Brown University publishes its first generative AI in teaching report amid an economics exam-integrity case
Brown University published its first generative AI in teaching report on July 7, 2026, as an economics class average fell from 96% to 48.6%.
Photo by Google DeepMind on Pexels
Brown University’s Generative AI in Teaching and Learning committee published its first report on July 7, 2026, setting out how the Providence, Rhode Island university should handle generative AI in teaching. The report draws on feedback from 105 faculty members, three-quarters of whom said they are concerned about students using AI to cheat. It landed in the same week that “Inside Higher Ed” reported a documented case in a Brown economics course, where a take-home midterm average of 96 percent fell to 48.6 percent once the final exam was moved in person.
The report and the classroom case frame one problem from two sides: institutional policy on one, and week-to-week assessment on the other. Together they show a university trying to write rules for a technology that its own faculty cannot reliably detect.
What Brown’s generative AI in teaching report recommends
The committee examined how generative AI is used across the university and set out formal recommendations for how Brown should adapt. Three-quarters of the 105 faculty members who gave feedback said they were concerned about students using AI to cheat. The same share reported the same concern in a 2025 nationwide survey by the American Association of Colleges and Universities.
Among its medium-term recommendations, the committee urged the university to amend the College Academic Code and the graduate student code to, in its words, “address GenAI realities and safeguard against misuse.” The report also asks faculty to de-emphasise punishment and to avoid highly restrictive rules, noting that “there is no way to check with 100 percent accuracy whether GenAI has been employed.” That tension, between enforcing clear limits and accepting that detection is unreliable, runs through the document.
The take-home exam that exposed the gap
The classroom case sits underneath that policy debate. Roberto Serrano, who has taught Welfare Economics and Social Choice Theory at Brown for nearly two decades, set a take-home midterm this spring for the first time. Enrolment had risen to 86 students, up from the 30 the course usually drew, an increase he attributes to the promised take-home format.
The midterm average came back at 96 percent. Historically, Serrano told “Inside Higher Ed”, the midterm average had ranged between 65 and 80 percent, and this exam was harder than in past years. He and his graders ran the paper through ChatGPT; the model returned answers that mirrored what many students had written, including a “contradiction argument” on a proof that a person would most obviously solve with a direct one.
With his dean’s approval, Serrano moved the final exam in person. After he told students he suspected AI use, 18 dropped the course and nine stayed enrolled but did not sit the final. The in-person final averaged 48.6 percent, which he described as a historic low; the average had never previously fallen below 65 percent. Three students scored zero. Serrano voided the midterm, reweighted the final to 80 percent of the grade, and lowered the pass line to 40 percent from 50. Nineteen students failed the course.
| Assessment (Welfare Economics, Brown) | Format | Class average |
|---|---|---|
| Midterm, spring 2026 | Take-home | 96% |
| Typical midterm, prior years | In-class | 65-80% |
| Final, spring 2026 | In-person | 48.6% |
| Lowest final average, prior years | In-person | 65% |
Source: course data submitted by Roberto Serrano, reported by “Inside Higher Ed”, 8 July 2026.
The course outcomes trace the same collapse in participation.
| Course outcome, spring 2026 | Students |
|---|---|
| Enrolled | 86 |
| Dropped after the midterm warning | 18 |
| Remained but did not sit the final | 9 |
| Scored zero on the final | 3 |
| Failed the course | 19 |
Source: “Inside Higher Ed”, 8 July 2026, from data submitted by Roberto Serrano.
What the case signals for AI and academic integrity
Serrano said Brown’s administrative response had been “meek.” After he went public in late June, the Standing Committee on the Academic Code asked him, through his department chair, to file individual complaints against each suspected student, with copies of their exams. A Brown spokesperson, Brian Clark, said the procedure for investigating allegations is the same whether one student or several are involved, and that the university had told the professor how the concerns “could be formally adjudicated.”
The workload of pursuing mass cases is itself a barrier, said Tricia Bertram Gallant, who directs the Academic Integrity Office at the University of California, San Diego. Faculty “don’t get rewarded for dealing with a 60-person case of cheating,” she told “Inside Higher Ed,” adding that their time is often not compensated. Her point echoes the committee’s caution against relying on detection and punishment.
The episode also mirrors a wider evidence base on assessment design. Winss Solutions has reported on how the OECD warns AI can raise grades but hurt learning when students lean on it during practice, and on how to implement AI in education in a responsible way. The Brown report points in the same direction: clearer codes, but paired with assessment that does not depend on catching the tool after the fact.
About the Brown committee and its response
Brown University, founded in 1764 and based in Providence, Rhode Island, convened its Generative AI in Teaching and Learning committee to review how the technology is used across the campus and to recommend a response. The July 7 document is the committee’s inaugural report. It surveyed 105 faculty, recommends amending the academic codes, and frames misuse as a risk to student learning and personal growth rather than only a disciplinary matter.
Brown’s handling of the economics case will test how those recommendations work in practice. The committee argues that “bright lines” around misuse should be enforced while accepting that no detector is fully reliable, a balance the university will have to strike as it decides what to do with a voided midterm and a course that failed a quarter of its students.
Sources: Brown University Office of the Provost; Inside Higher Ed
Featured image: photo by Google DeepMind on Pexels (free Pexels license).
Become a Sponsor
Our website is the heart of the mission of WINSS – it’s where we share updates, publish research, highlight community impact, and connect with supporters around the world. To keep this essential platform running, updated, and accessible, we rely on the generosity of you, who believe in our work.
We offer the option to sponsor monthly, or just once choosing the amount of your choice. If you run a company, please contact us via info@winssolutions.org.
I specialize in sustainability education, curriculum co-creation, and early-stage project strategy. At WINSS, I craft articles on sustainability, transformative AI, and related topics. When I’m not writing, you’ll find me chasing the perfect sushi roll, exploring cities around the globe, or unwinding with my dog Puffy — the world’s most loyal sidekick.
