AI in school: support, shortcut, or something in between?
By: Emilie-Verno Lavigne, Coordinator, Corporate Communications, OMERS
July 13, 2026
Artificial intelligence is now woven into daily life, including within our schools. A 2025 Actua report found that 9 in 10 Canadian youth aged 12 to 18 had used generative AI in some capacity. More than half reported using AI for homework, including translation work (79%), text creation (82%), and data analysis (75%).
The rapid development of generative AI over the past four years has given parents and caregivers little time to reflect on what it means for their children’s education. Many families are still forming their views, being uncertain about how AI can support learning, and where it may interfere.
The widespread use of AI to complete schoolwork raises important questions about how students learn. Most working-age Canadians developed their knowledge and expertise without AI. They memorized material, practiced skills, made mistakes, and worked through the difficult process of learning. That effortful process is central to building deep understanding.
The way students use AI matters just as much as whether they use it at all. When used thoughtfully, generative AI can support learning in the classroom, but when it replaces student effort, it may weaken the cognitive skills children need to develop.
How AI can support learning
To understand this balance, it helps to first look at where AI can genuinely support students and teachers.
Dr. Ying Xu, Assistant Professor at the Harvard Graduate School of Education, explains that generative AI can serve as a useful learning companion (Anderson, 2024). When AI tools are used to scaffold learning, students have been found to benefit significantly. For example, generative AI can provide practical homework help when students are stuck and explain academic concepts more thoroughly if they do not understand them. Like a human tutor, it can scaffold learning by generating prompts, hints, modeling, examples, and reflection phrases tailored to their competency levels. Based on student behavioural patterns and interaction data, AI tools can provide timely interventions and performance evaluations.
In essence, AI can become a personalized tutor for students in need. According to a 2026 Brookings report, it can be a valuable resource for teachers managing large classrooms, where supplying individualized support for all students is difficult. Similarly, generative AI can narrow achievement gaps by giving tailored support to children with physical and/or learning disabilities, such as dyslexia and speech or visual impairments. Students in economically disadvantaged situations who cannot afford individual tutoring can use an AI tool at little to no cost.
Generative AI is similarly helpful in bolstering students’ writing skills when used to assist their creative process. For instance, students struggling to formulate written thoughts or stuck in a writer’s block frequently turn to AI for inspiration. AI tools are also useful for correcting grammatical errors and organizing essay structures, ensuring that the student’s ideas are well-formatted. A 2023 study found that students receiving AI-generated essay feedback made greater improvements in their work compared to those who received none. With accessible, personalized feedback—often written at the same level as a teacher, such an outcome is not surprising.
Where AI can undermine learning
If the benefits of AI integration in the classroom are significant, many are wary of potential costs for children’s learning development. While research indicates that AI tools can scaffold learning in a developmentally sensitive manner, this is not typically how they are utilized outside of experimental contexts. As it currently stands, over half of teachers reported barriers to teaching AI in the classroom, with only 41% saying they taught their students how to use AI tools. Most youth are using AI without any structured guidance from educators.
Unregulated AI use to complete schoolwork creates two related problems. First, it offers children an easy way to submit work that may not reflect their own thinking. Detecting AI plagiarism remains difficult for both educators and AI detection tools, according to this study. Morever, a CopyLeaks report found that 63% of high school students and 46% of middle school students admitted unethical AI use, including academic dishonesty.
The second problem is less about whether students get caught and more about what they miss when AI “thinks” for them. If students offload school-related effort into an AI assistant, they may be preventing opportunities to develop problem-solving, concept formation, written, and imaginative skills, all of which are improved with increasingly challenging scholarly tasks.
Indeed, emerging research suggests that over-reliance on generative AI may hinder cognitive development and critical thinking skills. Tian & Zhang (2025) found that AI-dependent individuals engaged in less critical thinking than those who did not use AI, for two reasons: The LLM participants had the weakest brain connectivity of all groups, especially in regions associated with creative ideation, executive control and semantic integration. These writers also struggled to quote their own work, which tended to lack originality compared to the other groups. The conclusion was that AI essay-writers are preventing their brains from tapping into the deep and rich associative processing required for creative writing. While this study does not settle the question on its own, it points to a larger concern about what students may lose when they outsource the thinking process.
In the classroom, AI could enable an environment of “frictionless” learning, where students avoid mistakes, and by extension, the critical thought required to correct them. Effectively, researchsuggests that students’ self-regulation, agency, nuanced problem-solving, and capacity for ethical reflection were negatively affected by cognitive offloading to AI tools. In the absence of guardrails, deference to these tools creates passive and unengaged students.
What is happening in students’ brains when they offload work onto AI tools? Kosmyna et al.’s (2025) study on the neural consequences of AI-assisted essay writing provides potential answers. Researchers divided participants into three groups: large-language model (LLM), search engine, and brain-only. As participants wrote their essays, their brain activity was measured using EEG technology (a device tracking the brain’s electrical signal). Essays were then analyzed and scored by human teachers and an AI judge.
The brain-only group had the strongest brain connectivity (suggesting a high level of engagement between brain regions), followed by the search engine participants.
The LLM participants had the weakest brain connectivity of all groups, especially in regions associated with creative ideation, executive control, and semantic integration.
These writers also struggled to quote their own work, which tended to lack in originality compared to the other groups.
Researchers concluded that AI essay-writers are preventing their brains from tapping into the deep and rich associative processing required for creative writing.
While this study does not settle the question on its own, it points to a larger concern about what students may lose when they outsource the thinking process.
Teachers are most concerned about the future. In fact, 48% of educators surveyed thought that AI worsened their students’ critical and independent thinking, compared to 21% of youth. The opinion gap between students and teachers is not surprising. On one hand, students directly benefit from AI tools, seeing significant grade improvements and task acceleration. On the other hand, educators understand that students’ cognitive development requires effort, active problem-solving, and mistake-making. Young students, who often lack the oversight to recognize the importance of cognitive effort, risk facing long-term developmental consequences if AI is implemented too liberally in the classroom.
What to make of it?
The use of generative AI in the classroom brings both positives and negatives. If AI tools are designed with developmental age in mind, they could be excellent academic resources for children. In tutor mode, AI can help students by offering hints when they are stuck, explaining difficult concepts in simpler language, giving feedback on a draft after the student has already written it, or helping them organize their ideas without doing the work for them. In these cases, AI supports the learning process rather than replacing it. However, if implemented without restrictions, generative AI may become a surrogate student, allowing children to skip the effort, mistakes, and problem-solving needed to build strong cognitive skills.
There is no denying it—AI is here to stay, and no amount of restriction will stop students from using it. The issue is not whether to ban AI altogether, but how to manage its use. Guidelines and governance structures are required to inform and assist students, teachers, parents and school boards on how to use AI responsibly. Adults need to set the “rules of the game,” to ensure AI augments rather than harm our children. When it comes to their education and upbringing, there is no replacement for human judgment and oversight.
While governments and schools have their work cut out for them in establishing policies and programs suited for the new AI-enhanced world, parents cannot afford to sit on the sidelines. A first step for all families could be to reinforce the value of real-world experiences. Students need regular opportunities to gain experience without AI shortcuts. Activities promoting real-life engagement, such as sports, reading physical books, discussion, independent writing, and art, are all the more important today. These experiences continue to foster focus, critical thinking skills, patience, creativity, and a love for learning in children. The challenge for parents and educators is to make sure students learn how to use AI while still developing the skills they need to think for themselves.
Stories we think you'll like
The robot in your fantasy football league
5 min read
Do AI and digital health improve our well-being?
3 min read