Award winning economist and Brown University professor Roberto Serrano says he has detected what appears to be the largest AI cheating scandal in Ivy League history.
As Spanish newspaper El País reports, Serrano noticed red flags as soon as he looked at the scores of a March midterm exam for one of the classes he teaches, an advanced undergrad course in mathematical economics.
The take-home and closed-book exam — an “Honor Code” type of test Ivy League schools are known for — resulted in 40 out of 86 students scoring a perfect 100. The average score was an equally questionable 96 out of 100.
In other words, it’s not a stretch to assume students gave in to the temptation to ask an AI chatbot for answers, particularly in the confines of their own homes without a teaching assistant looking over their shoulder. In Serrano’s testing, that appeared to be the case.
“Some answers contained unusual passages that coincided with results obtained after running the questions through ChatGPT,” Serrano told El País.
Perhaps most tellingly, the average score of an in-person final, which accounted for half of the final grade of the class, was an abysmal 48 out of 100. Of the 27 students who didn’t even bother to show up for the test, 22 had scored a 100 during the midterm exam, providing plenty of credence to Serrano’s theory.
“The empirical evidence of fraud is overwhelming,” he told the paper.
The incident highlights just how pervasive the use of AI has become in the classroom. Even students at highly reputable Ivy League schools are resorting to the tools to cheaply score high grades — even when doing so directly contradicts an honor code they all swore to uphold.
Compounding the concerning development, literacy and numeracy rates have taken a major hit over the last couple of years. College professors warn that we’re hitting a crisis point as incoming students barely have a middle-school level understanding of math and other subjects.
Some professors are lamenting that they’ve quickly become “plagiarism cops,” whose main job it is to root out AI-facilitated cheating instead of actually teaching. It’s a cat-and-mouse game greatly complicated by rapidly improving tech that’s making cheating harder to spot.
At the same time, experts warn that the use of the tools is destroying their students’ ability to think critically as they become hopelessly dependent on the tech.
It shouldn’t come as a surprise that Serrano has decided to stop giving take-home exams altogether.
A similar story is playing out at other Ivy League schools. As The Atlantic reported last month, Princeton recently stopped a 133-year-old “Honor Code” tradition involving professors leaving the room when students, who sign a pledge not to cheat, take their final exams.
Thanks to the dramatic uptick of AI use and academic dishonesty in the classroom, Princeton did away with the tradition altogether.
“There’s an air of people cheating on take-homes and people just using ChatGPT,” Princeton senior and former chair of the Honor Committee Nadia Makuc told The Atlantic. “As long as people think there is more cheating, it encourages more cheating.”
Beyond the loss of integrity, AI cheating is eroding the trust between students and educators, while the latter worry about the deteriorating value of a college degree.
“If we no longer defend truth and decency and honesty, then what kind of credibility are we going to have as academics?” Serrano told El País.
More on AI cheating: Princeton in Shambles Over AI Cheating