On 26th August 2022, Discord user Sincarnate shared his latest achievement on a Discord group: winning the annual Colorado State Fair Fine Arts Competition under the category of Digital Arts / Digitally Manipulated Photography. The man behind the avatar is John Allen, and he has become the world’s first artist who has won a fine arts competition by utilising Midjourney, an image-generating artificial intelligence (AI) which takes in a written prompt as a parameter. The artwork is titled “Theatre D’opera Spatial” (Space Opera Theatre).
Parallels to Photography
The New York Times was the first news agency to pick up on the news, and the writer of the article, Kevin Roose, has rightfully pointed out that the controversy surrounding Mr Allen’s artwork, is reminiscent of the advent of photography in the 19th century. The comparison, however, can be quite jarring, considering that the invention of photography seems far behind us, to the point where we all carry a camera within our smartphones. Somehow, it seems that the corridor of history has wound itself towards a familiar door.
We know from art history that the birth of photography coincided with the early beginnings of the Impressionist movement, and one could guess that the camera’s accuracy in replicating the scenery on metal plates must have been affirming to the impressionists that many assumptions about art, including the idea that art had to be defined stubbornly by an approximation to reality, should be challenged. The difference between visiting history and living in it, however, is the uncertainty of how art will react, or should react, and how it might evolve from here.
Understanding the Backlash
In the meantime, the discussion surrounding AI-generated art continues. One of the main criticisms of AI-generated art is that relying on artificial intelligence is a form of cheating. In allowing verbal descriptions to be the only input needed to generate stunning artworks, the fine arts have transformed from a coalescence of visual language, to one that is linguistic. We might have lived with the idea that a picture speaks a thousand words, but now it seems it takes less than a handful to create one. There is an instinctive sense of betrayal that arises from that shift. Man-made art is not born from miracles; it is not born from the sudden conjuring of an idea. How many years would an artist spend drawing and painting, to finally replicate the vision that they saw in their heads? Now, two words of a prompt, like “space western”, could shorten two years of practice into two minutes, perhaps even less. By narrowing the time and effort it takes to create a piece of art, image-generating AIs can seem like a shortcut to artistry.
Since the effort needed to create an artwork is seemingly diminished, these artificial intelligences instigate a sense of anxiety in artists. It is not out of the realm of possibility that these technologies could replace graphic designers, digital animators, and even freelance artists. It is difficult to avoid the irony here. Artificial intelligence has often been argued to release human workers from dangerous and menial jobs, allowing them to pursue more creative and higher-skilled careers. In fact, robots are steadily replacing workers in the manufacturing industry. So what is the point of artificial intelligence encroaching on the domain of art, a domain which we have always assumed to be inherently human?
Surprisingly, the idea of art creation being a computational problem is one older than the invention of the computer itself. A whole century before Alan Turing’s pioneering idea of the Universal Turing Machine, which has been widely accepted as the basis for the first computer, Ada Lovelace wrote the world’s first complex computer programme. More commonly known as Note G, it is an appendix Lovelace authored which contained her description of an algorithm which could calculate Bernoulli numbers recursively. Almost two hundred years before our confrontations with AI-generated art, Lovelace had already imagined that a computer could possibly orchestrate music from inputs.
“Supposing that the fundamental relations of pitched sounds in the science of harmony and of musical composition were susceptible of such expression and adaptations, the engine might combose elaborate and scientific pieces of music of any degree of complexity or extent.”
Extract of Ada Lovelace’s notes in Taylor’s Scientific Memoirs, August 1843
There is almost a sense of predestination in learning about this piece of history, as if artificial intelligence was always going to intrude upon the art world at some point.
/imagine [prompt] a silver lining
Much, however, can still be said in regard to the allegations made against image-generating AIs. There is still the possibility that these technologies could heighten our standards for art, perhaps even challenge our current understanding of how an artwork should look. Since the time taken to create a piece of digital art is heavily reduced, artists could create highly detailed artworks by using platforms like Midjourney to generate detailed backgrounds, upon which they could further add details to their liking (an example). Yet, that is not to say that details and approximations to reality are what make a good artwork. The late Mark Rothko’s works were abstract and intended to give a spiritual experience. In his personal statement, he writes:
I insist upon the equal existence of the world engendered in the mind and the world engendered by God outside of it.
An extract of Mark Rothko’s Personal Statement, 1945
To add more details than we ever have, or to choose to lose more details in our artworks moving forward, could all pose a new direction which contemporary art could take (although of course, forecasting is hardly ever right.)
Furthermore, the creation of new styles would be more highly valued. Even though you could recreate images in the style of known artists or movements, these artworks would be restricted to those styles. Ideating a new style is still a boundary that image-generating AIs have not crossed.
Final Thoughts
These are all new possibilities that could unfold in response to image-generating AIs. What happens from here—only time will tell. One thing is clear though: the advent of image-generating artificial intelligences has not killed art. It is still possible to disagree with Mr Allen’s claim that “art is dead, dude”.
Regardless of how phenomenally beautiful AI-generated art has become, art is an act of creation that is not specific to a particular form of intelligence. The existence of image-generating AIs does not nullify the existence of current human artists. Mr Allen won a prize at a fine arts competition, there is no doubt about that, but does every artist who entered and failed to win a prize surrender their desire to continue making meaningful artworks? Even if image-generating AIs come to dominate, there will still be a portion of the internet, or an exhibition in the city, that continues to display the culmination of an artist’s efforts to derive beauty from the plain.
Further Readings:
Dall-E: Creating Images From Text