Skip to content

Self-learning games

Can you teach a computer to play?

Self-learning games lets students use Flappy Bird and a racing car to explore how a computer learns from examples, rewards and selection.

Start with Flappy BirdExplore the racing car
Flappy Bird · Racing car
ExamplesRewardsGenerations

Two games

From one button to your own circuit.

Start here

Flappy Bird

Can you beat an untrained AI model? Then let the computer watch you play. Collect examples, train your model and let it practise further with rewards and penalties. Who wins when you play again?

Start with Flappy Bird

Racing car

Drive yourself and build your own circuit with straights and bends. Combine provided tracks and your circuits in a practice series. Let the cars race: the best models form the basis for new generations. What would Darwin think?

Explore the racing car

The learning path

From your examples to new generations.

  1. 01 · Flappy Bird

    Play yourself.

    When do you flap? Try to beat an untrained AI model.

  2. 02 · Flappy Bird

    Demonstrate.

    Let the computer watch. Train and test with one recording, three recordings and then more.

  3. 03 · Flappy Bird

    Let it practise.

    Choose a reward and a penalty. Let several birds practise at once and speed up training.

  4. 04 · Flappy Bird

    Who wins now?

    Play against the starting model, the model that learned from you and the model that practised itself.

  5. 05 · Racing car

    Build your circuit.

    Drive the car, build and save your own track. Combine circuits in a practice series.

  6. 06 · Racing car

    Race generations.

    Let the cars drive the same practice series. Select the best models and test their offspring on another track.

For teachers

Practical for the classroom.

For students aged 13 to 15. Start with one button in Flappy Bird and continue with the racing car. Students play, predict, train and compare. No coding is required.

  • Open the website on a laptop. A keyboard is enough for the exercises.
  • Alternate playing and discussion. Work alone or in pairs.
  • How does performance change with more examples? How do rewards guide learning?
  • Why do the cars not all drive alike? What changes from generation to generation?
  • Work stays in the browser. Save a file to continue on another device.

AI literacy

What do students practise?

This project addresses these goals from the OECD and European Commission AI Literacy Framework:

  • Explain how it works.

    Students follow how examples and rewards change a model and how new generations of cars emerge.

    Engage with AI 2 · Shape AI 1

  • Choose examples deliberately.

    Collect one, three and more recordings. Retrain each time and investigate the effect of the number and choice of examples.

    Shape AI 3

  • Test fairly.

    Compare the starting model, the model that learned from you and the model that practised itself with the same openings. Test the car on another track too.

    Shape AI 2

  • Keep judging for yourself.

    Compare performance before and after practice. Decide whether to try extra examples, more training or a different reward.

    Manage AI 4
View the framework · OECD / European Union, 2026

Place in the learning path

From platforms to projects.

The platforms build computational thinking. The projects use that thinking in a cross-curricular assignment in which students build, measure, test and conclude.

Learning computational thinkingfrom blocks to code, from code to research
Platforms
Blocklyvisual step-by-step thinking.JavaScriptlogic in real code.HTMLstructuring webpages.
DodonaPython in the Classroom with tracking.DelphiPractice Python in the browser.
STEaM projects
Smart PlantsArduino, soil moisture, calibration and data.Robotic handArduino, sensors, servos and movement.WindMicro:bit, energy, voltage and current.Smart binAI, image recognition, data and microcontrollers.Self-learning gamesGetting started with machine learningParticulate matterArduino, air quality, data and poster session.

Privacy and licence

Privacy

The games and model training run in your browser. Your examples, models, answers and results are stored locally on this device while browser storage is available. They are not sent to a project server. You can import or export a file yourself through the platform. This project uses no first-party analytics or advertising cookies.

Using a shared device? Export your work if you want to keep it, then delete your experiment through “Mijn experiment” (My experiment). Local work can be lost if browser storage is cleared or unavailable.

GitHub Pages records visitors’ IP addresses for security purposes. External links are subject to the provider’s privacy policy.

Read how GitHub Pages processes technical visitor data.

Licence

Original project material by Robbe Wulgaert may be used, copied and adapted free of charge for education, provided “Robbe Wulgaert – AI in de Klas” is clearly credited, with a link to the original project where reasonably practical. Commercial sale, paid distribution, sublicensing or repackaging as a commercial product requires prior written permission. Other uses outside education also require prior written permission. Third-party components remain subject to their own licences.

Read the full licence terms. View sources and credits. Read the included software licence.

Just get started.

Start with Flappy Bird