Conceptual Mapping of Artist Interaction with Generative AI Systems

The rapid diffusion of generative models into studios, galleries and creative industries has outpaced the conceptual vocabulary available for describing what happens when an artist works with a machine. Public debate remains polarised between the claim that artificial intelligence is no more than a new brush and the claim that it is an autonomous author, yet neither position accounts for the range of practices observable in contemporary work. This paper addresses that gap through a structured conceptual review of scholarship in computational creativity, art history, empirical aesthetics and the law of authorship. Rather than reporting new measurements, the study synthesises existing work into an analytical instrument. Two contributions are offered. First, a four-mode taxonomy characterises the machine in artistic practice as tool, as medium, as collaborator and as subject-matter, distinguishing the modes by the locus of creative intention rather than by the underlying technology. Second, a tri- axial framework analyses a work along the axes of agency, authorship and reception, showing that most disagreements about so-called AI art are disagreements about which axis is under discussion. The framework is applied to documented cases, from early rule-based systems through adversarial image generation to current text-to-image practice, and to the consolidating legal position that copyright requires human authorship. The analysis indicates that machine autonomy and human authorship are not a zero-sum pair: the works that are technically most autonomous frequently demand the greatest curatorial and conceptual labour. Implications for artistic education, curation and cultural policy are discussed and an agenda for empirical research is proposed.

کلیدواژه ها:

artificial intelligence ، generative art ، computational creativity ، authorship ، human-machine collaboration ، aesthetics