Artistic creation has traditionally been thought to be a uniquely human ability. Recent advancements in artificial intelligence (AI), however, have enabled algorithms to create art that is nearly indistinguishable from human artwork.
Existing research suggests that people have a bias against AI artwork but cannot accurately identify it in blind comparisons. The current study extends this investigation to examine the esthetic judgment factors differentiating human and machine art.
Results: indicate that people are unable to accurately identify artwork source but prefer human art and experience more positive emotions in response to human artwork. The esthetic judgment factors differentiating human & machine-generated art were all related to positive emotionality.
This finding has several implications for this research area and limitation and avenues for future research are discussed.
[Keywords: esthetics, esthetic judgment, computational creativity, artificial intelligence art]
…The machine-generated images were created using 5 different algorithms (ie. Ryan Murdock’s Aleph2Image [A2I] and The Big Sleep, Katherine Crowson’s Diffusion Model [DD; et al 2022], Justin Bennington’s S2ML, and DALL·E Mini [Dem] hosted on HuggingFace) with text prompts based on the titles of the human images (ie. “moonlight on the beach” and “lakeshore with reeds”). To reduce the total set of images to 20 (10 human and 10 machine) for the pilot study, the authors used a priori decision rules to standardize the image selection process. Finally, the 20 images were formatted for similar sizing (ie. 512 px along the smallest dimension).
See Also: