Professor Roberto Serrano reveals evidence of AI-assisted cheating after moving exams in-person. #
Brown University economics professor Roberto Serrano says he has uncovered what he believes is evidence of mass AI-assisted cheating in his own classroom, after moving his final exam from a take-home format into a supervised, in-person sitting.
The average score across his 86-student course collapsed from 96 out of 100 on the take-home midterm to just 48 once students sat comparable material in a monitored room.
The Numbers Behind the Drop #
Serrano's Welfare Economics and Social Choice Theory course usually attracts a small cohort. In past editions, he said, enrolment topped out at 30 students and was sometimes as low as eight.
This spring, 86 students enrolled, a jump he has linked to his decision to allow a take-home midterm. Forty of them scored a perfect 100.
Serrano told the class the results were 'extraordinary,' noting the course had historically averaged between 65 and 80 per cent on exams he considered easier than the one he set this year. He said he made the exam deliberately harder, specifically because students would have unlimited time to complete it at home.
Serrano Gave Students a Chance to 'Prove Him Wrong' #
Suspicious of the scores, Serrano told his class he believed the midterm reflected 'massive cheating,' but offered them a way to challenge that conclusion. If the final exam's grade distribution looked similar to the midterm's, he said, the midterm would stand. If not, he would declare it "null and void" and reweight the final accordingly.
The catch was that the final would be sat in person. Twenty-seven of the 86 students dropped the course once they learned this, most of whom, Serrano said, had scored 100 on the midterm.
Of the 59 students who stayed and sat the final, 19 failed the course once the midterm was voided as promised. Serrano included one question from the midterm on the final: the class had performed 'beautifully' on it the first time, but the average score on the same question in person was just 10 per cent.
'The Empirical Evidence Is Overwhelming' #
Serrano has taken his findings to Brown's Standing Committee on the Academic Code, telling the committee the scale of the issue, around 50 students, meant 'there's a systemic failure that needs to be addressed from the top.'
He said he forwarded his communications with students to the provost and dean of the college and initially received 'complete silence.' The dean eventually responded, he said, calling the episode 'a wake-up call.'
Serrano, who has taught economics at Brown for 34 years, pushed back on a Brown Daily Herald quote from Associate Dean of the College for the Academic Code Love Wallace, who wrote that code violations are 'almost never' malicious and often come from 'a split-second decision.' Serrano called that framing implausible given the take-home exam had no time pressure: 'You cannot seriously tell me this was a split-second decision... many of those students cheated all the way to 100.'
He said he told his class directly that anyone who had simply asked an AI chatbot to complete the exam was 'showing me you're totally irrelevant.'
Brown's Own Data Shows Widespread AI Use #
The scandal has emerged as Brown's own Generative AI in Teaching and Learning Committee published its final report, drawn from feedback submitted by more than 100 faculty members and a university-wide survey.
The committee's report found that 56 per cent of undergraduate respondents and 67 per cent of graduate and medical students said they used generative AI tools daily or weekly. Large majorities also reported concerns about the impact on their own learning and a fear of losing cognitive capacity.
The report recommended amending Brown's academic codes to 'address GenAI realities and safeguard against misuse,' arguing the codes should speak to both integrity and the potential harm AI poses to student learning.
Serrano said he does not want to 'demonize' AI as a technology, but believes universities must draw firmer lines around its use in assessment. 'We cannot afford to have a society in which a significant fraction of our best young minds think that cheating is okay,' he said. 'That leads to a declining society, to a failed society... We cannot choose to become idiots.'
His case lands as universities across the US grapple with how to assess students honestly in the AI era, from Princeton scrapping a 133-year-old unsupervised exam tradition to widening faculty anxiety documented in Brown's own report.
© Copyright IBTimes 2025. All rights reserved.