Child’s First AI Conversation — Lesson Plan
Companion to a child’s first AI conversation: a facilitator’s script. Print this, or keep it open on a second screen while you run the session.
Before you start
- Using an adult’s own account, not a new account created for the child.
- The chosen tool and account path are allowed by the product’s current age rules, and no school name, address, photo, or other identifying detail will be typed in.
- 15-20 minutes set aside, with the session ending on the closing reflection, not fizzling out.
Step-by-step facilitator card
1. Name the tool (2 min)
Say, adjusted to age: “This makes answers from patterns in information. It can sound sure and still be wrong.” Ask the child to explain or demonstrate the idea in an accessible way — for example with their own words, a picture, an example, or a choice question. Verbal repetition is not required and does not establish readiness by itself.
2. Ask something real together (3 min)
Use a harmless, age-appropriate question whose answer can be checked from a suitable source. The child’s genuine question is useful only if it meets those conditions. Type it, read the answer aloud, and ask: “does this match what you already know?“
3. The false-answer test (5 min)
Prefer a question you can check from a book beside you, or turn off web/code tools for this step. Do not rely only on “word count this paste” or “news from yesterday” — tools often solve those. If the model is right, say so and try a second trick rather than pretending it failed. Compare a wrong answer to the true one. Say plainly: “It got that wrong, and it didn’t sound any less sure than when it was right.”
Question you used: ____________________________________________
What it got wrong: ____________________________________________
4. The source check (4 min)
Ask: “how could we find out if it’s actually right?” Do the check together — an age-appropriate primary or authoritative source, a suitable reference book, or a knowledgeable adult who can show the basis for the answer.
How you checked it: ____________________________________________
5. The privacy pause (3 min)
Ask: “if you wanted to ask it something about you — your school, your friends, where we live — what should you do first?” Practice with a hypothetical school-project question and have the child name which details to leave out.
6. Closing reflection (2-3 min)
Ask, in the child’s own words:
- What is this thing good at?
- What is it bad at, or where does it need checking?
- What would you never type into it without asking me first?
Their answers: ____________________________________________
Age-band adjustments
| Age band | Who types | Adjustment |
|---|---|---|
| ~6 | Adult | Adult narrates the false-answer test out loud; child watches and answers closing questions simply |
| ~10 | Child, adult watching | Child predicts confidence before checking; attempts source check with light guidance |
| ~13+ | Child drives | Child picks the trick question themselves; explains what they noticed instead of being told |
An adult still approves the question and source for suitability. If the output becomes sexual, violent, self-harm-related, hateful, manipulative, or otherwise harmful, stop the session, do not share the output, and tell a responsible adult; use appropriate safeguarding or emergency support if anyone may be in immediate danger.
After the session
- Repeat Steps 3-5 again within the next month — one session does not make the habit permanent.
- Decide separately, using the parent AI-literacy age matrix, whether and when this child gets an account of their own.
- If the child later has repeated independent access, revisit AI literacy by age for the supervision level that band expects.