Several art generation systems use "genetic algorithms" to implement an image evolution process. Technically, such systems fall under the optimization approach: they search a large space of possibilities to find the items which best fulfill a certain evaluation criterion. Many systems of this sort, however, do not implement a formal evaluation criterion, but use an "oracle" instead: human spectators are allowed to vote for what they like best. Supposedly, this "supervised learning" process allows the system to acquire the (complex but ultimately boring) concept of "conventional beauty".

Note that genetic algorithms do not model the process of natural evolution. In natural evolution, there is no given evaluation criterion which can be applied to individual organisms. The survival of an organism crucially depends on all the other organisms (of the same and other species) in its environment. To model natural evolution would be to model a whole environment with interacting (and mutating and procreating) organisms.
    

Links

David Eck: Eaters.

Erwin Driessens & Maria Verstappen: Breed, 1999.

Craig Reynolds: Evolutionary Computation and its application to art and design

Cyberdelia

Erik Borra's AA&AI Links.

References

Karl Sims: Swimming, creeping, jumping and fighting creatures. [VPRO video-tape. Archives Jochem van der Spek, Eric Vreedenburgh, UvA New Media.]

Stephen Todd and William Latham: Evolutionary Art and Computers. London: Academic Press, 1992.

   

Shape-grammar algorithms using evolutionary processes

J. Cagan and W. J. Mitchell, "Optimally directed shape generation by shape annealing," Environment and Planning B: Planning and Design 20 (1993): 5-12.

Knight, T. W. (1989). Transformations of De Stijl art: the paintings of Georges Vantongerloo and Fritz Glarner, Environment and Planning B 16: 51-98.

T. W. Knight, "Languages of designs: from known to new," Environment and Planning B 8 (1981): 213- 238.

T. W. Knight, "Transformations of languages of designs," Environment and Planning B 10 (1983): Part 1 125-128, Part 2: 129-154, Part 3:155-177;

T. W. Knight, Transformations in Design. Cambridge University Press, Cambridge, 1994.

Terry Knight's Shape Grammar Site

Schnier, T. & Gero, J.S. (1996) "Learning genetic representations as alternative to hand-coded shape grammars," AI in Design196.

Art generation by majority vote

Links

Ken Goldberg: TeleActor

Discussion

The fallacy of taking the statistics of appreciation as a criterion for art is demonstrated by Komar and Melamid: the "most wanted paintings" and the "least wanted paintings" are equally interesting.

JoAnn Wypijewski (ed.): Painting by Numbers. Komar and Melamid's Scientific Guide to Art. New York: Farrar Straus Giroux, 1997.

Material for further discussion:

Ray Kurzweil's CyberPoet;
A. Michael Noll's Mondriaan simulation.

Related issues: Images vs. Models; the vacuousness of the "Turing Test".

The Turing Test Page
Turing Test links
Alan Turing