Toronto students have spent a year asking to pull AI from a mandatory course, but the university has barely budged

By The Eyeopener · October 6, 2026.

At Toronto Metropolitan University (TMU), a group of first-year Creative School students wrote a letter a year ago to dean Natalie Álvarez complaining about the use of artificial intelligence in CSE100: Wicked Problems and Creative Intervention, a required course for several of the school's degree programs. Twelve months later, according to The Eyeopener, the university's student newspaper, the signatories are asking whether anything has really changed.

The letter, co-written by photography student Zoë Wolfe and carrying 35 signatures, drew on its own survey of 52 Creative School students: 40% said they felt they were "not learning anything" in the course, and 36.5% said they did not feel valued as students. The anonymous testimonies gathered were harsh: "the course is useless to us, and a slap in the face to artists," wrote one respondent.

CSE100 is the first in a series of courses dubbed "Creative School Experience," designed by professor Owais Lightwala together with journalist and academic Ramona Pringle after several years of development, and launched in the fall of 2025 with the stated ambition of teaching students in creative programs to apply their "superpowers" to the world's big problems. The conflict is not with that idea, but with how AI was implemented inside it. In its first run, the course included a tutor chatbot trained on the syllabus and the assignment prompts, to which students could upload drafts for feedback before submitting. After the complaints, that chatbot disappeared from the fall 2026 syllabus.

What did survive, in both runs, was the use of AI "personas" — simulated characters — in an interview-practice exercise. The Creative School justified the decision on practical, not pedagogical, grounds: with more than 600 students enrolled, sending them out to interview real people would require research ethics approvals that are unworkable at that scale; without the simulation, they say, students would get no interview practice at all.

The institutional response has been uneven. After the letter, the school formed a focus group to gather feedback and went as far as offering Wolfe a paid position working on the course redesign; the dean's office also promised $25 to focus group participants, a payment that, according to Wolfe, never arrived. The school attributed the non-payment to "an internal processing error" and says it began compensating participants with $50 as of September 28. In parallel, the report from TMU's own Leadership Task Force on Generative AI, published the week before this story, cites the CSE courses specifically as an example of integrating AI skills into the curriculum, and warns of the risk of cognitive offloading: students handing over to AI the work of thinking and solving problems for them, something that requires, according to the report, "educational scaffolding" and explicit training in academic integrity. The fall 2026 syllabus added, as a third learning outcome, that students must "critically engage with the ethical and sustainability dimensions" of AI use, but Paige Connolly, a first-year student currently enrolled, says that part of the course has not been taught yet.

This is a small story — a campus newspaper, one specific course, thirty-five signatures — but it describes with precision a pattern that will repeat on hundreds of campuses over the coming years. AI entered CSE100 before the pedagogy to teach it: first came the chatbot and the simulations, and only afterwards, under student pressure, the commitment to explain what the technology is and how to use it with judgment. The university's own report acknowledges as much when it names cognitive offloading as a risk to be prevented, not as something already solved.

The most revealing part is not the anger over the chatbot — which the school pulled at the first serious warning — but what survived the protest: the AI "personas" for interview practice, kept for a purely logistical reason (ethics committee bureaucracy does not scale to 600 students), not a pedagogical one. That is where the underlying pattern shows: many AI decisions inside institutions are born not of educational conviction but of operational constraints, and are then dressed up as innovation. And when the university on top of that promises $25 in compensation and fails to pay it for months, citing an "internal error," the cost is not only financial: it is the credibility of everything else the institution says about its own AI.

Our reading is that the clash at TMU is not a story about whether AI is useful for teaching creativity — it can be, and the students themselves acknowledge that learning to use it with judgment is part of their training — but about sequence: deploying first and explaining later inverts the order that gets technology adopted with confidence instead of resistance. In education, unlike other sectors where AI already shows clear productivity gains, the asset at stake is students' confidence that their effort and their judgment still matter. Universities that invest in designing that trust before switching the tool on — with AI literacy, real consent and promises kept — will build curricula that hold up over the long term; those that use AI as a compliance checkbox to look "competent" in an institutional report will only multiply the skepticism already taking root in a generation of students who, as Connolly puts it, chose a creative program precisely to exercise what badly implemented AI threatens to atrophy.