Caltech Math Professor Stops Grading Homework Because AI Does It
Caltech professor Yi Ni grades math homework only on completion because AI completes it undetectably, forcing syllabus changes across the elite STEM school.

A professor with 17 years at Caltech has stopped grading homework for accuracy because, as he puts it, virtually every student can now complete it with AI. Yi Ni, teaching an introductory math class at the Pasadena university on Friday, said he grades problem sets only on completion — an admission that assessment has quietly broken in one of the most selective STEM programs in the United States.
"There is a problem in the education process," Ni said. "Virtually all the homework in college math, you can use AI."
The shift matters beyond one classroom. Chatbots, once mocked for failing to count the r's in "strawberry," now perform math at the level of a graduate student, according to the report. That capability sits directly under a core requirement: the class Ni teaches is mandatory for every Caltech undergraduate, regardless of major.
What does this mean for students' market position?
For an undergraduate weighing a math degree, the signal is uncomfortable. Business Insider has reported that some students are reconsidering whether to study math at all, given that AI tools now match graduate-level performance in the field. The skill that a math degree certifies is the same skill employers can increasingly buy from software.
Inside the lecture hall, however, the scene looked analog. Ni used no laptop and no projector, filling eight blackboards with chalk during the Friday session. Students took notes on iPads; some arrived late; a few had no note-taking materials; a couple closed their eyes for extended periods. No student was visibly prompting a chatbot.
Can professors even detect AI homework?
Here the skeptic's question gets a sober answer: no. Ni said he cannot tell whether an assigned problem set was completed with AI. Unlike AI-generated writing — typically an obvious mix of short sentences, em dashes and rhetorical questions — math is pure numbers, the specialty of these programs. The grading change is not a policy choice so much as a surrender to undetectability.
One student, sophomore math major Sathvik Redrouthu, who took the class last year, said he avoided AI because he wanted to understand the concepts. Using it on homework, he said, is like "shooting yourself in the foot" once an in-person exam arrives.
"I personally think that if you leave AI out, then it pushes the average student to learn properly," Redrouthu said. He added that students who are not interested in math will use AI and get away with it — until the exam, at least.
What happens next?
Caltech has no unified guidance on how math professors should plan around AI. Some of Redrouthu's classes this year carry an explicit caveat on the syllabus: "AI allowed." "Syllabuses are changing," he said. As institutions revise requirements one course at a time, the burden of proving competence is shifting from graded homework toward proctored exams — and toward students themselves.
Source: feeds.businessinsider.com
More from Sophie Lindqvist
Show full bio
Staff writer covering regulation, policy and their impact on Job market data and labor economics at Path & Pay.
583 articles