The whole exercise on one page: the world children already live in, the gap that keeps AI feeling like magic, and a simple bridge that crosses it — with proof that it works.
Two things are quietly true at the same time, and the whole idea rests on putting them together.
Phones unlock on a face, videos are recommended, voices answer questions, pictures are generated. For today's child, AI is ordinary — just unexplained.
CT & AI for Grades 3–8 wants every child to be AI-ready — to understand the idea, not just to become a future coder.
Children already sort, guess from clues, learn from examples and fix mistakes. That is the very thinking AI uses. The raw material is in the room.
They meet AI as a finished tool: the output arrives, but the thinking behind it is invisible. So it looks mysterious, or like a flawless robot. And the usual fixes make it worse.
The standard route opens with definitions, formulas and programming. That loses Grades 3–8 on day one — and most schools have no devices and no AI specialist.
Even good lessons explain your brain — how you recognise a face or guess a word — and then stop. They never show the machine doing the same thing.
Every lesson follows one simple loop, and abstraction is only ever named after the child has lived it.
Play it — sort, guess, train, draw.
“How did you know?”
Watch the computer do the same thing.
Now the word: feature, model, layer…
That third step is the bridge. Every one of the six modules carries a “You ↔ The Machine” parallel — for example:
recognise a blurry dog from just a few clues, by matching it to dogs you’ve seen before.
runs a neural network trained on thousands of photos — it fires layer by layer and scores the guess: “dog, 92%.”
One thread runs from the playground to the language model: observe → find patterns → compress → use them to classify, predict or generate.
With a computer (the interactive site) or with paper and movement (cards, drawing, role-play — no computer). Low-cost, every learning style, and it works fully offline.
120 children wrote the same short test (out of 20) before and after three weeks. The section that wasn’t taught barely moved; both taught sections improved far more.
Average marks gained, out of 20 · sample marks, shown to illustrate the comparison · see the full report →