Brown University Professor Exposes Mass AI Cheating After Exam Scores Plunge From 96% to 48%

"We cannot choose to become idiots": The AI cheating scandal roiling Brown University

Ivy League college students are, by definition, intelligent. They don't need to use generative AI to cheat on exams; they could just learn the material. But they also tend to be competitive, ambitious, and overscheduled, so AI can look like an easy shortcut that makes more time in their lives for things that can't be done by a chatbot. When the pressure is on, which approach do they choose? A new scandal at Brown University reveals that huge numbers of these students are likely to cheat. Record scores A recent survey of Princeton students found that 29.9 percent admitted to cheating with AI on at least one exam or assignment. But the recent situation at Brown gives us a better sense of what this kind of cheating looks like in one particular class -- and just how much it may be substituting for actual learning. And we know all this because the blind economics professor at the center of it all, Roberto Serrano, is not letting it go. In just the last week, Serrano -- who was born in Spain -- has told his story to El País and Inside Higher Ed, which have both run significant pieces on the scandal. The story that Serrano told them begins in December 2025, when a gunman attacked Brown's campus and killed two people, including one who had recently introduced herself to Serrano. Shaken by the experience, Serrano decided that his spring 2026 section of the quite difficult ECON 1170 would allow take-home exams for both the midterm and the final. Suddenly, the course received an influx of students. El País has the story: The course... typically attracts few students, but very good ones. [Serrano] has never had more than 30 students enrolled at a time, and on some occasions he had only eight. This semester, probably because of the new evaluation system, 86 students signed up for the class. The results of the midterm exam, which was administered on March 5, were extraordinary, with an average score of 96 out of 100. Forty students scored a perfect 100. This was indeed extraordinary, because as Serrano told Inside Higher Ed, "Historically the average grade in the midterm of this course has ranged between 65 and 80 [percent], and this exam was harder than the exams I wrote in the past, because... take-home is an opportunity to challenge the class a little bit more, given that you're giving the students unlimited time." Beyond the numbers, many of the answers, even when correct, felt slightly off. They had a "very convoluted style," Serrano said. When he and his grad students ran the exam questions through ChatGPT, they received similar results. A suspicious Serrano decided that he would make the final exam in-person; he would see if students did similarly well on it. He emailed his class, telling them, "I am not declaring [the midterm] void for now. I am going to give the class a chance to prove me wrong. That is, if the distribution of the final exam is roughly similar to the distribution of the midterm, I will count the midterm. Otherwise, which is of course what I expect to happen, I will declare the midterm void and reweigh the final accordingly." Eighteen students suddenly dropped the course, while nine others didn't even attend the final exam. Of those 27 students, El País noted, "22 had scored a perfect 100 in the midterm exam." Among those who took the test, the average score plunged -- from 96 all the way down to 48. A failed society? The professor was horrified by what appeared to be massive cheating in his course -- cheating that was preventing most of the students from learning the material. Serrano comes across as someone with no inclination to coddle elite students. His attitude may be traceable in part to his own childhood, in which he went blind from retinal dystrophy at age 17 and had to make a choice about what the rest of his life would look like. From El País: After a short-lived crisis, he decided [blindness] would not stop him. He learned Braille, and his excellent academic record opened up the doors of Harvard. "Of course it affects my life, but one shouldn't over-dramatize. We economists understand reality as a set of people responding to optimization problems with restrictions. I view my disease simply as one more restriction that I have to deal with, and I optimize based on that," he says. As a university, Brown is grappling with hard questions about AI use at the moment. It recently released a provost-led report (PDF) on "Generative AI in Teaching and Learning," which found that it's not just professors who have concerns. Even though "56 percent of undergraduate respondents [at Brown] and 67 percent of graduate and medical student respondents reported intentionally using GenAI tools daily or weekly," the report notes that large majorities of students also have "concerns about the impact of GenAI use on their learning" and a "fear of negative consequences for their cognitive capacity." Serrano shares those concerns, and he wants universities as a whole to stand up for human thought. That's why he's not letting this story go, despite what he contends is a fairly tepid reaction from Brown administrators. "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."

