The Story of AI
Robots did not pop up all at once. It took people nearly eighty years, one small idea at a time. Climb into the time machine!
The Big Question
Long before phones and tablets, a clever man called Alan Turing asked a strange question. Could a machine ever think? Back then machines could only add up numbers. People laughed! But Alan wrote down a plan for a thinking machine. That plan started the whole story.
What did Alan Turing wonder about machines?
👀 For grown-ups: what really happened here?👀 For grown-ups
Turing's 1950 paper "Computing Machinery and Intelligence" sidestepped the unanswerable question "can machines think?" and replaced it with a test: could a machine hold a conversation well enough that a person could not tell it apart from another person? That swap — behaviour instead of inner experience — still shapes how AI is judged today.
The whole story, scene by scene
How was AI invented? This free history tells children aged 4–8 the real answer in 11 scenes, from the question Alan Turing asked in 1950 to the AI helpers they talk to today. Every scene reads aloud, earns one big word, and carries a note for grown-ups with the history behind it.
1950 — The Big Question
Big word: Machine — a thing people build to do a job for them
Long before phones and tablets, a clever man called Alan Turing asked a strange question. Could a machine ever think? Back then machines could only add up numbers. People laughed! But Alan wrote down a plan for a thinking machine. That plan started the whole story.
👀 For grown-ups: what really happened here?
Turing's 1950 paper "Computing Machinery and Intelligence" sidestepped the unanswerable question "can machines think?" and replaced it with a test: could a machine hold a conversation well enough that a person could not tell it apart from another person? That swap — behaviour instead of inner experience — still shapes how AI is judged today.
1956 — The Naming Party
Big word: Artificial Intelligence — a machine made by people that can learn and solve problems
Six years later, some scientists met at a big house for the whole summer. They wanted to build Alan's thinking machine. But first it needed a name! They picked two words: Artificial Intelligence. Everybody just says A.I. for short.
👀 For grown-ups: what really happened here?
The 1956 Dartmouth Summer Research Project coined the term "artificial intelligence". The organisers believed a couple of months of work by ten people might crack it. It took about seventy years and a planet's worth of computing power instead — a useful reminder of how badly experts can misjudge these timelines.
1966 — The Copycat Computer
Big word: Chatbot — a computer program you can type messages to
A computer called Eliza was the very first chatbot. Eliza knew one sneaky trick. She turned your own words into a question and handed them straight back to you! People chatted to her for hours. Some thought she really understood them. She did not. It was all copying.
👀 For grown-ups: what really happened here?
ELIZA, written by Joseph Weizenbaum, was a few hundred lines of pattern matching with no understanding at all. Weizenbaum was alarmed by how quickly people confided in it — the "ELIZA effect", our habit of reading a mind into any system that produces fluent language. It is exactly the trap modern chatbots set, at enormous scale.
1972 — The First Game Against a Computer
Big word: Rules — the little "if this happens, do that" steps a robot follows
In 1972 people crowded into a new kind of room called an arcade. On a screen, two little bats chased a bouncing ball. For the very first time, you could play a game against a computer! The computer never got bored and never missed. It just watched the ball and moved its bat — tiny rules, unbelievably fast.
👀 For grown-ups: what really happened here?
Pong paddles were pure reactive rules: mirror the ball’s horizontal position, track its vertical one, and move one step per frame. That sense–decide–act loop is the skeleton under arcade enemies, game bots and self-driving cars. No learning, no planning — just fast, simple rules.
1997 — The Chess Battle
Big word: Search — trying loads of choices to find the best one
A computer as big as a fridge, called Deep Blue, played chess against the best chess player on the whole planet. And Deep Blue won! It was not clever like a person though. It just checked a mountain of moves, unbelievably fast, and picked the best one.
👀 For grown-ups: what really happened here?
Deep Blue beat world champion Garry Kasparov by brute-force search plus hand-written chess rules — no learning involved. It is the clearest example of a machine reaching a human-level result by a completely inhuman method, which is why "it beat a person at X" tells you much less than it sounds like it does.
2002 — The Robot That Cleans the Floor
Big word: Robot — a machine that senses the world and does things by itself
In 2002, a round robot called Roomba waddled into real homes. It does not have a map. It just rolls along, feels the carpet, and turns when it bumps into a wall. It is not clever — it is very, very patient. It keeps sweeping until it has been everywhere, then finds its charger and parks itself.
👀 For grown-ups: what really happened here?
The Roomba (iRobot, 2002) is a simple autonomous agent: bump and cliff sensors plus a randomised coverage pattern. It makes millions of tiny sense–decide–act choices with no understanding of "clean" or "done" — a person manages the goal, the machine manages the grime.
2011 — A Voice in Your Pocket
Big word: Voice — sound made by people that a robot can learn to recognise
In 2011, a new thing inside a phone could understand your VOICE. You talked, and it turned your words into writing, worked out what you wanted, and answered you out loud. It was the first time lots of people talked to a computer like it was a friend. Talking is one trick in two halves: hearing your words, and choosing its own.
👀 For grown-ups: what really happened here?
Consumer voice assistants like Siri (2011) are a pipeline: an acoustic model turns audio into text, a slot-filling parser works out intent, and text-to-speech answers back. Voice is not one skill but several trained systems chained together — the same "hearing" and "talking" halves a child can actually practise with Robot Miles.
2012 — Robots Open Their Eyes
Big word: Training — showing a computer thousands of examples until it gets good at guessing
For ages, computers were terrible at looking at pictures. Then people tried something new. They showed a computer millions of photos and told it what was in each one. Cat. Dog. Bus. Banana. After millions of tries, it started getting them right. That is called training.
👀 For grown-ups: what really happened here?
In 2012 a neural network called AlexNet won the ImageNet contest by a huge margin, and computer vision changed overnight. Nothing about the maths was new — what was new was lots of labelled photos and graphics cards fast enough to train on them. Modern AI mostly runs on that same recipe: old ideas, enormous data, enormous compute.
2022 — Robots Learn to Chat
Big word: Predict — to guess what comes next
Then came the biggest surprise. People fed computers almost every book, story and web page ever written. The computer learned which words like to sit next to each other. Now it can write you a story or answer your question — by guessing the next word, over and over, super fast.
👀 For grown-ups: what really happened here?
Large language models are next-token predictors trained on internet-scale text. Everything they appear to know is a side effect of getting very good at that one prediction task. They have no beliefs and no way to check themselves, which is why they can be fluent and confidently wrong in the same sentence.
Today — Helpers Everywhere
Big word: Helper — something that does a job to help a person, not instead of them
AI is not stuck in a laboratory any more. It is all around you, quietly helping. It is not alive. It does not have feelings, and it does not want anything. It is a very fast helper — and there is always a person in charge of it.
👀 For grown-ups: what really happened here?
Each of these is narrow AI: a system trained for one job that has no idea the other jobs exist. The important safety point for children is the last one — deployed AI always sits inside a human process, and a person remains accountable for what it does.
Next — Your Turn
Big word: Choice — deciding what should happen — always a job for a person
Nobody knows the end of this story, because it has not happened yet. The people who build the next bit are kids like you. Remember: AI only ever does what people teach it. So the big question is not "what can AI do?" The big question is "what should we ask it to do?"
👀 For grown-ups: what really happened here?
Ending on agency rather than prediction is deliberate. Children hear a lot about what AI will do to them; far less about the fact that every dataset, rule and deployment decision was made by a person. That framing is what makes the ethics questions feel answerable instead of frightening.
Next, play the hands-on AI missions, build a robot in Train Your Robot, or find lesson plans and printables on the parents and teachers page.