I absolutely love genetic algorithms.

At university we evolved pilots for the classic moon lander game.

The genome was a collection of instructions like turn left, turn right, accelerate, decelerate.

Each generation you test all of them in the game then “breed” the most successful. Essentially take half of eachs genome and apply some mutations.

Eventually you end up with one that just manoeuvres perfectly onto the landing pad.

It’s just fascinating watching them gradually get better.

We also played around with NEAT which uses evolutionary algorithms to create neural networks by changing their structure and weights.

Just a really awesome topic.