A renowned UC Berkeley mathematician has abandoned grading homework for correctness and scrapped online tests, citing rampant artificial intelligence use by students , The New York Post reported.
Richard Borcherds, winner of the prestigious Fields Medal in 1998, told The California Post that advanced AI has made it nearly impossible to distinguish human submissions from chatbot-generated answers.
"I now base grades mainly on in-person exams with no electronic devices allowed," Borcherds said. "Online exams are now a waste of time as several students just use AI to answer the questions."
While he cannot always definitively prove AI usage, Borcherds noted that average scores in online assessments have surged suspiciously.
His observations are backed by recent academic research. A 2024 study in the journal Discover Education found that introductory statistics exam scores rose significantly when tests were moved online following the launch of ChatGPT.
Similarly, research published in Teaching of Psychology revealed that unmonitored online students scored significantly higher on final exams than their in-person counterparts. Researchers concluded that remote tests inflate scores and fail to reflect a student's true abilities.
Borcherds noted that while early AI tools gave themselves away with characteristic mathematical errors, rapid technological advancements have rendered such mistakes rare.
Consequently, his students now receive homework credit merely for completion. Borcherds argued that grading for correctness penalises honesty by adding a "huge incentive" to cheat, and that paying teaching assistants to mark AI-generated answers is a waste of time and money.
With UC Berkeley finding AI-detection software like Turnitin unreliable for campus-wide adoption, individual professors have been left to devise their own policies.
Despite the academic disruption, Borcherds remains philosophical. "If the worst effect of AI on society is that it is harder to assess homework, then I will be very relieved," he said.
Staff reporter