Berkeley, California – Richard Borcherds, a Fields Medal-winning mathematician at UC Berkeley, has adjusted his approach to grading student homework. He now evaluates submissions based on completion rather than accuracy after observing widespread use of artificial intelligence tools by students. The change reflects a direct response to how AI can produce high-scoring work with limited original effort from the student.
A Fields Medalist Confronts Classroom Realities
Borcherds holds one of the highest honors in mathematics, yet he faces the same classroom pressures as many instructors today. Students increasingly turn to AI systems that can solve complex problems and generate polished answers in moments. This capability has altered the traditional purpose of homework as a measure of individual understanding. The professor’s adjustment keeps the assignment process intact while removing the incentive for perfect scores achieved through external assistance. By focusing on whether work is submitted, he maintains a record of participation without attempting to verify every step of the solution process. The policy acknowledges that distinguishing AI-generated content from student work has become impractical in many cases.
Why Completion Replaced Correctness
Traditional grading rewarded precise answers and clear reasoning. When AI can deliver both at a high level, that reward structure loses its meaning for assessing personal mastery. Borcherds recognized that students could achieve top marks without demonstrating the skills the course aims to build. Grading solely on completion sidesteps the need to police tool usage. It still requires students to engage with the material enough to produce a finished product. The approach preserves some accountability while accepting that full verification of originality is no longer feasible for routine assignments.
What the Change Signals for Higher Education
The decision by a prominent mathematician highlights a practical adaptation rather than a broad institutional overhaul. Other faculty members may face similar questions about how to maintain academic standards when powerful tools are readily available. Borcherds’ method offers one concrete example of responding without abandoning homework entirely. The shift also underscores the speed at which AI has entered academic settings. Instructors who once designed assignments to test understanding now weigh whether those designs still serve their original purpose. Borcherds’ choice keeps the focus on consistent student effort even as the tools available to learners continue to evolve.
AI Disclaimer: This article was created with the assistance of AI tools and reviewed by a human editor.