Wovi

How to teach AI to kids · Ages 9–13

Teach one visible idea.
Let evidence do the explaining.

You do not need an AI lecture or a pile of new tools. In this free 20-minute lesson, a learner predicts what a small system will do, changes one thing, reruns the same test and explains the result.

Use the 20-minute lesson Compare both projects

Wovi adult guide · Updated August 13, 2026 · Free browser activities

ONE LEARNING LOOP

Ask before you explain

1 What do you predict?

2 What evidence did the system use?

3 What one thing will you change?

4 What should a person still decide?

LEARNERAges 9–13No prior AI or coding
ADULTParent or educatorNo technical expertise needed
TIME20 minutesOne complete evidence loop
MATERIALSBrowser + four promptsNo account or download

A practical mini lesson

Four phases.
Twenty focused minutes.

Choose one activity. Keep the system and test cases fixed so the learner can connect a deliberate change with its consequence.

0–3 MIN

Predict

Show the goal and visible inputs. Ask what the system will do and why. Record the learner's words before running it.

Do not explain the answer yet.
3–9 MIN

Run and inspect

Play the first attempt. Ask which training example, score or rule produced the surprising result.

Point to evidence on the screen.
9–15 MIN

Change one thing

Add one missing example or strengthen one safety rule. Rerun exactly the same test and compare.

One change keeps cause visible.
15–20 MIN

Explain and limit

Ask why the result changed, which case might still fail and when a human should review the decision.

End with a limit, not “AI is solved.”

Choose one lesson route

Examples or goals?
Both reveal a different AI idea.

ROUTE A · MACHINE LEARNING

Teach with Reef Rescue

Best when you want to discuss how examples shape classification and why accuracy can hide different kinds of mistakes.

  • Ask which example is missing
  • Separate false alarms from missed risks
  • End with the need for human review
Open the free lesson Read the concept guide →
ROUTE B · AI AGENTS + CODING

Teach with Agent Maze

Best when you want to discuss goals, reward tradeoffs, safeguards and how a visible rule maps into Python.

  • Predict the route before running
  • Find the underweighted safety rule
  • Distinguish fixed scoring from reinforcement learning
Open the free lesson Read the concept guide →

Teach responsible AI

Keep the learner curious.
Keep a person responsible.

Responsible use is not a warning added at the end. It appears in the questions asked throughout the activity.

  • What data or observation did the system use?
  • Which people or cases might be missing?
  • What harm follows from a false alarm or missed case?
  • When should the system stop for human review?
Read the current safety and privacy guide
WHAT TO LISTEN FOR

Evidence of understanding

Instead of: “The AI wanted to take the shortcut.”

Listen for: “The shortcut had the highest score because the storm penalty was too small.”

Then ask: “Would that rule work in every new map?”

Precise language helps children avoid treating a small system as a person or an oracle.

Continue the learning path

One lesson can open
several deeper routes.

AI projects for kidsUse both activities as complete predict-build-test-explain projects →AI coding and Python for kidsMove from visible agent rules to variables, conditions and classes →AI literacy curriculumSee how computer foundations, models, agents and responsible evaluation connect →

Teaching AI, clearly explained

Questions from parents and educators

How should I teach AI to a child?

Start with one visible decision. Ask the child to predict an outcome, change one example or rule, test the same case and explain the result. This builds a mental model before introducing more vocabulary or tools.

What age is this AI lesson for?

The current Wovi lesson routes are designed first for ages 9–13. Adults should adjust reading support, discussion depth and session length for the individual learner.

Do I need to know AI or Python first?

No. Each activity supplies a fixed scenario, visible evidence and discussion prompts. The adult's role is to ask what changed and why, not to provide a technical lecture or debug open-ended code.

Can I teach AI to kids for free?

Yes. Agent Maze and Reef Rescue are free browser prototypes with no required account. The 20-minute lesson structure on this page uses either activity and requires only a way to note a prediction and explanation.

How do I teach responsible AI use?

Keep a person responsible for the goal and final decision. Ask what data or observations the system used, which cases might be missing, who could be affected by an error and when the system should stop for human review.

Founding families

Want more ready-to-use AI lessons?

Join the adult early-access list for new lesson worlds, learning notes and optional educator or family research.

Join early access Free to join · Parent email only · Unsubscribe anytime