Eyes can tell: Assessment of implicit attitudes toward AI art

Iperception. 2023 Oct 30;14(5):20416695231209846. doi: 10.1177/20416695231209846. eCollection 2023 Sep-Oct.

Abstract

Advances in artificial intelligence (AI) have significantly improved the abilities of machines. Human-unique abilities, such as art creation, are now being challenged by AI. Recent studies have investigated and compared people's attitudes toward human-made and AI-generated artworks. These results suggest that a negative bias may exist toward the latter. However, none of these previous studies has examined the extent of this bias. In this study, we investigate whether a bias against AI art can be found at an implicit level. Viewers' attitudes toward AI art were assessed using eye-tracking measures and subjective aesthetic evaluations. Visual attention and aesthetic judgments were compared between artworks categorized as human-made and AI-made. The results showed that although it was difficult for individuals to identify AI-generated artwork, they exhibited an implicit prejudice against AI art. Participants looked longer at paintings that they thought were made by humans. No significant effect of categorization of paintings was found in subjective evaluations. These findings suggest that although human and AI art may be perceived as having similar aesthetic values, an implicit negative bias toward AI art exists. Although AI can now perform creative tasks, artistic creativity is still considered a human prerogative.

Keywords: AI art; aesthetic evaluation; creativity; eye-tracking; visual attention.