Raising Kids in the Age of AI

Recently I came across a WSJ article that asked a simple question: what do AI executives tell their own kids about the jobs of the future?

The most interesting part was not a prediction about which careers will survive. It was the posture behind the advice.

Metacognitive skills will be very important—flexibility, adaptability, experimentation, thinking critically, being able to challenge things. Developing critical-thinking skills requires friction, doing things that are hard, doing deep thinking. — Jaime Teevan (Chief scientist and technical fellow, Microsoft, and trustee, Yale University)

For a long time, the script felt stable: work hard, learn math, learn to write, build discipline, become good at something useful, and the world will reward you for it. That script no longer feels complete. Not because effort or knowledge matter less, but because the shape of value is shifting. If AI can increasingly explain, summarize, draft, code, brainstorm, and tutor, then the question is no longer just how to help our kids produce better answers. It is how to help them build a mind when answers are cheap.

That question feels harder because parenting was already harder before AI entered the picture. Jonathan Haidt’s The Anxious Generation argues that childhood had already been reshaped by the move from a play-based childhood to a phone-based childhood, with real consequences for mental health, attention, and social development. Parents were already trying to protect childhood from too much screen time, too much passive consumption, and too much phone-mediated life. AI adds another layer.

Now the challenge is not only protecting attention, but also protecting authorship, struggle, and the slow process by which children turn confusion into understanding.

That is why Jaime Teevan’s point about friction feels so important. Some friction is how a child grows. A child trying to find the right words is not just completing an assignment. A child sitting with confusion in math is not just being slowed down. A child revising a weak paragraph is not just producing better work. In those moments, they are building patience, ownership, and trust in their own ability to work through difficulty.

I have already practiced this with my daughter during homework, letting her sit with the hard part a little longer instead of stepping in right away. Now I need to make sure AI does not make it too easy to skip that process.

This is why metacognition matters so much. As I wrote in my last post, metacognition is thinking about your thinking. For a child, that means being able to ask: Do I actually understand this? What part is confusing me? Did I really try first? Why do I believe this answer? Am I reasoning, or am I just recognizing something that sounds right?

That mindset needs to be ingrained in how they think and how they approach every task.

In practice, that mostly comes down to sequence. Let the child try first. Let them feel the confusion first. Let them attempt the paragraph, the explanation, the idea, the problem. Then let AI help refine, expand, challenge, or clarify. The point is not to deny them useful tools, but to protect the part of development that comes from effort before assistance. First effort does more than produce output. It shows the child what they notice, where they hesitate, how they respond to frustration, and whether they can stay with something long enough to make it their own.

In The Anxious Generation, Haidt’s argument is not just that screens are distracting. It is that childhood needs certain conditions to develop well: play, independence, real-world interaction, and room to mature without being overwhelmed by phone-based life. AI does not replace that earlier concern. It compounds it. We are parenting after one major rewiring of childhood, and in the middle of another.

That makes parenting harder, asks more from us: more judgment, more restraint, and more willingness to tolerate our children’s frustration instead of rescuing them immediately. It asks us to care more about process, less about results.

It also changes what I want from education. I still want schools to teach fundamentals, and I still want my kids to read, write, reason, and learn real content. But producing correct answers or polished output is no longer enough. Education has to care more about whether a child can stay curious, notice what they do not understand, tolerate uncertainty, think independently, and keep their voice while using powerful tools. That also means schools need to insert necessary friction by changing how they measure outcomes, rewarding process, reasoning, revision, and ownership, not just results.

If AI makes intelligence easier to access, education has to care more about judgment. If AI makes writing easier to produce, education has to care more about authorship. If AI makes answers easier to get, education has to care more about whether the child is actually becoming someone who can think.

I have a deep, hopeful belief that humans ultimately like to be around and spend time with other humans — Daniela Amodei (President and co-founder, Anthropic)

Daniela’s comment matters for the same reason. Even in a world with AGI, children are still becoming people, not just workers. They still need friendship, conversation, play, character, belonging, and a sense of self. The social part of childhood simply becomes easier to neglect.

I want my kids to learn AI because it will be part of their world. But I do not want fluency with tools to come at the cost of independent thinking and real social development. I want their voice to remain theirs, their judgment to remain theirs, and their chance to become themselves to remain theirs.

That, more than any prediction about the jobs of the future, feels like the real parenting challenge of this moment.

Links

• WSJ article: What AI Executives Tell Their Own Kids About the Jobs of the Future
• The Anxious Generation by Jonathan Haidt: Amazon link