"AI Does All the Assignments, So What's the Point of Grading?"... Fields Medalist Professor Stops Grading for Correctness
Fields Medalist Professor Gives Credit Only for Submitting Assignments
"Grading AI-Assisted Assignments for Correctness Is Pointless"
AI Cheating Cases Continue to Emerge at South Korean Universities
A UC Berkeley mathematics professor who won the Fields Medal has stopped grading students' assignments based on whether their answers are correct. Many students have been using artificial intelligence (AI) to solve assignments and earn high scores, making it pointless to grade them on correctness, he said.
On October 25, the California Post, a sister publication of the New York Post, reported that Richard Borcherds, a professor in the Department of Mathematics at the University of California, Berkeley (UC Berkeley), said in a recent interview with the campus newspaper, The Daily Californian, that he had changed how he evaluates assignments.
"There's No Point in Checking Answers"... A Math Professor's Concerns in the Age of AI
Richard Borcherds, professor of mathematics at the University of California, Berkeley (UC Berkeley). Photo: UC Berkeley's official website.
View original imageProfessor Borcherds said he had watched students using AI consistently earn high marks on assignments. "Now that it's become difficult to prevent students from using AI, there's no point in paying graders to check whether their answers are correct," he said. He now gives students credit only for completing and submitting their assignments. Professor Borcherds said this approach "isn't very good," but added, "There isn't an alternative right now." Other assessments, including exams, remain unchanged.
Professor Borcherds received the Fields Medal, often called the "Nobel Prize of mathematics," in 1998. He said that just five years ago, mathematicians seemed relatively safe from advances in AI, but now he is unsure whether he should encourage students to pursue mathematics. "I have no idea what advice to give," Professor Borcherds said. "I asked ChatGPT the same question, and even ChatGPT said it didn't know what would happen in the future." He added that if AI continues to advance at its current pace, "in a few years, there may be nothing that human mathematicians do that AI can't do better."
In another interview with The Daily Californian last month, Professor Borcherds reportedly expressed surprise at AI's recent achievements in mathematics, saying they were at a level he had expected to take another 10 years to reach. At the time, he said, "If you're starting out in mathematics as a graduate student, AI will probably always be better at mathematics than you are." He was also reported to have said, in reference to the graduate-level abstract algebra course he teaches, that "AI would be a better instructor than he is."
AI's mathematical abilities have already reached the point where it can take on major unsolved problems. On September 8, OpenAI announced that one of its internal AI systems had found a solution to the Navier-Stokes equation problem, one of the seven Millennium Prize Problems. The company said about 10,000 AI agents produced a proof in 88 hours, which was then verified using Lean, a mathematical proof verification program. Terence Tao, a Fields Medal-winning professor at the University of California, Los Angeles (UCLA), called it an "amazing achievement." However, some have pointed out that the proof concerns a case involving external forces, and therefore differs in scope from the original problem, which carries a $1 million prize (about 1.3 billion won). The Clay Mathematics Institute, which set the problem, has not yet recognized it as solved.
35% Fail Rate and a Sharp Rise in Cheating... Universities Respond in Different Ways
The effects of AI use on academic performance are also showing up in the numbers at the university. According to The Daily Californian, in the spring 2026 semester, 35.3% of students enrolled in the university's introductory computer science course CS10 received failing grades (F), as did 10.6% of those in CS61A. Department guidelines set 7% as the threshold for the share of D and F grades in lower-division courses. Dan Garcia, who taught the courses, cited "a sharp rise in academic misconduct by students using AI" as the main reason. A survey of about 95,000 UC undergraduates also found that students who frequently used generative AI were more likely to say they had engaged in academic misconduct.
Responses vary by department and professor. UC Berkeley School of Law has banned the use of generative AI for most assignments and exams. At the university-wide level, however, instructors are asked to specify their AI-use policies in their syllabi, while individual professors determine what is allowed based on the level and subject area of each course. The university has also tested several tools for detecting AI use, but concluded that they were not adequate for adoption as a common campus-wide standard.
Instead, in its official guidance, the UC Berkeley Center for Teaching and Learning recommends redesigning assessments rather than relying solely on catching students using AI. Suggested approaches include asking students to explain the reasoning behind their answers and their solution process, or briefly explain their submitted work orally. The guidance also recommends that professors first try entering assignment questions into generative AI and, if AI can easily produce the correct answer, revise the questions so that students must think more deeply to solve them.
Hot Picks Today
"Two Spend 500,000 Won in a Flash, Yet Wait Two Hours?" ... 'Dreamland' Shunned by Kids and Foreign Tourists
- The Era of Getting 200 Million Won After 30 Years Is Over... A Japanese Bank's "Perpetual Department Head" Who Just Clocked In Cries Out
- Spain Hit by Its Worst Tragedy... Lee Kangin, Moreno and Real Madrid Join Soccer World in Mourning
- They Said, “Make It Your Home for Life” - Then Suddenly Kicked 200 Passengers Off... What Happened to the Around-the-World Cruise?
- "130 Rice Bowls Sold Out in a Day"... 'Unfiltered Face' Adds 8 Million Won to Night Market Sales: China's Bungeunwoo
Cases of academic misconduct involving AI have also continued to emerge at South Korean universities. In November 2025, evidence emerged that many of the about 600 students enrolled in Yonsei University's "Natural Language Processing (NLP) and ChatGPT" course had cheated by using ChatGPT and other tools during a midterm exam held online. The professor in charge announced that the midterm exam scores of students caught cheating would be recorded as zero. At Seoul National University, a student was caught using AI to cheat on a midterm exam in a general education course, and the exam was retaken. According to data the Ministry of Education submitted to the National Assembly, of the 224 cases of exam cheating detected by universities nationwide between 2020 and November 2025, four explicitly cited the use of ChatGPT as the reason. However, the figures include only cases directly identified by university administrations, so the actual scale may be larger.
© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.