UC Berkeley Math Professor Abandons Homework Grading as AI Tools Render Traditional Assessment Obsolete

Fields Medal Winner Abandons Traditional Homework Grading

Richard Borcherds, a renowned UC Berkeley mathematician and 1998 Fields Medal winner, has stopped grading homework for correctness, declaring it pointless in the age of AI in education2

2

. Instead, he awards full credit to anyone who submits assignments on time, admitting it's "not a very good system" but acknowledging he has yet to find a better alternative2

2

. The professor now bases grades mainly on in-person exams with no electronic devices allowed, noting that online exams have become "a waste of time as several students just use AI to answer the questions"5

5

. Borcherds observed that average scores increase significantly during online exams, a pattern confirmed by his classroom experience5

5

.

Research Confirms AI's Impact on Assessment Integrity

A 2024 study published in Discover Education analyzed exam scores from an introductory statistics course before and after ChatGPT arrived, finding that scores rose significantly when exams of similar difficulty moved from in-person to online formats5

5

. Another study in the journal Teaching of Psychology compared two sections of an undergraduate personality psychology course, with 153 students taking the final exam in person and 160 taking it online without monitoring5

5

. The online group scored significantly higher, and whether the exam was online or in person predicted final scores better than earlier course performance5

5

. These findings underscore the challenges professors face in maintaining academic integrity when AI tools like ChatGPT are readily available to students.

California Professors Adopt Old-School Tactics

Professors across California's elite institutions are fighting back against AI-assisted cheating with decidedly old-school weapons: paper and pencil3

3

. Morgan Polikoff, a professor of education at USC Rossier, said "the technology has been shoved down all our throats and we're all just dealing with the fallout"3

3

. He planned to adopt a policy similar to Borcherds', switching more graded undergraduate assignments to sit-down, paper-based exams3

3

. At UCLA, statistics lecturer Thomas Maierhofer has taken a hybrid approach, conducting all exams as in-person on-paper tests without phones or laptops while grading homework and labs on completion only3

3

. USC pays roughly $3.1 million annually for a ChatGPT Edu subscription covering 80,000 users, yet leaves AI rules up to individual instructors3

3

.

The Educator's Dilemma in Higher Education

A former Arizona State University professor described grading homework as an "infuriating exercise in futility" after AI tools proliferated following ChatGPT's late 2022 launch1

1

. The instructor suspected at least 90% of online students were using AI to complete all or part of their work1

1

. When confronting students about AI-generated assignments, the professor found that promises to stop using AI tools rarely held up1

1

. Arizona State University's current stance reads: "Our focus is on using AI not only to address educational challenges, but to do so in a way that respects and uplifts all learners"1

1

. UC Berkeley investigated AI detection tools including Turnitin and Rumi but found them unreliable enough that they opted against campus-wide adoption, leaving individual professors to set their own policies5

5

.

Divergent Views on AI's Role in Learning

At Stanford Law School, professor Julian Nyarko takes a more optimistic view, stating he's "skeptical of human capability, more so than overly enthusiastic about AI capabilities"3

3

. In a study Nyarko led, 16 law professors blindly rated answers to common student questions and preferred Google's AI over their own colleagues 75% of the time3

3

. However, he warned against irresponsible use: "There's a way to use AI that will harm students, which is basically, just write this paper for me or solve this question for me"3

3

. Many of his colleagues who previously assigned eight-hour take-home exams now feel they can't, fearing students will simply hand the work to AI3

3

. UCLA's Maierhofer is developing ALISSTAIR, a custom chatbot built to support student learning, viewing AI as a potential teaching tool when paired with rigorous in-person exams3

3

.

Implications for Critical-Thinking Skills and Future Careers

Borcherds, who five years ago believed mathematics was among the fields least likely to be disrupted by AI, now questions the future of his profession2

2

. When students express interest in pursuing math careers, he admits: "I have no idea what to advise. I even asked ChatGPT this exact question, and even ChatGPT said it didn't have any idea what was going to happen"2

2

. Students who outsource learning to AI risk losing the ability to develop critical-thinking skills and perform expertise with ease1

1

. USC's Polikoff emphasized the need for standardized and AI-proof measures, stating: "It's become increasingly difficult to know what anyone knows and can do, and that signal is important"3

3

. Despite the challenges, 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"5

5

.