Why I expelled AI from the classroom

Artificial intelligence (AI) is certainly in vogue these days. Within post-secondary institutions, it is rapidly reshaping the pedagogical landscape. Some academics maintain that AI enriches the student learning experience, whereas others believe it enhances critical thinking.

Tech enthusiasts regard AI as a vital component of the contemporary university. Ted Brodheim, CIO adviser for education at Zoom, suggests that AI has become “the backbone of today’s learning ecosystem” in higher education. Likewise, Evan Solomon, Canada’s first-ever minister of artificial intelligence and digital innovation, contends that AI is having a “transformative effect” on education curriculum.

Such high praise has convinced many educators to jump on the AI bandwagon; however, I’m not one of them. As an undergraduate sociology professor, I bucked the trend by banning AI for all classroom assignments. Here’s why.

In social sciences, students find it challenging enough to master the subject matter at hand. For instance, our classes discuss a wide variety of contentious issues, such as free speech, abortion, euthanasia and prostitution. When framing arguments, we also analyze important theories, principles and case studies. Providing space for AI detracts from other important priorities — chief among these is grasping the course content.

Also, students need to focus on basic skill development. When they spend countless hours reading, writing and interpreting texts, an evolution occurs. They organize their thoughts more clearly and defend their positions more persuasively. Consequently, their reasoning becomes increasingly sophisticated.

Improving cognitive functions — memory recall, concentration, attention to detail, reasoning — requires a concerted effort, and as any professor worth his or her salt knows, there is no substitute for hard work. Letting AI do all of the heavy lifting defeats the entire purpose of being in a university.

Yet students find AI indispensable because it produces a complete essay in seconds. In my classes, almost without exception, those who relied on AI to write papers failed their midterm exams. AI has become an economic lifeline. If they managed to fool the professor, these students would save themselves the time and money required to repeat the course.

But deep down, most understand why submitting AI-generated essays is wrong. As one university undergrad wrote in Maclean’s, “Delegating my work to a robot never appealed to me. Using it for school sounded like cheating.” Don’t take her word for it. The research supports her claim.

According to current data, approximately 25 to 35 per cent of students use AI to write their entire papers. Worse still, one survey found that 76 per cent of those who used ChatGPT for school assignments felt they were “somewhat” or “definitely” cheating.

Excuses for cheating — poor time management, language barrier, lack of expertise, fear of failing — have become easy justifications for flat-out dishonesty. Submitting an AI-generated paper still constitutes academic fraud. The student’s name is on the title page, but a computer program is the real author. This level of deceit makes education transactional and worthless.

The rise of AI has dire consequences for education. Political commentator Fareed Zakaria, author of In Defense of a Liberal Education, warns that students cannot sidestep the learning process. “The central virtue of a liberal education,” he argues, “is that it teaches you how to write, and writing makes you think.” When students delegate the task of writing to AI, their comprehension suffers. AI will only make declining literacy rates worse.

Students even risk developing a false sense of their true academic potential. Lorena A. Barba, a professor of mechanical and aerospace engineering at George Washington University, explains how AI creates the “illusion of competence.” She found that over-reliance on AI-generated answers leads to a “superficial and fleeting grasp of concepts, as students don’t actively engage with the material.”

Barba further explains how students “may feel competent after using AI, but this feeling can mask a lack of comprehension, leading to gaps in their learning.” In other words, no matter how intelligent they think they are, students who employ AI as a crutch are less knowledgeable.

Additional empirical evidence corroborates Barba’s conclusions. For example, findings in one study revealed “a significant negative correlation between frequent AI tool usage and critical thinking abilities.” Moreover, students who depend on ChatGPT for academic writing tasks become lazier thinkers.

Although AI allows undergrads to cut corners, cognitive offloading comes at a cost. No actual understanding occurs whenever an AI system spits out content that is quickly uploaded for a mark. By not engaging with scholarly research, students turn into passive spectators watching a text-generator sport.

Some students have referred to AI as “the modern approach to learning.” But not everything new is beneficial.

If AI dependency defines today’s education, call me old school. When the next group of students enters my class, they will be welcomed with open arms. However, AI will be shown the door.

Stuart Chambers, PhD, is a professor in the School of Sociological and Anthropological Studies at the University of Ottawa.