August Education Special: A Good Tech Childhood

Somewhere near you, this week, a parent is having the annual summer argument about screens. Possibly it is you. It is certainly, some days, me.

The argument usually presents itself as a choice between two childhoods. On one side, the tech-free childhood: wooden toys, boredom, long grass, a phone locked in a kitchen drawer until eighteen. It is a lovely picture, and the people advocating it mean well. On the other side, the childhood we fear: eight hours of algorithmic feeds, an essay outsourced to a chatbot, a mind going soft.

Both sides are answering the same badly posed question: are screens good for children?

Which is rather like asking whether cooking is good for you. It depends entirely on what you are cooking.

So I would like to offer a third option, perhaps the greatest achievement for us as parents: a good tech childhood. A childhood in which the tools of our time are shaped, deliberately, by the adults who know and love the child, to serve the child. We do not boast that our children have never seen a book. I look forward to the day nobody boasts that their children have never used AI.

This is newsletter number fifty, it is the summer holidays, and I am wearing my three hats at once for this one: the human capital economist, the CEO of an AI company, and the mother of three children who are, as I type, variously practising off-ice figure skating, building something suspicious in the garden, and listening to a self-made audiobook about cell division as a courtroom drama. Consider this my summer soap box, offered with warmth.

Socrates hated writing too

I was talking with someone this week who asked a fair question: why should we change just because AI has come along?

My answer is that whether we should or not, we do. Available tools and technologies change us at every turn, always have, and always will. One story I like about this is Socrates, who argued against writing itself. He worried that if people wrote things down, they would stop exercising their memories, and knowledge would become something you looked up rather than something you carried with you in your mind. He was right, by the way. Our memories did get worse. Ancient bards could recite epics for days; I struggle with a six-digit code between rooms. That is a real and serious loss, and Socrates saw it coming.

But look at what we got in the trade. Libraries. Science. Law. The ability to stand on the shoulders of everyone who thought before us instead of starting from scratch each generation. Writing changed us, and I think when we look back on that particular deal, it feels to me overwhelmingly worth it.

Every technology since has run the same bargain. We do not forage for our food, and we do not feel diminished by the supermarket. We do not spin our own cloth, make fire by friction, or navigate by the stars. And we don't mourn any of it, because those losses happened before our lifetimes.

Available technology changes what humans do and, in time, what humans are. That is not a corruption of the natural order but simply the way things are for humans and the tools they have been creating since the dawn of time.

So the question has never been "should technology change us?" It always has and it always will. The question, at every single turn, is the one an economist would ask: have we gained more than we have lost?

And here is where I have real sympathy with the worried parents, because for the first time in a while, a lot of people sense the answer might be no. The last big consumer technology wave, the attention economy of feeds and infinite scroll, took a great deal from our children and gave back very little. People are right to feel burned.

I want to be clear about where I stand on that, because it colours everything else in this piece: nobody is more furious about that trade than I am, and few parents are less laissez-faire about technology and their children than I am. None of my three has ever set foot on TikTok, because I cannot configure it to serve their interests rather than its own, and they will not until I can or they turn 18 - whichever comes first. I am not the parent who waves technology through. I am the parent who reads the terms.

But the lesson of that dire episode is not "refuse the next technology." The lesson is that the trade does not make itself. If we sit back and wait for the companies that build these tools to arrange a good deal for our children, we will be waiting a long time. No deal that suits our children will magically arrive on our laps. The trade is made good by us: by deliberate, thoughtful, expert use, in our own homes, on our own terms. Not the AI as an answer machine. The AI (and technology more widely) as a tool wielded with skill, by adults who decided what it was for, how to use it, and how to teach our children to use it.

And when the adults do decide, I can tell you from my own kitchen table and from the research: technology used well is absolutely, undeniably capable of engaging children in learning in a way nothing before it has managed. This was never a screens-or-no-screens question. It is a what-are-we-cooking question.

Why this one stings

There is a second reason this moment feels harder than the ones before it: the pace.

It used to take generations for a skill to fade. Navigating by the stars declined over centuries; nobody watched it vanish within their own parenting years. Now a skill can go from essential to optional inside a few months. And so we find ourselves mourning things that were part of our own childhoods: the handwritten note, the folded map, the orienteering badge. That ache is real, and it is worth respecting in ourselves and in each other. But it is nostalgia, and our nostalgia is a poor basis for someone else's education. Our children will not miss folding maps any more than we miss darning socks. What they will miss, if we get this wrong, is the six years everyone else spent learning to use the tools of their era while we were guarding the drawer.

