Universities Don't Have an AI Problem — They Have an Integrity Problem

In a recent opinion piece in The Globe and Mail, Anna Jahn, executive director of the Centre for Media, Technology and Democracy at McGill University's Max Bell School of Public Policy, made a blunt case to Canadian university leaders: stop treating artificial intelligence as a cheating problem, and start redesigning degrees around it.

Much of her diagnosis lands. Three years into the disruption, the main institutional response really has been enforcement — the “AI-proof” quiz, the plagiarism policy nobody can articulate. Her own children, she writes, are improvising their educations alone: one refuses to touch AI, the other uses it daily and wonders whether they are still learning anything at all. Jahn is right that this is a leadership failure, and right that most Canadian universities are not even trying.

Her prescription borrows from the historian Niall Ferguson: split education into the “cloister” — no technology, printed books, oral exams — and the “starship,” intensive training on the most capable AI systems, in roughly an 80-20 blend. It is an elegant model. It is also an architecture answer to an ethics question, and it skips the part that should worry us most: why are students cheating at all — and when did we stop treating that as the scandal?

Everywhere else, we still call it what it is

Outside the university, Canada has never treated cheating as a design flaw to be routed around. When candidates obtained leaked content for Ontario's bar licensing examinations, the Law Society voided results and treated it as a matter of professional standing — not as feedback on exam design. Medical regulators go further: a landmark New England Journal of Medicine study found that physicians disciplined by state boards were three times as likely to have a record of unprofessional behaviour in medical school. Integrity lapses are not a phase students grow out of. They are the single best predictor of the professional a student becomes.

The line society drew for lawyers, physicians, engineers and pharmacists was always the same one: the credential must mean that this named person carries this knowledge. Not their roommate. Not their browser plugin.

Some knowledge has to be in your head

There is a reason that line exists, and it is not nostalgia. In safety-critical work, knowledge is only useful if it is present in the moment. The worker who opens the wrong valve, the nurse reading a drug label, the supervisor deciding whether an excavation is safe to enter — none of them gets to pause the emergency and prompt a chatbot. Training for hazardous materials, dangerous goods transport and first aid exists precisely because the knowledge must arrive with the person. A credential earned by someone — or something — else is not a shortcut. It is a counterfeit, and eventually someone pays for it.

The students already know

Here is the detail the redesign debate keeps missing. In KPMG's 2025 survey of Canadian students, 73 per cent now use generative AI for schoolwork — and six in ten say they feel like they are cheating when they do. Two-thirds say they are not learning as much. An MIT Media Lab study of essay writing called the result “cognitive debt”. Students' consciences are working fine. What is missing is an institution willing to stand behind that conscience with a clear ethos: this is the line, here is why it matters, and here is what happens when you cross it.

Other societies build that ethos deliberately. Japan teaches dotoku — moral education as a formal school subject — from the first year of elementary school, before technical instruction hardens into habit. Aristotle made the same point twenty-three centuries ago: we become just by doing just acts. Integrity is not a personality trait. It is a practice, and practices can be taught, required and enforced.

What this looked like in practice

A composite drawn from real patterns we see in workplace training. A Canadian industrial employer audited its safety training records and found what universities are now finding: certificates technically “earned,” with no way to prove who actually did the learning — logins shared, quizzes completed by AI in a background tab. The fix was not banning technology. It was restoring integrity to the credential: real assessment, identity verified at the moment it counts, consequences applied. Full disclosure: this is the problem Gardril was built to solve, so we have a stake in this argument — it is also why we know the mechanics. Our guide to buying cheat-resistant training covers what that looks like in detail.

The sharper version of Jahn's call

So, a respectful amendment to Jahn's challenge to university presidents. Pilot the cloister and the starship — but the defining question of your tenure is not the ratio between them. It is whether your degree still certifies that the named graduate knows what it claims. AI did not create the cheating crisis; it industrialized one we had stopped policing. Redesign the degrees. But teach integrity as an ethos, verify the person, and restore the consequence — the way every licensed profession in Canada already does.