A U.S. prof suspects most of his students used AI to cheat. It's a growing concern here, too

It's a startling caseful sparking click in donnish circles.

After a lifelessly on-campus shot last December, a professor switched to take-home exams for the first time, hoping to alleviate stress for his students.

Roberto Serrano's advanced undergraduate economics course's typically small enrollment skyrocketed and students scored an average of 96 per cent on his mid-term — higher than any previous cohort, with nearly half earning a perfect score. Many submitted answers remarkably similar to what his marking assistants found after running the exam through ChatGPT.

A suspicious Serrano flipped the final back to an in-person format and vowed to void the mid-term if those results prove inconsistent with the final. Enrollment dropped, with the remaining students scoring an average of 48 per cent on that last exam.

Cheating is a perennial struggle for universities and colleges, yet amid rising student usage of generative AI, stories like the one out of Brown University are becoming more common — including in Canada, according to educators and post-secondary staffers.

Work is needed from students, instructors and institutions alike, they say, to spell out what academic integrity means and emphasize its value in this moment of technological disruption.

When gen AI and specifically ChatGPT burst onto the scene in 2022, there was a steep rise in referrals of students involved in academic misconduct to UBC Okanagan's Academic Integrity Matters (AIM) program.

The program's work includes teaching students school policies and the importance of academic integrity, explained Rina Garcia Chua, who manages the program.

Depending on the circumstances of an "academic integrity violation case or process, we are there as an intervention," she said from Kelowna, B.C.

AI cheating runs wild on campus

"Most are students who are confused. There's miscommunication. They're overwhelmed," she said, noting that most tend to be first-year students.

After students complete a course, since updated to incorporate generative AI, "we have a one-on-one meeting with them ... And process all of these learnings together."

Students who cheat are generally motivated by pressure, rationalization and opportunity, says Rahul Kumar, an associate professor of education at Brock University in St. Catharines, Ont., who researches academic integrity and AI.

The rising student usage of gen AI is making that path easier to take than before, he said. Stressed over a difficult class or heavily weighted exam, for instance, a student may justify an AI shortcut having seen classmates already using it.

Is using AI in university cheating? We asked students

Allyson Miller, director of the Academic Integrity Office at Toronto Metropolitan University, is even more concerned by a recent increase in repeat offenders this past school year.

"Either a student is not learning from the first instance or they're getting away with it so frequently that it's worth playing the odds," she said.

Miller's team at TMU is prepping a tougher message to come next term, underlining the stronger penalties and consequences facing those caught making repeated violations.

"It's really spelling out, 'This is what you're risking. This is your future that you're taking chances with,'" Miller said.

"Students are not looking forward and understanding the implications of these decisions — it's not just a zero on an assignment.... It's suspension, which leads to people losing student visas, people losing opportunities to stay in the country, people losing job offers."

At UBCO, the number of student referrals to AIM have recently started to fall, says Garcia Chua.

She credits the team's partnering with faculties to build more tailored academic integrity resources — weaving in specific references, say, for first-year nursing, biology or chemistry students — and emphasizing the need for instructors to spell out if or when AI tools may be used, and being clear about their own usage.

Garcia Chua thinks students' engagement with and perception of AI are evolving, pointing to a recent survey of student respondents across both UBC campuses.

"They understand the implications... And they're actually asking for more lessons outside of gen AI," she said.

Brock University's Kumar believes instructors should be teaching students responsible and ethical AI use, while also underlining the value of "hard, tedious work, which doesn't pay off immediately, but will pay off in the long run."

Most importantly, educators must design secure assessments to accurately measure what students actually know, he says, versus what they can easily find by using gen AI.

Some instructors are adapting already, he noted, turning to methods like pen-and-paper assignments, in-person testing, oral exams and defences, or opting for mini-assessments throughout a course.

Canadian universities grapple with evaluating students amid AI cheating fears

But institutions must adequately support this work, he added, since these methods are often more time- and labour-intensive — all while instructors are juggling job cuts, fewer teaching assistants, ballooning class sizes and may also be teaching both in-person and online formats.

Institutions must "provide resources, build capacities, make systemic changes," he said. "Failing that, the reputation of higher education will take a big hit."

Robin Whitaker, president of the Canadian Association of University Teachers, laments what she calls a "customer service model of thinking" about post-secondary, where cash-strapped schools treat students as "a revenue stream" and prioritize particular outcomes, like top grades or scoring specific jobs upon graduation.

If students are treated as such — as they grapple with rising tuition, cost of living pressures, work and school, Whitaker says — it can create a context where taking AI shortcuts don't seem so bad.

She believes, however, that giving time and space to the student-teacher connection is key to conversations about academic integrity.

"Fundamentally, human learning thrives when there is that connection," Whitaker, an associate professor of anthropology at Memorial University, said from St. John's.

Miller, from TMU, also decries a transactional view of post-secondary as a place students pay tuition for a piece of paper to get a job. She thinks universities have a responsibility to help students grow as human beings.

She points out that people tapping AI to bypass foundational learning aren't what employers are looking for, anyway.

"Employers want people who can use AI in a way that is not going to destroy their company," she said. "To leverage these tools to make them better professionals, not as crutches to get them through being able to do the job."

Senior Digital Writer

Add some “good” to your morning and evening.

---

---