Last April, three UNB students presented at the Student Systems-Thinking Symposium — Changemakers of the Future and asked a deceptively simple question: Why are so many students using generative AI? What followed revealed something much deeper than technology alone.

“We never took sides saying AI is good or bad,” says Colton Kammerer, a fourth-year business student and two-time captain of UNB’s men’s hockey team. “It was more about understanding what is motivating students to use it and why it continues to rise.”

Colton worked alongside fifth-year entrepreneurship student Sohan Kobiri and fourth-year mechanical engineering student Aditya Raval in TME 3513: Introduction to Systems Thinking, taught by Dr. Kush Bubbar. Their project, Why Students Use GenAI in Canadian Undergraduate Education: A Problem of Incentives, Not Just Technology, earned first place at the symposium, securing the team a spot at the Map the System Global Competition at Oxford University this July.

Looking beyond the tool

At the outset, the team believed their project would focus on assessment design in the age of
AI. That changed quickly. “We started with a solutions-focused mindset,” Sohan says. “We thought we had it figured out. But as we started exploring, Dr. Bubbar really challenged our view and showed us things that we didn’t really see.”

Systems thinking is a discipline that emphasizes relationships and underlying structures. It prompted the students to step back from surface-level debates about cheating and the tools and policies designed to control it.

Instead, they examined the broader system that shapes student behaviour. This includes grading practices, assessment design, instructor workload, inconsistent AI policies and deeply rooted beliefs about what success in university looks like.

The idea of cognitive debt

As the project evolved, the team introduced a concept that resonated strongly with both students and faculty: cognitive debt. “Cognitive debt occurs when students repeatedly offload their thinking to AI in ways that make work a lot easier in the short term, but weaken true understanding in the long term,” Colton says. “Part of our report was that although that offloading may be okay in some settings, in a learning environment, it could have very negative consequences.”

Through surveys, interviews and informal conversations with peers, the students heard a recurring theme: many students felt disconnected from what they were learning, even while achieving strong grades. “I asked one of my friends if he remembered what he learned in first or second year,” Aditya says. “He said, ‘I’ll be honest, I don’t remember anything.’”

The students were careful not to frame AI as the sole cause.

“AI came in as such a disruption in the system, a system already led by grades,” Aditya adds. “All the students surveyed felt they had to mindlessly offload their real learning onto AI and that’s how the term cognitive debt came up.”

Teaching conditions matter

One of the clearest insights to emerge from their project was the link between teaching conditions, student motivation and AI use.

“We did find a very clear causal link,” Colton says. “Sometimes teachers underestimate how much their energy, attitude, the way they motivate their class has an impact on whether students use AI or not.”

To Colton, Aditya and Sohan, engagement and course design shape student behaviour. In their research, the team found that many instructors want to respond more thoughtfully to AI but face real constraints, including large class sizes, heavy workloads and limited institutional support for teaching development. In that environment, things like bans or AI detection tools often backfire.

“You can’t trust a detection tool,” says Colton. “It’s gotten to a point where you can actually get yourself into legal trouble as a prof if you try to rely on a detection tool to claim that people are plagiarizing.” Instead of standardized rules applied across an institution, the students argued for course-level, ecological responses that align AI use with specific learning goals.

“Every course has different learning outcomes,” Sohan says. “When you standardize course design, you take the expertise out of the instructor because they know best; they’re experts. So, supporting them where they come up with their own ecological response in the system rather than you just trying to tell them what they should do.”

Learning made visible

If the project reinforced anything for the team, it was the impact of course design. “Dr. Bubbar wanted students to come in with their own ideas,” Sohan says. “The whole idea is that you evolve through the course. Dr. Bubbar really challenged our view and showed us things that we didn’t really see.”

The class allowed AI for specific tasks, such as interview transcription or research discovery, while designing assessments that required deep engagement, synthesis and reflection. “The way he designed the course... we couldn’t have gone through this course with just using AI,” adds Colton. “We could use AI in some surface-level things within the course, but to really be able to find those motivations and deeper-rooted problems, it would have been impossible with AI.”

That experience also changed how the students approached their own learning. “I came out of this course more confused than I ever was,” admits Aditya. “And I think that was Dr. Bubbar’s intent. He taught us how to question things. This course taught us a lot about thinking about the problem itself and not jumping to solutions or conclusions, mapping out the whole problem before we even get into anything.”

From a poster to Oxford

At the April symposium, the team presented their work to faculty members, politicians, social innovation professionals, academic leaders and administrators from across the university. The conversations were often challenging, but intentional. “We are trying to challenge people’s perspective on a very hot topic,” Sohan says. “We’re just trying to show a perspective that you’ve never seen. If we can give people a new perspective, they can think about it, and that pushes the system towards change, and for me, that’s the whole purpose of this project.”

What comes next

For Colton, graduation means professional hockey before transitioning into his next chapter. Sohan recently accepted a role with a Y Combinator-backed startup and plans to continue applying systems thinking to entrepreneurial challenges. Aditya has another year and a half at UNB and hopes to carry these insights into his senior design project. All three see their project as unfinished. “There will be a delay in this system,” Colton says. “It’s part of systems thinking. You show where delays are, and over the next five to 10 years, we’ll see if there is a consequence of that or not.” For alumni reading this story — many of whom are navigating AI in their own workplaces — the takeaway is familiar and timely: technology alone doesn’t change behaviour. Systems do.