The Bias in Creativity Evaluation of Artificial Intelligence

Abstract

Drawing from folk psychology theories, we examine when and how the producer’s identity as artificial intelligence (AI) or human influences creativity evaluations. Four experimental studies demonstrated that people generally discount the creativity of apparent AI (vs. human) production, both directly and because of reduced effort perceptions. However, this effect is not ubiquitous. Evaluators who believe that creativity is a fixed (vs. malleable) attribute are less likely to discount the creativity of productions presented as made by AI rather than humans. Moreover, the evaluation bias is found when production is framed as artistic but not when it is framed as commercial. Thus, the bias in creativity evaluations toward AI (vs. human) production is situated and depends on both internal (i.e., lay theories) and external (i.e., production domain) influences.