Is anyone else here interested in this subject?

For those who don't know, the concept uses the model of evolution as a means of programming the solution to a problem. It has very wide appeal for A.I. and neural networks.

The basic way it works is to start with an encoding system ("chromosone sequences") for behavioral components and functions, fill it up with random values, then start to play the artificial selection game.

Example:
1. Generate 100 chromosone sequences at random, each representing a program that can be run.
2. Let them all run, and evaluate how well they perform the desired task.
3. Take the two best results and use them as parents for the next generation.
4. Use those two to generate 100 more sequences using mutation and crossover.
5. Repeat until you have a "good" solution.

I found this site to be a good introduction to the concepts. I've been using this idea to code a game/educational program that generates 3D artificial life-forms that learn to adapt intelligently to their environment. It's kind of scary once it starts to actually work!