Evolved Virtual Creatures by Karl Sims

This is an early example of art built on artificial neural networks, although the networks were evolved by genetic algorithms rather than trained on data. The work stands as one of the most significant links from the rule-based AI of Cohen to “Modern AI art,” relying on the evaluation of trained data. The work showcased the evolution of virtual block creatures through simulated Darwinian processes. These creatures were tested on their abilities to perform tasks such as swimming, jumping, and competing. The most successful ones had their virtual genes copied, combined,and mutated to create a new generation. This iterative process led to the emergence of increasingly adept creatures. The project was a continuation and further demonstration of the potential of evolutionary algorithms in artificial life, demonstrating how virtual entities could evolve complex behaviors over successive generations. Sims' work remains influential in the fields of computer graphics and artificial life along with pioneers Yoichioro Kawaguchi, William Latham and Christa Sommerer.