Generative AI and the Arts: “art is essentially social”
I’ve read a recent essay by Ted Chiang in the New Yorker (my free article for the month!), on why generative AI does not make art. Butchering and simplifying, but basically — art is created by thousands of decisions, and, including their intentions, all these decisions separate true art from AI-generated art. I sort of agree with his essay, but I think stronger arguments may be made. I’d like to expand the conversation into a consideration of art’s relationship to society, as this strengthens the distinctions between art and AI-generated work, and has implications for what’s to come in the future.
The following is also a more considered reply to my own initial hot takes on the effect of generative AI on art. When generative AI was starting to take off — with the introduction of Dall-E —, I quickly extrapolated 10 directions which would follow in the aftermath. My musings at the time were instinctive and admittedly a bit glib, so I wish to fill out my thoughts more roundly here. I am led along these thoughts from past thinkers like Theodor Adorno and Arnold Hauser. As well as from my experiences watching the field of art change and develop today.
Chiang takes as a baseline a prosaic definition of art, that art is the result of a lot of choices. So for example, a 10000 word story might be the result of 10000 decisions, and these decisions have some intentions behind them. While I agree that decisions and intentions are crucial, I’ll lay out a different definition of art. The type of art I’m talking about is that activity which self-consciously puts itself forward as part of the larger conversation with other artworks we consider part of the history of art. This way of describing art may seem self-referential, but this I think is necessary. The self-consciousness is key, because it gets at the importance of intention and reflection in the activity of art. In this sense, art is always — in part, not entirely — a response to other artwork, putting itself in some relation to the constellation of other artworks. Also at the same time, in setting itself forward as art, it puts itself in relation to what is not art. This somewhat circular definition brings to the fore a consideration of what art is and is not, so allows a comparison to the products of generative AI.
This is not meant to be a philosophical treatise, so I don’t wish to walk down many boring alleyways, but laying it out this way clarifies that art is not simply within the individual (a set of decisions), nor is it simply a collection of objects from history. This conversation is dynamic and is prompted by pressures, both individual and societal. We see how art, and our understandings of art, have changed over history. It’s a continual conversation within a social environment, suffused by its time, and giving back to the culture at large. Part of it is within the artist, part of it is without.
Not all of these considerations need to be in the mind of the artist when making the work. Nevertheless, “art is essentially social”, and what Adorno means by this, is that artworks “await their interpretation.” By inviting reflection on its relation with history and other art, art asks to be interpreted, and awaits this as part of a fulfillment of its making. This speaks of a deep need within an artwork, to be completed by its reception and interpretation. And also why an artwork which does not demand anything from us falls into the background, and falls away from the conversation.
Unlike the view of a mountain range, art asks us to consider it in relation to the other things we have created, the world in which we have been cultivated, the future we envision for ourselves. There is a human point-of-view inherent in the work. This is even the case if an artwork is generated automatically somehow — the originary impulse still resides in the human intention and action. At the same time, the artwork resides in a social environment, so its interpretation is also to be discovered in a reflection on society, and not simply on the decisions of the individual artist.
This way of viewing art as “essentially social” risks a certain dryness in interpretation, possibly emphasizing its sociological characteristics over its sensual qualities. But there exists a more vital form of artistic interpretation. There is an interplay, in the gestation of a work, between the thousands of decisions making up a work, and the techniques employed in its making. The viewer is able to enter into this interplay through the evidence of making. In being absorbed in the making of the artwork, the viewer retraces the decisions of its genesis, the thoughts of the artist. The artwork becomes a set of potentials — it could have gone one way, or another. What may play out when absorbing a work is a virtual re-enactment of the making of the work. The ‘gesture’ in art is more than the manual act. It encompasses intention, projection, and decision. It is a distinct pleasure for the viewer to imagine being in the artist’s shoes.
