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 final exam in person after unusually high take-home midterm scores made him suspect AI-assisted cheating. The class had 86 students; after the in-person final was announced, 18 dropped the course and nine did not show up for the exam, leaving 59 test takers. The midterm average was 96%, despite Serrano saying earlier midterms usually averaged 65% to 80% and this one was designed to be harder.
Nauti's Take
The real signal is not that students may have used AI. That was predictable.
The sharper gap is 96% at home versus 48.6% in the exam room. If institutions want to assess knowledge, they need formats where thinking becomes visible: oral checks, process evidence, in-person components, version history.
AI bans without assessment redesign are mostly theatre.
Briefingshow
The case does not prove every individual cheating allegation, but it shows how fragile take-home exams have become without clear AI rules. Universities can no longer treat AI misuse as a small conduct issue. They need assessment designs, policies, and bulk investigation procedures that match the scale of the problem.
Sources