The bell that kept ringing

Which brings me to school, and the logical fallacy called the appeal to tradition: the idea that because something has always been done this way, this must be the right way.

Mass schooling as we know it is a Victorian invention, designed with real skill for the world that built it. The industrial economy needed workers who could tell the time, arrive punctually, read, write and count well enough to be useful, and come when the bell rang. So we built an institution that produced exactly that, in batches, by age, to a timetable. For its moment, it was a triumph.

But strip education back to its essentials and it has only ever been one thing: well-meaning adults using every available tool to equip young people to go out and succeed in the world in front of them. That is what the Victorians were doing. The bell, the rows, the timetable were not sacred; they were simply the best available tools for the world of 1870.

So the constructive question, the one I would put to every school leader, policymaker and parent this summer, is the same one the Victorians answered so well for their era: are we using every available tool, resource and possibility to equip our young people for the world they will actually live in? Because I struggle to see how any education system can stand in front of the greatest aggregation of human knowledge ever assembled, a patient tutor in every pocket, in every language, at any hour, and take pride in not using it.

Ignoring that it exists does not protect our children from the world they live in. It just sends them into it unprepared. And they are already in it: the share of teenagers using ChatGPT for schoolwork doubled in a single year, and 92% of UK university students now use AI in their studies. Banning it is not a strategy. Teaching them to use it well is.

The world of 1870 needed punctual factory workers, and school produced them. The world of 2026 needs critical thinkers, innovators, people who can learn, unlearn and relearn at pace using every tool available. Reading, writing and numeracy remain the bedrock; nothing higher gets built without them. But the ten hours spent on what could take one? That was a feature of the factory, not a feature of learning.

There is no glamour in taking ten hours over something that wants to take one. I put my washing in the machine every morning. I have never once felt I cheated the laundry.

Two hours, and the other six

A network of schools in the United States runs its entire academic curriculum in about two hours a day, using AI-guided, mastery-based learning, one child at a time, at each child's own level. Two hours. The children reportedly perform in the top percentiles nationally. I see this at play in my kids using well-tuned AI every day.

You can question the model, and people do. But let's think about that number for a moment: if the academics of a school day can be done, done well, in two hours, then the most interesting question in education is suddenly an economist's question: what is the opportunity cost of the other six?

What could a child do with six returned hours a day? Play, properly and at length. Sport. Cooking. Building things. Boredom, which is where imagination lives. Friendship, face to face. Reading for the love of it. Starting the small ridiculous project that becomes the big serious one.

My son learned a large share of his primary maths on a games-based app, half an hour a day, giggling. The result was not less childhood. It was more: the rest of the morning was outside. The efficiency paid for the childhood.

We know this trade from our own lives. The washing machine gave women back their afternoons; the question was only ever what we did with the hours it returned. Childhood is about to be handed the same gift, and the families and schools who think hardest about the other six hours will give their children something extraordinary.

One librarian, a thousand children

Everyone agrees that the way a child falls in love with reading is the right book at the right time. One book, meeting one child exactly where their curiosity already lives, at exactly the level that stretches without defeating them. Ask any lifelong reader and they can probably name theirs.

Now look at how we currently deliver that magic. One human librarian, where schools have one at all, serving hundreds or thousands of children, with no time to know any of them. And the results are in: the National Literacy Trust's most recent survey found children's enjoyment of reading at its lowest level since records began two decades ago, with only around one in three children saying they enjoy reading in their free time. Whatever we are romanticising when we defend the current arrangement, it is not working. Children are not falling in love with reading under the system we have.

So here is a tool - thank you Claude - that can know every book on your shelves and every interest your child has, and make the match, at their pace, every single week. The right book at the right time, for every child rather than the lucky few. Why on earth would we not do that? My daughter's AI sent her from a Stephen Fry chapter on Aphrodite to page 142 of the art book on our bottom shelf, Botticelli's Birth of Venus, with a question about what the painter saw that Fry missed. She moved between rooms like it was a scavenger hunt. That is not screen time. That is a librarian for every child, at last, and a child engaging with reading, art, and the bookshelf.

And if you want the proof this scales beyond one bookish household: a World Bank trial in Nigeria found six weeks with a well-set-up AI tutor and a teacher in the room produced learning gains equal to up to two years of school, for about forty-eight dollars a child, with the biggest gains going to the girls who started furthest behind. Configured well, this is the most level playing field education has ever had.

