Bias in perception of art produced by artificial intelligence

Abstract

Some perceive AI as encroaching on human capacities, perchance affecting their perceptions of AI. Applying Schema theory, this study examines whether and to what extent human judges are biased in ascribing “creativity” to the artwork of artificial intelligence. To begin to examine whether an artist’s perceived identity (i.e., as an artificially intelligent agent (AI) or human) might produce different reactions in judging the same artwork, participants blocked by subject characteristics (e.g., graduate, undergraduate) were randomly assigned to one of two types of focus groups. The groups differed only in that they were led to believe the same artwork was the product of a human or the product of an AI artist, triggering very different identity schemas about the artist. Participants were asked to (a) define art; whether AI can (b) make art and (c) if given art (which was the same in all groups) is “art.” Both focus groups types similarly associated art with expression, creativity, providing a message, and broadly conceived the boundaries of what was art. They differed in that the “Human Artist group” had an interactive view of art (perceiving, providing feedback) but the “AI Artist group” said any sense (visual, auditory, etc.) stimulus was art. The “Human Artist group” thought AI could make art while the “AI Artist group” disagreed. Both types of focus groups believed that the given art piece is art. Implications of findings for applications and theory are discussed.