From: Tomi Slotte Dufva
Subject: The Death of AI Art
Format: A̶S̶C̶I̶I̶ HTML+js+god nows what the webpages use these days
Average Reading Time:Time will tell.
Signal to Noise Ratio: Only known by algorithms
Genre: provocation (copycat)
The recent trend with technology is to bet heavily on Artificial Intelligence. From automating tiresome factory jobs, AI is now moving into more cognition demanding jobs. Not only truck or taxi drivers are threatened, but doctors, nurses, and many others as well. Similarly, art and design worlds are preparing for the new intelligent era.
Behind these near-future visions lies a plethora of lobbying, visions, fantasies, and misunderstandings. From the techno-fantasies of singularity afforded by strong AI, to the more down to earth optimization of the workforce for maximum profit, these visions utilize and modify vocabulary usually reserved to humans and some other mammals: AI is creative; AI is intelligent; AI learns, and so on. In doing so, it completely realigns the meanings of these words. These new realigned concepts then both further the misunderstanding and cement the new concepts into our collective discourse.
Some of the AI visions are most likely going to happen. In fact, we already have AI doctors, nurses, companion for senior citizens and nannies, to name a few. Similarly, the art world is purportedly being disrupted by AI, as AI seems to be able to churn both realistic and dreamlike imageries, astonishing spectators with its capabilities. The current situation bears a resemblance to the late nineties convergence of computers, communication, and television. In 1996 Lev Manovich wrote his polemic piece on the death of the computer art. Manovich writes:
“Lots of people talk about the coming convergence of computers,
communication and television. This convergence will probably happen. In fact, judging from the new models of personal computers which are clearly being positioned as consumer electronic devices (incorporating answering machine and TV cards in them), it is indeed well underway.
Those of us who work with digital art often debate another convergence
– the convergence between the art world and the computer art world. I
recently came to the conclusion that this particular convergence will
NOT happen. Below are the reasons why.”
Manovich then goes to depict two art worlds, that of the contemporary art world, which he names the Duchamp-land and that of the new media art world -the Turing-land. The basic gist (I am surely not doing the right to the original text, so please go and read it if you will) of it is that the modus operandi of these two art worlds are so different that they will never meet. Duchamp-land operates with an ironical, self-referential focus on the subject; it is interested in the content, and it often has complicated meanings.
Turing-land, instead of content, has gravitated towards the allure of new technology. As such, the focus remains uncomplicated, simple, and lacks irony. Although, Manovich acknowledges that there are contemporary artists in Duchamp-land that do use the new technology, but concludes:
“What we should not expect from Turing-land is art which will be accepted in Duchamp-land. Duchamp-land wants art, not research into new aesthetic possibilities of new media.”
This divergence between content (and its multiple meanings) and fascination for the new aesthetic possibilities enabled by new technology is something we are currently seeing in the world of AI art. I would argue that the world of AI art, as it is now, is too fascinated by the veneer of new algorithms, drowned in the concepts of navigating the latent space, machine teaching, machine learning, and artificial creativity, and thus can never integrate with the current art world. It is too focused on the surface, without much thought on the content -or being able to look that aesthetic ironically. For instance, GAN images are more akin to a new Photoshop filter, that is able to produce obscure random images, than a revolutionary new way of creative thinking. Jonathan Jones summed this up for an exhibition critique for Guardian: (With the excerpt I do not mean in any way to focus on Klingemann; I find some of his work very intriguing, Jones’s text is here to illustrate an aspect common in many AI-related works -that they lack content we usually assume contemporary art has. Whether or not content is needed for art is, of course, open for debate.)
“Mario Klingemann’s piece Circuit Training makes the process by which machines learn explicit. Klingemann creates art using a programme that can generate images and then criticise its own work. You can join in by deciding which images are “interesting” as human faces morphed by the machine flow by on screen. It’s one of the most boring works of art I’ve ever experienced. The mutant faces are not meaningful or significant in any way. There’s clearly no more “intelligence” behind them than in a photocopier that accidentally produces “interesting” degradations.”
In fact, the current AI related exhibitions seem almost silly: who in this day and age would seriously, non-ironically, come up with an oil color exhibition, or an exhibition focused wholly on the new abilities of synthetic fiber brushes (they are so much more intelligent brushes than the animal fur ones!)
However, seriously, the reason I am having trouble with the AI movement is the lack of in-depth perspectives, critique, and the possibility to see the subject ironically. The way I interpret irony bears a resemblance to how Timothy Morton describes irony in the context of ecological awareness:
There’s various different types of irony. One type of irony involves achieving a kind of escape velocity from the phenomena that you are finding ironic. And another type of irony involves realising that you’re caught in these phenomena in some way that’s inextricable. And that’s the kind of irony that I’m talking about. So ecological awareness’ irony is downwards in direction, rather than upwards. We like to think in our anthropocentric way, that irony means that you transcended something, but actually what it means is that you’ve realised that you’re stuck in something and you have this kind of uncanny awareness of that, and there’s not much you can do about that feeling of stuckness.
Such an ironic view on AI and art would allow for more alternative perspectives, that are not stuck on the latest enhancements in algorithm design, or on getting even faster and efficient GPU's and CPU's. Instead, such work could start to question the paths forward from much more extensive plain — moving away from the lethargic "explorations" of the latent space to rethinking the concepts of machine thinking and creativity. From toying with post-anthropocentric ideas with "thinking machines" that are as inspiring as a desk in open office space, to moving the thinking to consider what does posthuman, post-anthropocentric views look from the machine side of things.
All in all, I do not mean AI art would be categorically uninteresting; instead, the current attitude and comprehension of it are - if it remains as such AI art will be a short-lived fad.