Today, we have a technical talk from Adam M. Smith of the expressive intelligence studio at University of California, Santa Cruz. He makes the claim in his slides, which apparently outline his Ph.D. research project, that a machine can become a creative entity that designs a game. To achieve this goal, he aims at developing an “intelligent designer” (a computational agent) that turns experience into communicable knowledge. Intelligence in this scenario comes from experience, which he describes as knowledge production based on fixed knowledge of past designs and expanding knowledge of discoverable actions. The design activity in games is reduced then to its very basics, being an action that produces rule systems and the necessary logic behind those. Thus, the more basic a rule system is, the more complex the actions resulting from human interaction with it might be.
His take on creativity as a rational pursuit of curiosity, resulting in a surprise, is quite interesting (and debatable). Using a design grammar or a set of structural elements of games, the game design agent could be ready to perform his task. The only thing I am left to wonder is how to make the game really exciting (appealing to human emotions — for this a library of information about emotional perception of game design elements could be highly useful) and feel not only “playable” but much like a game from a real designer. The slides are definitely thought provoking and well worth looking through.
Togelius, J. and Schmidhuber, J. (2008) An Experiment in Automatic Game Design. IEEE CIG 2008.
Smith, A., Romero, M., Pousman, Z., Mateas, M. (March, 2008) Tableau Machine: A Creative Alien Presence. AAAI Spring Symposium on Creative Intelligent Systems.
Nelson, M.J. and Mateas, M. (2009) A requirements analysis for videogame design support tools. In Proceedings of the 4th International Conference on the Foundations of Digital Games (FDG 2009), Orlando, Florida, USA.