Creative problem solving and automated discovery : an analysis of psychological and AI research

Abstract

Since creativity is the ability to produce something novel and unexpected, it has always fascinated people. Consequently, efforts have been made in AI to invent creative computer programs. At the same time much effort was spent in psychology to analyze the foundations of human creative behaviour. However, until now efforts in AI to produce creative programs have been largely independent from psychological research. In this study, we try to combine both fields of research. First, we give a short summary of the main results of psychological research on creativity. Based on these results we propose a model of the creative process that emphasizes its information processing aspects. Then we describe AI approaches to the implementation of the various components of this model and contrast them with the results of psychological research. As a result we will not only reveal weaknesses of current AI systems hindering them in achieving creativity, but we will also make plausible suggestions - based on psychological research - for overcoming these weaknesses.