An art competition at a US state fair stirred the entire art world last September. The winning entry, ‘Théâtre D’opéra Spatial’, a work by Jason Allen created through Midjourney, sparked conversations online (mostly on the side of disapproval and/or alarm) about the usage of Artificial Intelligence (AI) to create art. In a rather simplistic and demoralized conclusion to the emergence of this technology, Jason Allen encouraged fellow artists to accept it because for him, “Art is dead. It’s over. A.I. won. Humans lost.”
Much has already been said about the topic, especially from artists whose livelihoods are imperiled by this phenomenon in a lot of ways. In this piece, I will not rehash the usual criticisms, but instead offer a different perspective on the situation.
My idea is that AI art is a symptom of our postmodern condition. Through the lens of Fredric Jameson’s ‘Postmodernism, or The Cultural Logic of Late Capitalism’, I will attempt to critique AI art, in consideration of the fact that his text is seminal in understanding the attributes of the postmodern world.
A semi-semiotic explanation of AI-produced art
Removing technical nuance, basic semiotics can provide a simple explanation of how AI can produce images. Simply put, these are machines that have no concept of the signifier and the signified. Through a method called deep learning, they are then trained to match the signifier to the signified by using images from the internet (a point of ethical contention from artists whose works are uploaded online, with or without their consent). Once they have learned specific variables to correctly identify a sign, they take note of these as dimensions in their “brain”, more specifically, in their latent space. Each image-making AI is different from the other because they have different developers feeding them with different image datasets, affecting how they distinguish signs (in a semiotic sense). In short, each AI artist (if I may call them that) has their own imagination, their unique latent space.
A user first interacts with these AI by inputting a text prompt that they want an image of. The machine then internalizes the signified (input text) using the learnings stored in their latent space and produces a signifier (output image) using a process called ‘diffusion’. That term is where the actual name of these AI systems comes from: they are formally called ‘diffusion models’. Aside from latent spaces, these systems are different from each other because of how their diffusion models are engineered. Known examples of these models are DALL-E 2, Stable Diffusion, and Midjourney – the last of which is what Jason Allen used to produce his winning artwork.
To my understanding, the diffusion model is the paintbrush of the AI as it is the mechanism that transforms the information from the latent space – their brain, their imagination – to something identifiable by humans.
Applying pastiche and simulacrum
Despite being written in 1984, Jameson’s definition of pastiche resonates with the matter at hand in the 20th century, providing a possible entry point for interpreting this contemporary mode of manufacturing art or images. He described that in a postmodern cultural production,
“the producers of culture have nowhere to turn but to the past: the imitation of dead styles, speech through all the masks and voices stored up in the imaginary museum of a now global culture.”
He also related pastiche to the architectural term ‘historicism’ which
“randomly and without principle but with gusto cannibalizes all the architectural styles of the past and combines them in overstimulating ensembles.”
Akin to artists looking at the past for reference or inspiration to concoct a new creation, these diffusion models are also digesting existing images, albeit at a much different scale compared to the human mind, and seemingly larger, faster, and more powerful in terms of potential. Its current unrestrictedness, legally and morally, is a ground for the general distress towards these machinations.
While its algorithm has been determined by lines of code – thought and typed by a human being – the exact operation of the latent space and diffusion remains partially abstract, like a semi-black box. We may understand the concept, but we cannot truly know how it fully works internally. As such, its image production can be seen as a “random cannibalization of styles”, a pastiche. This can also be linked to simulacrum, in which the resulting object is a copy of something that has never existed, of a phantom in the latent space.
Quoting Marxist theorist Guy Debord, Jameson extended that the culture of simulacrum is
“a society of which … ‘the image has become the final form of commodity reification.’”
Enter capitalism.
Capitalistic tendency of postmodernity
There are several specific arguments already by artists and thinkers against the usage of AI for production, which I will not go into detail anymore. As an agent in the art world, as a consumer and observer rather than a producer, my apprehension towards this hotly debated technology stems from the natural tendency of avarice in a capitalistic world, resulting in a paradoxical progress: bountiful but unproductive. In an art market that is already largely driven by sales – a stagnated one, in my opinion – how else can art organically flourish if another wealth-generating apparatus enters the scene?
When I read the section on capitalism in Jameson’s text, I felt that my misgivings are parroted, especially in the part where he suggested that
“our faulty representations of some immense communicational and computer network are themselves but a distorted figuration of something even deeper, namely the whole world system of present-day multinational capitalism.”
The general uncertainty and unclarity towards the enormity of AI is reminiscent of the sublime, an aesthetic concept forwarded by Edmund Burke in the 17th century, which Jameson summarized as
“an experience bordering on terror, the fitful glimpse, in astonishment, stupor and awe, of what was so enormous as to crush human life altogether: a description then refine by Kant to include the question of representation … but also of the limits of figuration and the incapacity of the human mind to give representation to such enormous forces.”
With that, AI art provokes a postmodern sublimity.
Depthlessness and uncanny valley
Also shown as the cover of the book compiling Jameson’s postmodernist texts, Andy Warhol’s ‘Diamond Dust Shoes’ (1980) was used to illustrate the depthless quality of postmodern pieces, using the modernist work Vincent Van Gogh’s ‘A Pair of Boots’ (1887) as a point of comparison. He described Warhol’s shoes as “a random collection of dead objects, hanging together on the canvas like so many turnips” with a “deathly quality” conferred by the “photographic/negative” (a then-novel technology that was treated with the same trepidation as AI now). Aside from Van Gogh, he also referenced Edvard Munch’s ‘The Scream’ (1893) to demonstrate the rootedness of modernist works in the self and its expression, a doorway to their interpretation. On the other hand, the postmodern allows freedom from these modern enclosures, entailing the loss of an emotional anchor for explication, marking it as
“a liberation from every other kind of feeling as well, since there is no longer a self present to do the feeling.”
In essence, that concept – which he forwarded as the ‘waning of affect’ – is relevant to works produced by AI. Most especially that the creator, like its creation, is also arguably bereft of a present self. It runs on a coded algorithm, its bones are bits and bytes of electrical information. To me, this is what induces an undercurrent of unease when looking at AI art, a sense of being in a liminal space between human and computer, a feeling called uncanny valley.
However, diffusion models are continuously growing. We are only at the cusp of its infancy. With its unimaginable speed to learn from human-generated images and art, soon enough, it will reach a point of obscuring its artificiality, of generating authenticity as if it was created by a human hand.
A postmodern approach to a postmodern issue
So now, how do we approach the developing phenomenon of the usage of AI as a means to generate art? Jameson simplified the position of Marx towards capitalism, which can be extended to postmodernism, by saying:
“We are, somehow, to lift our minds to a point at which it is possible to understand that capitalism is at one and the same time the best thing that has ever happened to the human race, and the worst.”
Ending with a question rather than a specific solution, in a similar spirit, I also summon my textual inspiration, Jameson, who asked*:
“Can we in fact identify some ‘moment of truth’ within the more evident ‘moments of falsehood’ of postmodern culture?”
* He also answered it in his text, a conclusion I clearly did not follow.
This was written for Art Criticism (Art Stud 255) course in UP Diliman, 1st Semester AY 22-23.