Skip to content

AI learning games for young learners

Why did AI
do that?

Group signals, shape a story, and check a rescue route with Dogo. Explore how AI learns, creates, and makes mistakes by trying things for yourself.

MISSION 07 · The Story Projector

Gameplay preview

Change a story clue. See how context shapes what comes next.

Try this mission

9 interactive missions to explore nowFrom how AI learns to generation and fact-checking.All released missions are currently open to explore.

See all missions

Our approach & sources

Play built on carefully checked ideas.

Read our approach & sources

Explore the ideas behind each mission, the sources we use, and the limits of each teaching model.

Pricing & exploring

Try a mission together.

All released missions are free to explore for now.

Price when purchases open

International
US$3.99

One-time payment. No subscription.

Purchases are not open yet

For parents

Before you begin

What is Mindogo, and what can children learn?

Interactive games for exploring how AI learns, generates, and makes mistakes—no coding required. Children group examples, change stories, and check evidence, practising how to compare results and explain their decisions.
See all missions

Who is it for? Should a parent join in?

Curious young learners who can read short instructions and enjoy trying things out. Play the first mission together to see if it fits. Then let your child make the decisions, helping with reading or controls when needed.

What device do we need?

Missions work on phones, tablets, and computers. A larger screen makes it easier to compare maps, evidence, and results.

Do we need an account? How is progress saved?

You can play without signing up. Progress saves in the current browser, with optional sync through a parent account. Clearing site data or changing browsers or devices may make local progress unavailable.

Do the games use real large language models?

They use controlled, repeatable teaching models that show key principles, rather than every mechanism of a commercial model. Children do not need to send chat messages to an external AI service.

How does pricing work?

All released missions are free to explore for now; purchases are not open yet. When sales open, Missions 01–05 will stay free. The remaining released content will be sold as the pack listed at purchase: ¥19.9 in mainland China or US$3.99 outside mainland China, paid once with no subscription. You can request a full refund within 14 days of payment.
Full purchase terms

Reference / Stanford CS229
CS229 Lecture Notes

Stanford University · Course notes

CS229 Machine Learning

CS229 Lecture Notes
Tengyu Ma · Andrew Ng

Stanford University · CS229
Version cited: August 23, 2026

Visit the Stanford course website

What the source supports

The notes describe clustering as grouping similar examples using features of the data, without answer labels. Chapter 10 introduces k-means as one way to do this.

How it informs the missions

In Starlight Mixer, children tune the resonance and observe similar signals forming clusters. In the first act of Homebound Rescue, they compare rhythms, paths, and waveforms to group unlabeled signals into channels. Both make grouping by similarity something children can observe and explain.

Limits of the reference

Starlight, resonance, and the grouping rules are teaching choices, not a complete implementation of the chapter’s algorithm. The notes support the underlying idea, not measured learning outcomes in children or endorsement by Stanford University or the authors.