‘We cannot choose to become idiots’: a Brown professor’s proof of mass AI cheating

What happened

Roberto Serrano, an economics professor at Brown University, detected a massive surge in AI-aided cheating during his course. His evidence comes from the dramatic spike in average scores on a take-home midterm, which hit 96 out of 100. When the final exam was held in person, the average score collapsed to 48. Serrano argues that the only reasonable explanation is that most students relied on AI tools extensively during the unsupervised exam.

Why it matters

This case provides concrete proof that AI-assisted cheating is already widespread and skews remote assessments. Educators and institutions must recognize that traditional take-home exams can no longer reliably measure student understanding without safeguards against AI usage. For businesses and organizations employing remote testing or skills assessments, it signals that trust in unmonitored work will degrade unless new verification or proctoring methods are adopted. The incident also pressures education technology companies to develop AI-detection tools that keep pace with AI’s ability to generate sophisticated responses.

What to watch next

Expect universities and certification bodies to increasingly embrace in-person or monitored assessments to hold down cheating risks. Look for more research quantifying AI’s impact on grading and academic integrity. AI-detection software providers will be under pressure to improve accuracy and avoid false positives. Meanwhile, educators will have to rethink assessment design, balancing the convenience of remote testing against the growing threat of AI-enabled fraud. This case serves as a clear warning that remote evaluation without strong controls is vulnerable and potentially meaningless in the age of AI.

AI Quick Briefs Editorial Desk