Professor's chart exposes the scale of AI cheating in college exams

Serving tech enthusiasts for over 25 years. TechSpot means tech analysis and advice you can trust. Facepalm: The endemic scale of AI-assisted cheating among students over the past few years is well documented, but few visual representations illustrate the problem as well as a recent chart from a Brown University professor. As concerning as the graph is, the university's muted response to it is potentially even more worrying. When Brown University economics professor Robert Serrano suspected that most of his students had used AI to cheat on a take-home midterm exam, he decided to compare their scores with those from an in-person final. The result was perhaps one of the clearest examples of generative AI's effect on academia. The reason Serrano allowed the take-home midterms is both reasonable and a sobering sign of the world American students currently inhabit: Some students were uncomfortable taking exams in class after a gunman killed two students and injured nine others at the university last December. However, Serrano soon noticed signs that something was amiss. When 86 students signed up for his welfare economics course that semester, compared to the usual 30, he suspected that many had taken it because he allowed take-home exams, providing them the opportunity to use ChatGPT or similar AI tools. The professor's suspicions were just about confirmed when the midterm scores averaged 96%. This was far above the typical range of 65% to 80% despite Serrano's efforts to make the test more difficult to account for the students' unlimited time. While many professors have attempted to counter AI cheating with AI-based tools, Serrano understood AI-proofing often produces false positives and negatives. Students at other universities have sued their schools after being falsely accused of using AI, and a study from last year indicated that the practice simply trains students to write against AI detectors, often by using AI. Instead, Serrano employed a more straightforward solution - he ran his midterm through ChatGPT. Unsurprisingly, the chatbot's answers closely resembled the ones his students submitted, employing similar language and reasoning. When the professor announced that the final exam would be in person, 18 students dropped the course, and another nine did not show up for the exam. The remaining 59 students' scores averaged 48.6%. Prior averages had never fallen below 65%. Three students scored 0%, and charting the results revealed that most fell more than 30 points behind their midterm scores. The chart suggests that only two students probably took the tests without AI, one of whom was likely the top scorer with 95.5% on the midterm and 95% on the final. Interestingly, the other student who appears innocent scored just 55% on the midterm and 59% on the final. They posted the lowest midterm score and were the only student whose score improved on the final. Serrano subsequently dropped the midterm scores and made the final count for 80% of the students' final grade. Normally, he would have passed everyone who scored 50% or higher, but in this case, the professor lowered the threshold to 40%. However, Serrano received no response after submitting the data to Brown University's Standing Committee on the Academic Code. After he went public with the incident, the committee asked for individual complaints against each student and copies of their exams, but Serrano suspected that they would simply run them through AI-proofers. Experts suggest that the problem will persist as long as no one is paid to ensure student integrity. While Brown University is still investigating the issue, it remains unclear whether anyone will find a better deterrent to AI cheating than analog and in-person testing.

A Brown professor's proof of mass AI cheating

A Brown professor let his class sit take-home exams. The average hit 96. Then he made the final in-person, and the score crashed to 48. The gap, he says, is a warning about what AI is doing to learning. An economics professor at Brown University suspects most of his class cheated with AI, and he has the numbers to make the case. Roberto Serrano watched his take-home midterm average hit 96 out of 100. When he switched the final to an in-person test, the average fell to 48. He has taken the story public, telling El País and Inside Higher Ed that he will not let it go. The take-home format came out of tragedy. After a gunman killed two students on campus last December, many said they felt anxious sitting exams in a room full of people. Serrano offered take-home midterm and final papers to ease that. The irony stings: the one time in decades he relaxed the rules, much of the class cheated. The numbers that gave it away Serrano's course, ECON 1170, is an advanced undergraduate economics class that usually draws a small, strong group. He had never taught more than 30 students, and once had just eight. This term, 86 signed up. The new take-home format likely drew them in. The midterm results were, in his word, extraordinary. The class averaged 96, and 40 students scored a perfect 100. The historical average for the course sits between 65 and 80, and this exam was harder than usual. Take-home, Serrano reasoned, was a chance to push the class further, given the unlimited time. The answers themselves felt off. Many correct ones carried a "very convoluted style". When Serrano and his graders fed the questions to ChatGPT, they got similar results back. The test he set to prove it So he set a trap. He told the class the final would be in-person, and that he would compare the two distributions. If they matched, he would keep the midterm. If not, he would void it and reweight the final. The response spoke for itself. Eighteen students dropped the course, and nine more skipped the final. Of those 27, El País noted, 22 had scored a perfect 100 on the midterm. Among the students who did sit the exam, the average dropped from 96 to 48. By Serrano's count, at least 50 students cheated on the midterm, and he calls the evidence overwhelming. A wider reckoning Brown is not alone. A recent survey of Princeton students found that 29.9 per cent admitted to cheating on at least one exam or assignment, most of it using AI. Schools have spent two years chasing detection tools and rethinking how they test at all. Students feel the strain too. Brown's own provost-led report found that most undergraduates use generative AI weekly or daily. Yet large majorities also worry about the effect on their learning, and fear what it does to their "cognitive capacity". That worry sits alongside a broader shift, as AI reshapes who gets hired and even how people think and write. Serrano frames it in the starkest terms. Why it matters "We cannot afford to have a society in which a significant fraction of our best young minds think that cheating is okay," Serrano told Inside Higher Ed. "That leads to a declining society, to a failed society. We cannot choose to become idiots." His experiment is small, one class, one term. But it turns a fuzzy worry into a hard number. Take the AI away, and half the apparent knowledge goes with it. That is the figure universities now have to sit with.

