Caltech math professor grades homework on completion, not accuracy, citing AI use
Caltech professor Yi Ni changed homework grading in a required math course from accuracy to completion, citing widespread AI use. The report shows no unified campus guidance, leaving policy to individual instructors. It matters because problem sets have been a core evidence of learning in STEM, and AI can solve assigned problems at scale.
Professor Yi Ni, who has taught at Caltech for 17 years, told Business Insider that he no longer grades introductory math homework on accuracy because students use AI to complete it. He now grades only on completion. The class is a core requirement at Caltech, meaning all students must take it. Ni said virtually all college math homework can be done with AI. He can tell when writing is AI-generated, but not when math problem sets are.
The evidence is direct but narrow: Ni's grading change, a classroom observation, and student comments. Business Insider reported seeing students taking notes on iPads, not prompting chatbots during lecture. A math major, Sathvik Redrouthu, said he did not use AI because he wanted to understand concepts and that relying on it would fail on in-person exams. Redrouthu said students not interested in math would probably use AI and pass homework until exams. The source gives no institutional data on AI use rates.
Caltech has no unified guidance for math professors on AI, according to the report. That absence shifts policy to individual instructors, creating inconsistent classroom rules. Some syllabuses now explicitly allow AI, Redrouthu said. The pattern matters because assessment has long relied on graded problem sets as proof of practice. When homework no longer measures accuracy, the primary signal of learning shifts to other assessments, such as exams or supervised work. Ni's move shows a top-tier STEM school confronting that decision in real time.
The change affects students who need math credentials for engineering, science, and research careers, and it affects professors who must preserve trust in grades. In a required course, completion credit means grades no longer provide direct evidence of accuracy. That changes what homework signals to employers, graduate programs, and students themselves. The stakes are the reliability of assessment at a leading technical university when AI can solve assigned problems at scale.
Students and professors at Caltech are directly affected because a required math course no longer uses accuracy in homework as evidence of learning. The change affects students who need math credentials and professors who must preserve grade reliability.
Sources (1)
- [1]Caltech professor no longer grades homework's because everyone uses AI(https://www.businessinsider.com/ai-math-class-university-homework-caltech-careers-2026-11)