AI in schools: the questions we cannot avoid

If we are not careful, discussion about AI in schools can quickly become reduced to a binary choice between embracing it and banning it. However, neither position addresses the more difficult educational questions its growing use presents.

The most important question may not be whether pupils should use AI, but what thinking we need them to do for themselves. Where might AI help, where might it hinder?

Schools increasingly need to consider:

• When might AI support learning, and when might it bypass the learning a task was intended to develop?
  

• In the age of AI, how confidently can homework and coursework demonstrate a pupil’s independent capability? What changes might be needed to both in light of this?
  

• What forms of assessment will provide the most reliable and valid evidence of what pupils know and can do?
  

• How do we help pupils recognise inaccurate output without the subject knowledge needed to question it?
  

• What responsibility do schools and teachers have to teach pupils how to use AI intelligently and responsibly?

  
  

A restriction may establish an appropriate boundary within school, but it cannot be the whole educational response. Pupils will encounter and use AI beyond the classroom.

We therefore need to prepare them to understand not only its possibilities, but also its limitations and potential effects on their learning.

At ResearchED 2026, AI was a recurring theme through many of the sessions I attended.

• 

• 

• 

Subsequently I've spent a few days reflecting on points made during the day and tying together threads drawn from the various sessions. I have written a Substack article drawing together points discussed on knowledge, memory, cognitive offloading, homework, feedback and assessment, my reflections on those and the many questions I came away with.

Read Who Is Doing the Thinking? on Substack → (Who Is Doing the Thinking? - Rob Chambers)