In Canada, 73 per cent of students rely on generative AI for schoolwork, up from 52 per cent just two years ago.JEFF PACHOUD/AFP/Getty Images
Anna Jahn is executive director of the Centre for Media, Technology and Democracy at McGill University’s Max Bell School of Public Policy.
My two children are undergraduates, one at McGill, one at the University of Toronto. They have made opposite bets on AI. One refuses to touch it, convinced that real learning only happens without the machine. The other, who uses it daily and carries a quiet mix of shame and worry, wonders: Am I still learning anything at all? Neither has received meaningful guidance from their university. They are improvising their education alone.
Every student now faces this choice, and most have already made it. In Britain, 95 per cent of undergraduates now use generative AI, and 94 per cent use it for assessed work. In Canada, 73 per cent of students rely on it for schoolwork, up from 52 per cent just two years ago. This is not a wave that is coming. It has arrived, and it is nearly total.
Opinion: AI and the death of the university essay
Yet most universities still treat AI as a cheating problem. I hear this at departmental meetings where we debate how to prohibit AI in our own courses, and when colleagues announce they have finally designed the “AI-proof” quiz. Three years into the biggest disruption higher education has faced in a century, our main response has been enforcement. Even the boldest institutional moves stop at procurement. Oxford made headlines last fall as the first British university to give ChatGPT to every student and staff member. That is a subscription and a training program. It is not a rethink of what a degree is for.
Students understand the stakes better than their institutions do. Nearly half of Canadian students say their critical thinking has deteriorated since they started using AI. The research supports their instinct. An MIT Media Lab study measured brain activity in students writing essays; those using AI showed the weakest neural connectivity, and many could not quote from their own essay minutes after finishing it. The researchers called it “cognitive debt.” Students feel that debt accumulating. They are asking for help carrying it, and their universities are responding with poorly articulated plagiarism policies.
The historian Niall Ferguson believes two educational models will dominate, which he calls “the cloister and the starship.” In the cloister, there’s no technology at all: small seminars of 15 students, printed books, handwritten and oral exams, and the slow, effortful development of the capacities no machine can supply – critical thinking, argument, judgment. In the starship, the opposite: intensive, constantly updated training in mastering the most capable AI systems, the ones graduates will use every day of their working lives.
What is key is that the two modes cannot be blended, because then you get what we have now: AI quietly doing the thinking in courses designed for humans. Instead, a deliberate separation of cloister and starship means we can protect deep learning while taking new AI tools seriously enough to teach them properly. My guess is that the right balance is roughly 80-per-cent cloister, 20-per cent starship, though each discipline will have to find its own mix. And it’s important not to become overly nostalgic, as the mass lecture model has been failing students long before ChatGPT arrived.
Choosing a direction shouldn’t be left to individual professors tinkering on their own, which is what is happening now. Achieving the right balance requires redesigning entire degrees, assessments, timetables and hiring – decisions only presidents, provosts and boards can make. So far, I see little evidence that Canadian university leadership grasps the scale of the task.
Opinion: The value of human writing in an age of AI
Here is what worries me most. Mr. Ferguson is building his mixed model at the University of Austin, a private institution for a select few. Elite universities have the money for small seminars, oral examinations and cutting-edge AI training. Publicly funded universities, which educate the vast majority of Canadian students, do not, and most are not even trying. The danger is not that universities change too fast. It is that only the rich ones change at all. That would make deep thinking a luxury good and leave a generation at most public institutions to negotiate the most consequential technology of their lifetime alone.
So here is my call to university leaders: Treat this as the defining question of your tenure. Pick two programs and pilot the split model this September. Put a redesign plan before your board within the year and tell provincial governments what it costs. Above all, stop policing and start leading. Students are waiting to hear what their degree is for in the age of AI, and the future of higher education may depend on us getting it right.