At Caltech, mathematics professor Yi Ni still teaches in a traditional style, filling blackboards with formulas and problems, but AI has changed how he evaluates students. Because AI tools can now solve much of the homework assigned in college math - and he cannot reliably tell whether students used them - Ni grades problem sets for completion rather than correctness. He says the shift exposes a broader challenge for education: take-home work may no longer show what students understand. The class is a core requirement for all Caltech students, and the campus has no unified policy directing math faculty on how to respond.
A sophomore math major who took Ni’s course said he avoided AI because working through problems helped him learn and prepare for in-person exams. He argued that students who use AI to bypass homework may struggle when they must solve problems themselves, though those uninterested in math may rely on it and still benefit until exam time. During a visit to the lecture, students appeared to take notes, ask questions, or disengage; none were seen prompting AI. Meanwhile, policies are evolving unevenly: some courses now explicitly permit AI, and syllabuses are changing. The account captures a tension between AI’s ability to produce correct mathematical answers and the continuing need for students to develop their own understanding.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.