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

What happened

Richard Borcherds, a famous math professor at UC Berkeley, no longer grades homework for correctness. He saw many students use AI to get top scores. Now he only grades whether a student turned the homework in. He says this is "not a very good system," but he has no better option right now.

Key facts

• Borcherds is a Fields medalist, which is one of the top prizes in math.
• He spoke in a sit-down interview on Tuesday.
• OpenAI said AI agents apparently solved the Navier-Stokes existence and smoothness problem, one of math's hardest unsolved problems.
• Campus AI detection tools, including Turnitin and Rumi, were found not good enough to use across the whole campus.

Background

AI tools like ChatGPT can now solve many math problems. Schools are trying to decide how students should use them. At UC Berkeley, each professor decides the AI rules for their own class. The campus suggests that teachers write a clear AI policy in their course syllabus, which is the document that explains a class's rules and plans.

Why it matters

Borcherds worries that AI may soon do math better than people. This makes it hard for him to advise students about careers in math. He also says schools move slowly, so change may take a few years. His story shows how AI is already changing how students learn and how teachers check their work.

What to watch

• Whether other UC Berkeley professors change how they grade homework.
• Whether the Navier-Stokes claim by OpenAI is confirmed by other mathematicians.
• Whether schools find better ways to detect AI use.
• How the AI Hub's sample AI policies are used in course syllabi.