Artworks are perceived as less creative and aesthetically valuable if they are labeled as AI-made
2023/04/26


A series of four experiments has revealed that the same artwork is preferred less and perceived as less creative and awe-inducing when participants are told that it is made by artificial intelligence (AI). These effects were stronger among individuals who believe that creativity is a uniquely human characteristic. The study was published in Computers in Human Behavior.

Artificial intelligence is a term used to describe computer systems that are able to perform tasks normally requiring human intelligence. These tasks include visual perception, speech recognition, translation between languages, decision-making and others. Artificial intelligence or AI has seen explosive development in the first decades of the 21st century. The domain of tasks performed by AI has also increased to include fields that used to be considered exclusively human. This includes artistic creativity.

“AI has not only created paintings that look like the work of renowned masters like Rembrandt, but has also created original artistic styles that have been sold in auctions at high prices. Moreover, AI has composed original songs and music scores, has written poetry, and has designed whole cities and houses,” wrote study author Kobe Millet and his colleagues.

These developments made researchers curious about how people respond to AI-made art. How do AI-made artistic creations relate to people’s beliefs about the human nature? These questions are important because they have the potential to challenge people’s beliefs about the defining features of humankind. They also challenge the security of the view that humans have a unique position in the world.

Millet and his colleagues wanted to explore how and if telling people that a piece of art is made by AI affects their perception and experience with such work. Their expectation is that people will tend to be biased against AI artwork creations and see them as worse than those created by humans, regardless of the features of the art piece.

They conducted a series of 4 experiments. In all experiments participants were shown two pieces of art and told that one was made by a human and the other by an AI system (although both were either AI or human made depending on the experiment). They were then asked to indicate for which of the two pieces they experienced more awe.

In addition to this, experiments 1, 2 and 4 asked participants to indicate which piece they saw as more creative. Experiment 4 asked them which they would prefer to buy. Experiment 3 also assessed anthropocentric creativity, the belief that creativity is a uniquely human characteristic and its role in assessments participants made.

Participants of experiment 1 were 206 Dutch students (31% female) and they evaluated two AI-produced music pieces. In experiment 2, 298 U.K. residents recruited online on Prolific evaluated two paintings, both of which were produced by a human. Experiment 3 again used paintings, but these were made by an AI system called AICAN.

Participants were 404 UK residents recruited via Prolific. In this, study, researchers also measured anthropocentric creativity beliefs using a 5-item scale they made themselves. Experiment 4 tested the findings of all previous studies together on a group of 800 UK residents recruited via Prolific. The pieces of art used in this experiment were art reproductions – posters.

Results showed that participants reported less awe for the piece of music that was labeled as AI-made compared to the piece labeled as human-made (Experiment 1). Researchers conclude that the difference in evaluations is solely due to the information about the source (AI vs. human) and independent of the art content. In experiment 2 participants reported less awe for the painting labeled as AI made. This painting was also perceived as less creative.

Participants of experiment 3 again reported less awe for the painting labeled as AI-made (although both paintings were AI-made!). However, further analysis showed that this difference is completely created by participants holding anthropocentric creativity beliefs. In other words, bias against AI was present only in participants who scored high on these beliefs.

Results of experiment 4 were consistent with those of the previous 3 studies. Participants of this study reported less awe and less creativity for the poster labeled as AI-made. The bias against the AI was more pronounced in participants high on anthropocentric creativity beliefs. Finally, participants indicated that they would be less likely to buy the poster labeled as AI-made.

“Results of four experiments (including one large-scale preregistered experiment comprehensively testing all hypotheses) show that people display a negative bias against AI-made art across various forms of art,” the researchers concluded. “Merely labeling a work of art as AI-made (vs. human-made) is enough to shift people’s preferences toward the human-made one.

“This bias is expressed both in a cognitive (reduced perception of creativity) and emotional (reduced experience of awe) manner, while the former accounts for the effect on the latter. This suggests assigning lower creative value to AI art results in more restricted emotional responses in terms of awe. Importantly, these effects are more pronounced for people who endorse anthropocentric creativity beliefs more strongly.”

These studies make an important contribution to scientific knowledge about how AI-artwork is perceived by the general population. However, it also has limitations that need to be taken into account. Notably, the reasons why people with stronger anthropocentric creativity beliefs derogate AI-made art remained unknown. Additionally, all study participants come from two western, highly developed countries. It is possible that results would not be the same on samples from different populations.

The study, “Defending humankind: Anthropocentric bias in the appreciation of AI art”, was authored by Kobe Millet, Florian Buehler, Guanzhong Du, and Michail D. Kokkoris.

© PsyPost