Martina Rau, how should AI be used in children’s education

There are also tasks that we carry out with what we call "cognitive tools". No one would expect me to manually analyse data anymore, for example - it's clear that we have software for that. But I still need to understand how the software works and how to use it properly. ils

Cognitive tools aren't necessarily digital. If I intentionally leave the bread knife out on the kitchen counter as a reminder to prepare my son's morning snack, the knife serves as a memory aid at that moment. Humans are natural and enthusiastic users of cognitive tools that reduce the strain on our memory. The fact that these aids are increasingly powered by AI is not cause for concern in itself. The question is rather: How can I use AI as a tool to improve my learning? I can use AI to create practice materials. If I want to expand my vocabulary, AI can extract the difficult words from a text that I'm reading and turn them into flashcards. That frees up time that I can use for vocabulary drills.

I can also use AI to get a clearer explanation of a concept or to check whether I have understood it correctly. That said, researchers are familiar with a phenomenon called "the illusion of knowing", whereby people read an AI-generated explanation and come away thinking they have grasped a principle - when in fact they have not. True understanding means being able to explain something in your own words, put it into practice and transfer the concept to a new problem, all on one's own. This requires a healthy amount of scepticism and discipline, because there's a natural tendency to overestimate ourselves. I therefore recommend that children first try to formulate as many explanations as possible themselves, and only then use AI to check their answers.

To help AI identify errors and offer students meaningful feedback, it helps to direct it to particular sources, such as course materials or reputable websites that it can use as a reference point. Instances of AI getting things wrong are worth discussing in class. Teachers can use these errors and hallucinations to drive home the point that AI cannot be blindly trusted. Everyone benefits from this kind of discussion, and engaging with the material helps the lesson stick in students' minds.

If we see children trying to avoid difficult cognitive tasks by using AI, things start to become problematic. We wouldn't turn a blind eye to cheat sheets under their desk, and this is no different. In this case, I'd need to have an honest discussion with the class - and if necessary, simply ban the use of AI. Whether it's digital or analogue, people who don't put in the cognitive work won't benefit in the end.

I think it is a good idea for parents to share with their children about how they use AI in their own lives, for example on the job. Where does it help, and where does it fall short? Parents are able to interpret and contextualize AI output because they built up knowledge before the dawn of the AI era. This can be incredibly valuable for children.

As a society, we're currently in an exploratory phase when it comes to AI. We have fears and hopes, so why not talk about these things with our children? Good teaching motivates children and connects with their everyday lives and the world they inhabit. Embrace AI as a genuine topic in the classroom, and the school wins in the end - and so do the children it sends out into the world.