AIQ
AIQ
Part of the Neural Networks lesson guide. Teaching a different grade? 🔧 Builder (8–10) · 💻 Hacker (11–14) · ⚡ Architect (15–18)
Gather the class in a circle on the floor if you can — this lesson works best with a little wiggle room. Start by tapping your own head and asking:
"Everybody point to your head! What's in there that helps you think, and blink, and even giggle?" Let a few kids shout "Brain!" Nod big.
Then say, in your own words, the app's opening line: "Inside AI — that's the smart computer helper — there's something that copies how YOUR brain works! But it's not really a brain. It's lots and lots of tiny math helpers, passing notes to each other!"
Hold up four fingers and count off the four words the app shows on screen when this lesson opens: Network 🕸️, Brain 🧠, Layers 📊, and Signals ⚡. Say each word, have the class echo it back, and do a tiny action for each — spread your fingers like a web for "Network," tap your head for "Brain," stack your hands like pancakes for "Layers," and do a quick jazz-hands zap for "Signals."
Ask: "Do you think a computer has a real brain inside it, like you do?" Take a few guesses without correcting yet — you'll come back to this exact question at the end of the lesson, so it's fine to let the mystery sit for now. Say: "Let's go find out what's really inside!"
One more warm-up game before you dive in: say "I'm going to whisper a secret to the first friend in a line, and we'll see what happens by the time it gets to the end!" Pick five volunteers, whisper a short silly word (like "banana" or "wiggle") to the first child, and have them whisper it down the line to the last child, who says it out loud. It probably won't come out exactly right, and that's the fun part — say "See how a message changes a little bit as it passes from friend to friend? That's kind of like what happens inside AI, except the friends are tiny math helpers, and they're trying to help the message get MORE right instead of sillier!"
Walk the class through the lesson's three scenes. Keep every explanation short, physical, and full of "let's pretend."
Say: "Your brain is full of tiny helpers called neurons. There are SO many of them — grown-up scientists counted about 86 billion! That's way more than all the people in the whole world put together." Pause for the gasp. "These tiny helpers pass zap-zap-zap signals to each other, superfast, and that's how you think and move and laugh."
Then: "AI's brain helpers are inspired by your brain, but they're not real. They're pretend copies — tiny math machines. Instead of a real zap, they just pass along a number. And each one only does ONE small job." Ask: "If you only had ONE small job, what job would you want?" Let a few kids answer silly jobs (expect things like "eating snacks" or "petting the dog") — laugh with them, then say: "A math helper's job is much more boring than that — it just takes a number in, does a tiny bit of math, and passes a new number out!"
Do a tiny "be the math helper" warm-up: hold up two fingers and say "Let's pretend this is my number — 2!" Whisper a made-up rule to one volunteer, like "add 1," and have them shout the new number. Then whisper "add 1" to the next volunteer, who takes that new number and adds one more. After three or four volunteers, point out: "See? Each helper only did one tiny job — add one little number — but by passing it along, we ended up somewhere totally different from where we started! That's the whole secret of ALL of AI's brain helpers, just with different tiny math jobs each time."
Then talk about connections: "Each little helper is joined to the next by a connection — like a string between two cups. Some connections are extra important, and some are barely important at all." Hold up a string loosely and say "this connection matters just a little," then pull a different string taut and say "this one matters a LOT." Explain: "AI calls that importance a helper's 'weight' — grown-ups will hear this word a lot when people talk about AI, so now you already know it means how important a connection is!"
Line up four or five volunteers shoulder to shoulder, facing the class, to be "layers." Say: "Imagine you're all math helpers standing in a line, and I show the first friend a picture of a cat." Whisper "cat" to the first student. "The first friend's job is just to notice the picture — that's the input layer." Have that student pass a clap down the line.
"The friends in the middle are the hidden layers — their job is to notice little clues. Does it have pointy ears? Does it have whiskers? Does it have a tail?" Have each middle student say one clue out loud as the clap passes to them.
"The very last friend is the output layer — their job is to shout the final answer!" Have the last student shout "It's a CAT!" Ask the class: "What do you think the last friend does if they're not totally sure?" (Expected response, with a nudge: "Say how sure they are" — connect this to the app saying things like "95% sure it's a cat.")
Pause here and check for understanding before moving on: point to each volunteer in your line in turn and ask the rest of the class "What was this friend's job again?" Only move to Scene 3 once most hands go up confidently for each one — noticing the picture, spotting clues, and shouting the answer.
Say: "Now let's pretend the last friend guesses wrong — they shout 'It's a DOG!' but it was really a cat." Ask: "Uh oh — what should happen now?" Take guesses.
Explain: "Here's the surprising part. The 'oops' travels BACKWARDS down the whole line of friends, and each friend adjusts their job just a tiny, tiny bit so next time they'll do better." Have your line of volunteers physically turn around and pass an "oops" tap backward, person by person, back to the first friend. "This tiny fixing happens millions and millions of times, until the helpers get really, really good at guessing right."
Have the volunteers sit down and summarize together as a class: math helpers pass a guess forward, and if it's wrong, the fix travels backward. That's the whole secret.
Do a quick round with a new picture to check understanding: whisper "ball" to a new first volunteer and run the whole line again, forward guess and all. If the class shouts the right answer, celebrate loudly — "You just did what AI does millions and millions of times!" If a "wrong" guess happens on purpose, walk through the backward "oops" pass one more time so every child sees the full forward-then-backward cycle at least twice before moving on.
Sit the class back in a circle for this part. There are no wrong answers here — the goal is just getting kids to say the ideas out loud in their own words.
Close by returning to the opening question: "So — does a computer have a real brain inside it?" Let the class answer together: "No! It has tiny math helpers, inspired by a brain, but not a real one." Celebrate that they figured out the mystery.
Extension activity — Human Neural Network Relay: Split the class into two or three lines of five to six students each, standing shoulder to shoulder like in the main activity. Whisper a simple picture idea to the first student in each line (a dog, a ball, a tree). Each student in the middle passes along the clap-signal and adds one made-up "clue word" out loud. The last student in the line shouts their final guess. Then run it in reverse: have the class pretend the guess was wrong, and pass an "oops" tap backward through the same line, with each student calling out one thing they'd "try differently next time." Repeat with new lines and new secret pictures so every child gets a turn as the first, middle, or last "helper."
If time allows, finish with a quick drawing prompt: have each child draw their own "brain helper" — a silly little creature whose only job is one tiny math task, like "add 1" or "say if it's round." Ask a few kids to show their drawing and explain the one job their helper does. This keeps the idea that AI is built from many tiny, simple jobs working together, rather than one all-knowing brain, fresh in their minds after the lesson ends.