Many Canadian educators began this school year with little to no guidance on how to manage AI use in their classrooms.
Just four provinces have formal AI guidelines for educators: British Columbia, New Brunswick, Newfoundland and Labrador and Quebec. Provinces like Manitoba are still working on their own.
Provinces are responsible for establishing the environments students learn in, and yet half have been quiet on one of the most disruptive technologies to enter classrooms since the internet.
Other jurisdictions show a more prudent approach.
After considering parent feedback, New York City announced it was imposing a moratorium on AI use in public schools for the 2026-27 year. According to a policy published by Chancellor Kamar Samuels of NYC Public Schools, this includes “strong limits” for elementary students, including a ban on any “student-facing AI,” while high school students will have “limited, guided opportunities to learn about and use generative AI for specific approved purposes.”
In early September, Los Angeles passed an even more restrictive policy for its public schools, imposing a district-wide moratorium on generative AI use by students on all school devices for this school year.
These approaches are time-limited and prudent. However, they are not a long-term answer.
In our upcoming book Beyond School Choice, Professor Josh Brake of Harvey Mudd College in California argues that an AI “ban” can buy time, but it cannot answer the deeper questions.
Educators need to approach AI with a view to the purposes of education itself, asking how it will serve the formation of the whole person so that students can flourish and contribute to the common good.
Just because technology is new and transformative doesn’t mean that it is good for learning. In fact, the skills that students actually need to develop in order to use AI tools may well not be shaped by that technology at all.
“When we build ever more complicated systems [or policies] to try and force students to do what we want them to, the message they receive is not that the work is worth doing for its own sake, but rather that we don’t trust that they can do it honestly,” Brake said.
In the absence of guidance from provincial governments, there is a risk education unions will steer the conversation.
Various reports from the Canadian Teachers’ Federation, a national alliance of teachers’ organizations, emphasize the risks that AI will undermine their members’ autonomy and the need for human judgment in the classroom. They insist technology must support their workers. They also express concerns that educators are being left behind.
This raises concerns that AI approaches could be shaped by what unions, teachers and education workers will and will not do, rather than by what leads to the best learning outcomes.
We need more than piecemeal approaches to AI guidance.
Policies should be developed in consultation with educators, parents and students. And they must start from key, learner-focused questions: What educational purpose does this tool serve? What evidence shows it advances that purpose? What human capacity might its use strengthen or weaken?
Until we can answer those questions, restraint is a policy choice in its own right.
New York has decided that, for younger students, the answer to whether AI belongs in the classroom is, for now, “no.”
Provinces here should be prepared to follow suit.
