The Sheffield Press

US News

Brown professor says AI cheating turns classrooms into a failed society

By Darren Ryding ·
Brown professor says AI cheating turns classrooms into a failed society

Brown University economics professor Roberto Serrano turned his spring 2026 final in ECON 1170, Welfare Economics and Social Choice Theory, into an in-person exam after a take-home midterm produced results that did not fit the course’s long record of harder grading. He reviewed answers that echoed ChatGPT-style reasoning.

Serrano had used a take-home, closed-book midterm for the first time in nearly two decades teaching the course. He said the format was part of an effort to accommodate students unsettled after the Dec. 13, 2025 shooting at Brown, when two students were killed and nine others were injured. The class, which usually enrolled fewer than 30 students, swelled to 86 this semester, and Serrano said the take-home structure partly drove the jump.

The average was 96 percent, the median was 98 percent, and 40 of the 86 students scored 100 percent. Serrano said historical averages for the exam had generally fallen between about 65 percent and 80 percent, or 65 percent and 85 percent. Some submissions included oddly convoluted proof strategies that matched AI-generated reasoning rather than the work he expected from students.

AI-generated illustration
AI-generated illustration

After the midterm, Serrano told students he suspected many had used AI to cheat. That change carried immediate consequences: more than a dozen students dropped the course and many more failed once the final could no longer be outsourced to a chatbot. Serrano says he has conclusive evidence that at least 50 students cheated on the March midterm exam.

Serrano described the administration’s reaction as “meek.” The Academic Code Committee later called the episode a “wake-up call,” while Brown’s own cheating panel has no standardized method for handling AI misuse or other misconduct.

Brown University — Wikimedia Commons
Marco Almbauer via Wikimedia Commons (Public domain)

Assistant professor Bobby Pakzad-Hurson said he no longer trusts take-home exams after seeing what happened in Serrano’s class, and other economics faculty said AI appears far more dangerous in homework and take-home settings than in-person tests.

US newsBrown