UC Berkeley professor and Fields medalist Richard Borcherds has stopped grading student homework for accuracy, citing the widespread use of artificial intelligence tools among his mathematics pupils.
Key Points
- Richard Borcherds now grades homework based solely on completion because AI makes verifying individual student work for correctness impractical.
- The decision follows OpenAI’s recent progress in using AI agents to address the complex Navier-Stokes existence and smoothness mathematical problem.
- UC Berkeley has tested AI detection software like Turnitin and Rumi but concluded these tools are currently inadequate for campus-wide standardization.
- The university leaves AI policy to individual faculty discretion, providing resources through the campus AI Hub to help instructors draft syllabus statements.
- Borcherds expressed uncertainty regarding the future of mathematics as a career path, noting that AI advancements are rapidly outpacing human capabilities.