Autonomous Intentionality in Computationally Creative Systems

Abstract

We consider the problem of building an artificial system that exhibits creativity. We begin by noting that the attribution of creativity is influenced by perceptions about the system and that perhaps the most difficult hurdles to overcome in this respect are those of system intentionality and system autonomy. We discuss ways in which these difficulties might be addressed and cite examples from the DARCI system that already exhibit these characteristics to some extent. We also discuss the difficult problem of evaluation and present several useful methods, as well as examples of their application to results produced by DARCI.