Written with: Various devices | Keyboard | OnlyOffice | Windows 10
A quick preface:
I haven’t updated my Neocities in about five months. I started writing this post just over two weeks before uploading it. In the five-month period, I obsessively seeked out peace and quiet, a neutral state of mind which would last long enough for me to be productive. In the past two weeks, my life has seemed too chaotic to even think about writing. I’ve come to realise that this attempt to find “balance” is actually fairly unstable, and I might need more help than I’m currently asking for.
But more to the point of this post, there have been plenty of new developments since I began writing it. Some headlines suggest I have been too optimistic, that the future of the writers and artists of the world is much more bleak than I gave credit for. Others suggest that I’m merely fearmongering over nothing. Having acknowledged that, and being fully aware of the likelihood that any writing on a topic like this will likely age terribly, I want to upload it anyway. I’m not even going to proofread it again. I just want it to be known that I wrote four pages of something. Its value or lack thereof can’t be found unless it exists for others to read. Know that I live, and I write.
======
AI media creation has been in the news and social spheres pretty much non-stop for the past two months. I've been unsure how to think and feel about it.
On the one hand, this is happening almost exactly a year after the manufactured hype of the NFT fraud wave, and superficially this cultural moment shares a great deal with that. It's still an attempt by a handful of big tech vultures to bankroll a new investment category.
But where NFTs and crypto were comically vacant, a circus of non-production where nothing of value was made except the illusion of value, AI admittedly..."creates" "something". That is, a process occurs which results in a tangible product of sorts. It has limited but noteworthy real-world functions, and is already making good, to varying degrees, on its promise of "cutting out the middleman" in existing processes of creation. It creates copy writing, manufactures images, produces code, all at a cost-to-performance bargain of far fewer labour or training hours. It “does something”, like it or not.
But, like NFTs, the actual value being produced in this “new” process is obfuscated by silicon valley hype men with a vested interest in making numbers go up. Some proudly declare that the job of the “artist” - writer, illustrator, or musician - is doomed, because AI creations can (or soon will) replace it. There will simply be no point, supposedly, in paying artists to do what a "free" tool with access to a nebulously-obtained cloud of data can do, faster and just as well.
This concept is concerning in two different ways. First, and maybe more importantly, there are the practical implications for the large swathes of society who depend financially on an art or craft. These are people whose labour has, for the past few decades, been further and further devalued in capitalist society. They already struggle to find roles and career entry points which will allow them to survive off the back of a skill they have spent years – and often, exhorbitant tertiary education fees – honing for that exact purpose. Given that employers, producers, and anyone else in a position to pay them already have little respect for the artist’s abilities, the emergence of a tool which can “do the work” for cheap suggests an ever worsening situation for the artist – and by extension, a narrowing of who, societally, can actually afford to make art.
But the other half of the argument in favour of AI art is this: that if AI can create its own art, there is no point whatsoever to human art. This is idiotic for reasons that feel almost too obvious to spend time on. However, I still think that it can be a cause of genuine introspective anxiety for many artists, especially if we are facing a future where their work continues to be routinely dismissed. If the individual, personal creation of art is no longer economically viable, why pursue it at all? This is not a question of “can I spent my time making art”, but “should I even bother?” I think by focusing on this question, we can glean a better sense of (a) what value AI content creation actually has; (b) its difference from traditional, human-created media; and (c) how it might affect our existing systems of value.
These three points could be the basis of a much longer essay, but for now I want to write down one point which I think touches on all three of those topics. That is: despite the hype, even a layperson is able to notice an important difference between AI and human-created media. I firmly believe that this divide is not something only art critics and aesthetes are aware of; there is a fundamental, even obvious distinction, in form and in function, between these kinds of media. What’s more, this difference is so fundamental to the actual nature of AI content generation that it will never be bridged, at least until the “AI” producing art reaches the point of actual personhood (at which point we are no longer talking about today’s “AI art”, but rather the artistic expression of an entirely new category of being - but I digress).
Maybe an example will illustrate my point. Early in the AI hype wave, my dad started excitiedly showing me examples of a new subgenre of YouTube audiovisual content: “music videos” in which each lyric of a song had been fed into an AI image generator, producing elaborate visual artworks which, figuratively or literally, mirrored said lyric. On the one hand, this is exactly the kind of thing which internet artists have perceived as a threat, either to their already unfairly low status and perceived value in the current culture, or to their actual livelihoods, or both. A computer performing a role that a paid, trained human artist could do just as well, if not easily better? An outrage.
But consider again the premise of these videos: each lyric, interpreted in isolation, becomes a separate, distinct image, imagined entirely out of its original context. Obviously a human artist could take this on as an entertaining creative exercise, and could easily produce something far better. But most human artists would prefer to spend their time on a music video which actually has a consistent through-logic, a narrative, an artistic interplay with the entire original work, being the song. In other words, a human would create something that results from their understanding of the existing work.
