Evaluating machine creativity

摘要

Can a computer be creative? And what can we learn about our own creativity from studying computational creativity? My research offers a comprehensive and practically useful investigation into how to evaluate the level of creativity demonstrated by computational systems. How should something so subjective as creativity be measured? I argue that it is most productive to treat creativity as a collection of inter-related factors such as originality, value and productivity, which are more tightly defined and therefore more amenable to measurement. Potential factors are being derived from empirical studies examining a wide variety of our writings on creativity. These will be tested in a simulation of a creative environment: the best performing factors will be applied to evaluation of existing creative systems, in comparison to assessments made by human judges. The aim of this is to identify key components for creativity, giving insight into how to approach the evaluation and improvement of computational systems and also towards human creativity as well.