Children who can think with AI, not for it.
AI Playground is a 10-week after-school program in AI literacy and creative technology for elementary students. Designed against dependency. Every project starts on paper. Every child leaves with something they made, choices they defended, and the habit of asking where the AI got it wrong.
AI should strengthen a child's thinking, not replace it.
Most "AI for kids" offerings make one of two mistakes: they avoid AI entirely, or they hand children a chatbot. We do neither. AI is in the room — used carefully, by design — because children will encounter it whether we like it or not. The question is whether they encounter it well.
Every project begins on paper. Children sketch, brainstorm, and explain their idea before any AI is opened. When AI is used, our instructors model how to ask a clear question, how to spot a confidently wrong answer, and how to improve a weak response. The work that comes out of class is the child's — done with AI as one of several tools, not produced by AI.
We are explicit with children about authorship. In our rooms, students don't say "AI made this." They say "I made this. AI helped."
Five stages we teach every child to follow — every time they touch AI.
Think
Pencil before keyboard.
Ask
A clear request with five parts: audience, role, task, constraints, format.
Check
Where's your source? Is the answer right?
Improve
What's one change that helps the most?
Explain
In your own voice. What did you decide?
Seven rules. Memorized. Recited weekly.
Children learn the Safety Code in Week 1 and recite it at the start of every session. It is the spine of the program — the smallest thing every parent can ask about and every child can perform.
- 01Think first.Sketch or say the idea before asking the AI.
- 02Keep private things private.No real names, schools, addresses, phones, or photos in any AI tool — ever.
- 03Check the answer.AI can be confidently wrong. Look for proof.
- 04Be the author.AI helps. The student decides.
- 05Ask for help.If something feels weird, tell the instructor right away.
- 06Respect the room.Kind voices, careful hands, eyes up at the signal.
- 07Leave it better.Devices closed. Table reset. Ready for the next group.
Each cohort follows the same shape. Each child finishes with a project they can show.
What AI is and isn't. The Safety Code. How AI 'learns' from examples.
Prompting as communication. Strong vs. weak prompts. Five-part prompt template.
AI mistakes and hallucinations. How to catch them. Why they happen.
Plan on paper. Build. Test. Improve. Iterate. Journal every human decision.
Revision, presentation skills, a full rehearsal — with the artifact live.
A parent-facing gallery. Each child presents what they made, what they chose, and what AI got wrong.
How we know your child is thinking, not copying.
We measure five habits, week by week, child by child:
- Authorship index — the share of journal entries that contain "I decided / I chose / I disagreed."
- Hallucination catches — AI errors the child spotted across the term.
- Prompt-on-paper rate — the share of sessions where the child sketched before typing.
- AI-attribution rate — the share of talks that name where AI helped and where AI got it wrong.
- Refusal rate — the times the child chose not to use AI on a step.
Cohorts trending toward dependency show drops in the first three. Cohorts trending toward thoughtfulness climb across all five. Parents see the trend in their child's midpoint and final progress reports.
The questions you'd ask if we were sitting across the kitchen table.
Will my child become dependent on AI?+
The curriculum is built specifically to prevent that. Every project begins on paper and we grade for the human choices your child made. By Week 10 your child can identify an AI hallucination, explain why it happened, and decide what to do instead.
Will my child be chatting freely with an AI?+
No. K–3 students do not type into the AI at all — the instructor pilots the AI on a projected screen. Grades 4–6 students may type into our school-approved sandbox under live instructor monitoring. We do not give students an unrestricted chatbot.
What information do you collect?+
From parents — basic contact info. From children — first name and grade only. We never type your child's full name, school, address, phone, photo, or family details into any AI tool. The sandbox we use does not train on student input.
How much screen time is involved?+
K–3 screens are on for at most 15 minutes of a 1-hour session. Grades 4–6 screens are on for at most 25 minutes. The rest is conversation, sketching, building, presenting, and movement.
What if AI gives an incorrect answer in class?+
Good — we use those moments. A whole unit is about spotting and fixing confidently wrong AI answers. Children leave the program able to identify a hallucination, explain why it's wrong, and decide what to do instead.
Is this coding, or just playing with a chatbot?+
Neither, exactly. It's AI literacy combined with creative building. Grades 4–6 do touch some structured no-code building — a quiz, a small site, an app idea. It is not a code-syntax class.
How is this different from regular computer time?+
It is curriculum-based: written lesson plans, learning objectives, vocabulary, assessments, a final parent showcase, and a journal each student keeps documenting their human decisions.
Ready to see what a careful AI program looks like in your room?
Schedule a 30-minute discovery call. We'll walk through the curriculum, the safety system, the showcase, and what a launch at your school would look like.