The images produced by an AI are an imitation of understanding; information is extracted from a data pool, spliced, massaged by algorithms into something that feels like it makes sense. And in the case of these videos, this lack of true understanding is actually a boon. By asking a machine which cannot understand the original work to produce something based on its particular kind of “understanding”, we see something…new. Not the most original work, but still something which a human artist would be less likely to produce.
To the limited extent in which AI "sees" the world, and to the limited extent that it "creates art", it does both things differently to humans, and therefore has a fundamentally different skillset. There are things that AI can excel at which humans can't - largely due to AI's very limitations - just as there are aspects of human art which AI can never truly emulate.
AI is a tool with specific uses, and indeed many human artists have already realised this. Holly Herndon and Amnesia Scanner, for instance, use AI to create visual supplements to their musical work (among other things). A similar use-case to the videos my dad would send me, although in this case involving the creative input and oversight of accomplished, interesting visual and musical artists. For one thing, the difference that their human input provides is striking: their work is fascinating, sometimes haunting, sometimes beautiful. What's more, their work in this field predated the AI hype wave by several years.
Humans are humans, and are ultimately much better suited to the job of exploring the human experience through art than an algorithm would be. While not everyone might be able to articulate this difference, and certain tech bros with a vested interest in pouring money into AI might claim otherwise, I genuinely believe that everyone is at least subconsciously aware of this.
The looming anxiety behind that defence, though, is the question of what happens when AI is falsely presented as human art. I'm not talking strictly about "deepfake"-driven misinformation, which is a whole other discussion, but some of these points might be relevant to it. This concern is not just one of artist exploitation - of creative producers directing the flow of money away from hard-working, old-fashioned artists who are both deserving and in need. It also concerns all artistic images, including those made by humans: if AI art becomes "good enough" (whatever that means) to be indistinguishable from human art, how can we trust that any image is actually human-made? How can we suffuse any image at all with value, when any given one could have been pumped out by an algorithm?
I have a few responses to this. One brief one is that this feels more like a problem for the Western traditional art brain, one which values "authenticity" - a certificate of originality, guaranteed proof that a particular painting is the original, physical artefact rather than a forgery - over the content and even process behind the image itself. In reality, it is impossible to "waste" the value that we give onto images purely due to a question of authenticity. By way of analogy, imagine praying at the grave of a loved one every day for 10 years, only to find out that the wrong body had been buried in the their plot. Were those prayers "wasted"? No, because a prayer is not a transaction based on physical goods; it's an abstract process involving an abstract "object", being the memory (or soul, if you like) of the deceased. It has many "purposes", but one of those is the creation and maintaining of value. Likewise, I believe that "mistakenly" attributing human value to an AI image is no source of crisis.
Even so, I believe the responses we have to human and AI art are fundamentally different. Whatever the results of a blind "taste test" of human versus AI images might reveal, humans can and do create things which only humans can create. There is a "certain something" we, as artists and audiences, discover through art alone, an abstract and undefinable *meaningfulness* which we are able to imbue in creation. We will always seek out and find it. And this quest for meaningfulness results in art which is inimitably human.
My earnest hope for the near future of art is that, in the face of the "threat" of AI, artists recognise what does make their work unique, and pour their efforts into nurturing this. Because we can understand the human experience better than any electronic process - because we live through joy, tragedy, love, anxiety, etc ad infinite - we are uniquely equipped to pay tribute to and find meaning in these things through art. It’s unlikely that producer-types, those who actually wield the power to fund and make visible these works, will have the same realisation. That’s why artists and audiences need to fight for the work of artists, just as they already do. AI image generation wasn’t the first excuse not to pay artists, and it won’t be the last.
Arguably, this challenge - to create something that is distinct from mere natural processes, to eke out signatures of our own existence - has always been with humans, ever since we first started creating art. Consider, more recently, the art of Andy Goldsworthy, the English sculptor who uses the natural world as both material and art gallery. These impermanent, fleeting installations are composed only of what he finds in the outdoors: piles of leaves, rocks, dirt, sand. Normally they are quickly eroded by wind or water. But by creating patterns in these objects, by seeing the world through human eyes, by making space for his own process of creation among the infinite processes which already exist within nature, he *does something*. And anyone looking at his work, regardless of artistic or critical education, can recognise that his works are human, despite how minimal his intervention in the natural environment sometimes is.
Maybe this has all been an overlong statement of the obvious. As I said earlier, I do believe that everyone knows this, at least subconsciously, but for artists in particular, this is tiresome old news. Most don't need to be told that their work has value; they're already busy enough trying to convey this to the key-holders of capitalist society, who have been routinely devaluing art for decades. Still, AI has (limited) usefulness as a tool, not just for "content farms", but for artists themselves. It can do things artists can't, but it cannot do all that artists do.
And furthermore, I think that AI, for all its flaws and dangers, offers a new opportunity to remind ourselves why we make art in the first place. It is a reminder of what art actually *does*, and what artists *do* in creating it. The more practical concern is finding a way to both create art authentically and survive in this current stage of history. But if we lose sight of what art is and does, we suffer another kind of death, one where we ourselves forget the value of what we do. In such times of uncertainty, these questions become more, not less, relevant.