The First Generation Raised With AI Is Already Here

The question is no longer whether children will use artificial intelligence. It is what happens when intelligence itself becomes part of the environment in which they grow up.

A child gets stuck on a homework question.

Until very recently, there were only so many places to go next. Ask a parent. Ask a teacher the following morning. Look through a textbook. Search the internet. Call a friend. Or sit there for a while and try to figure it out.

That last option mattered more than we probably realized.

Today, there is another possibility. The child can open a chatbot and type: I don't understand this. Explain it to me.

Seconds later, something answers.

It doesn't get impatient. It doesn't have somewhere else to be. It can explain the same idea five different ways. If asked, it can simplify the vocabulary, create an example, test the child afterward and offer encouragement.

Or it can simply do the assignment.

The distinction between those two outcomes may turn out to be one of the most important questions in childhood over the next decade.

Because the first generation raised with artificial intelligence isn't coming.

It is already here.

In June, Common Sense Media surveyed more than 1,200 American children between nine and seventeen. Eighty-six percent had used AI. Nearly a quarter were already using it every day. More than half of AI users had asked it for advice about their health or body, and more than a third had discussed feelings or personal problems with it. Among children who had used AI for emotional conversations, one in four said that AI sometimes understood them better than most people did.

UNICEF is seeing the same shift internationally. Data released this summer from ten countries suggest that children are adopting AI at rates more than three times faster than adults in some places. Millions are already using it for schoolwork; others are asking it about things that worry them. UNICEF described the situation in unusually stark terms: a generation is effectively growing up inside an experiment whose developmental consequences we are only beginning to understand.

And yet much of the adult conversation still seems stuck on a narrower question:

Are kids using AI to cheat?

Of course some are.

But I think cheating may eventually look like one of the least interesting things AI changed about childhood.

The disappearing space between question and answer

Children do something adults rarely get credit for anymore: they spend an enormous amount of time not knowing things.

They don't know why the moon follows the car.

They don't know how to spell necessary.

They don't know why their friend stopped speaking to them at lunch.

They don't know how to begin the first paragraph of an essay.

They don't know what they are good at, what other people think of them, or quite what to make of themselves.

Growing up has always involved moving through this enormous landscape of uncertainty.

And adults have built institutions around helping children navigate it. Families. Schools. Libraries. Coaches. Religious communities. Friends. Older siblings. Eventually, the wider world.

AI introduces something genuinely new into that landscape.

It can stand between the question and nearly every traditional source of an answer.

That is why I don't think of generative AI simply as another technology children happen to use. Television competed for their attention. Search engines reorganized access to information. Smartphones placed a network in their pockets.

Generative AI can participate in the act of interpretation itself.

A search engine can show a teenager ten pages about heartbreak.

A chatbot can say, Here's what I think is happening between you and your friend.

That is a different kind of technology.

Childhood needs a certain amount of friction

We normally treat friction as something to eliminate.

Good technology is fast. Convenient. Intuitive. It gets us from intention to result with as little resistance as possible.

That makes perfect sense for adults trying to book a flight or summarize meeting notes.

I'm less convinced it always makes sense for a developing mind.

A child wrestling with a sentence is doing more than inefficiently producing words. He is discovering how thoughts fit together.

A student staring at a difficult math problem is not merely waiting for the correct answer. She is learning what to do when the answer doesn't immediately appear.

A teenager trying to decide whether an argument makes sense is building judgment.

Even boredom has developmental work hidden inside it. Left without immediate stimulation, children invent games, bother their siblings, stare out windows, make things, daydream and occasionally discover something worth thinking about.

I think we need a better name for this.

Call it developmental friction: the small difficulties, delays, uncertainties and failures through which capacities are built.

Not all friction is valuable. A terrible textbook does not build character. Neither does making information unnecessarily inaccessible.

But some friction is the exercise.

You don't strengthen a muscle by designing a machine that lifts every weight for you.

And we may soon need to learn the cognitive equivalent.

Children already see the trade-off

This month, another Common Sense Media survey produced a finding I find more interesting than the usual warnings about cheating.

Seventy percent of teenagers surveyed said they use AI for schoolwork. Among AI users, 38 percent said AI caused them to generate fewer of their own ideas. Thirty-nine percent said they sometimes felt they were missing out on learning when AI completed assignments.

In other words, teenagers themselves can feel the trade-off.

At the same time, two-thirds said AI helps them understand their schoolwork. This is why simple arguments about banning or embracing the technology are inadequate. It can genuinely teach while also making it remarkably easy not to learn.

A student who asks AI:

Write my argument about Macbeth.

and a student who asks:

Here is my argument about Macbeth. What is the strongest objection to it?

are technically doing the same thing.

They are both "using AI."

Developmentally, they may be doing almost opposite things.

One is outsourcing the difficult part of thinking.

