An artificial neural network (ANN) is a piece of software that is designed to simulate the way a human brain analyzes and processes information. Because it simulates a brain, it is able to recognise patterns and act upon those patterns with it's own decisions. Neural networks are part of the foundation of artificial intelligence (AI).
We tried to make the term "artificial intelligence" more accessible by creating a game that is played by multiple individual neural network entities represented by blue creatures.
The AI starts without any knowledge of the simulation. It does not know where to go, what the world looks like, what a game is and what it's goal is. Over time, and after multiple generations, the AI will have learned to recognise certain terain patterns and how to act upon them.
The further the AI moves from the left to the right side of the level, the more score it will receive, ergo the better it performs.
Using genetic algorythms, the AI will reproduce after all batches have tried their best to get as far as possible in the level and a new generation will be generated, inheriting DNA from the previous generation which will then be slightly mutated to achieve (biological) evolution.
Usually simulations are ran with an increased speed. Which means that the simulation could run 100 times faster than in this simulation which would also result in the AI learning 100 times faster. This is often the reason that AI is described as a scary technology because it can learn with incredible inhuman speed.
We run this simulation in real time in order to achieve a smooth simulation that runs more easily.