People Struggle to Identify AI From Human Art But Prefer Works Made by Humans, Study Finds

A recent study from Bowling Green State University reveals that people struggle to differentiate between art created by artificial intelligence (AI) and human artists.

According to Tech Xplore, while generative AI has advanced to create visually comparable artwork, the research discovered that people still maintain a subsurface preference for genuine human art.

AI vs. Human Arts

The study, led by Andrew Samo, a doctoral candidate in industrial and organizational psychology and distinguished research professor Dr. Scott Highhouse, focused on people's ability to identify the source of images created by either AI or human artists.

Despite advancements in AI art, participants were unable to consistently distinguish between the two, often making correct identifications only slightly more than half the time.

Samo and Highhouse designed the study to eliminate bias by not informing participants that AI would generate some art. Instead, participants were told they would view a series of images and asked to rate them based on aesthetic judgment factors without being aware of the AI involvement.

The findings indicated that, on average, participants correctly identified the source of the artwork approximately 50-60% of the time, and their confidence in their guesses was notably low. Despite the difficulty in differentiation, participants consistently expressed a more positive emotional response to human-generated art.

Four key aesthetic judgment factors - self-reflection, attraction, nostalgia, and amusement -accounted for most of the variance in participants' preferences. Human-made art scored higher in these factors, indicating a stronger emotional connection.

'Uncanny Valley Effect'

Interestingly, participants could not articulate why they preferred human art, even though their emotional responses suggested a clear bias.

The study proposed the possibility of the "uncanny valley effect," where subtle differences in AI-created art, imperceptible consciously, might be sensed subconsciously.

The researchers emphasized the evolving capabilities of generative AI models, which can now produce art, music, poetry, prose, and text that closely resemble human creations.

While initially thought to be limited to repetitive or non-original tasks, the potential applications of generative AI are expanding. With advancing generative AI models, the research prompts inquiries into AI's psychological repercussions and societal influences, particularly as these models seamlessly integrate into daily existence.

The researchers recognize the imperative for continual investigation to comprehend the evolving dynamics between humans and AI, particularly in the context of creativity.

"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," Samo said in a statement. "If you redid this, I'm not sure if people would be able to tell the differences at all."

The study's findings were published in the journal Psychology of Aesthetics, Creativity, and the Arts.