This interplay between an artist’s decisions and the making of the work, allows for spontaneity to enter an artwork. “Spontaneity amid the involuntary is the vital element of art.” For Adorno, the spontaneous subject is a wellspring of “what it stores within itself”, and what’s stored are the sedimentations of a life of an experiencing subject. Namely, these sedimentations could be an artwork seen last week, or a conversation had yesterday, or the fallout from a breakup. These spontaneous moments found within the work can be giddy moments for the viewer, as they can speak eloquently of a shared sense of surprise or discovery.
Artworks are infused with the spirit that things could be different, that one decision could have gone another way. This is why art is connected with our notions of freedom, in action and in words. The glimmers of potential shimmering through art separate it from other objects. Art differs from other societal activities in that it does not have the same practical use. We easily see the fruits of manual labor or human services, and more practical fields produce discoveries which create new tools, new efficiencies. Art does not have the same goals, and this gives it access to our more personal desires. By giving free play to our own desires, we enter into the field of imagination, and of visions for other possibilities and worlds different from what we perceive. This delving within ourselves at the same time connects us to each other, as we project our personal desires into a larger world, into a vision for a society we want to live in, because we all share the same human abilities to want, to communicate, to interpret, to understand. Art provides this kind of conversation.
As well, artists have always played a special role in society. Significantly for the last couple centuries, artists have epitomized the role of a mobile element in society. The starving artist in the garret is the next day fêted by the aristocracy. The potentials of the artist’s making of artwork is writ larger as the potentials for human movement within society. Especially since the rise of the bourgeoisie, art is intimately tied to the possibilities of societal mobility, to considerations of freedom of action. Art-making asks what would we do if we could do anything. What is at play, what is up for grabs? This intimate connection with freedom (of action, of movement) gives art its utopian aspect. Within an artwork, we ask ourselves, what would do here if we were free to do anything? Within society, art ask us, what would we do in our lives if we were free to do anything?
This leads back to the nominal topic, to the new presence of AI in our lives. In the type of art which matters to us, generative AI is not playing the same game, so we won’t find the same pleasures and the same questioning we find with art. There is not the same reliving of art-making, nor the experiencing of an artistic comportment. If there is sedimentation within a work of generative AI, it is of data collection and pattern-matching. Listening to a piece of music composed by AI, we may see the ways in which it sounds like an original music creation; we may recognize that it fulfills a satisfaction that we experience when listening to other past works, that it matches our expectations of what a work in a particular genre might have. It will likely even surprise us in making certain ‘leaps’, playing with our expectations in a way that good music often does. So art created by generative AI will satisfy one side of the equation, making music to listen to while at the dentist, or a book to pass the time with in an airport. But it would miss an essential element of art, which is its demand to and from society.
That said, it could be that AI may have the eventual ability to participate in the conversation of art. At its current stage, generative AI is not there, but I could envision an eventual ‘experiencing subject’ that’s not human. But to me, more likely would be an incorporation of AI with our selves. Rather than asking a robot to create a digital painting, we may simply be accessing our internal link to an onboard computer that augments our perceptions and actions with multiple possibilities. We would continue to be art-making creatures, but possibly supercharged with greater capacities for access and techniques. And the resulting artwork would be eloquent, but not entirely of the artist's thoughts and intentions.
Whether AI can make art may be a ‘small potatoes’ question compared to a more hopeful inquiry — one that follows naturally from the considerations above — whether it could lead to a future where more of us participate in the conversation of art. The promise of AI is that it will free us from lives based on our labor, whether manual or analytical. In such a picture, the question which follows is: What would each of us do? A future where more people are able to ask this question of themselves could lead to a sea-change in art-making, broadening our conceptions of artists and art-making. Art-making would no longer be the playground of the well-off. There could be many worlds of art, including activities not traditionally incorporated. Adorno offers a glimpse of this enticing future: “Ultimately, perhaps, even carpets, ornaments, all nonfigural things longingly await their interpretation.”