An Ivy League professor suspected AI cheating, so he decided to fight back

Roberto Serrano was shocked when he saw the results of the midterm exam in his advanced mathematical economics class at Brown University last semester: The average score was 96, when in the past, it had ranged from the 60s to the 80s. Nearly half of the students this year got a perfect score of 100. When he and his teaching assistants ran the exam through a large language model, ChatGPT, it gave an odd, convoluted process for solving one problem rather than a straightforward direct proof. So did numerous students. He had decided, after the deadly classroom shooting at the school in December, to allow take-home exams for the first time. And he had just learned that those can no longer be trusted to measure student learning, even at an Ivy League school. "I think it's basically impossible," Serrano said, "to come up with an alternative explanation beyond massive cheating to explain the data." The results in Serrano's class show that higher education is still struggling with a really basic question: As students get better at using artificial intelligence, how can faculty ensure that students are learning? When ChatGPT was released in 2022, some educators worried that students would use it to cheat, but others said the large language models (LLMs) couldn't write a compelling essay or solve difficult math problems. Now they can. (The Washington Post has a content partnership with OpenAI, the creator of ChatGPT.) And new devices are complicating efforts to test students. There are innocent-looking calculators with cameras and generative AI. ("Snap. Solve. Done." is one device's slogan.) Eyeglasses embedded with cameras and AI can provide answers to tests without professors having any idea it's happening. "The reality is that students are going to be using AI, whether you know about it or not," said Laurent Lessard, an associate professor of mechanical and industrial engineering at Northeastern University. Lessard, who served on an AI assessment task force -- something many universities have -- said the technology is moving so fast that professors have to adapt quickly or they won't have any understanding of its capabilities and how students are using it. He said AI could successfully complete a semester-long college course in about two hours. In many ways, universities are embracing AI, with researchers using it to supercharge their work in some fields, faculty designing tutoring apps to help students grapple with tough problems, and school officials adding courses and degrees related to AI to prepare students for the fast-changing workforce. But the problem remains: For students who know they could get a better grade if they used an LLM and know they'll be competing for jobs and admissions against classmates who are using it, the incentive to turn to such a quick and easy tool could be powerful, Lessard said; that's why so many faculty members are trying to rethink how they teach, how they test and what they grade. Some schools are resurrecting or considering, old-school practices such as blue books and oral exams. Stanford and Princeton universities began allowing and, in Princeton's case requiring, proctored exams this year despite their honor codes. The University of Chicago's law school is piloting a ban on devices in its first-year core classrooms. And some professors have stopped assigning, or grading, homework; emphasized the learning process over the final product; and added more time for in-person discussions. A majority of teens, 59 percent, think using AI to cheat is a regular occurrence at their school, according to a February report from the Pew Research Center. Experts say there is widespread use of AI among college students. At Brown, a committee that has been studying AI use for much of this academic year released a report last week that recommended several changes, including examining academic codes to ensure that academic integrity remains a core standard. The university is looking into what happened in Serrano's Welfare Economics and Social Choice Theory class. "Brown treats every allegation of academic integrity with the utmost seriousness," said Brian Clark, a spokesman for the university. Serrano, a professor of economics, knows that learning can be a struggle: At 17, while growing up in Madrid, he began to go blind. He considered not continuing his education. Instead, he learned to use a cane and read Braille. In college, his father would help him reconstruct the notes he had scrawled in class so that Serrano could transcribe them into Braille. After earning his doctorate from Harvard University, he began teaching at Brown. Many people face far more difficult circumstances, he said, but "it certainly gave me the clear idea that learning and succeeding doesn't come without effort." Serrano was surprised when 86 students signed up for his course this year, about triple the typical enrollment. Now he wonders if that was because the syllabus made clear the midterm and final would be take-home exams. After the midterm, Serrano told his class that it appeared there had been widespread cheating, despite students having signed an academic integrity pledge when taking the test. The response was silence, he said. He also told them he was giving them a chance to prove him wrong. After he switched the final exam to an in-person, three-hour test, 27 students dropped the class, he said. Twenty-two of them had gotten a perfect score on the midterm. The average score on the final was a 48.6. Given the results, he told the class he was voiding the midterm. And he asked them, if they chose to use AI on an exam: "Why are you here? Why are you attending a university?" The happy news, Serrano said, was that the course still had a very strong group of students who showed a good command of the challenging material in the final exam. The highest grade on the final was a 95. That student also got a 95 on the midterm. After he told his story to El País, a Spanish newspaper, he heard from hundreds of people, at all kinds of schools, most of them confirming his conviction that concern about AI in the classroom is widespread. There has always been cheating in academia, and new technologies have often brought new challenges. But AI in its various forms came so suddenly -- like a tsunami, he said -- it caught educators unprepared to not only prevent cheating, but also to assess whether students are actually learning. Now they're responding, he said. "With enough people thinking, we should all figure out better solutions."