AI and 'art'
AI is an interesting term. Covering everything from simple decision trees (where you can see and understand exactly how the system makes decisions at each point in the decision-making chain) to crazy black-box systems (where you know what goes in and what comes out, but can’t get an understanding of what’s actually happening in-between.) Being so broad, its actually quite difficult to define; Douglas Adams1
once described technology as "stuff that doesn't work yet", and AI has a similar problem. The AI Effect says that as soon as AI successfully solves a problem, that solution method is no longer within the domain of AI
. An AI chatbot based on decision trees is "just" a decision tree - not AI. ChatGPT is "just" a Large Language Model - not something that actually understands what you're asking it (or even how its responding to a prompt.)
The cleverer AI gets - or rather, the cleverer it appears to get - the more impressive it seems. But seeing words appear that look like ‘real’, natural language, or images that ‘look good’ and fit the prompt you fed into the system is one thing; having an ‘application’ or a reason to do that in the first place is another. Or to put it another way; I can see what it does and how to use it, but it isn’t quite as clear what its actually for.
“Art” is always an interesting term. Its also an incredibly difficult thing to define - even anti-art, a movement which aims to reject existing definitions of “art”, is still a form of art. We use the same word to describe the painting of the Sistine Chapel that we use to describe a toddler's first experiments with a paintbrush. And again, the question of what its actually for is something different.
I think a lot of what makes 'art' comes down to intent.
But is it art?
On the wall behind my desk at home, I have a copy of Magritte’s “The Treachery of Images” (the famous painting of a pipe with “this is not a pipe” below it, written in French.) With this piece of art, Magritte tries to tell us something about the human condition; the way we see “a pipe”, and the words “this is not a pipe” creates a sense of cognitive dissonance (“How can you say “this is not a pipe” when it obviously is a pipe?”), which is only resolved by the realisation that it isn’t a pipe, just an image of a pipe; you can’t smoke it, or stuff it with tobacco, or gesticulate with it when you’re making a point. You could make a similar statement with words (eg. “The map is not the territory”), but the point of art - I think - is to express something that goes beyond words. (A lot of Magritte’s work was specifically around the similarities and differences in how we process and interpret words and images - see "Les Mots et les images" for an example.)
The copy of it on my wall is - I would argue - a piece of art in itself. Is it Magritte’s artwork? Not really - its a cheap copy; a printout of a fairly low-resolution image I found on the internet, printed on a poorly looked-after inkjet printer in need of a clean, on cheap paper. A far cry from the 60x80 inch oil-on-canvas original. There are stripes where the printer ink heads obviously needed cleaning when I printed it out. The paper is unframed, yellowing slightly, and stuck up with bits of blu-tac. In the background of my video calls, its supposed to be there as a personal reminder that data about the world is not the same as the world; just a representation of a particular point of view. But its also a private joke; I could easily have ordered a decent print, or properly cleaned the printer before printing it, or used some of the better quality paper that I have in my cupboard, or put it in a frame. I could have got out some paint and a paintbrush and made my own version of it that way. But I didn't want it to be 'the best' copy of the original painting; I wanted it to be an obviously cheap copy, put up in a place where, in a post-pandemic world, I’m seen by dozens, if not hundreds of people. In that context- on display alongside some of my children’s (and wife’s) lockdown artwork, it means something slightly different; on top of what Magritte wanted to say, I want it to say something about intellectual property, about ‘appropriating’ someone’s work, about how we use cultural artefacts as an expression of identity.
(I can't make my mind up whether or not I should add a little card next to it; maybe the addition of a small card with the carefully laid out text saying something like Scott Thompson, 2020: “Ceci n’est pas une Magritte”. Inkjet print on paper, 8.3" x 11.7" would turn it into 'real' art. Maybe it needs to be in a 'real' gallery for that to count...)
I’m not saying that its good art, or even necessarily an original idea, but I do think it qualifies as “art” - at least as much as, say, a piece of stencilled graffiti, or an unmade bed or a urinal placed in a gallery environment.
When something like AI is used to create “a picture in the style of...” or “a poem in the style of...”, what comes out is something in a given style of painting, or the style of writing. A painting might have the kind of colours, composition and brush strokes that the artist is most associated with; the poem might have the same kind of rhythm, use the same kind of words etc. To me, the point of ‘art’ is to say something more than what you see on the surface; the difference between a photograph and a painting isn’t that the photograph is a ‘better’ reflection of the image the artist saw with their eyes (which it is), but that the painting is a ‘better’ reflection of the image the artist saw in their mind.
