while True: learn()

while True: learn() is a puzzle game about machine learning, neural networks, big data and AI, framed around understanding your cat. You play as a coder whose cat is brilliant at machine learning but cannot understand human language. With the cat’s help, you solve puzzles to learn how machine learning works and build a cat-to-human translator.

Solve machine learning puzzles

Each puzzle introduces a real technique in historical order, from 1960s expert systems to recurrent neural networks that predict what comes next. You build flowcharts by dragging ready-made blocks onto a board and wiring them together, then press Release to watch data flow through your solution. The blocks include expert systems, decision trees, perceptrons and neural networks. Short notes explain what each block does in the game and what its real-world counterpart does, with links to articles, videos and courses for further reading.

You sort data by shape and colour while working within limits on accuracy, speed and cost. Bronze medals are usually easier to earn; gold requires a more efficient approach, trading some accuracy for speed or returning with better hardware. You can save a solution and reuse it as a single block in later jobs.

Work as a freelancer

You take freelance contracts inspired by real problems machine learning has addressed, including a self-driving car piloted by your cat. You read emails from unusual clients and can socialize on forums. The game also lets you gamble on startups and, when you feel confident, become a startup CTO: your schemes then face the market, where you can cash out at the top or lose everything.

Upgrade your life

Freelance work earns money for faster hardware, which helps your schemes run more quickly. You can also buy a smartphone, a geeky figurine, outfits for your cat and even an aloe plant.

The game requires no coding experience, and you do not write code in it. It offers an introduction to what machine learning terms mean, what the ideas are for and why they mattered, rather than a course.