AI and Art: a Provisional Taxonomy

Abstract

This paper addresses the need for greater terminological clarity in the still-fluid field of AI-mediated artistic practices. It proposes a provisional taxonomy intended not as a definitive classification, but as an orienting analytical framework that stabilizes key terms while discarding others that generate conceptual confusion. The discussion contributes to a broader effort to move beyond attributive accounts of artificial intelligence – particularly those that seek to assess whether AI is creative or an artificial author – and instead situates it within the relational dynamics and concrete contexts of its use. Against the widespread notion of “art made by AI,” I argue that this conception is perpetuated by the use of certain terms commonly employed to describe AI-mediated art, which are instead always the result of complex forms of human–machine interaction. To begin with, the concept of “artificial art” is problematized, as it misleadingly suggests a form of non-human authorship in light of an ill-defined and unstable natural/artificial dichotomy. The paper instead adopts “AI art” as an umbrella term, understood not as art produced by artificial intelligence, but as a set of artistic practices in which AI systems play a constitutive role both in the production process and in determining the salient features of the artwork. Particular attention is given to generative AI art, understood here as a specific artistic genre based on artistic production assisted by generative systems, conceived neither as mere tools nor as creative collaborators, but as exhibiting a form of functional autonomy within the production process.