It turns out that people intuitively prefer human paintings to AI-generated paintings, and AI works may be falling into the uncanny valley phenomenon
In an experiment in which participants were asked to compare paintings generated by the image generation AI with paintings drawn by humans, it was confirmed that even if participants did not know which painting was which, the human painting was rated higher and perceived as more familiar.
Artificial intelligence and art: Identifying the aesthetic judgment factors that distinguish human- and machine-generated artwork.
Research finds people struggle to identify AI from human art, but prefer human-made works
https://techxplore.com/news/2023-12-people-struggle-ai-human-art.html
The impetus for this research came from Andrew Samo, a doctoral candidate at Bowling Green State University, and Professor Scott Highhouse, who wondered: 'As generative AI models continue to improve, will humans be able to distinguish between AI-generated and human-drawn images?'
Previous research has shown that people are biased toward AI artwork, but it was unclear what would happen if people were to compare the work of both humans and AI without any preconceived notions when they could no longer distinguish between the two.
To answer this question, Samo and his colleagues conducted an experiment in which participants were shown a series of paintings without being told that they contained AI-generated images, and then rated them using a psychometric test developed to quantify artistic emotions and experiences.
They found that even though participants were able to correctly identify the source of the artwork just over half the time—about the same accuracy as flipping a coin—they consistently perceived the human-created artwork as more positive.
The psychological assessment included 30 to 50 aesthetic judgment criteria, but four main areas showed significant differences between the human and AI works. Specifically, the human-created artworks received higher scores for '
Interestingly, when asked why they felt this way, the participants couldn't answer, which Samo and his colleagues believe is probably because their brains were detecting subtle differences between the AI and human work.
Regarding these results, Samo explained, 'One possible explanation is the uncanny valley effect, which is when an attempt to appear human is subtly off. Even if the overall result looks good, there may be small inconsistencies in the visual work or creative story created by AI that only the human subconscious can detect.'
The authors of this paper, Samo and Highhouse, are investigating the possibilities of generative AI. Generative AI has made rapid progress since Samo and his colleagues conducted their experiment and published their results in a paper. Samo commented, 'Some new AI models are now capable of generating high-quality images that are faithful to the real world. It would be interesting to repeat this experiment using such AI.'
Related Posts:
