Experts disagree on how much AI can hurt–or help–children in Canada

As governments move to ban AI in classrooms, experts in the GTA scramble to weigh the pros and cons. 

Modern artificial intelligence (AI), best known for iterations like GenAI, which collects and relays data; Large Language Models (LLMs), such as ChatGPT; and Agentic AI, which manages tasks, is sweeping through the global landscape like wildfire. 

While many state that this heralds a new tech boom, with hectares of GTA real estate being set aside for massive data centres required to power the tech, not all are convinced it deserves to be as omnipresent as it is shaping up to be. 

Recent KPMG data showed that 79 per cent of Canadians are concerned about the negative outcomes of unchecked AI applications, with a supporting Leger poll showing more than two-thirds of Canadians support banning access to AI chatbots like ChatGPT for children under 16.

This sentiment has entrenched itself on Parliament Hill, as last month, the Carney administration piggybacked potential bans on AI use for youth alongside potential restrictions on social media.

Certain provincial powers have even taken matters into their own hands; as recently as last week, Manitoba Premier Wab Kinew announced plans for a full-scale ban on social media and AI chatbots for provincial youth.

Even as notions in the federal sphere are more whispers than policy, this hasn’t stopped those in the world of Canadian tech from getting worried. 

Tovi Grossman, Co-founder of AXL, a massive Toronto-based venture enterprise laying the groundwork for Canadian companies building on AI, is one of the foremost experts on AI integration in the country. 

Operating from the ground up, Grossman, who is also an Associate Professor at the University of Toronto (U of T), has recently been touting that — if AI is curtailed in youth development — it may hinder a generation of Canadians as they enter the job market. 

“It would be doing a disservice to students by not allowing them to use AI tools responsibly, because then, as they eventually enter the job market, there’s almost no question that they will be exposed to these tools and will be expected to use them, and use them well,” Grossman told INsauga.com.

Grossman, currently playing double-duty in the private sector and as a professor, is seeing how this technology is stitching itself into both companies and campuses, further fuelling his belief that if there isn’t some AI exposure early on, it’ll do nothing but hamstring the workers of tomorrow. 

However, even Grossman knows that the technology needs guardrails, as in its nascent state, it has seen a surge of errors in the current market, such as last year’s headline-making fiasco, where a ChatGPT update made the program so agreeable and sycophantic to its end user that it encouraged them to engage in actions harmful to themselves or others. 

Such unignorable botches, in turn, lay the groundwork for how Grossman believes these technologies should integrate into the modern classroom, while remaining heavily resigned to a task-oriented structure. 

Currently, U of T uses a platform called LearnAid, which uses AI to help students locate sources and manage course content without just spitting solutions back at them.

“We designed a tool where they can ask questions, but it doesn’t give them answers; it guides them towards solutions so they can still do the thinking for themselves,’ says Grossman. 

While this is relegated to a post-secondary setting, Grossman believes that similar AI applications should be integrated into public schools, laying the groundwork for a hypothesized future. However, even with borrowed guardrails from platforms like LearnAid, the question of whether critical thinking is being replaced by an overreliance on developing minds remains a major point of contention. 

Dr. Feng Ji, an associate professor in the Department of Applied Psychology and Human Development at U of T, has been monitoring the impact of AI on the day-to-day lives of young children. 

Echoing sentiments of other industry professionals, Ji harkens to the psychological touchstone known as ‘cognitive offloading,’ where, no matter its form, AI always runs a significant risk of replacing key developmental steps in a child or adolescent’s life. 

“From a developmental and psychometric perspective, the key issue is not simply whether students use AI, but what mental process is being supported or replaced. If AI replaces thinking, it is harmful. If AI is used after students have made their own attempt, to compare reasoning, receive feedback, or explore alternatives, it can be useful,” Ji told INsauga.com via email. 

While this threat is something those in education are trying to keep up with (to mixed results), Ji knows that at this point, the digital genie is far past any hope of being crammed back into the bottle. 

As a result, Ji staunchly believes that while strong foundations in reading, writing, reasoning, social interaction, and self-regulation are fundamental for the brain of any developing child, you can’t put a bag over a kid’s head and tell them to look away from a technology that’s everywhere. 

Grossman echoes this sentiment by proposing a marriage of modern digital literacy (computers, smartphones, and social media) and AI literacy, packaging them as parts of a whole and regulating and restricting them accordingly.

However, therein lies the rub: legislators are already looking to ban AI and lumping it in with social media wholesale.

When acknowledging them as two sides of the same coin, experts like Grossman and Ji believe they share very different spaces in the minds of Canadian youth. 

“A world with limited exposure and meaningful education can absolutely coexist. In fact, for younger students, that may be preferable. They can learn what AI is, what it can and cannot do, and how to question its outputs without relying on it as a substitute for their own thinking,” says Ji. 

Grossman further details that education will fortify young minds to be aware of AI’s pitfalls — more than its shortcuts — citing ‘digital hallucinations,’ a prevalent issue within LLMs, where, given the sheer amount of data being generated, they have a tendency to spit out incorrect or fabricated information. 

“It’s important to learn what these ‘hallucinations’ are, and how to spot them,” says Grossman. “It’s about being deeply involved and leveraging my own experiences within the system — that is the end goal of AI literacy.”

Grossman cited a study conducted at the U of T, where two groups of high-school-aged students were tasked with using AI in different ways to prepare for a test. Upon completion, it was revealed that those who used AI as a study tool and did not just copy/paste the answers it generated had much more success, further lending credence to his hypothesis. 

If this approach is set in stone, even Ji believes it can have practical applications, stating, “ If AI replaces thinking, it is harmful. If AI is used after students have made their own attempt, to compare reasoning, receive feedback, or explore alternatives, it can be useful.”

However, despite the tug-of-war between policymakers, tech entrepreneurs, and behavioural scientists, all fighting over the classroom, one fundamental component of a child’s life is being overlooked — what happens at home. 

In a perfect scenario, where the collective braintrust creates a perfectly regulated program of AI use in the classroom that has no cracks whatsoever, there is still the domain of a child’s life at home, far out of reach of policymakers and educators. 

When confronted with the prospect of living rooms versus classrooms, Ji noted that the rapid pace of technological development, plus potential apathy towards restrictions at home, could cause disarray in a child’s developing brain. 

“If the gap between school and home is too large, students may learn to hide their use rather than reflect on it. Students with more unrestricted access at home may also gain advantages in assignments, while students with less access may be disadvantaged. This could worsen educational inequality,” says Ji. 

Even Grossman, when looking at AI use at home, remains cautious of its impact on children’s development, stating that certain AI platforms can take something as simple as a family photo and turn it into a colouring page, which on its surface may seem innocent enough, but could still hinder precious neurons firing off in the creative centre of a child’s brain.

“For a six-year-old, it is likely better for them to colour a page themselves than get handed something by an AI system. This could then double back to the potential for parents’ overreliance as well, asking an LLM, ‘give me something creative to do with my kids this weekend,’ instead of sitting down with their children and figuring something out,” says Grossman.

While the boogeyman of at-home regulation falls to parents, Ji believes there are already vital steps to be taken when drafting AI programs for young children in a classroom setting that should also be applied at home.

This includes limits on anthropomorphic design (AI bots appearing as cute animals, etc.), strong privacy protections, oversight, and age-appropriate use.

Until then, Ji remains steadfast that any current societal issues impacting adults with AI should be applied tenfold when considering integration with young minds, concluding that “It [AI] should be framed as a tool, not as a friend, therapist, or replacement for human relationships.”