Attribution of Creative Skills to Human and Artificial Artists in Alethic and Epistemic Possible Worlds

The advancement of artificial intelligence systems in the realm of computational creativity, particularly applied in visual arts and music, calls for an investigation into the status of these systems as creative entities. Through an online survey involving a sample of N = 64 participants, we investigated how both human-created and artificial intelligence-generated artworks can influence users’ creative judgment, emotional evaluation, and the perceived pleasantness of the artworks. We used 18 artworks (nine created by humans and nine created by the DALL-E system), providing both true and false information about the artworks’ authorship. We analyzed three possible worlds, an alethic world without authorship information, an epistemic world with either correct or incorrect authorship information, and a world with access to both alethic and epistemic worlds, where a comparison between ground and believed truth can be drawn. Furthermore, we administered the Big-Five personality test to understand if specific personality traits could influence the aesthetic judgments of subjects. Our findings include: (a) creative judgments significantly correlate with both perceived pleasantness and emotional intensity across all possible worlds; (b) individuals tend to consider human-created artworks more creative only when they possess information about the authorship (possible world-epistemic and possible world-alethic-epistemic); (c) artificially generated artworks evoke more intense and positive emotions when people have no information about authorship (possible world-alethic); and (d) extroverted participants tend to find artworks generated by artificial systems more appealing when lacking information about the artworks’ provenance.