Suspecting AI cheating, Ivy League prof ordered an in-person final; scores fell 50%
The setup
- Ivy League students don’t need AI to cheat — they could learn the material — but competitiveness, ambition and overscheduling make AI look like an easy shortcut.
- A scandal at Brown University suggests large numbers of these students are likely cheating.
- Context: a recent survey of Princeton students found 29.9% admitted to cheating with AI on at least one exam or assignment. The Brown case shows what that looks like in a single class — and how much it may substitute for actual learning.
- At the center is Roberto Serrano, a blind economics professor (born in Spain), who has told his story to El País and Inside Higher Ed and refuses to let it go.
Origin of the policy change
- The story begins in December 2025, when a gunman attacked Brown’s campus and killed two people — including someone who had recently introduced herself to Serrano.
- Shaken, Serrano decided his spring 2026 section of the difficult ECON 1170 would allow take-home exams for both midterm and final.
The suspicious results
- The course typically attracts few but strong students — never more than 30, sometimes as few as 8. This semester 86 students signed up, likely because of the new evaluation system.
- The March 5 midterm average was 96/100, with 40 students scoring a perfect 100 — extraordinary, given the course’s historical midterm average of 65–80 and the fact that this exam was harder than past ones (take-home gives unlimited time).
- Many answers, even correct ones, had a “very convoluted style.” When Serrano and his grad students ran the questions through ChatGPT, they got similar results.
The in-person test
- Serrano made the final exam in-person, telling students he wasn’t voiding the midterm yet: if the final’s distribution roughly matched the midterm’s, he’d count the midterm; otherwise he’d void it and reweight the final.
- 18 students dropped the course and 9 more didn’t attend the final. Of those 27, 22 had scored a perfect 100 on the midterm.
- Among those who took the in-person final, the average plunged from 96 to 48.
”A failed society?”
- Serrano was horrified by what looked like massive cheating that prevented most students from learning the material. He shows no inclination to coddle elite students.
- His attitude may trace to his own life: he went blind from retinal dystrophy at 17, learned Braille, and reached Harvard. As an economist he frames blindness as “one more restriction… and I optimize based on that.”
- Brown’s institutional context: a provost-led report on “Generative AI in Teaching and Learning” found 56% of undergrad and 67% of grad/medical respondents intentionally use GenAI daily or weekly — yet large majorities also worry about its impact on their learning and fear “negative consequences for their cognitive capacity.”
- Serrano wants universities to stand up for human thought, despite what he calls a tepid reaction from Brown administrators.
- His warning: “We cannot afford to have a society in which a significant fraction of our best young minds think that cheating is okay… That leads to a declining society, to a failed society.” And: “We cannot choose to become idiots.”