Recent advancements in artificial intelligence (AI) art have sparked much conversation on the nature of art and our perception of it. The capacity for creativity is often thought of as a defining trait that sets humans apart from other animals or machines. However, growing proficiency of AI-generated art has challenged this idea and begun to worry artists that their job may be taken over by machines.
Norberto Grzywacz studies the neuroscience of aesthetics, researching what people find appealing in art and how these opinions change as a function of culture. In his 2019 article, “Is Beauty in the Eye of the Beholder or an Objective Truth? A Neuroscientific Answer”, Grzywacz and colleagues discuss the neuroscientific and evolutionary aspects of the philosophical question: is beauty subjective or objective? The scientific answer to this question is that aesthetic preferences have both subjective and objective mechanisms. The processing fluency theory explains a mechanism of objective beauty, stating that stimuli will have a greater aesthetic response when its properties are easier for the perceiver to process. In this way, AI-generated art could be equally if not more appealing. AI art works by using machine learning to create a new piece of art based on the properties given or shown to the AI. If fed the properties that are easiest to process, or most fluid, the AI should be able to create the most aesthetically pleasing piece of art.
However, there is more than contributes to aesthetic appeal. The subjective beauty mechanisms in our brain alter our aesthetic preferences based on our experiences and environment, or, in other words, our culture. What each person values is greatly influenced by their relationships with others. Grzywacz uses a simulation to model this influence. He starts out with points randomly spread across a spectrum of aesthetic preferences and with each iteration, the influence of surrounding points changes the location of each point. Eventually, points form ‘islands’ of aesthetic preferences. This is a polarization of values often found in culture and further points to culture as the biggest variable in aesthetic preference. It seems that this aspect is hard to pin down for AI art, as most pieces produced are either simplified to objective beauty, or way too specified to niche preferences. The fluid nature of preferences and ‘islands’ also makes it difficult for an AI to keep up without human oversight.
In conclusion, while AI art may be aesthetically pleasing, it is prone to lacking in the culturally aware human touch that creates polarization and social value in art. As culture is constantly evolving, so too is the aesthetic preferences of its people. Until further machine learning advancements are made, AI will always need an artist to tell it what properties and values to use. In the future, we may find that our most famous artists are actually coders.
Works Cited
Aleem, Hassan, et al. “Is Beauty in the Eye of the Beholder or an Objective Truth? A Neuroscientific Answer.” Springer Series on Bio- and Neurosystems, 16 Nov. 2019, pp. 101–110., https://doi.org/10.1007/978-3-030-24326-5_11.
Elgammal, Ahmed. “AI Is Blurring the Definition of Artist.” American Scientist, 2019, https://www.americanscientist.org/article/ai-is-blurring-the-definition-of-artist.
Grzywacz, Norberto. “How the Brain Causes Social Polarization: The Example of Aesthetic Values.” Neuroscience Seminar. 2 Feb. 2023, Loyola University Chicago, Loyola University Chicago.
No comments:
Post a Comment