A mathematics professor at the University of California, Berkeley, has stopped grading homework for accuracy, arguing that widespread artificial intelligence use has made traditional evaluations increasingly meaningless. Richard Borcherds, a 1998 Fields Medal recipient, now awards homework credit based on completion rather than correctness after observing students consistently achieve high scores with AI assistance. Borcherds acknowledges the approach is imperfect but maintains that grading AI-generated answers offers little meaningful assessment of students’ mathematical abilities. The decision exposes a growing challenge for American higher education as universities struggle to distinguish genuine academic achievement from machine-produced results. Berkeley has declined to adopt standardized AI-detection software because of reliability concerns, leaving individual instructors responsible for establishing classroom policies. Meanwhile, Borcherds warns that rapidly advancing AI could eventually outperform human mathematicians, raising questions about the future of mathematics education and the value of traditional academic credentials.
Key Takeaways
• Academic Standards Under Pressure: Berkeley professor Richard Borcherds has abandoned correctness-based homework grading because AI-assisted assignments no longer reliably demonstrate students’ independent mathematical abilities.
• Universities Lack Effective Safeguards: Berkeley has found available AI-detection systems insufficiently reliable for campuswide adoption, leaving professors to establish their own rules while increasingly relying on supervised examinations.
• Mathematics Faces an Uncertain Future: Borcherds warns that continued AI advances could surpass human mathematical capabilities, threatening established teaching methods, academic assessment, and traditional career expectations.
In-Depth
Artificial intelligence is forcing American universities to confront an uncomfortable question: What is a college education worth when students can outsource their intellectual work to machines?
At the University of California, Berkeley, mathematics professor Richard Borcherds has stopped grading homework for correctness, citing widespread student reliance on AI. The Fields Medal-winning mathematician now awards credit for completed assignments, acknowledging that the arrangement falls short of an effective academic standard.
Borcherds argues that grading accuracy encourages students to use AI rather than independently solve problems. His decision highlights a fundamental weakness in traditional coursework: Correct answers no longer necessarily demonstrate comprehension, discipline, or individual achievement.
The problem extends beyond one classroom. Berkeley has examined AI-detection technologies but found them insufficiently reliable for standardized implementation. Consequently, instructors retain considerable discretion over classroom restrictions, creating inconsistent expectations across courses.
Borcherds has reportedly emphasized supervised, in-person examinations as a more dependable measure of student knowledge. Such assessments could become increasingly important as universities attempt to preserve the distinction between genuine proficiency and technological assistance.
The implications extend into mathematical research itself. Borcherds has warned that continued AI development could eventually leave human mathematicians unable to outperform increasingly sophisticated systems, complicating his ability to recommend mathematics as a career.
For American higher education, the central issue is accountability. Universities charge substantial tuition while promising students marketable knowledge and demonstrated competence. If institutions cannot reliably establish whether students possess those abilities, the credibility of their credentials comes into question.
AI may transform education, but technological convenience cannot replace the responsibility to ensure students actually learn.
Sources
• https://hoodline.com/2026/10/berkeley-math-professor-stops-grading-homework-amid-student-ai-use/
• https://www.theeducationmagazine.com/news/uc-berkeley-professor-stops-grading-homework/