Using AI to Generate Visual Art: Do Individual Differences in Creativity Predict AI-Assisted Art Quality?
William Orwig Lucas Bellaiche Sarah Spooner Anh Vo Zia Baig Anya Ragnhildstveit Daniel L. Schacter Nathaniel Barr Paul Seli a Harvard Universityb Duke Universityc Pneuma Science,Draper,UT,USAd Miami Universitye University of Waterloof University of Cambridgeg Sheridan College
DOI: https://doi.org/10.1080/10400419.2024.2440691
IF: 2.032
2024-12-13
Creativity Research Journal
Abstract:As artificial intelligence (AI) advances in the realm of generative art, a critical question emerges: does human creativity matter? That is, do more-creative people produce more-creative AI-assisted artwork? To explore this, we conducted an online, pre-registered study in which we measured individual differences in creativity through two divergent-thinking tasks: The Alternate Uses Task and the Divergent Associations Task. Separately, participants produced creative wordsets for a hypothetical AI-art generator, which we then input into DALL-E to generate images. A group of trained raters independently assessed these images for creativity. Results revealed that both DAT performance and semantic diversity of the wordsets positively associated with the creativity of the AI-assisted images, suggesting that individuals with stronger divergent-thinking skills, and those who generated more-creative wordsets, tended to inspire more-creative AI-assisted artwork. Mediation analyses supported this conclusion by demonstrating a significant pathway between individual creative ability and AI-art creativity, mediated by semantic diversity. However, while our models yielded significant results, the effect sizes were modest, suggesting that the relationship between individual creative ability and AI-assisted creative outputs is relatively small. Taken together, these results suggest that while individual creativity appears to contribute to the quality of AI-assisted artwork, its influence may be relatively limited.
psychology, multidisciplinary, educational