Abstract

In this article, I consider whether, in using AI to make images, human agents initiate computer processes that can in themselves be called creative. To answer this question, I compare a particular kind of creativity found in non-AI, or traditional, art—what I call “rule-bound creativity”—with the use of deep learning in generative AI. The reason I do this is because the history of the longest-standing and most serious objection—let’s call it the “Standard Objection”—to creative computers rests on the observation that, by definition, computers run in a strictly rule-bound way. The case of rule-bound creativity in non-AI art shows that the mere presence and following of rules does not prevent creativity. Given their unique functionality, Generative AI systems can give the appearance of creativity to an unprecedented degree. This is because, in addition to producing items that are genuinely new, these systems can give the appearance of agency. In the end, however, this is not enough to overcome the Standard Objection. Rule-bound production can be creative but only in a way that requires agency. The rules that define generative AI are rules for modeling certain functions of agency; they are not, and cannot be, rules for instantiating agency.