The three-minute difference

Now the part that decides everything, and I want to unpack it properly because I use the word "configure" a great deal and I should not assume everyone knows what I mean.

Every AI tool arrives with default settings, chosen by the company that built it. On those defaults it is eager to please, quick to hand over answers, and generous with praise whether or not the work deserves it. Those defaults suit the company; an agreeable product is a popular product. They do not suit your child.

Configuration means changing that. Inside ChatGPT, Claude or Gemini there is a settings box, usually called personalisation or custom instructions, where you can type, in ordinary English, standing orders that shape every conversation the AI ever has with your child. You do not need any technical skill. You are writing a brief, the way you would brief a new tutor on their first day.

Ours says, roughly: put this user's wellbeing and quality of educational experience above all other priorities. Never give the answer; ask the question that helps her find it herself using Socratic method. Your priority is to engage her in learning, and challenge her thinking, not to please her. When you state a fact, link a credible source. Praise specific things when they are earned, and say so plainly when work is below her best. Connect what she is doing today to what she read last week. Be proactive in engaging her in learning using her interests as hooks.

That takes about three minutes to type, and it is the whole difference between the two summers.

In one house, the AI on factory settings writes the holiday homework while the child watches and their brain turns to mush; the child learns, lesson by lesson, that thinking is something you can outsource, and the research on unguided AI use in schools confirms the damage, with students who leaned on an unconfigured chatbot scoring markedly worse once it was taken away.

In the other house, the same underlying technology, briefed by a parent in three minutes, refuses to hand over answers, asks the next question instead, shows its sources, and stretches the child a little further each day. Same tool, opposite childhoods. The difference was never the machine; it was the three minutes an adult spent telling it who it works for.

Is cooking good for us? Well it depends what we're cooking.

Press-ups with Claude

A few things running in our house this August, offered parent to parent; take what is useful.

Languages over breakfast: voice mode propped against the toaster, ten minutes of French or Spanish conversation while the toast browns. No audience, no embarrassment, a hundred reps by September.

Knowledge in their ears: any chapter, any topic, turned into a fifteen-minute podcast for the car, the beach, the bike ride. Screens off, curiosity on.

The holes game: teach them the one prompt that matters, "test me on what I think I know, and find the holes." That shift, from being taught to teaching themselves, is the skill of the next fifty years, and it is learnable at ten.

Build the idea: if your child has an idea for an app, a story, a business, this is the first summer in history they can actually make it, at the kitchen table, for pocket money. Let them. The lesson is not technical; the lesson is that they are allowed to imagine, plan, and build.

And my favourite: my ten-year-old's daily circuit. Every morning Claude sets him a challenge that mixes the physical and the mental: read a chapter, then a scavenger hunt around the garden for something from the reading, twenty press-ups, a timed run, a question to answer at the finish line. He attacks it like a mission. His sisters would loathe it, and most of his class might too, and that is precisely the point: it was built for one boy who needs to move while he thinks, and for him it means a summer of being active and reading daily, at the same time, happily. What's not to like about that?

[I'm also running a daily version of this over on Instagram this summer, one household use a day, so come and borrow from it there

Notice what made the circuit work, though, because it is the thread running through this newsletter. It was not the technology itself, alone. It was a parent who knows the boy, deciding what the technology should do for him. The problem with AI and children was never the tool; the problem, and the opportunity, is using it thoughtfully. That is a human challenge, not a technology one. It happens, not coincidentally, to be the thesis of my entire working life: the value of AI has never lived in the tool or the access alone. It lives in the human capability wrapped around it, at the kitchen table exactly as much as in the boardroom.

The soap box, folded away

None of this is an argument against schools, and it is emphatically not an argument against teachers, who do the most human work there is and are irreplaceable in it. It is instead an argument against the appeal to tradition: against defending the ten hours because they have always been ten, and the bell because there has always been a bell.

Our great-grandparents built an education system with real courage for the world in front of them. The way we honour them is to do the same for ours: protect the foundations fiercely, hand our children the tools of their time, teach them to drive rather than be driven, and then, with the hours the machines give back, return to them the oldest curriculum there is. Grass. Boredom. Books. Each other.

That is a good tech childhood, and I would like us to be able to say we gave our children one, out loud, as the achievement it is.

This is issue fifty. Thank you for reading, arguing, forwarding and replying for the forty-nine before it; the replies are the best part of my week. If this one would be useful to another parent mid-screens-argument, or to anyone who thinks about what school is for, do send it on; that is exactly what it is for.

And tell me, too: what would your child do with the other six hours?