AI generated art challenges the definition of art itself. Recent experimentation with artificial intelligence has led to various intriguing projects and works of art that bring up many questions regarding what it means to create art. Most notably, in December 2018, British auction house Christie’s garnered national attention with the sale of Portrait of Edmond Belamy, a work of AI art created by Obvious, a Paris-based collective, for $432,000. The painting was created by an algorithm that was fed 15,000 portraits and programmed to generate an image from the input data, using the method GAN that stands for “generative adversarial network”. The final image isn’t a photorealistic portrait created from scratch, yet its sweeping brush strokes have a mystical, poetic impression about them, and it wouldn’t be unreasonable to assume it was painted by a human.
Similar efforts have been going on in other forms—Microsoft has developed a drawing bot that creates images based on text descriptions. It is also “intelligent” enough to add details and fill in the blanks—a quality that can be compared to creativity. Microsoft suggests that the machine could create images that would serve as starting points for painters or interior designers, or even produce animated films from screenplays with greater computing power in the future.
At Rutgers University’s Art and Artificial Intelligence Lab, researchers are working on CAN, a system that they hope to be “creative” rather than “generative,” that would have the ability to produce something new and different from its input data – work that wouldn’t belong to any specific artistic style. Most of its output artwork is abstract, but the very possibility of a computer algorithm being able to consistently create original visuals contradicts our understanding of the skill involved in art, encroaching on the territory of what we call art-making.
In fact, these researchers showed the output of the art algorithm to humans and asked if they could tell the difference between human-generated art and the output of the CAN computer, and their results showed very little difference in the responses of the subjects. If AI can create images indistinguishable from human-made works, are these pieces “art”? In order to begin to answer that question, we have to first ask ourselves—what is art?
Art has been notoriously difficult to define. Even dictionary definitions vary—Cambridge Dictionary, defines art as “the making of object, images, music, etc. that are beautiful or that express feelings,” while Merriam-Webster addresses the process of making art, defining it as “the conscious use of skill and creative imagination especially in the production of aesthetic objects.” In the end, defining art is often left up to the discretion of the individual viewer. To many, art may simply be a visual image with an aesthetic intent. In this sense, any AI created image that aesthetically appeals to some would certainly be a work of art. However, I would encourage a view of art as something that has an intent to express emotion, is based on experience, or says something broader about the world. If art is defined as such, then these are qualities that are intrinsically impossible for AI to have. In this sense, what AI generates may not be art at all—it is simply an output of data completely dependent on an algorithm that was written by humans.
This brings to mind the following question—who is the maker of this art? Is it fair to say that it was made by AI? Humans dictate the result by giving instructions and input to the computer. Still, the AI is analyzing the data, making calculations, connections, and inferences that its creators likely wouldn’t. Many have concluded that AI-based art is a partnership between a human and a computer, and that both have played an essential role in the creation of the final product. Furthermore, the computer brought no intention or purpose to the creation of the work; unlike a human, a computer doesn’t feel compelled to create or produce on its own account. It analyzes data and creates pixels that are appropriate based on the information it has learned. Through considering it in these terms, there is no comparison between computer-generated art and the creative and conceptual process that a human artist follows to express themselves visually.
In many ineffable ways, AI-generated art feels unnatural. Art is something so human and expressive that it seems to be one of the few things a computer cannot replace. AI art is a fascinating application of advancing technology and is certainly valid as its own medium. However, I can’t bring myself to place these two concepts in the same category. While AI may be able to produce a convincingly similar final product, the real value behind art comes from its meaning, intent, and relation to human experience.
https://drawingbot.azurewebsites.net/
https://news.artnet.com/art-world/rutgers-artificial-intelligence-art-1019066
Hero image: Photo by Michael Dziedzic on Unsplash
