How to talk to your child about AI

Whether or not you've ever discussed it, your child already has a theory of what AI is. They built it themselves, out of movie robots, playground rumors, and whatever happened the first time they typed a question into a chat box. The common versions: it's a magic answer machine that knows everything; it's a robot friend; it's the cheating thing we're not supposed to use. All three are wrong, and each one leads a kid somewhere you don't want them to go — the oracle gets believed, the friend gets confided in, the contraband gets used in secret.

The good news: you don't need a computer science degree to give them a better theory. You need one honest idea, delivered at the right altitude for their age.

The one idea to land

Here it is: AI is a machine that has read almost everything, and is very good at guessing what words should come next. It sounds sure of itself even when it's wrong. And it was made by a company — it's a product, not a person.

Every important habit falls out of that one idea. Check what it tells you, because good guessers still guess wrong. Don't treat it as a friend, because it can't actually know you. And remember someone built it, with their own goals — which is a thing worth knowing about everything on a screen, not just AI.

Ages 6–8: make it a game

At this age, skip the mechanics entirely and aim for one gut-level lesson: the confident voice can be wrong. The best way isn't explaining — it's catching the AI in a mistake together, about something your family knows better than any machine.

Ask it about your street, grandma's dumpling recipe, your dog's personality. Watch it answer smoothly and get it wrong. Laugh about it. A seven-year-old who has personally caught the computer being confidently wrong about dumplings is inoculated in a way no lecture achieves.

“It read millions of books, so it's a really good guesser. But guessers get things wrong — want to try to trick it?”

Ages 9–12: the second-source rule

This is when AI starts showing up in homework, so the conversation gets practical. Two habits to build now, while they still think your rules are mostly reasonable:

• Trust, then verify. Anything important the AI tells you gets checked somewhere else — a book, a real site, a teacher, you. Not because AI is bad, but because that's what smart people do with any single source.
• Helpers don't do the work. The line between AI-that-explains and AI-that-does-it-for-you, drawn together, out loud. (We wrote a whole guide on where that line goes.)

“If it told you something wrong and you put it in your project, whose mistake is it — yours or the computer's?”

That question usually kicks off a genuinely good argument. Let it run. A kid who has argued about responsibility for an AI's mistake has thought harder about AI than most adults.

Teens: ask them to teach you

By the teenage years, lecturing is dead on arrival — and honestly, they may know the tools better than you do. So flip it. Ask what they use AI for. Ask them to show you their best tricks. Sincere curiosity gets you into rooms that authority can't open.

Then, inside that conversation, raise the two things teens genuinely need an adult to say out loud. First: what you type into an AI goes to a company. Before you tell it something private, it's worth a beat of “would I be fine with this sitting in a stranger's database?” Second, and more delicate: an AI is an always-available listener that never disagrees, never gets tired of you, and never needs anything back. That can feel like friendship. It's missing what makes friendship work — a real friend can be changed by knowing you, can push back, can be there in a way that costs them something. If AI ever starts feeling like the easiest relationship in their life, that's worth noticing, gently.

“What's something you've figured out about these tools that most adults don't get?”

Keep it a running conversation

None of this is one Big Talk. It's a dozen small ones — at dinner, in the car, over a homework question. A few that reliably go somewhere: Do you think it actually understands you, or just seems to? What's something it would be wrong about? If you could ask it anything and nobody would ever know, what would you ask? That last one usually earns a suspicious look. Ask it anyway. Even unanswered, it plants the right question in their head — what am I telling this thing, and why?