Imagine a lego set that constructs itself into the image on the box. You open the container and pour out the pieces and then by some technological wizardry the little plastic blocks pivot and connect into a finished set. Would it still be a Lego set? Is it not the proposition of Lego an opportunity to build something? Perhaps something of your own design?
In recent months we have witnessed the emergence of imagery generated by Artificial Intelligence (AI). These peculiar pictures have bubbled to the surface of every social media stream, blog, and forum that features art. Similar technology has popped up in every creative field, from music to writing to the performing arts. As someone who makes art for a living – who assembles his own Lego – I have some thoughts.
My name is Luke Oram. I am an artist and illustrator working in the music, publishing and entertainment industries. I have been making a living on my art skills for long enough to have insight into the theory and methodology of art. I can foresee how AI generation tools will relate to my practice and to our shared visual culture, which is the turbulent stream on which all artists sail haphazardly.
For as long as I’ve been any kind of artist (from before I learned to read or write if the juvenile scribbles are proof) there has been discourse around technology and art. How does the power of computers impact what we do as artists? Where is the unique thing we do that makes art unique and worthwhile? What are the ends of the mass democratising power of software?
Through all of the innovations and adaptations of technology in the past there remained a sense of the authorial hand behind the software. I could feel it in the works of others and see my own intent propagate through whatever tools I could get my hands on. This is important because I believe technology has been overwhelmingly positive for art and image production. I don’t fetishise the methodology of the past or place much stock in received notions of proper practice. I’m happy to embrace whatever systems I find available to me in order to create the works I have in mind. I do not think I am a luddite.
Despite this pro-technology outlook I think this emergent phenomenon, AI generated imagery, is bad for art and for artists. Not only is it bad, it’s bad in a number of ways. Let’s investigate what I mean by saying this.
The first argument against the widespread use of AI generated imagery seems so self-evident as to not require stating. I shall state it anyway: I sell art. I paint and draw images for clients. It’s art, of course, but it’s also work and how I support myself and make my living.
An artist pushes their little boat out into the stream and hopes that things eventually work out; these days they call that precarious. Art and its industry must respond to this reality.
An artist’s ability to create art is reliant on their ability to support themselves, to make a living. Fine artists are able to do this by selling their work directly and are less directly threatened by these emerging technologies. Most of us are not fine artists. Most of us rely on commercial clients commissioning work from us. If those clients cease to commission work from us our ability to create art collapses. If it is not possible to make a living making art then people will not pursue art as a career, accelerating and solidifying the process of automation.
Visual imagination has a particular way of operating and it does so in specific spaces, at least when it’s manifested into a finished work of art. Those of us who make art must contend within that space. In real terms this means that our vast works of talent, effort and skill must compete with scrawls created by photobashers, tracers, plagiarists; scrawls lacking all talent, effort, and skill. These scrawls can, and do, supersede the work of skilled artists. It happens all the time. They outcompete us on price or on malleability to corporate interference in the creative process. We compete with such work because we must; there is simply no alternative if we wish to operate professionally within the spaces where there is work to be had.
That’s the deal.
We can invest in the development of our skills and knowledge, study at the most prestigious institutions, publish work for the hottest clients but, no matter the investment, the study, or the success that we have, we will always be casting our work into the stream of shared visual culture hoping that its vaguely-charted currents might carry the work past a few more eyes. We hope that those eyes might belong to someone who needs or wants our work for their own projects, businesses, or pleasure.
I am also a set of eyes diligently peering into the surf for whatever it may bring me even as I hope for a wave to carry my own little boat out into that lauded stream. The very first time I encountered a generated image with the full understanding of how it was produced, I immediately grasped that the computer would close whatever gap existed between the best artist and itself in a witheringly small scrump of time.
A technology that can produce work that can compete with my own while being cheaper is going to reduce the value of my work. A tireless algorithmic image maker who requires neither payment nor patience nor praise is going to undermine a comparatively slow, comparatively particular, somewhat mortal artist such as myself. This represents a disaster for my art practice.
But the disaster does not end with my art practice alone. There is a broader impact, on art and on the stream of shared visual culture.
Many AI proponents seem interested only in learning what words to type in order to get similar results to what they have already observed. This projection of ownership into the realm of initial ideas became apparent very early on, with comments beneath publicly posted AI images quickly progressing from the seemingly innocent, “Oh, cool! What prompts did you use?” to the borderline paranoid (and utterly ridiculous and contemptible), “You stole my prompts.”
There is a need to project authorship onto a system specifically designed to eliminate it. Sailing an automatic boat out to sea we still claim to be its captain. Are we not the owners of the automatic Lego?
The culture forming around the use of AI generated imagery and art automation is attempting to position the user as a legitimate artist (a ‘Whisperer’ in current parlance). This is reflected in the seeking of prompt ownership. Ironically, users of AI generation technology are demanding the same protections for which we, actual artists, were accused of ‘gatekeeping’ for demanding.
