Thinking with AI
Every kid at CHIP directs AI. This area teaches how to ask it well, how to check what it says, and how it learns, so the kid stays the one who decides.
What they learn
- The machine does what you said, not what you meant. Kids learn to give role, context and format, and to fix a request by changing a word.
- Catching mistakes. Spotting when an answer is invented, and iterating until it is right.
- How an AI learns. Kids train a small AI with their own drawings, give it an honest exam and find out why it gets things wrong.
- Check before you share. Stop for ten seconds, look where a photo came from, read other sources, and be wary of traps.
How they learn it
Is it true or is it a trick?
An invented case to practice: stop before sharing, look closely with a magnifier, find where a photo came from and give a verdict with proof.
- 1Hold the button for ten seconds while you look at the photo, before deciding what to do.
- 2Use the magnifier: find at least three odd things.
- 3Search for the photo and pick the oldest one: that is the secret.
- 4Give a verdict backed by at least two proofs, then defend it to your table.
What they build
A sealed verdict with proofs and a detective credential, or your own AI trained with your drawings and used in a game.
The AI prompts studio runs an open AI on the kid's own computer, and nothing is sent outside.
Skills it develops
How it grows with age
Is it true or is it a trick? Teach the computer your drawings. The same request, asked two ways.
The leak that never happened: edited chat screenshots, a voice cloned with AI and phishing traps, in a case marked as fiction.
The viral news story, with misleading charts and the limits of AI detectors, and My detector: train one, measure it and find its bias.
Tools they use
Real tools, with one line on what your kid does with each.
- IA CHIP · an open model (Qwen) running in the browser
Learns to ask an AI well: role, context, format, catching hallucinations and iterating. It runs on their own computer and sends nothing out.
- Brain studio
Trains their own AI with drawings or sentences and tests it: mistakes, bias and an honest «I don't know».
- Fake Detectives
Checks what they see online: stop before sharing, read laterally, trace images and spot AI clues.
Every outside tool is used with a workshop account. A kid never creates or uses an account of their own.
Let's talk
about your kid.
Request an interview. We will write to set it up, and your family goes first on the list for the first cohort.
Request an interview