The other is adding resistance to it.

That distinction matters far more than whether an AI tool appears somewhere in the process.

We may be teaching children a new form of dependence without noticing it

There is another possibility that deserves more attention.

For most of history, expertise was attached to people.

The person who knew something had a face, a reputation, limitations and often a relationship with you.

A teacher could be wrong.

A father might say, "I don't know."

A physician could explain why she believed one diagnosis was more likely than another.

A friend giving relationship advice brought her own history into the conversation.

Children gradually learned that knowledge comes from somewhere.

AI makes that provenance less visible.

An answer arrives fluent, immediate and strangely detached from the human chain that produced the knowledge behind it.

That creates a developmental challenge that goes beyond misinformation.

Children may need to learn something previous generations rarely had to practice deliberately:

how to live with an answer without automatically surrendering judgment to the thing that produced it.

That sounds straightforward. It isn't.

Adults struggle with it too.

UNICEF's updated guidance on AI and children now treats transparency, explainability, children's development and AI literacy as core requirements of child-centred AI—not peripheral technical concerns. It also specifically addresses the emergence of AI companions, recognizing that artificial intelligence is moving into social and emotional territory, not merely educational software.

This may become one of the defining literacies of the next generation.

Not simply knowing how to prompt a machine.

Knowing when not to ask it.

The most consequential AI decisions may happen at the kitchen table

Much of today's AI debate understandably focuses on governments and corporations.

What should companies be allowed to build?

What data should they collect?

Should schools permit AI?

How should platforms protect minors?

Those questions matter enormously. Children cannot be expected to defend themselves against systems deliberately engineered by organizations with vastly greater technical and economic power.

But there is another layer of AI governance that will happen much closer to home.

It will happen when a child says, "Can I use AI for this?"

And an adult has to decide what this actually is.

Using AI to explain photosynthesis may be different from using it to write the lab report.

Using AI to practice French may be different from having it compose the assignment.

Asking for information about anxiety may be different from allowing a chatbot to become the primary place a lonely teenager goes to feel understood.

The interesting boundary is not AI versus no AI.

It is assistance versus substitution.

What is this technology helping the child become capable of doing?

And what is it quietly making unnecessary for the child to learn?

Those questions are harder than setting a screen-time limit. They require us to think about what childhood is for.

Every civilization decides what children must learn to do for themselves

That has always been true.

A child growing up in a hunter-gatherer society had to acquire abilities that many adults today could not replicate. Industrial societies reorganized childhood around clocks, classrooms, literacy and increasingly specialized knowledge. The digital era made information retrieval almost effortless and created entirely new demands around attention and media literacy.

Every environment cultivates some capacities and allows others to atrophy.

AI will do the same.

Perhaps children who grow up with it will become extraordinarily good at asking questions, synthesizing information and collaborating with nonhuman intelligence.

Perhaps some forms of rote learning will finally deserve to disappear.

Perhaps individualized AI tutors will give children who currently lack access to excellent teachers opportunities that would once have been unimaginable.

Those possibilities are real.

But progress rarely arrives without exchanging one set of abilities for another.

So the interesting question is not whether children will be better or worse because of AI.

Better at what?

Worse at what?

And which abilities are so important to human agency that we should preserve the effort required to develop them, even when a machine can make that effort unnecessary?

I keep returning to a simple example.

Imagine two children facing a difficult problem.

The first asks AI for the answer.

The second struggles with it for ten minutes, produces a bad answer, realizes why it is bad, tries again, and then asks AI to challenge the result.

The second child has used more artificial intelligence.

But the first may have surrendered more of his own.

That is the paradox adults will increasingly have to navigate.

The first AI generation

We tend to describe generations by the technologies they encountered: the television generation, the internet generation, the smartphone generation.

But artificial intelligence feels different to me.

Those technologies largely changed what surrounded childhood.

AI has the potential to participate in how children think, create, decide and understand themselves.

Which means the most important question may not be how much AI children use.

It may be what remains when the AI is removed.

Can the child still form an argument?

Can she tolerate not knowing?

Can he create before asking what has already been created?

Can she recognize when an authoritative answer deserves skepticism?

Can he sit with a difficult emotion long enough to decide which human being he should trust with it?

Can they imagine something before requesting ten possibilities from a machine?

These aren't nostalgic questions. I don't want children raised for a world that no longer exists.

Quite the opposite.

Children growing up now will probably need to become far more sophisticated users of artificial intelligence than most adults are today.

But sophistication may eventually mean something different from fluency.

Perhaps the truly AI-literate person will know not only how to extract extraordinary capability from a machine, but also which parts of being human are valuable precisely because they remain inefficient, uncertain and difficult.

The first generation raised with AI is already among us.

We don't yet know exactly what the technology will make possible for them.

The more urgent question is what we want them to remain capable of doing without it.