Chatbot TAs, instant coding: Here’s how these teachers weave AI into their classrooms

Faced with the reality that the majority Canadian students are using generative AI for school workMore Teachers are bringing artificial intelligence into their university classrooms, setting clear rules and encouraging students to use it responsibly and with a critical eye.

It’s forcing instructors to rethink how they teach and evaluate students – from the outside. Concerns about academic integrity – Institutions leave decisions about the use of AI to individual faculty.

These university professors explain how they are incorporating AI into their courses and how they are guiding students to learn what is expected of them.

an ai teaching assistant

Antonello Callimasi makes it a priority to respond to students’ questions promptly – devoting four blocks of time every day to do so. But when a Université du Québec à Montréal accounting professor is unavailable (or students are hesitant to reach out directly), Bobby joins him.

Bobby is an AI agent “teaching assistant” that Callimasi created last year for one of his courses by training ChatGPT on hundreds of assignments, presentations, notes and recorded lectures he had prepared over the years. Accessible 24/7, Bobby’s responses come directly from Callimasi’s content and refer students back to him for further clarification.

“He’s able to summarize the material. He’s able to create sample exams. He’s able to answer specific questions,” Callimasi said.

The agent can also handle sophisticated requests, he said.

For example, a student who used Bobby last semester requested to review past conversations to find out what had caused him the most trouble. The results pointed him to areas needing additional study before the final exam—which Callimasi has kept old school: paper and pencil, no instruments or equipment allowed.

He said, “It’s a learning tool, but you can’t rely on Bobby to do his job.”

‘Feedback Dialogue’ with AI

Political science professor Joseph Wong has long used weekly readings to expose first-year students to new ideas and diverse perspectives for his seminars in Munk One, a small-group graduate program at the University of Toronto that explores global public policy making.

However, ever since generative AI has entered the public sphere, it has reimagined how to engage its students in the same productive mental struggle.

For example, their last assignment used to be a magazine-long feature article proposing a solution to a global challenge. Students now create a three-minute TikTok video, along with a written reflection about making it.

Similarly, the traditional response letters that they previously handed out with the weekly readings are out; They have now become “feedback conversations” with AI. After finishing reading, students submit their back-and-forth discussion with the AI ​​agent: a chat where they ideally delve deeper into what made them special to them.

This lets students practice effective AI prompting, Wong says, but like the earlier response letter, these interactions can also highlight how deeply students are engaging with the content — or even their frustration if the chatbot conversation goes awry.

“I’m assessing to what extent they are now wrestling with not only reading, but also AI with their conversation partner,” Wong said.

Nevertheless, he keeps track of students’ analytical and critical progress.

“As teachers, we have to constantly remind them of the value of learning the content and the value of learning those skills.”

Connecting students to AI

Sidney Shapiro has been using AI and machine learning for a decade, but the tools available today are helping him quickly adapt lessons to make them more engaging and relevant to students.

Five years ago, a programming course basically required students to “watch me type (code) for two and a half hours,” recalled Shapiro, an assistant professor of business analytics who teaches in both the business school and the computer science department at the University of Lethbridge.

Recently, in the same course, they used generative AI to develop code based on students’ improvised suggestions—for a medical clinic for superheroes, for cows wearing colorful sweaters—and were able to zip through a series of “hilarious” exercises for creative, memorable learning.

In another example, when he used ChatGPT to translate that day’s slide deck into a “squidby toilet, Ohio-type conversation,” he turned the tables on a disengaged student, Shapiro said, whom he then presented.

“This student told me (it) was the hardest class he had ever taken, and it was so embarrassing — but he remembered everything, so it worked out perfectly.”

Each semester, Shapiro is tinkering with the use of automation and AI (as allowed by his institution’s policies), but he’s sticking to some basics, including emphasizing foundational skills like reading and thinking critically, learning to write concisely, and being transparent about AI use.

He said, “I don’t think students need to be experts in everything anymore because there are AI assistants that do a lot of things, but they still need critical thinking…so they can recognize when there’s a trap, when there’s something that doesn’t make sense and how to fix it.”

If teachers pretend AI doesn’t exist, “the reality is that a lot of people will use it anyway, but they will feel guilty about it and won’t learn how to use it properly.”

evaluation, reconsideration of objectives

Maggie McDonnell uses AI in every course she teaches. For example, students in their graduate, professional writing program at Concordia University start by researching the benefits and drawbacks of AI use in different industries, and also work with them to set AI policy for all of their operations.

Meanwhile, at the Université de Sherbrooke’s Master Teacher Program, he and his colleagues have rattled off an annotated bibliography assignment for an upcoming course—”one of those things that AI can do in a minute and do really well.” Instead, they have chosen another way for students to demonstrate their research, synthesis, and communication skills (and if they use AI for this, they have to disclose it).

McDonnell says some students may still use AI as a shortcut, especially in general or required courses they don’t see value in. But in the career-oriented classes she teaches, she finds that students see the relevance of engaging with AI effectively and ethically. They are also fine when she bans AI for certain tasks.

Still, incorporating this new technology means McDonnell has to regularly rethink how she’s assessing students, rethink learning objectives and “come up with new things all the time” – recalling the challenge and excitement she felt when she started teaching 25 years ago.

“There will always be ways for people to get around what you’re trying to ask them to do, so do we want to be enforcers and police officers or do we want to find other ways to engage (students) so we can get the same learning?” He said.

“Part of the challenge for us as teachers is to take a step back and say ‘Is what I’m asking them to do … that’s important or how they get (there) that’s important?'”