Can AI Create with Flair? Artistic Creativity and Artificial Intelligence

Abstract Can Artificial Intelligence (AI) be creative? In order to address this question, we need to understand creativity before moving to apply our chosen definition to the assessment of AI. With this in mind, this article focuses on one component of one definition of creativity: flair. Flair is Berys Gaut’s addition to value and originality, the components typically included in accounts of creativity. Flair is the basis of Gaut’s rejection of AI creativity; but is he right to reject this possibility on these grounds? In this article I argue that flair is possible for AI. I begin by outlining the flair condition. I then turn to assess whether flair is possible for AI, arguing that it is potentially achievable. I will then consider Gaut’s objection to the possibility of AI creativity through flair. I argue that if we are happy to accept that skill is sufficient for flair in place of understanding, then flair is possible for AI. However, there remain two issues, which I will then turn to examine. The first issue is to clarify the relationship between skill and understanding in order to establish whether they can indeed be pulled apart in flair. The second issue is to consider whether flair is inextricably linked to agency, and thus is not possible for AI. I conclude that flair is possible for AI systems to achieve if we do not require understanding, suggesting that we can instead rely upon skill to help delineate creative processes in non-agents.