Instead of a wellspring of opportunity in this technology I see a fetish for automation that is packaged with the same dismissive irony as other “disruptive tech” products. When asked how it will benefit me as an artist, proponents of AI generated imagery either tell me that it is inevitable or that it can help me come up with ideas. This proposal, to a trained artist, is abyssal. Coming up with ideas is one of the most rewarding and enjoyable aspects of art; it is not a problem or a drudgery that can be solved by automation but a core responsibility of the artist and a source of joy.
Those who tout the inevitability of AI generated imagery may not be able to answer the question of how it will improve my work or my life but they are correct in the most basic possible use of the term. It is here, and here to stay. What does that mean for art, from my perspective? Where does my understanding of art diverge from that of the AI proponents?
The project of building a unified and universally accepted understanding of art is a doomed project. Art transcends the view and language of any one person or school of thought. Because I cannot understand the entirety of what art is I judge the artistic merit of how a tool or technology is being used – and in whose interests.
Art is, for me, not just mundane technique. I have dedicated most of my life and my entire adulthood to the meaningful practice of art. There are skills associated with the production of art and we sell those skills and their products. We do not, however, make art to satisfy the demand for its products, as do the AI companies. We do that part of it to make a living and support the rest but it is not the whole of what we do. Reducing the scope of art to its transactions throws the baby out and keeps the bathwater.
So what is art? One collides with this a multitude of times in primary education alone, then more and more if one pursues the study of art. At the level of higher art education this question features thousands of texts with increasingly baroque language and deepening implications for one to inexpertly stumble through picking out threads. The maturing sense of purpose doesn’t excavate youthful excitement; instead it builds upon it.
I don’t have the ability to ask my young self what art is but, if I did, I think he’d respond with something like, “It’s fun!”. A child’s answer and a truthful answer.
At the most profound and universal level art is an experience of making – what my imagined younger self calls fun. That comes before the experience of appreciation, both in terms of how we learn about art and how we develop the ability to appraise it. As children we are handed poster paints and crayons years before we are taken to a gallery and instructed to think really hard about what we see there and, when that does happen, we look upon the storied works within the gallery and judge them via our practical exposure to the difficulty of making. The frustration at one’s inability to produce wonderful objects at the level of those masters is the soil in which determination takes root. A strange plant with precious fruit. While that fruit is vital in the nourishment it gives to those who eat it it is also how the tree perpetuates itself.
I’ve spoken to philosopher friends in order to have them explain to me why I don’t like AI generated imagery. They tell me that there is a deep and irreconcilable difference between desire and jouissance (enjoyment). The meaning of desire is obvious: it’s the wish for something, most often something pleasurable. Jouissance pertains more to the enjoyment of a thing; this is not always easy or wholly pleasurable. Our jouissance often rests on the sacrifice we made in the process of producing or acquiring the object. Hobbies produce jouissance because we have to spend a lot of time and thought on them, whereas desires can be filled instantaneously.
Consider the difference between wanting and receiving an ice cream (desire), and learning how to make it (jouissance); the latter feels somehow deeper and more satisfying, even if there is no qualitative difference between the two desserts produced in the end.
The machines that technological progress brings to use can very plainly meet the desire for images. That’s why all the early talk of the low quality of the images produced, of how such low quality images were nothing to worry about, was naught but pablum.
Whatever desire for images exists this process will satiate and it will do so incessantly. That’s the reality of how we deploy these technologies directly into our culture; located publicly on some easily accessible website with no warning or detailed information. By the time anyone notices the products of a new technology it is already functioning and influential. Already warping around itself the culture that created it. Before most of us had heard of AI generated imagery many more were already fixated on the things it might do for them. Mostly in terms of generating them a lot of easy, quick cash.
The problem here is that the desire can never truly be satisfied. The object of desire is quickly revealed to have been a representation; incomplete; facsimile; symbol. The acquisition of such a symbol tends to intensify and reinforce the desire more often than not.
“I want Darth Vader fighting Sonic on Mars, in the style of Munch.” They prompt.
AI image generation can serve only to heighten this desire as it spews an endless sea of images that can never quite meet their prompts. The symbol is mismatched. The facsimile falls flat. They will get their pictures of cross-franchise superhero weddings but this will intensify their demand for more such dross. This can be evidenced in the language used by those most readily drawn to the technology: descriptions of how amazing it will be to be able to see any image they desire, specially formulated near-instantly. As if symbolic mathematics can create joy in people who are not academic algebrists.
Will the near-instant and vast vista of varied visions which flood back to the user actually be what they need?
The machine does not struggle. It thinks nothing and feels nothing and thus can mean nothing. AI generation lacks the component of participation and this will never truly create jouissance in the prompt-giver.
Think again of that set of self-assembling Lego. Remember the strange and uninstructed things that you made from Lego as a child? Surely that experience, that expression, is valuable to the participant. How do we serve that experience once we have flattened all potential out of the experience.
Imagine that with the fascinating gimmick of Automatic Lego comes the market response of hyperfocus and oversaturation. Stores have to make a decision about how much shelf space to give the older product now that this new one is getting so much attention. Sure, you could buy your kids the classic lego. But do you want them to be mocked at school? Can you bear their accusations of getting them ‘the wrong thing’?
