Roberto Serrano suspected something was wrong the moment he graded the midterm. The Brown University economics professor had opened his advanced course on welfare economics and social choice theory to take-home exams after a campus shooting. Enrollment jumped from the usual 30 students to 86. The average score came in at 96 out of 100. Dozens earned perfect marks. The exam had been harder than previous versions.
But the answers felt off. They carried a convoluted style that didn’t match how his students usually wrote. Serrano and his teaching assistants fed the questions into ChatGPT. The outputs looked strikingly similar. He shared his concerns in a message to the class. Many had used artificial intelligence on the midterm, he believed. To give them a chance to prove otherwise, he switched the final exam to in-person only.
The results spoke volumes. Eighteen students dropped the course. Nine more skipped the final. Among those who showed up, the class average crashed to 48. The chart that plots each student’s performance tells the story in stark lines. High 90s on the midterm often turned into 50s or worse on the final. One student scored 95.5 then 95. Another went from 55 to 59. The pattern left little doubt in Serrano’s mind.
“The cost of cheating has basically gone down to zero,” he told Business Insider. “It’s very easy for students to succumb to the temptation.”
This wasn’t an isolated case of a few students cutting corners. Serrano has conclusive evidence that at least 50 students cheated on the March midterm exam. It marks one of the largest known incidents of its kind at an Ivy League school. And it arrives at a moment when universities across the country grapple with the same problem. AI tools have made it simple to generate essays, solve problems, and mimic human reasoning at scale.
Serrano, who went blind at age 17 from retinal dystrophy, learned Braille and earned his way to Harvard. He views challenges as optimization problems with constraints. His own path informs his stance. He sees this scandal as more than academic dishonesty. It points to something deeper about how the next generation approaches knowledge.
“We cannot afford to have a society in which a significant fraction of our best young minds think that cheating is okay,” he told Inside Higher Ed. “That leads to a declining society, to a failed society. We cannot choose to become idiots.”
The words carry weight. They come from a professor who has taught the material for years. Historical midterm averages in his course ranged from 65 to 80 percent. This version was tougher because it allowed unlimited time. Yet scores soared. The final, taken under supervision with no outside help, exposed the gap. Only a handful of students maintained strong performance. Many top performers from the midterm simply disappeared.
Brown’s response has struck Serrano as tepid. He submitted his evidence to the university’s Standing Committee on the Academic Code back in May. For weeks he heard nothing. After the story broke this week, administrators asked him to file individual complaints against each suspected student, complete with copies of their exams. The process feels cumbersome for a case this large. Faculty members already shoulder heavy loads. Adjudicating dozens of cases brings no extra compensation or time relief.
A spokesperson for the university pushed back on claims of inaction. Brian E. Clark, vice president for news and strategic campus communications, said Brown treats every allegation of academic integrity violations with the utmost seriousness. The committee would move forward according to its procedures. Multiple leaders had reached out to Serrano. Yet the professor describes the overall reaction from leadership as meek. A recent provost-led report on generative AI in teaching and learning at Brown found that 56 percent of undergraduates use such tools daily or weekly. Large majorities of students expressed concerns about the impact on their learning and feared negative effects on their cognitive capacity. Seventy-five percent of professors share worries about cheating.
The report recommends amending the academic code to address AI misuse directly. It calls for clear bright lines on what counts as unacceptable. It suggests de-emphasizing punishment in favor of open dialogue. Serrano agrees that norms are shifting. But he insists on holding the line. His decision to void the midterm and reweight the final toward 80 percent of the grade led to 19 students failing the course. The passing threshold dropped to 40 percent from the usual 50. The outcome reflects the reality he uncovered.
News of the scandal spread quickly. It hit at a time when similar stories emerge from campuses nationwide. A Princeton survey found nearly 30 percent of students admitted to using AI for cheating on at least one assignment. Data from other schools shows usage rates climbing toward 80 percent in some recent unpublished studies. Professors experiment with new assignment designs that resist AI assistance. Some move entirely to oral exams or in-class work. Others ban AI outright and accept the enforcement headache.
Serrano plans to eliminate take-home exams and homework from his grading in the future. He has received hundreds of emails from alumni, colleagues, and others. The attention has left him a bit overwhelmed. Yet he sees value in the conversation it has sparked. Paul Graham, co-founder of Y Combinator, shared thoughts on the matter on X. Staffers from Google DeepMind weighed in as well. The tech community has taken notice because the implications stretch far beyond the classroom.
Students who rely on AI may enter the workforce without the deep understanding their grades suggest. Employers already question whether recent graduates can think critically or solve problems independently. One consistent low scorer on both exams earned Serrano’s admiration for integrity. The student who performed strongly on both without apparent AI help stood out as the kind of person he would hire. Integrity matters. It always has. But the tools available today make it easier than ever to fake it.
Brown’s own admissions policy prohibits applicants from using AI in any form. That rule remains aspirational when it comes to current students. The university’s AI committee acknowledged that enforcement alone won’t solve the issue. Dialogue about responsible use must play a role. Still, Serrano draws a firm boundary. Cheating at this scale hollows out education. It trains minds to outsource thinking rather than develop it.
The December 2025 shooting that prompted the original policy change adds another layer. A gunman killed two people on campus, one of whom had recently introduced herself to Serrano. The tragedy shook everyone. The decision to offer flexibility made sense in the moment. It also created the conditions for widespread abuse. Serrano doesn’t regret trying to accommodate his students. He does regret what it revealed about their choices.
Recent coverage has amplified the story. Ars Technica highlighted how the final exam average fell by half. It noted the 27 students who either dropped or skipped the test, 22 of whom had scored perfectly on the midterm. El Pais reported on the case in late June, calling it the biggest known scandal at Brown. Fortune framed it around the professor’s blunt assessment that humanity has chosen to become idiots. The coverage converges on the same point. This incident isn’t an anomaly. It reflects a broader failure in how institutions and individuals respond to powerful new technology.
Professors now face a choice. They can design assessments that AI cannot easily complete. They can accept some level of assistance and focus on process over product. Or they can return to proctored, closed-book formats that limit outside influence. Serrano has made his decision. No more take-homes. No more homework that can be outsourced. The final exam format worked as a diagnostic tool. It showed who had mastered the material and who had not.
Yet the larger question lingers. What happens when an entire generation learns to bypass the hard work of thinking? Serrano’s warning isn’t hyperbolic. Societies built on knowledge and expertise cannot thrive if their best minds opt out of acquiring either. The chart from his class should serve as evidence. Scores don’t lie when conditions change. Students who dropped or failed under supervision revealed their dependence on external help.
Universities will continue to study the problem. Brown’s report offers a starting point with its call for updated policies and honest conversation. Other schools experiment with honor codes that explicitly address AI. Some deploy detection software, though those tools remain imperfect and easily gamed. The fundamental issue goes beyond technology. It touches on values. On what society expects from its educated class. On whether learning still matters when answers come cheap.
Serrano refuses to lower the bar. His course demands real engagement with complex ideas about welfare and choice. Those concepts don’t yield to quick prompts. They require time, reflection, and intellectual effort. The students who earned their grades the old-fashioned way proved it can be done. The rest showed what happens when temptation wins. The scandal at Brown won’t be the last. But it offers a clear signal. Pay attention. Adjust. Or watch the foundations erode one perfect score at a time.
WebProNews is an iEntry Publication