How do I teach my kid about AI when I do not understand it myself?

You do not need to understand how a neural network works to teach your child good judgment about one. You need to model the behavior you want: checking, doubting, and saying "I do not know" without embarrassment. Not knowing is not the obstacle here. Pretending to know is.

The thing you actually have that they do not

Your teenager can probably drive the tools faster than you. What they usually lack is the instinct that a confident source can be wrong, which most adults acquire slowly and painfully through experience — a builder's quote, a news story they later saw corrected, a doctor who turned out to be behind the literature.

That instinct is the whole subject. The technical layer is a detail, and it changes every eighteen months anyway.

Four things you can do this week

Ask it something you already know the answer to. Sit down together and ask a model about your own job, your own town, a family event, a hobby you have done for twenty years. Watch what happens. You will find errors, and you will find them at a level of confidence indistinguishable from the correct parts. That five-minute demonstration teaches more than any lecture about hallucination, because your child watched you catch it.

Ask for sources and then check one. Pick a citation and look for it. Sometimes it is real. Sometimes the authors are real, the journal is real, and the paper does not exist. That is a memorable afternoon.

Ask the same question twice in fresh sessions. Notice that the answer changes. Ask your child what that tells them about how the thing works. They will usually get there on their own, and an idea they arrived at is worth ten they were told.

Say "I do not know, let us find out" and then actually find out. This is the entire method. It is also the thing most adults will not do in front of their children.

Three questions that work at any age

These are worth having ready when AI comes up at dinner, in a film, or in an argument about homework.

How would we check that? Applies to a chatbot answer, a video, a screenshot, a friend's claim about what the school said.

Who would be responsible if that were wrong? Moves the conversation from technology to accountability, which is where it belongs.

What did it not say? Models produce plausible middles. They are much worse at telling you what is missing, and noticing absence is a high-order skill.

What not to do

Do not make it a scare topic. A teenager who has been told AI is dangerous, and who then uses it and finds it helpful, concludes you do not know what you are talking about — and stops bringing you the harder questions.

Do not outsource it entirely to the school. Most schools are still writing their policy, and the policy is about conduct rather than judgment.

Do not fake competence. If you bluff about how it works, you will be corrected by a fourteen-year-old who read something, and you will lose the standing you need for the conversations that matter more.

If you want something structured

There is a point where "ask good questions at dinner" runs out, usually when your child starts using these tools for real work. That is what a course is for. ZEROTH is built so that the adult does not have to be the expert: you install it once, hand out the reading, and the material does the teaching while you stay the person who asks the questions. World 1 is free and takes about an evening — including for you.