Pretty soon Automatic Lego is the only option readily available to buy and eventually it’s the ubiquitous lego product. Legacy Lego becomes a footnote, a memory. You begin to miss the simpler, truer enjoyment of building your own creation out of small plastic blocks, so you find some way of defeating the automatic mechanism and voila!
But something has changed. The pastime of building something now requires a deliberate sabotage of the product as it comes out of the box. A defiance of the intended operation. The potential is now obfuscated behind function, rules, and simple expectations. It resides now outside the prescribed interaction with the object and for that reason becomes less and less easy to engage with over time.
The ordinary becomes luxury by dint of this process. Indeed, it may even become abnormal; a source of curiosity, anger, or ridicule.
It will be an irrecoverable loss, a breaking of traditions of craft that stretch back into the earliest prehistory. As it becomes impossible to make a living in craft or in art, as those roles are increasingly consigned to machines, these jouissance-producing skills will be lost.
The space which art and craft once occupied, where ideas were realised, has become an immediate nothing; a place that no longer exists. A shore to which we cannot return.
So I am saying, “Technology BAD”. I am a luddite after all, right? Not at all. I simply believe in using the right tool for the right job. In my view the right use of machines here would be to create something that brings more people into the practice of making art, that facilitates the development of passion rather than burying it. There are undeniable benefits to the emergence of AI art technology.
The entire arc so far has about it a damp and cloying funk of boring inevitability, familiar and predictable. It is possible and necessary to dream of better directions for AI technology. It’s here right now and will only become more prevalent and inevitable as these systems accelerate. I do not wish to see my practice be reduced to typing words and waiting for machinery to deliver the results, assembling my Lego for me.
I do not think the place of AI is in replacing the craft of the artist but I do foresee a possible role as defender of the rights of artists. A technology that can understand patterns in images well enough to recreate them must be capable identifying when someone has copied my work without permission and it must be able to determine between important violations of my rights (commercial use without permission) and unimportant, personal, or transformative use (non-profit, fan art, review).
There are AI-driven platforms for that purpose. I have no idea how to make one. I paint monsters and grim reapers for a living! But I do want more people to experience making art, not fewer. I want a culture that grows from participation and enriches itself. A technology that isolates us as totally individual consumers will do the exact opposite; who among the most ardent AI enthusiasts is truly interested in the images produced for other users? AI generated solipsism.
It isn’t hard to imagine AI being deployed to alleviate certain difficult obstacles the artist faces. AI technology is already in use for this purpose in the AI upscaling of video game graphics, for example, which spares game developers thousands of hours of remastering time and keep games playable long after the systems they were designed for have fallen into obsolescence. It is easy to envision a newly connected network of creators using AI to defeat the varied barriers of language and culture that divide people.
How wonderful a machine that encourages you to enjoy making something instead of trying to give you what you want in that instant! It sounds far-fetched because we typically imagine machinery to be labour-saving or force-multiplying because we have designed them that way and the norms which inform the design practice are so powerfully ubiquitous as to be invisible.
Having looked at the potential benefits of AI technology in art it must be said that I am not seeing the emergence of a new collaborative community intent on working with this technology to create something positive or additive. The potential for that, though unproven, may well exist. But it’s not what I’m seeing. What I’m seeing is individual people and corporations attempting to capitalise on this technology and wearing the laurels of an artist without ever lifting a finger to actually achieve that status.
As I have been writing this article I have also begun work on a new project which happens to be a highly enjoyable image that illustrates the world of a role playing game. Going through my typical process of engaging with the client and creating rough concepts and sketches feels familiar to me, comforting. It feels like good work, worthwhile, meaningful even though many – even within the art world – might consider such a project a piece of trivial entertainment.
As with so many paintings I have created for clients in the past, accommodating the client’s needs offers me obstacles and problems. These arise in the space between my ideas and theirs and a good proportion of an illustrator’s work is solving and resolving these issues. There is a noble struggle to that process that produces satisfaction in all involved, not least myself as the artist.
I doubt that the finished piece will be seen by more than a few thousand people. What is important to me in this moment, here and now, is the simple pleasure of feeling a sense of continuum. The jouissance of participating in a human tradition that has an ancient history and, though now somewhat perturbed, an undecided future. Despite the emergence of technologies, like AI generated imagery, that will hurt artists I remain optimistic that the future of art, though untold, is a bright one. Through persecution, poverty, disability, and madness artists have persisted. The earliest of our ancestors to even resemble us in shape or mind created art on the walls of the caves in which they lived. Should the species reduce itself to stone age cave dwellings artists will again scrawl on the walls with whatever is at hand. Because what artists really do, that core part of it that cannot be wholly explained in any single attempt, will continue to matter so long as humans draw breath. Even after that, should other thinking beings one day find our works.
How it will matter, and to whom, is what is at stake. I argue against a future in which AI art makes the career of an artist untenable. I desire a future where we artists, who have shared so much joy, can continue to make a living off of this wonderful endeavour. My jouissance is found in pursuing this endeavour with my own hands and eyes and I don’t want to see that jouissance fade from the world.
Luke Oram