How it was built3 of 11

Where AI learned everything

Everything an AI can do, it learned from things people made. Most of those people were never asked and were never paid.

How old is your kid?

How to explain it, by age

Same idea every time. What changes is how much of the machinery a kid can hold, so pick the version that fits and go up a band when it stops being interesting.

Ages 4-6

Use the copying idea she already has. "The computer learned to draw by looking at millions of pictures other people drew. Nobody asked those people if that was okay first." She already knows about asking before you use someone's things. This is the same rule, bigger.

Try this

Ask her to draw something, then pretend to copy her style and call it yours. Watch the objection arrive. Then say that is what a lot of artists feel about this, and leave it there.

Listen for

Any version of "you have to ask first". That is the entire concept at this age.

If she already has more access than most. If she has used an image generator, ask her whose drawings she thinks it learned from.

Ages 7-9

Add the scale and the people. "It learned by reading books, websites, and looking at art made by real people, millions of them. Most of them did not know it was happening, and most were not paid. Some are pretty upset about it. Some think it is fine because that is how people learn too, by looking at other people's work."

Try this

Generate a picture in the style of an illustrator she loves. Then look up that illustrator, see their actual work, and ask her what she thinks they would say about it. Let her land wherever she lands.

Listen for

Her holding both sides at once. That artists are upset and that people also learn by copying. Not resolving it is the correct outcome.

If she already has more access than most. Ask what she thinks a fair arrangement would look like, and take the answer seriously.

Ages 10-12

Introduce the actual argument. "Companies used enormous amounts of writing, art, music and video to build these systems, generally without asking. Their position is that this is like a person learning from everything they have read. The other position is that a company making billions from other people's work without paying them is straightforwardly unfair. Courts are deciding this right now."

Try this

Have her argue the company's side out loud, properly, then argue the artists' side. Making her build both cases is the exercise. She will find the company's argument better than she expected, which is uncomfortable and useful.

Listen for

Her identifying where the analogy to human learning breaks down. Scale and commercial gain are the usual answers and both are good ones.

If she already has more access than most. Introduce the fact that some of the training data was itself scraped illegally, and that "we already built it" has become an argument in its own right.

Ages 13+

Go to the labour underneath. Beyond the training data, these systems were shaped by large numbers of low-paid workers, often in Kenya, the Philippines and India, rating outputs and filtering the worst material on the internet so the model would not repeat it. That work is genuinely traumatic and largely invisible. Any honest account of how these things were built includes it.

Try this

Have him research who does content moderation and data labelling for one major AI company, and what they are paid. The gap between the public story and the supply chain is the lesson.

Listen for

Him connecting this to other supply chains he already knows about. Clothing and electronics are the obvious parallels and the reasoning transfers.

If she already has more access than most. Ask what he would require of a company before calling its model ethically built, and whether any current one would meet that bar.

Why children get this faster than adults

Every child has been taught, repeatedly, that you ask before using someone else’s things and that copying someone’s work and calling it yours is wrong. Those rules are enforced daily at school.

So when you explain that a company built a product by absorbing millions of people’s work without asking, they do not need the argument explained. They already have the framework. Their instinct is usually sharper than the adult version, which tends to arrive pre-tired.

Let them disagree with you

There is a real argument on the other side, and a child who only hears one half will be poorly equipped the first time they meet the other.

Humans do learn by absorbing other people’s work. Every writer read first. The question is whether doing it mechanically, at enormous scale, for commercial gain, is the same act or a different one. That is a genuinely hard question and smart people land in different places.

Say so. A kid who knows this is contested will keep thinking about it. A kid who was handed a verdict will stop.

Misconceptions to address

That the model contains copies of the things it read. Mostly it does not. It holds patterns learned from them, which is why the argument about whether that counts as copying is genuinely difficult rather than obvious.
That everything on the internet is free to use because it is public. Visible and permitted are different things, which is a rule most children already understand from school.
That this is a settled legal question. It is being fought in courts right now and the answers are not in yet.
Adults get this wrong by assuming someone must have checked. In many cases nobody did.

Running this with a class

Have every student contribute one drawing to a class collection, then announce that you will use the collection to produce new drawings in the class style, which you will then put your own name on. Let the reaction happen. The discussion afterwards writes itself, and students arrive at the fair-use and consent arguments on their own.

Last reviewed August 2026