Is it OK if children use Chat GPT to help with homework?

There is a big difference between a child asking a generative AI tool “How did the dinosaurs die out?” and them asking it “How do you solve this Maths puzzle?”

The first question (about dinosaurs) is about information retrieval and theory curation.
⇢ The key educational risk is making sure children have the AI literacy to evaluate the accuracy and sources of the information given.

The second question (about the Maths puzzle) is looking for an answer to a task that had learning value for the child.
⇢ The key educational risk is that this type of usage potentially bypasses learning while still producing an answer.

There is pretty robust evidence that children and teens can produce better academic work when assisted by AI. For example, students who use ChatGPT to assist them in writing an essay produce higher quality essays.

But have they learnt the skill of essay-writing or learnt about the topic – and does that matter? (After all, most adults do not write essays as part of daily life.)

One problem is that these students have potentially avoided what we call ‘productive struggle’ – when we get stuck on a problem and have to spend time working our way through it in a way that constructs understanding.

Productive struggle is an experience that is potentially incredibly useful in life (and to society).

When generative AI helps students learn (e.g. understand concepts or refine skills), that’s one thing. When it enables shortcuts that bypass children’s learning, that’s another.

Learning happens in the process not in the outcome.

I’m not suggesting we ban AI from education (it’s way too late for that thought!) or pretend students are not using it. But we do need to do some serious thinking on how to “harness its potential to enhance education while mitigating the downside” (Ethan Mollick‘s words) and not leave AI usage in education to chance – or to market forces.

We can’t pretend all AI usage is the same when it comes to children’s development and wellbeing. We need to do the work now to understand this better and identify appropriate guardrails.

In the meantime, thinking about how a child’s AI use might be adding to or diminishing a child’s learning process, is probably a good benchmark for parents to use.

Interested in this topic? You’ll love my keynote talk on ‘How to raise future-ready kids in the age of AI’.