I'm Julien, dad of Romane (6) and Meryl (3). I document our experiments to help them discover artificial intelligence with paper, pencils, and plenty of failed attempts.
Every article tells the story of an activity actually done with my kids. Photos and reactions included.
AI can be explained with bottle caps, a hand-drawn target, and a secret notebook. Screens can wait.
Sometimes it works, sometimes Meryl knocks everything over. We tell both — that's how you learn.
We played at aiming for a target, adjusting after each throw. Then we talked about how an AI does exactly that: it misses, gets told how it missed, and adjusts. Over and over, until it works.
6-10 yrsWe tried writing a letter between two imaginary characters. Romane chose the flamingo, writing back to Lili the rabbit. Simple enough, until we had to figure out how Lili had written at all, given she has no thumb.
3-7 yrsWe emptied out a random pile of objects and Romane made her own categories: 'small things', 'for building'. Then she did two piles with a secret criterion I couldn't crack. Meryl sorted by what he liked, what he quite liked, and what he didn't like much.
“It's like learning to ride a bike, but longer.”
Romane, 6 — after seventeen missed throws
A tested activity, a real kid's reaction, zero jargon. Tailored to your child's age.