answers for accuracy seem pointless. He also told The Daily Californian he is uncertain wha
AI.info Team · Written with AI · fact-checked ·
Richard Borcherds, a UC Berkeley professor and Fields medalist, no longer grades his students’ homework for correctness. After seeing students use artificial intelligence to receive consistently high scores, he said he saw “no point” in paying readers to grade homework for accuracy. Students who submit their work now receive credit for completing it, whether or not their answers are correct.
Borcherds acknowledged that the new arrangement is not a satisfactory answer to the problem. He told The Daily Californian that it is “not a very good system,” but that he has no alternative at present. His change concerns how homework is graded; it does not mean students no longer receive assignments.
“If AI continues to improve at present rate, then, in a few years, there’ll be nothing human mathematicians can do that AI can’t do better.”
Richard Borcherds, UC Berkeley professor and Fields medalist, speaking to The Daily Californian
Uncertainty about the future of mathematics
Borcherds also said he no longer knows whether to encourage students to enter the math department. Five years ago, he said, a career as a mathematician seemed among the safest from the AI boom. Asked what advice he would give now, he said he had no idea, adding that he had asked ChatGPT the same question and it also could not predict what would happen.
The interview followed an announcement by OpenAI about AI agents and the Navier-Stokes existence and smoothness problem, which the newspaper described as one of mathematics’ most difficult unsolved problems. Borcherds said he would not be surprised if inertia in the academic system meant that little changed in the math department for a couple of years despite developments in AI.
He compared that institutional inertia with the continued publication of physical paper journals, which he described as having been “utterly pointless” for two decades. But he also distinguished AI’s potential effects on mathematics from earlier technological shifts: “There have been plenty of technical breakthroughs in the past — hand weaving was put out of business by looms, and horses were put out of business by motor cars, and so on,” he said. “This is the first time anything like this has happened in mathematics, though.”
Course policies remain instructor-specific
The Daily Californian reported that UC Berkeley’s Academic Senate asks faculty to include course expectations about AI use in their syllabi. The paper cited an email from Anne Marie Richard, assistant vice provost for undergraduate education and chief academic technology officer, who said policies can vary with a course’s level and focus. The article also reported that campus trials of AI-detection tools, including Turnitin and Rumi, found them inadequate for standardization.
Berkeley’s Center for Teaching & Learning likewise presents sample course policies spanning required, limited and prohibited uses of generative AI, and encourages instructors to adapt expectations to their course context. Its guidance describes course policies as tied to learning goals rather than as one rule for every class.
Borcherds’s decision is a response to his experience grading homework, not a university-wide change in assessment policy. His account leaves open how instructors should evaluate work when students can use AI, while his comments about the future of mathematics reflect his own uncertainty about what the technology may mean for the field.