16 Comments
User's avatar
Marine G.'s avatar

Très marrant et pédagogique !

En revanche dans votre jeu on peut marquer contre son camp ET faire des points. ;.)

On est en train de créer un jeu de course star wars utra-rapide mais on a déjà appris que 1/ Claude code super vite 2/ fait beaucoup d'erreurs (durée du jeu : 3 secondes au lieu de 3 minutes 😅) 3/ est persuadé qu'il corrige son erreur mais en fait non. 😁

Jean-Paul Paoli's avatar

Hahaha oui alors effectivement il est pas bien fini le basket chess 🏀 il faudrait d’autres séances

Et on se rend compte aussi qu’il y a qqs bugs des que ça devient un peu complexe il faut passer à d’autres systèmes … peut être que c’est un métier finalement :)

Jean-Paul Paoli's avatar

Et sinon mettre opus dessus sur Claude payant / Gpt5 ça aide

Marine G.'s avatar

Oui ouf finalement, on n'a pas créé un super jeu en 10 minutes :)

Anastasia | ModernMomPlaybook's avatar

This is fantastic! I love how you captured the difference between how kids and adults approach AI. Kids see it as a playful collaborator, not something to be revered or over-engineered. It’s such a powerful reminder that learning doesn’t have to be “perfect” or formal.

Also yes to this: "MIT's Mitchel Resnick found that kids who create with technology develop fundamentally different mental models than those who just consume it."

Jean-Paul Paoli's avatar

Exactly. To be honest they weren’t impressed at all … why all the fuss haha

Success Louis's avatar

This is beautiful 🔥

Such an awesome way to play around with AI and build cool stuff.

Eva Keiffenheim MSc's avatar

What a cool idea. Thanks for sharing. I love that you didn't take away the iteration and struggle parts from kids' learning journey.

Jean-Paul Paoli's avatar

Not sure I understand. You mean you would have expected something around some difficulty in prompting ? There wasn’t any ! I mean prompting could have been better but Ai are clever enough to interpret now. My kids are 9 and 7 btw so not 4 and 5 - which could have been a different experience

Eva Keiffenheim MSc's avatar

Your entire article is a case study in creating what learning scientists call "desirable difficulties" (as you likely know already).

I wasn't suggesting the difficulty was in getting the AI to understand them. The difficulty was in your children getting to a full and complete understanding of their own idea. The AI acted as a mirror, instantly reflecting the flaws and ambiguities in their thinking.

Every time the AI produced a bug, it wasn't a failure of the AI's interpretive ability but an opportunity for your children to strengthen their ability to think systematically.

That is the iteration and struggle I was celebrating, and you were wise to let it unfold without interference.

Jean-Paul Paoli's avatar

Yes exactly that was my point and experience - but didn’t know the theory behind ! So Thanks for the explanation !

Karen Spinner's avatar

Sounds like fun! I’m going to try this with my own kids!

Jean-Paul Paoli's avatar

You should definitely have some fun ! keep us posted !

Bruno's avatar

What would be the ideal age for kids to be involved in such interactions according to your experience?

Jean-Paul Paoli's avatar

Well mine were 9 and 7 when we did this but may be you can start earlier … you need to be able to articulate the idea of a game though so I think it depends on the kids but not sure a lot know how to do that before 6