Turn-based evolution in a simplified model of artistic creative process

Abstract

Evolutionary computation has often been presented as a possible model for creativity in computers. In this paper, evolution is discussed in the light of a theoretical model of human artistic process, recently presented by the author. Some crucial differences between human artistic creativity and natural evolution are observed and discussed, also in the light of other creative processes occurring in nature. As a tractable way to overcome these limitations, a new kind of evolutionary implementation of creativity is proposed, based on a simplified version of the previously presented model, and the results of initial experiments are presented and discussed. Artistic creativity is here modeled as an iterated turn-based process, alternating between a conceptual representation and a material representation of the work-to-be. Evolutionary computation is proposed as a heuristic solution to the principal steps in this process, translating back and forth between the two kinds of representation. Those steps are: implementation, going from concept to material form, and re-conceptualization, forming a new conceptual representation based on the current material form. The advantages and disadvantages of this approach are discussed, and concluding from the initial experiments, it is a very promising path, well worth further exploration.