NEAT – Art
Second Year Experimental Games Project (BA)
NEAT Art is an experiment on teaching a self-learning algorithm to create images, based on “Neuro-Evolution through Augmenting Topologies” (a technique conceived by Ken Stanley in 2002, full paper). Opposed to other instances of AI producing creative works, where existing data sets are used, the NEAT Art AI is trained by live input from passersby. The artistic quality of the current state is rated from 1 to 10, which in turn is used to evolve the observed generation of single species of the AI. The game also allows for quick experimentation and simulation of more complex AIs as well as the possibilities to observe the creation process and all relations between in- and outputs.
Tagline NEAT Art is an experiment on the evolvement process of artificial intelligence in an atypical field: Being creative.
Genre 2D Art AI
Team Solo Project: Tom Schildhauer