Session 1 — What is AI?

Duration: 60 min · Format: live online

What you'll learn: by the end, you can explain what AI is, point to AI you use every day, and describe how a computer learns from examples.

Soft skill focus — Curiosity

Today you'll also grow Curiosity. Every "how does my phone know my face?" question is curiosity doing the work AI can't do for us.

Try this: when you wonder how something like face unlock works, don't jump straight to "it's AI" — stop and come up with two or three of your own theories about how it might work first.

Think about: What is one thing about AI you now really want to find out?

What you'll need

• A device that can open Quick, Draw! in a web browser.
• Paper and a pencil.

Hook

Let's think about a few questions:

• How does a phone know it's your face and unlocks just for you?
• How does a game enemy chase you around the screen?
• How does a video app always know what you'll like next?

Have a guess at each one. The secret behind all of these is one thing — AI. Today you'll find out what that really means.

Teach — What does "AI" mean?

Here's what those two letters mean:

• AI = Artificial Intelligence.
• Artificial = made by people (not natural).
• Intelligence = being smart — learning, deciding, solving problems.
• So AI = a computer that can do smart things — a little like a brain we build.

⚠ Watch for the #1 mix-up: it's easy to think AI is alive or thinks and feels like a person. It isn't — a computer is not alive and has no feelings; it just gets very good at spotting patterns and following examples.

Look at this diagram — it shows places you already meet AI:

Left to right: voice assistants · video suggestions · face unlock · smart game characters · map directions · language translation.

Which of these did you use today? Can you name one that isn't in the picture?

Teach — How does a computer learn?

Computers can learn from examples — this is called Machine Learning. Think of teaching a robot to recognise a cat, and follow this diagram:

Here are the three steps:

1. Show examples — we give it lots of pictures, each labelled ("this is a cat", "this is a dog").
2. It finds patterns — it notices cats have pointy ears and whiskers and slowly learns.
3. It makes a guess — show a brand-new picture and it predicts: "Cat!"

The big idea: the more good examples it sees, the better it guesses.

Think about it: if you only showed it 2 cats, would it be good or bad at guessing? (Bad — too few examples to find a reliable pattern.)

Activity

Activity 1 — Quick, Draw! (≈10 min). Open Quick, Draw! and play a few rounds. It learned from millions of drawings other kids made.

• As you play, notice: When did it guess right? When did it get confused? Why?

Activity 2 — Be the AI (≈10 min, no computer). Find a partner. One of you secretly picks a rule (e.g. "things that are round") and shows 5 examples; the other spots the pattern and predicts the next one. Then swap.

That's exactly what an AI does — find the pattern from examples, then make a guess.

Check yourself

Try to answer these, then check yourself after the arrow.

1. What does AI stand for? → Artificial Intelligence — "made by people" + "being smart".
2. How does a computer learn to recognise a cat? → From lots of labelled examples; it finds patterns (Machine Learning).
3. True or False: an AI thinks exactly like a human brain. → False — it doesn't feel or think like us; it's just good at spotting patterns from examples.

Wrap-up

• Finish this sentence in your own words: "AI is…"
• Try this at home — Pattern Detective: find 3 things at home or school that use AI. For one of them, guess what examples it might have learned from. Bring the list to Session 2.

Tips & extra challenges

• Watch out: it's tempting to think "AI is alive / thinks like us." Really it's just pattern-spotting from examples.
• Want more? Try this: learn three "grown-up" words — model (the "brain" the computer builds from examples), training (the teaching part, when it studies the examples), and label (the correct answer written on each example). Then invent a made-up AI (a "homework-checker", a "lost-cat finder") and describe what examples it would need, what labels those examples would carry, and one time it might get confused. Finish by naming 3 real-world problems an AI could help solve.

Vocabulary

Resources

• Quick, Draw! — draw and watch AI guess (main demo).
• Teachable Machine — you'll train your own model in Session 3.
• Machine Learning for Kids — friendly projects for extension.
• AI Experiments by Google — short demos of what AI can do.

Practice set

Practise on your own — the answer comes after the arrow.

1. What do the two words in "AI" stand for, and what does each mean? → Artificial = made by people; Intelligence = being smart (learning, deciding, solving). Together: a computer that does smart things.
2. Name three places you have met AI today. → Any reasonable answer, e.g. voice assistant, video suggestions, face unlock, map directions, translation, a smart game character.
3. A phone unlocks only for its owner's face. What examples do you think it learned from? → Lots of pictures of the owner's face (labelled "this is the owner") so it learns that face's pattern.
4. True or False: an AI that recognises cats also understands that cats are alive and like to nap. → False — it only matches patterns in pictures; it has no idea what a cat really is or feels.
5. You want an AI to tell apples from bananas but you only show it 2 apples and 2 bananas. What will probably go wrong, and how do you fix it? → Too few examples to find a reliable pattern, so it guesses badly; fix by showing many more labelled examples of each.
6. A friend says "AI is a robot brain that thinks and feels like us." What is wrong with that, and how would you correct it? → It is not alive and has no feelings; it is just very good at spotting patterns and following examples.

Going deeper (optional)

• The "garbage in, garbage out" idea. Imagine every "cat" picture you showed the computer happened to have a red sofa in the background. What might the AI actually be learning — cats, or red sofas? It could learn the wrong pattern and then call a red cushion a "cat". The big idea: an AI is only as good as the examples it learns from, so the examples must be clear and varied. In Session 3 you'll record examples from different angles and backgrounds for exactly this reason.
• Guessing with a percentage. A real AI rarely says just "cat" — it says something like "I'm 90% sure it's a cat." Picture two guesses ("cat: 95%" vs "cat: 52%"): which one is the AI confident about, and which is almost a coin-flip? This is the word confidence, which you'll see for real on the bars in Session 3.

Common mistakes & fixes

• Mistake: thinking AI is alive or has feelings. → Fix: remember the anchor — a computer is not alive; it only spots patterns in examples. It's "smart, but not alive."
• Mistake: believing AI "just knows" things by magic, with no learning. → Fix: go back through Show examples → Find patterns → Make a guess; nothing happens without examples first.
• Mistake: assuming one or two examples are enough to learn from. → Fix: think of the "only 2 cats" question — more good examples means better, more reliable guesses.
• Mistake: calling every computer or gadget "AI" (a calculator, a light switch). → Fix: ask "does it learn from examples and make guesses?" A calculator just follows fixed rules; that is not AI.
• Mistake: mixing up the guess (prediction) with the examples (data). → Fix: look at the diagram — examples/data go in, the prediction comes out; they are different ends of the same process.

What's next

Session 2 — Patterns Everywhere: you become a pattern-spotter and teach the computer to sort things into groups.