And I think that’s the fundamental problem with talking about AI-generated “art”; the AI is certainly doing clever, amazing things. It can create pieces of text that are impressive to read, and beautiful images to look at. It can stimulate emotional reactions. But the “art” - if there is any - isn’t really in the output, but the prompts fed into the system by a human being with an idea of what they want to come out of it. There is no “image in the artists mind”, because there is no artist. Just a network of classifications of what does or does not fit with the prompt before the machine randomly generates a new set of variations to be judged against the same criteria until the algorithm - which has no real understanding beneath a shallow surface level - is satisfied.
…Or is it 'Content'?
There’s a whole industry around “content”; there's stuff we want to share with our friends, the stuff we want to share with our communities, the stuff we want to share with the world that all feeds into the 'content libraries' of online platforms, sorted and filtered by their algorithms to populate other people’s newsfeeds. Then there's the stuff that gets created by brands in the hope that it will be shared by people, the stuff that gets created by brands that they pay to be put in front of people, the stuff that gets created just to populate a channel and feed the algorithms with the kind of data signals that lead to more people seeing the content ("Like for yes, comment for no!").
I have no doubt at all that AI-generated content is going to be a paradigm shift in the future of "online media" (that is, the social networks that are less and less about the people you have 'relationships' with... and TikTok.) Probably many more industries besides.
But “art”? No; in the same way that the Fairlight sampler/synthesiser/sequencer led to new ways that musicians could make music, or that photography led to new ways that people could make pictures, AI is leading to new ways that artists can create art.
But its not a revolution; its just another tool. In the same way that people still want to hear an orchestra in a hall, or a brass band marching as part of a state funeral, or a bunch of teenagers with guitars making a racket in a dark and sweaty arena, I don’t believe AI-generated “art” is going to disrupt anything in the world of “art”. Its a new tool, the way that Photoshop was once a new tool for graphic designers, acrylic paint was once a new tool for painters, or the sampler/drum machine/electric guitar/multi-track recorder was once a new tool for musicians. A fully-equipped music studio can make noises, but it can’t make music without someone having some sort of idea of what to do.
The looser we are with what we mean by "art" (as opposed to "content"), the harder it becomes to understand its true power. Likewise, unless we get a bit more specific about what we're talking about when we say "AI", the harder it is to really understand it. We've been here before, with "digital" and "data", and more recently "metaverse".
There's a Marshall McLuhan quote that I think is quite illuminating on the topic- on the subject of media analysis he said;
Our conventional response to all media, namely that it is how they are used that counts, is the numb stance of the technological idiot. For the “content” of a medium is like the juicy piece of meat carried by the burglar to distract the watchdog of the mind. […] The serious artist is the only person able to encounter technology with impunity, just because he is an expert aware of the changes in sense perception
- Marshall McLuhan in Understanding Media: The Extensions of Man, 1964; p.18
Its possible to read McLuhan's words and understand what he was trying to say from interpreting the language alone. Its a little easier when you read it in context, and understand that McLuhan uses the terms "media" and "technology" almost interchangeably.
Technology changes your relationship with the world; a world full of guns is clearly a more dangerous world than a world without guns, but a world where you have a gun (I expect) feels safer than a world where your guns are taken away from you. (At least, thats what I interpret from reading tweets about gun rights…) So, AI technology is changing the world - in different ways for those who use it, and for those who don't/won't/can't.
Its easier still to understand McLuhan's point when you understand that he was an English professor and very well read, and familiar with T.S. Eliot;
The chief use of the ‘meaning’ of a poem, in the ordinary sense, may be (for here again I am speaking of some kinds of poetry and not all) to satisfy one habit of the reader, to keep his mind diverted and quiet, while the poem does its work upon him: much as the imaginary burglar is always provided with a nice piece of meat for the house-dog..
- T.S. Eliot — ‘The Use of Poetry and the Use of Criticism’, 1933
'Art' is what 'does its work' on its audience, while 'content' is merely the juicy piece of meat. And AI - or at least, the kind of AI that is built around a large language model (which understands what words 'fit' where, without ever really understanding what those words truly mean to a reader) - is no more capable of creating anything more than 'content' than a paintbrush is.
(At least for now.)