Breaking

Fields Medal Mathematician Abandons Homework Grading as AI Outpaces Student Problem-Solving

October 27, 2026 at 12:28 EDT

Foundation Models
Education

Richard Borcherds, a Fields Medal-winning mathematician at UC Berkeley, has stopped grading homework assignments for correctness, instead awarding credit based solely on completion. Borcherds, who teaches MATH 250A, a graduate-level abstract algebra course, made the shift after observing that widespread student reliance on artificial intelligence tools has rendered traditional homework assessment ineffective.

October 2026 · Education

AI Transforms University Grading

As artificial intelligence tools match student problem-solving, a leading mathematics professor eliminates correctness-based grading. Universities nationwide redesign assessment to adapt.

Assessment Crisis

61%

False positive rate of AI detection software for non-native English speakers

100%

Credit awarded by Borcherds for homework submission, regardless of correctness

Universities Adapting Evaluation Methods

→

Oral Examinations — One-on-one questioning to verify student understanding in real-time

→

In-Class Handwritten Work — Eliminating remote access to AI tools during assessments

→

Milestone Projects — Semester-long assignments with supervised checkpoints

→

Varied Formats — Presentations, debates, demonstrations, and hands-on work alongside writing

The Core Problem

Traditional homework—once a reliable way to assess student learning—has become ineffective when AI systems match or exceed student problem-solving ability. Without reliable detection, grading for correctness penalizes honest work and incentivizes AI use. The result: educators are abandoning homework-based grading entirely and redesigning how learning is evaluated.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.