A 66‑year‑old mathematics professor at UC Berkeley has stopped grading homework for correctness, arguing that artificial intelligence now solves most problems for students. The change highlights how AI is reshaping assessment even at world‑class universities.

He told The Daily Californian that AI can now solve mathematical problems so well that he sees “no point” in paying graders to check the accuracy of students’ assignments. As a result, he gives credit simply for completing the work, regardless of correctness.

In an interview with The California Post, the professor explained that grading for correctness creates a huge incentive for students to use AI instead of solving problems themselves, and it penalises honest effort. Because online exams can be gamed with AI, he now relies mainly on in‑person examinations where electronic devices are prohibited.

The professor is a Fields Medalist, having received the award in 1998 for his work on the “monstrous moonshine” conjecture. He teaches at UC Berkeley, which the 2026‑2027 U.S. News Best Global Universities ranking places seventh in the world.

A UC Berkeley study published in May examined more than 500,000 university grades awarded between 2018 and 2025. The research found that courses with assignments easy for AI—such as writing and coding—saw sharp increases in top grades after ChatGPT became widely available.

These developments show that powerful AI tools are forcing even elite institutions to rethink traditional grading. By shifting credit to assignment completion and emphasizing in‑person exams, the professor aims to preserve genuine learning in a rapidly changing landscape.