Parenting in the age of AI

Our friends at NASCEE recently sent out an invitation to the first seminar on AI in Basic Education, as part of the Lefa AI initiative.

As an evaluator, I am extremely interested in this topic. As a parent of school-going children, I am perhaps even more fascinated by it. Although “fascinated” may not be quite the right word. It is the kind of fascination where you lean in because you really want to see what happens next, while at the same time keeping one hand firmly over your heart in case what happens next gives it a shock.

AI in education raises all sorts of practical questions about teaching, assessment, access and learning. But in our house, some of the questions already feel much more immediate. What should children still learn to do for themselves? When does using AI become a useful extension of their abilities, and when does it quietly remove the struggle through which those abilities are developed? And perhaps most difficult of all: how do we explain to children why they should spend time learning something when a machine can produce a polished version in seconds?

I do not have tidy answers. But watching my own children use these tools has made the questions much harder to ignore.

A few days ago, my youngest showed me an English dialogue she had written for school, and I was genuinely surprised. It was funny. The characters had personalities. The dialogue moved. There was timing and a sense of scene. And she had done it on her own.

I would love to say this came from years of reading. Heavens knows it did not. Reading is an ongoing struggle in our house. My best guess is that she has absorbed enough stories through television and other media to understand, almost intuitively, how characters behave and how dialogue works. That in itself is interesting. We often speak about creativity as though it emerges from nowhere, but of course it does not. Human beings are pattern learners. We absorb stories, rhythms, phrases, gestures and conventions from the culture around us, and then we recombine them into something that feels like our own.

My son presents almost the opposite problem. Writing does not come easily to him, and he is very willing to offload cognitive work. But he is also quite inventive about it. For his school entrepreneur day, he decided he needed a Wheel of Fortune-style spinner for a game, so he used ChatGPT to programme one. He also gets somebody else to prepare accurate study notes, then feeds those notes into an AI system that turns them into flashcards, memory prompts and other study tools.

Part of me thinks: clever. Another part of me thinks: oh dear. Because he may become very good at orchestrating tools without ever becoming very good at writing. What does the future look like? Must he necessarily be good at writing? 

And I am beginning to think that this is where the AI debate becomes much harder when we are talking about children. As adults, cognitive offloading is often sensible. I use a calculator because I already understand arithmetic. I use spellcheck because I can spell well enough to know when it is wrong. I can ask AI to tighten a paragraph because I have spent decades learning how paragraphs work.

But what happens when we offload the work before the underlying capability has been built?

A child learns to write by writing bad sentences, then slightly better ones, then awkward paragraphs, and eventually paragraphs that begin to say what they intended. The inefficiency is not a bug. It is the learning process. AI can remove a great deal of that struggle, which sounds wonderful until we remember that the struggle may be exactly what develops the capability.

So the question is not simply whether my son is “cheating”. It is more uncomfortable than that. What experiences does he still need to have in order to become someone who can think and write for himself? If AI consistently occupies that space before the skill is established, he may become very good at directing writing without ever becoming very good at writing. Those are not the same thing.

And then there is my daughter. What happens when she spends two hours creating something funny and original, only to discover that somebody else can type a prompt into an LLM and produce something equally polished in three seconds? I worry not only about skill loss, but about motivation. Why bother mastering something when the output itself can be reproduced almost instantly?

That question troubles me because creativity is not valuable only because it produces a poem, photograph, song or school dialogue. There is value in becoming the person who can make it. The capability matters. The eye matters. The judgement matters. The satisfaction of being able to do something matters.

She is also becoming interested in photography, and perhaps one day AI will extend what she can create. I do not necessarily object to that. If she develops an eye for composition, light, framing and emotion, AI may become an expressive tool. But if the technology does the aesthetic work before she has had the chance to develop those instincts herself, then something else may be happening.

I increasingly find myself returning to one question:

What human capability is the technology extending, and what capability is it bypassing?

Perhaps that is the distinction I need as a parent. AI after mastery may be an extraordinary tool. AI before mastery may sometimes quietly remove the very practice through which mastery develops.

And unfortunately, “because you still have to learn it first” is going to be a much harder argument to make to our children than it ever was to us.