Canada’s post-secondary institutions are looking for new ways to assess students as they respond to fears about AI being used to cheat on exams.
During the COVID-19 pandemic, most university exams were moved online. Then came generative AI tools like ChatGPT, capable of producing essays and answering complex questions in seconds.
In the U.S., reports of rampant AI cheating led to an explosion in sales of “blue books” used for old-fashioned pen-and-paper exams this school year.
In Canada, some professors are making a similar move amid widespread reports of AI cheating, while others are testing out oral exams or finding ways to incorporate AI. Six in 10 Canadian students said they use generative AI for their schoolwork, according to an October 2024 study from KPMG in Canada.
“We are definitely in a moment of transition with a lot of our assessments,” said Karsten Mundel, co-chair of the University of Alberta’s AI Steering Committee.
Mundel speaks with his students about his expectations around AI. If they use it for brainstorming, he asks them to explain their process and the prompts they used so he can see how they led to the final product.
He takes an optimistic view of this new challenge, saying AI has reinvigorated conversations about what academic integrity means in the current day.
“I get worried when AI in any educational context gets boiled down to this tool of cheating,” he said.
“I think it’s an exciting time right now because of the transformations that it will bring, and to really help us get at the core of what skills we’re trying to teach.”
At his school, Mundel says, there’s an increase in handwritten exams, as well as new approaches that incorporate oral exams and assignments that use AI and then have students reflect on their AI use.
He says going back to pen and paper isn’t necessarily the best solution, and acknowledges some students have complained about the change.
“We don’t have the skills anymore — universally, at least — to hand-write long-form things. And so that’s a learning curve for our students, and for the instructors who have to read.”
Many post-secondary students today have grown up working primarily on electronic devices and don’t have as much experience with writing by hand in university. For example in Ontario, learning cursive in elementary school was made optional in 2006, though the provincial government made it mandatory again in recent years.
Katie Tamsett, vice-president, academic, of the U of A’s student union, says concerns of cheating using AI have to be balanced with the fact that the technology is being used in the real world.
“As students, we’re seeing that in the workforce, AI is being used. And so when we’re doing our courses in university, we want to be seeing that AI is being incorporated as a tool.”
Tamsett says the student union is in ongoing conversations with the university about how to develop best practices around AI.
University of Toronto Students’ Union president Melani Vevecka says her experience with pen-and-paper exams has been largely positive, but says they can be a barrier for students with anxiety or learning disabilities.
“Part of the challenge to accommodate everybody is figuring out what kind of assessments will hold value in a world where students can probably generate a decent essay within a few minutes,” she said.
Vevecka understands the pitfalls of relying on AI, and says she knows some students have used it to cheat.
But she also says it’s been helpful in her studies, like, for example, generating practice questions ahead of a final exam.
She feels universities’ responses to it have in some cases been “overly reactionary.”
What Vevecka would like to see is more of a focus on clarity and education around AI, “rather than vague restrictions or punitive suspicion, which is kind of something that most academics are trying to do.”
“I think that universities should be creating academic cultures where students are empowered to think critically about the tools that they use, and where trust is preserved through transparency and not just surveillance.”
Universities grapple with making AI a teaching vs. cheating tool
Jennifer Figner, provost and vice-president, academic, at the British Columbia Institute of Technology, says the move to in-person exams is a trend, but one she views as being “fear-based” — and a route her school is encouraging professors not to take.
“What really we should be doing is challenging ourselves to figure out, how do you incorporate AI into testing or into assessment, rather than trying to work around it by going back to pencils and paper and stuff that we did in 1970?” she said.
On the other hand, Figner says, the pandemic coinciding with generative AI created an environment where cheating became so easy that not doing it could put students at a disadvantage.
Software that detects AI cheating is imperfect, so she also worries about students being wrongfully penalized.
And oral exams can be “far more labour-intensive and time-consuming” than having all students take an exam at once.
Figner says AI is ultimately going to force the entire education sector to “totally revamp” the way students are assessed and evaluated.
Christina Hendricks, academic director at the University of British Columbia’s Centre for Teaching, Learning and Technology, does handwritten exams for finals in her philosophy classes.
But some UBC professors are sticking to computers, doing in-class exams with supervision to deter cheating. Some are done in a lab where the only thing students can access is the exam, and the rest of the computer is locked.
In some disciplines, she’s heard of instructors assigning infographics, slides or videos to get around AI — but now all those things are also easily done with AI tools.
Her centre helps instructors take small steps to change their assessment setups over time.
In the long term, Hendricks agrees that universities will have to completely overhaul their assessment strategies.
“I think that there’s going to be these reflective, existential questions for some faculty,” she said.
“What are we teaching? What do students need to know in their future lives, as people in an AI world and for their careers, and how do we adjust our activities and assessments to match that?”
