UC Berkeley mathematics professor Richard Borcherds has stopped grading homework for correctness after concluding that widespread use of artificial intelligence has undermined the value of traditional grading. Students can now use AI tools to produce strong answers and rack up top marks, as reported by The New York Post.

Borcherds, who received the prestigious Fields Medal in 1998, said the old grading system creates a powerful incentive for students to let artificial intelligence do the work. Meanwhile, students who complete assignments honestly can find themselves penalized.

“Student use of AI for homework is now widespread,” Borcherds told The California Post. “Grading for correctness just adds a huge incentive for them to use AI for homework instead of doing it themselves, and penalizes students for honesty.”

I think what people are getting upset about is they’re slowly realizing that they are dumb and somehow blaming this on AI, even though it’s not really AI’s fault.” - Richard Borcherds https://t.co/msT8AX5gKT

— Michael Jackson (@WorkMJ) October 20, 2026

Borcherds also said paying graders to evaluate answers produced by artificial intelligence wastes both time and money. Under his current approach, students receive credit primarily for completing assignments rather than producing correct answers, though he acknowledged that he does not know of a better solution.

Gun Laws Have Changed! Protect yourself by knowing where you can and cannot carry your firearm, so you can travel with confidence.

The professor said distinguishing between student work and material generated by artificial intelligence has become increasingly difficult. The occasional telltale mistakes once left by AI systems are also becoming less common as the technology improves.

“There is no easy way to tell the difference between human and AI generated work,” Borcherds told The California Post. “Occasionally one could tell it is AI because AI made a few characteristic errors, but AI has now improved so much that it rarely makes obvious mistakes.”

The growing ability of AI tools to solve complex academic problems is raising questions about how universities can assess student work and prepare future mathematicians. Borcherds addressed those broader concerns in an interview with the Daily Californian, the student newspaper at UC Berkeley.

“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,” Borcherds told the paper. His warning reaches well beyond a stack of suspiciously polished homework assignments.

Borcherds said he was uncertain whether aspiring mathematicians should continue pursuing the field. He noted that just five years ago, mathematics appeared to offer comparatively strong protection from the technological disruption affecting other professions.

“I have no idea what to advise. I even asked ChatGPT this exact question, and even ChatGPT said it didn’t have any idea what was going to happen,” he told the student newspaper.

The professor also suggested universities might not immediately adjust to the rapid development of artificial intelligence because academic institutions often resist change. He compared the situation to academic journals that still publish physical editions despite having lost much of their purpose, in his view, decades ago.

“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,” Borcherds said. “This is the first time anything like this has happened in mathematics, though.”

With no better answer available, Borcherds now bases grades mainly on exams administered in person, where students cannot use electronic devices. He described online exams as “a waste of time” because some students use artificial intelligence to answer the questions.

Borcherds said he cannot identify which students are using AI, but he has observed that average scores rise substantially during online exams. That leaves instructors facing the same basic problem that wrecked traditional homework grading: impressive results no longer necessarily reveal who actually did the work.

UC Berkeley has responded by establishing guidance that requires faculty members to address AI use in their courses. Anne Marie Richard, the university’s assistant vice provost for undergraduate education and chief academic technology officer, said the Berkeley Division of the Academic Senate asks instructors to include statements in their syllabi explaining course expectations and rules for AI tools.

The university also considered AI detection software, including Turnitin and Rumi, but determined that the programs were not reliable enough for a standardized solution across the campus, according to the Daily Californian. Individual instructors therefore remain responsible for setting AI policies in their own classrooms.

For Borcherds, unreliable homework may be among the smaller consequences of artificial intelligence. “If the worst effect of AI on society is that it is harder to assess homework then I will be very relieved,” he told The California Post.