I think there is a lot of value in adaptive algorithms, and I absolutely love watching something evolve. Genetically generated contraptions like these cars or these ones are very cool, but I'm even more interested in genetic AI like learning to walk, path-optimizing neural nets, and other such ideas.

I've played around with neural networks, probabilistic graphical models, and trying to evolve randomly generated code to a solution, but I'm still trying to find a way to emulate the elegance of something like Conway's game of life using genetic methods.

Anyone else interested in the topic? Maybe we can get a project going to build a simple platform. Or maybe someone knows of a platform that we can build on?

A 'pretty' visual representation of the system and/or organisms is very important to me, and I'd like to play around with genetically altering both physical parameters as well as AI. I'm pretty open-minded here; my main goal is to learn.

I vote python for the primary language, but I have experience with others and will gladly change my mind if there is good reason to.

EDIT/UPDATE: Alright, well... after thinking a bit, here is my idea (that I'm still willing to change if someone has another one): let's take an existing implementation of Conway's game of life, and just add a little AI to each living cell. To make this work, I think we need timesteps in between progression of the game; within these sub-timesteps, cells are free to move around to place themselves in the best possible locations before Conway's rules hit the environment. Inheritance of AI properties can then just follow the game's spawning rules. If we do it right, we should end up with some cells that will organize themselves into patterns like the game's oscillators, stable shapes, replicators, or guns.

UPDATE 2: I've set up the repo; head over there to check it out.