Artificial intelligence has become deeply embedded in college coursework across the country, and new research from Cornell University suggests higher education may be unprepared for how quickly the technology is reshaping learning — and cheating.
A new study led by Cornell researchers found that more than one-third of college students regularly use generative AI tools like OpenAI’s ChatGPT to complete assignments, while roughly 9% admitted to using the technology to cheat.
The findings come from survey responses collected from more than 95,000 students at 20 public research universities nationwide during the 2023-24 academic year, making it one of the largest studies yet examining how students use AI in higher education.
Researchers said the results point to an urgent need for colleges and universities to rethink how students are evaluated.
“Assessment reform is necessary and urgent,” said Rene Kizilcec, associate professor of information science at Cornell and director of the Future of Learning Lab. “The fact that students are misusing GenAI is a problem for assessment validity, and that’s a problem for the credibility of university credentials.”
The study, published Thursday in the journal Science, found AI use varies widely by academic discipline. About 62% of computer science students reported using generative AI regularly, compared to 24% of arts students.
Researchers also identified demographic differences. Male students reported significantly higher AI usage rates than female students, while underrepresented minority students reported lower rates of use compared to white and Asian students. Researchers warned those gaps could create long-term inequities as AI tools become more advanced and expensive.
The study also found that students who used AI daily were substantially more likely to use it for cheating purposes. Among daily users, 26% admitted to cheating with AI tools, compared to 7% among students who used the technology monthly.
To measure cheating rates more accurately, researchers used a survey technique designed to encourage honest responses without requiring students to directly identify themselves as cheating.
The findings arrive as colleges and universities nationwide continue struggling to define acceptable AI use inside classrooms. Some institutions have banned AI-assisted work outright, while others are moving toward integrating the technology into coursework and professional training.
Cornell researchers suggested several possible responses, including a return to more tightly controlled testing environments, clearer classroom policies around AI use and redesigned assessments that incorporate AI tools while still measuring critical thinking and professional skills.
Researchers also warned that universities risk widening educational disparities if students do not have equal access to AI tools or AI literacy training.
“As we expect GenAI use among students to only grow, for better and worse, we also expect that GenAI misuse will grow, which is concerning,” Kizilcec said.
FingerLakes1.com is the region’s leading all-digital news publication. The company was founded in 1998 and has been keeping residents informed for more than two decades. Have a lead? Send it to news@fingerlakes1.com.