What can my child actually make with AI?
Images, video and audio, custom assistants that follow rules they wrote, automations, and working apps from a plain-language description. That list is real rather than aspirational. It is the Creation unit of the program, lesson by lesson. What matters more than the list is that every item is something a child directs rather than requests.
Free to read. No account, nothing to sign up for.
The short list, and what each one actually involves
Images. Describing something precisely enough that it comes back the way you pictured it, which sounds easy and is the whole skill. Then directing it: style, mood, light, composition, the way an art director does.
Video and audio. Shorter pieces, but the same job: deciding what it should feel like and getting the tool there, rather than accepting what it offers first.
A custom assistant. The child decides what it is for, who it is for, what it must never do, and writes those rules themselves. This is usually the one that surprises parents, because the interesting part is not technical, it is deciding what the thing should be.
Automations and agents. Making a sequence of steps happen without being asked each time, and finding out where that goes wrong.
A working app, described in plain language and built from the description. Not a toy demo, but something with a purpose they chose.
Why "made" and "generated" are different words
Everything above can be done in one lazy step or twenty deliberate ones, and the list does not distinguish between them. "My kid made an app with AI" covers both a child who described a real problem and iterated until it worked, and a child who accepted the first thing that compiled.
The difference shows in whether they can say why it is like that. Not how it works, but why it is that way and not another way. Anything on this list is worth doing when the answer is specific, and is just tool-tourism when it is not.
So the useful question after "what can they make" is "what did they decide".
Start with one thing, not a tour
The common failure is trying all of it in a weekend. Six tools, six first results, nothing finished, and a child who now believes they have "done AI".
One project, carried to something they would show somebody, teaches more than the tour, because everything worth learning here lives in the second half, after the first result and before the thing is actually good. A tour never reaches that half.
Images are usually the best starting point: the feedback is immediate and obvious, and a child can see straight away that their description was vague.
A first project worth trying this week
Ask them to make a cover for a book they have read, with one condition: someone else who has read it should recognise it with the title removed.
That condition does the teaching. Generic covers fail it immediately, and the only way past is to work out what specifically matters about that book, which is a question about the book rather than about AI. Then they have to get the tool to render that, which is where the describing gets good.
Half an hour, no account beyond whatever you already use, and by the end you will both know whether they are directing the thing or taking what it offers.
IF YOU WANT TO GO FURTHER
This answer comes from the program we teach: sixteen video lessons that take a student from “what even is this” to directing AI on purpose. Parents get their own access alongside their child, because it is hard to guide something you have not seen.
The first lesson is free to watch, no account needed. Start there before you decide anything.