Human Art Vs AI Art: Why We Prefer One Over The Other
People mostly like human-created art more than AI-generated one, a recent study published in the journal Cognitive Research: Principles and Implications suggests that narratives, as well as the perceived effort put into creating the art play an immense role in shaping people’s judgment, but what is even art?
“Art is the queen of all sciences communicating knowledge to all the generations of the world”
— Leonardo Da Vinci
Throughout history, art has been considered an activity exclusive to humans, it’s a tool for us to transmit our emotions in a way that is intuitive and requires no words.
However, with advent of artificial intelligence all of that is changing, art no longer requires individuals who lived through painful or joyful experiences for it to be created, in fact, it doesn’t require any experiences to be lived whatsoever, a machine can produce a piece of art that is indistinguishable from human-made art in a matter of seconds, so was Da Vinci mistaken when he said:
“Where the spirit does not work with the hand, there is no art”
— Leonardo Da Vinci
The study’s aim was to find out if people actually have a preference for human art, and if so, to identify the factors that contribute to this preference. For that purpose, the researchers designed two studies.
Study 1
The first study involved 149 participants from Amazon’s Mechanical Turk, they were shown 30 AI-generated paintings, each labeled as either “human-created” or “AI-created”. The participants evaluated each artwork using criteria such as liking, beauty, profundity, and worth. The research team deliberately used AI-generated artworks exclusively for this phase to maintain a controlled and comparable set of stimuli.
The study’s findings showed a notable preference among participants for artworks labeled as “human-created” rather than “AI-created”, despite all the artworks being made by AI. This preference was consistent across different criteria used to judge the artworks, such as liking, beauty, profundity, and worth.
This preference was more substantial on the deeper aspects of art like meaning and worth, as for more surface-level aspects like liking and beauty, the difference between human and AI labels was not as significant.
What can we learn from this
First, we recognize that art has two purposes:
- Providing sensational joy (either visual or auditory) to the person that experiences the art (like someone who looks at a painting, for example)
- It serves as a way to broadcast a message from the artist to the audience (like through what emotion the art makes us feel when we experience it)
So the findings of the study indicate that AI performs well in the first category, it can create beautiful art that is pleasurable to witness. But no so much in the second category, as it lacks the ability to convey deep meaning through its generated art. This should be relieving to people who fear that AI would steal their jobs in the field of art.
Study 2
For Study 2, the researches enlisted 148 participants from Prolific and used the same set of 30 AI-created paintings from Study 1. They also introduced new criteria like emotional impact, perceived narrativity, personal meaning, perceived effort, and estimated creation time. Additionally, participants underwent several questionnaires assessing their cognitive capabilities, empathy, attitudes toward AI, and perspectives on creativity.
What can we learn from this
Consistent with the earlier findings, researchers found that people indicated a preference for “human-created” art over “AI art”. In addition to that, they also noticed that narratives and perceived effort were crucial factors in the participants’ decision-making process.
An intriguing discovery arises in Study 2 when examining the moderating impacts. The results indicate that participants like and find AI-labeled art more beautiful the more they could create a narrative to accompany it. This highlights the role of narratives in enhancing the appreciation of art, particularly art created by AI, in surface-level aspects (beauty and liking).
It’s almost as if we go out of our way to make up a story that accompanies the AI-generated art, just so we can give it bonus points. However, more in-depth research is necessary to strengthen this assertion, especially since this analysis is based on correlations rather than causation.
Exploring future horizons
In the future, further research can expand on these discoveries by exploring specific aspects of AI-generated art. This could involve studying the influence of various AI generative models, participants’ familiarity with AI, and how knowing the actual creator of AI-produced art affects the perceptions and preferences of participants. Furthermore, the images used in the studies were chosen two years ago, and thanks to the constant evolution of AI generative models, newer art can be used to get more accurate results.
