Chatbot TAs, coding on the fly: Here's how these educators weave AI into their classrooms

Facing the reality that a majority of Canadian students are using generative AI for schoolwork, more educators are bringing artificial intelligence into their university classrooms, setting clear rules and encouraging students to use it responsibly — and with a critical eye.

That's forcing instructors to rethink how they teach and assess students since — outside of concerns about academic integrity — institutions tend to leave decisions about AI use to individual faculty.

These university professors explain how they're weaving AI into their courses and how they're guiding students to learn what they’re expected to.

Antonello Callimaci prioritizes answering student queries promptly — dedicating four blocks of time daily to do so. But when the Université du Québec à Montréal accounting professor is unavailable (or students are hesitant to reach out directly), Bobby's got his back.

Bobby is an AI agent "teaching assistant" that Callimaci built last year by training ChatGPT on the hundreds of assignments, presentations, notes and recorded lectures he's prepared over the years for one of his courses. Accessible 24/7, Bobby's responses come straight from Callimaci's content and also point students back to him for further clarification.

"He's able to summarize material. He's able to build sample exams. He is able to answer specific questions," Callimaci said.

The agent can handle sophisticated requests, too, he said.

For instance, a student who used Bobby all last semester requested it review past interactions to see what had given him the most trouble. The results pointed him to areas needing extra study before the final exam — which Callimaci has kept old-school: paper and pencil, no tools or devices allowed.

"It's a learning tool, but you cannot count on Bobby to do your work," he said.

Political science professor Joseph Wong has long used weekly readings to push first-year students to grapple with new ideas and diverse perspectives for his seminar in Munk One, a small-group undergraduate program at the University of Toronto that explores global public policymaking.

Since generative AI hit the public sphere, however, he's reimagined how to keep his students engaged in that same productive mental struggle.

For instance, his final assignment used to be a magazine-length feature article proposing a solution to a global challenge. Now students produce a three-minute TikTok video, accompanied by a written reflection about making it.

In a similar vein, the traditional reaction papers he previously assigned alongside weekly readings are out; they've now become "reaction dialogues" with AI. After finishing the readings, students submit their back-and-forth discussion with an AI agent: a chat where they've ideally delved deeper into what stood out for them.

It lets students practise effective AI prompting, Wong says, but like the earlier reaction paper, these interactions can also highlight how deeply students engaged with the material — or even their frustration if the chatbot conversation goes awry.