Is it still my child’s work if AI helped make it?
Sometimes yes, sometimes no, and the test is close to the one for homework: could they explain every decision in it, and would anyone be surprised to learn how it was made? A child who described a character twelve different ways until one matched what they had pictured made something. A child who typed "cool dragon" and posted the first result found something. The work is in the deciding, not the rendering.
Free to read. No account, nothing to sign up for.
What we teach: the tool renders; the taste is theirs.
The test, and the one way it differs from homework
The two questions are nearly the same. Could they explain every decision in it? Would anyone be surprised to learn how it was made? Pass both and it is theirs; fail the second and the problem is honesty rather than authorship.
The difference is what counts as a decision. On an essay, the decisions are the thinking. On a picture, a song, or a short film, the decisions are choices of taste: this composition and not that one, warmer light, a face that reads as tired rather than sad. Those are real decisions and they are hard to make well, which is why "I only wrote the prompt" undersells some work and flatters other work enormously.
A useful way to put it to a child: if somebody asked you to change one thing about this, would you know what you would change? Somebody who authored it always does.
Where the work actually happens
Watch the difference between two children asked for a book cover. The first types "fantasy book cover", takes the first image, and is finished in ninety seconds. The second starts with a mood, tries six versions, notices the castle keeps pulling the eye away from the figure, moves the figure forward, changes the palette twice because the first two felt like somebody else’s book, and stops when it matches the thing they had in their head before they started.
Both used the same tool for the same task and only one of them made anything. The second child rejected far more than they accepted, and that rejecting is the work. It is where their taste shows up.
This is why the number of prompts is a better signal than the quality of the final image. One prompt usually means the tool decided. Fifteen usually means the child did.
The genuinely grey middle
Plenty of cases do not resolve cleanly, and it is worth saying so rather than pretending the test is sharper than it is.
A child who had a clear idea, could not get the tool anywhere near it, and settled for something they quite like has made something, but less of it is theirs than they think, and being honest about that is more useful than a ruling. A child who stumbled into a result they never imagined and immediately recognised it as right has done something real too; recognising is a creative act, even when it is not planning.
What matters is that they can tell you which one happened. The children who get into trouble are not the ones in the grey middle, they are the ones who have not noticed there is a middle.
What to ask to find out
"What did you try that did not work?" Anybody who authored something has a list. Anybody who accepted the first result has nothing to say.
"Why this one?" The answer separates a choice from a default. "It looked cool" is a default; "the other ones made her look angry and she is supposed to be tired" is a choice.
"What would you change if you had another hour?" Authors always have an answer, usually a specific one, and often one they are slightly annoyed about.
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.