Artificial Intelligence Invades the Artworld, Part (1): What Is Art, Anyway?
Recent developments in artificial intelligence allow users to produce digital imagery by entering search terms into AI art generators, such as DALL-E 2, Midjourney, and Stable Diffusion. The generators produce images by using machine learning algorithms in response to a set of input parameters or search terms (Alex McFarland). In this series of posts, I will take up a set of interrelated philosophical questions about the use of this technology: (1) Is the imagery created by AI art generators art? (2) If so, who is the artist—the AI art generator or the user of the AI generator? (3) Is AI-generated art (hereafter, “AI art”) just as valuable and desirable as art created by human beings? (4) Do we have reason to fear that AI art generators will replace artists or make them obsolete?
Anyone familiar with the history of art, or the philosophy of art, will know that there is no clear consensus about the necessary and sufficient conditions for a work of art. Theorists have developed accounts of art based upon its imitative quality, its aesthetic merit, its ability to express or embody emotion, its acceptance by the artworld, and more. But there is no account of art that is universally accepted. Moreover, throughout the history of art, various challenges have emerged to widely accepted ways of understanding art; in some cases, these challenges have prompted theoretical revisions concerning our understanding of art and its function. Before the camera, it was widely believed that the function of art was mimesis, or imitation. An oil portrait, for example, would count as a work of art because it provides a representation, or imitation, of the sitter. Indeed, the goal of most painters would have been to provide an accurate representation of the sitter (while perhaps also capturing something of the sitter’s spirit or personality). But the development and popularization of the camera called this conception of art into question. Why bother to paint a portrait of someone if the camera could provide a perfect representation? This prompted artists and theorists to reconsider the nature and function of painting. Perhaps the goal of a painter should not (or not only) be to represent reality. According to this revised view, a cubist painting, such as Picasso’s Girl with Mandolin, would count as art, even though it does not simply imitate reality. If artists are not constrained by imitation, then Jackson Pollock’s Blue Poles, Carmen Herrara’s Flight of Colors, and Hilma af Klint’s Trees of Knowledge No. 2 Series would also count as art.
At the same time, the camera presented an existential threat to artists. It was not unreasonable to fear that artists would become obsolete. Indeed, Charles Baudelaire, writer and art critic, referred to photography as “art’s most mortal enemy” (as cited in Kevin Roose). Why would anyone hire a portrait painter if they could simply, and more affordably, take a photograph of the would-be sitter? I can only speculate here, but perhaps there would be more professional portrait painters if we were all without our cameras and iPhones. Be this as it may, there are nevertheless many professional and amateur painters and artists alive and working today. Art is not obsolete. The camera did not send painters to the grave.
Other challenges to our understanding of art include Marcel Duchamp’s Fountain. Marcel Duchamp entered his piece, Fountain, into an art show that advertised itself as being open to all forms of art. Fountain was not a traditional art piece—a drawing, painting, etching, collage, etc.; rather, it was a simple, unglamorous, run-of-the-mill urinal. Moreover, Duchamp did not build the urinal himself; instead, he found and signed it (under the pseudonym R. Mutt). This piece, one of Duchamp’s “readymades,” sparked controversy and again prompted artists and theorists to reconsider the nature of art. To this day, many consider Fountain to be the most important work of 20th-century art.
Still, not everyone agrees that Fountain is a work of art. While many artists and ivory tower intellectuals consider Fountain to be a work of art, others are dubious about its status as art. Some deny that Fountain is a work of art because works of art must, it is believed, be pretty, inspiring, or—minimally—created through the skill and craftsmanship of the artist. If Fountain is a work of art, then why isn’t anything and everything a work of art? “But that’s just the point,” a defender of Duchamp might respond. Duchamp taught us that anything and everything – even a urinal—can be viewed as a work of art.
Like the camera and Duchamp’s readymades, AI art generators ask us to consider the nature and function of art. According to some—the critics—AI art generators do not actually produce art because art, by its very concept, must be created by a human being with an intention. Defenders of AI art have responded by claiming that AI art generators are simply another tool in the artist’s toolbox. Just as some artists use cameras or Adobe Photoshop as part of their creative process, other artists use AI art generators. And while the AI art generator itself may lack an intention, the human user does not. The human user intends to create a work of art through their use of the AI art generator. Thus, the artist who uses the generator creates art.
The artworld has had to confront these issues recently in response to several prize-winning AI-generated art images. For example, Boris Eldagsen won, but refused, an award for an AI-generated image, The Electrician, which he submitted to a photography show and competition. He claimed not to care about the award. Instead, he wanted to spark a conversation about AI art. On his website, he explains, “I applied as a cheeky monkey, to find out if the [competitions] are prepared for AI images to enter. They are not.” And Jason M. Allen received a blue ribbon for his entry, Théâtre D’opéra Spatial, in the Colorado State Fair’s annual art competition for emerging digital artists. Allen created the piece using the popular AI generator, Midjourney. Importantly, he did not try to deceive anyone about his use of the technology and entered the piece under the name, “Jason M. Allen via Midjourney.” (Roose) Still, Allen received a great deal of backlash on social media from those who believed that he deceived the judges, cheated, or simply failed to produce art.
If a readymade is art, and if a photograph is art, why not also say that an image created using AI is art? If the artist intends the piece to be art, who is to say that they are wrong? Given that, as we have seen, there is so little agreement about what art is in the first place, I do not think that it is especially fruitful or productive to get bogged down by this question. I would imagine that on some accounts it is, and on other accounts it isn’t. I will not quixotically hope to win over my readers once and for all with my own pet theory of art. (I will save that task for a time when I have more energy for chasing windmills!) There are, however, other important and worthwhile questions that we might want to ask about the development of AI art. In the next post in this series, I will take up one of these questions: Assuming that AI generated imagery is art, who is the creator? The AI art generator or the human user?
Sources:
Boris Eldagsen (artist’s website). Accessed June 9, 2023.
https://www.eldagsen.com/sony-world-photography-awards-2023/.
McFarland, Alex. “Beginners Guide to AI Generators,” Unite AI. Updated on February 28, 2023. Accessed June 9, 2023. https://www.unite.ai/beginners-guide-to-ai-image-generators/#:~:text=AI%20image%20generators%20work%20by,3D%20models%20and%20game%20assets.
Roose, Kevin, “An A.I.-Generated Picture Won an Art Prize. Artists Aren’t Happy,” The New York Times, Sept. 2, 2022. Accessed June 6, 20223. https://www.nytimes.com/2022/09/02/technology/ai-artificial-intelligence-artists.html.