Professor suspected AI-powered cheating on take-home midterms, makes finals in-person — only two students scored within 10% of their midterm score
TL;DR
Brown professor Roberto Serrano moved the Econ 1170 final back in person after unusually strong take-home midterm results raised cheating concerns. Of 86 enrolled students, 18 dropped after the announcement and nine skipped the final. Among 59 remaining test takers, three scored zero. Only two students scored within 10 percentage points of their midterm result on the in-person final; only one did better than on the midterm.
Nauti's Take
This is the uncomfortable part of AI in education: not every suspicious answer proves cheating, but a 96 percent midterm average followed by a sharp in-person collapse is hard to hand-wave away. The old take-home model breaks when students can generate polished solution paths with little friction.
Fair exams need to either integrate AI openly or be designed so that individual reasoning is actually tested.
Briefingshow
The case shows how quickly take-home exams can lose signal in an AI-heavy environment. The core issue is not just catching cheating, but assessment design: if prompts can be solved with ChatGPT-like answers, they may measure tool access more than understanding. Universities need clearer rules and formats that make reasoning visible.
Sources