People have Difficulty Distinguishing AI from Human Art but Prefer Human-Made Works, According to a Study

People have a difficult time distinguishing between art that was created by artificial intelligence (AI) and art that was created by human artists, according to a study that was conducted not too long ago at Bowling Green State University.

An investigation conducted by Tech Xplore found that despite the fact that generative artificial intelligence has progressed to the point where it can produce artwork that is visually identical, individuals continue to have an underlying preference for genuine human art.

AI vs. Human Arts

Andrew Samo, a doctorate candidate in industrial and organizational psychology and a distinguished research professor Dr. Scott Highhouse, led the study that focused on people’s ability to identify the source of graphics generated by either artificial intelligence (AI) or human artists.

In spite of the progress that has been made in artificial intelligence art, the participants were unable to differentiate between the two in a consistent manner, and they frequently made correct identifications just slightly more than half of the time.

By not alerting the participants that artificial intelligence will be responsible for the creation of some art, Samo and Highhouse constructed the study to remove any possibility of bias. Participants were instead informed that they would be viewing a sequence of photos and asked to score them based on aesthetic judgment parameters. They were not made aware of the participation of artificial intelligence in the process.

The findings suggested that, on average, participants accurately recognized the source of the artwork roughly fifty to sixty percent of the time, and their confidence in their guesses was very low. A greater positive emotional response was regularly reported by participants when they were exposed to human-generated art, despite the fact that differentiation became impossible.

Self-reflection, attraction, nostalgia, and enjoyment were the four primary aesthetic judgment elements that accounted for the majority of the diversity in the preferences of the participants. Art that was created by humans received higher scores in these categories, which indicates a stronger emotional connection.

‘Uncanny Valley Effect’

It is interesting to note that the participants were unable to define the reasons why they favored human art, despite the fact that their emotional responses indicated a definite preference.

The research suggested that there is a chance of the “uncanny valley effect,” which is a phenomenon in which minor changes in artificial intelligence-created art that are not noticeable to the human eye may be perceived unconsciously.

The researchers highlighted the developing capabilities of generative artificial intelligence models, which are now capable of producing works of art, music, poetry, prose, and text that are strikingly similar to those produced by humans.

The possible uses of generative artificial intelligence are expanding, despite the fact that it was once assumed to be restricted to activities that were repetitive or not innovative. The findings raises investigations concerning the psychological implications and societal influences of artificial intelligence, particularly as these models incorporate themselves seamlessly into everyday life. This is especially true as generative AI models continue to advance.

Especially in the context of creativity, the researchers are aware of the importance of doing ongoing study in order to gain a better understanding of the ever-changing interactions that exist between humans and artificial intelligence.

“Some of these new models can generate images that are really high quality and high fidelity toward the actual world, so it’d be interesting to run this study again,” according to a statement issued by Samo. “If you redid this, I’m not sure if people would be able to tell the differences at all.”

It was in the journal Psychology of Aesthetics, Creativity, and the Arts that the findings of the study were submitted for publication.