Artificial intelligence is moving into classrooms faster than researchers can determine what it means for learning. Students are already using generative AI for schoolwork, teachers are encountering AI powered tools in their classrooms, and schools are making decisions about technologies whose educational benefits are still uncertain.

A new research feature from Children and Screens, AI in Education: What the Research Says, brings together researchers and educators to examine what we currently know about AI and learning. TLC Lab Director Dr. Adam Dubé is featured throughout the report, sharing findings from our research and discussing how schools can make more evidence based decisions about AI.

Students are using AI, but not simply to cheat

Much of the discussion about students and generative AI has focused on academic dishonesty. The research presents a more complicated picture. Dr. Dubé points to evidence that only a minority of students consistently use AI to cheat. Far more report using it to explain ideas, generate ideas, summarize texts, and edit their writing.

Those uses may sound productive, but they raise an important learning question. If students regularly use AI to perform skills they are still developing, they may complete the immediate task without getting the practice required to develop the underlying skill.

“We let these systems do it for us, and they didn’t give us the opportunity to practice doing that ourselves.” — Dr. Adam Dubé

The distinction is therefore not simply between using and not using AI. Educators need to consider what students are supposed to learn and whether AI supports that learning or does the important cognitive work for them.

Evidence needs to come before adoption

The rapid introduction of AI into education also reflects a much broader problem with educational technology: technologies are often purchased and deployed before schools have strong evidence that they improve learning.

Dr. Dubé argues that schools should be asking basic questions before adopting AI tools. What does the technology actually do? Was it designed with educators and learning experts? Is there independent evidence that it improves the outcomes it claims to improve? And, once it is being used, is anyone evaluating whether it is actually working?

“With generative AI in education, we can’t assume that it’s going to fix education. We have to demand evidence that it will work and that it is working once we buy it.” — Dr. Adam Dubé

This is especially important because current research does not show that AI is universally beneficial for learning. Dr. Dubé describes the evidence as mixed, with studies finding positive, negative, null, and mixed outcomes. Whether AI helps depends on the technology, how it is designed, what students do with it, and the learning goal.

Teachers need more than prompt training

The report also highlights the role of teachers in decisions about classroom AI. Technologies can be introduced at the school or district level, leaving individual educators with little choice over whether or how they enter their classrooms.

At the same time, much of the professional development surrounding generative AI has focused on how to operate the technology, including how to write prompts. Dr. Dubé argues that teachers instead need the knowledge to critically evaluate when AI should and should not be used.

“What we need going forward is for teachers to be taught about these systems, but they need to be taught so that they can be a critical judge of when it’s good to use these things and not good to use these things.” — Dr. Adam Dubé

That means putting educators back at the centre of decisions about educational technology and giving them the evidence and expertise needed to evaluate tools for their own students and learning objectives.

Start with what we know about learning

The report ultimately points toward a straightforward principle for navigating AI in education: start with learning, not technology.

Rather than asking how AI will transform education, schools can begin with what research already tells us about effective teaching and learning. Technology can then be evaluated according to whether it supports those principles.

“We know what good learning looks like, we know what good teaching looks like, and we have to ask how technology is reflecting that back at us.” — Dr. Adam Dubé

This approach is central to our work at the TLC Lab. As AI becomes increasingly embedded in the technologies children use, our research is focused on building the evidence needed to understand when technology supports learning, when it gets in the way, and how we can design and select better tools for students and teachers.

Read AI in Education: What the Research Says at Children and Screens