DALL-E and other AI artists offer only the imitation of art
I have long thought of art (including both literary art and art art) as one mind reaching out across the void and asking one poignant question: do you see it too? And the appreciation of art is of the same character; it is the mind of the viewer seeking an expression that validates or enhances what they see in the world.
Lovers can hold hands. Minds cannot touch each other in the same way. Art is how minds hold hands. By which analogy, AI art becomes a kind of sex doll. However lifelike you make it, there is no soul there, not acceptance, no affection, no loneliness reaching out to meet your loneliness.
This is a wonderful, thoughtful essay. One thing I'd like to suggest, though, is that taking DALL-E and other similar systems to their logical extreme -- where they take over and there is no original art and they are left with only themselves for input -- is a good thought experiment, but unlikely in real life. Much more likely is that these systems co-exist with and parasitize human creators constantly. A good analogy is Google Translate. Google Translate can't work without a planet full of multilingual individuals using their human brains and cultural knowledge to translate among living languages by their own efforts; Google Translate then comes along and siphons this data and devalues the work of translators by allowing anyone to "use google translate" without acknowledging (let alone paying) the humans behind the algorithms who did the actual linguistic work. But Google Translate has not made people stop translating. It just devalues the work of people who are translators, who were once a highly respected and admired group, much-needed in any areas of life, and well-remunerated in many of those roles. So artists and writers face the fate of translators, I think is a better analogy. Humans continue to do the work, and DALL-E or GPT-3 simply makes that work seem pointless, even those those AIs couldn't function without the humans they exploit.
To steelman the opposite view a bit, there's still intentionality in curation. A piece of AI art can be displayed in, say, a blog post, due to a human editor deciding that it fits well there, and we can recognize that as valuable. But... yeah, if that comes to pass, it would be a very limited view of art.
I have become more optimistic on this question than I was a month ago, though. If we truly value conscious, intentional art — and I agree that we do — then we'll quickly learn to devalue AI art. Many people like DALL-E right now because it's new and impressive, but there's a negative feedback loop at play. AI art will destroy its own value soon enough.
Thank you very much for this piece. The passion you put into it is special. (Wish I could word that sentence better!) My partner is executive director of a nonprofit facility that rents studios to artists and puts on exhibits, so I frequently interact with artists, from recognized to unrecognized. Also, I worked in advertising for a decade, so have interacted with commercial artists.
Your piece opens up the question of "what would society be like if there were no artists?" Even if we never saw their art, I think we would lose something by not having people being artists in our midst. We need people who are living a vision.
Perhaps thoughtful people will always be able to recognize human art. Not by looking at it, but by knowing the facts behind it. Right now, I'm looking at a Sierra Club calendar, a photo I took in the mountains, and an abstract oil painting of a redwood forest. Each connects me with nature, people and life, even though the calendar is the most remote from being art.
I read a very good book on the invention of printing and the socio-cultural changes it wrought -- not unlike what the digital world is doing to us now. In some ways, society recovered from the chaos brought on by the invention of printing, and perhaps will in the coming decades.
Perhaps artists themselves -- like Jackson Pollock and Andy Warhol helped bring on the AI-art problem, by the way they deconstructed the values you praise in your piece. As open-minded as I try to be, personally, I don't believe Warhol's most famous works are art, and I don't think Tolstoy or you would either, for all the reasons you both present. If they are art, they are only so by a few synapse-firings conceptually.
Might the questions you raise about art be part of a broader search for authenticity in life?
People nowadays often (justifiably) laugh derisively about the concept of NFTs, but might NFTs possibly become necessary solely to authenticate human-produced art and thus force a differentiation from AI-art? Even if for many other purposes they are merely a gimmick? This thought is just a brainstorm at this moment, not developed yet.
There are many valid points here, though I take issue with two statements. Firstly, the discounting of pareidolia as "illusion of art". As an artist in whose art practice pareidolia is a central theme, I urge you to further examine its function. For one, the recognition or projection (immediately concerted in perception) of a face in rock, cloud, or other natural elements, makes apparent how the "external" world is in fact battered and processed in the brain's "interior". In the case of the rock faces, what makes them art is not the individual rock and their likeness to a face, but the human discernment that picked these samples and compiled them into a gallery. It is the ensemble presentation (the framing, similar to framing a photographic composition) that divorces the work from the natural "unconscious" elements that sculpted them (like a collage or remix).
You write "The key difference being that a sculptor is conscious, but when it is instead the wind or rain or the deep fermentation of geological processes acting as sculptor, these things are not conscious. Lacking consciousness, they lack intentionality, and therefore their products lack meaning."
Are you implying the meaning is inherent in the work, not something divined/projected/perceived by the beholder? In a sense certain artists, such as Andy Goldworthy, attempt through effort to emulate what is effortless in nature. This brings me to my other issue, with regards to your insistence on the artist's "conscious" impulse/intent to create. I understand that you mean to infer that the living artist imbued the work with his life force or intention or uses the creative process to work out some other, living processes. But much art comes from an unconscious place, and works to wrestle that into the fore, which is exactly how it foregoes the limits of conscious control and may imbue the work with a transcendental quality. AI lacks struggle, which is what creates that quality we call "soul" or "duende". AI art, as you rightfully propose, may not be able to mirror this accurately (yet). However, it is currently a kind of divination, a means for the artist (or layman) to outsource his or her brainstorm, and generate ideas, to be further worked out in collaboration with the technology. And, like the illusion of compassion glimpsed in a recently-fed (and seemingly friendly) grizzly bear's eyes, we may yet catch a glimpse of what we are by projecting it into the mirror of what we are not.
Thank you for this thoughtful piece. Much I agree with in it.
One place I would push back:
"The response of AI-art enthusiasts, like those at the tech companies standing to make trillions off AI, or those who simply enjoy nerding out over new technological toys, will be to suggest that what matters in art is solely its extrinsic properties."
I work at a company standing to make a fortune off of AI, and I enjoy nerding out new technological toys, but my response is not that what matters in art is solely its extrinsic properties. My response is that, if you actually dive into the creation of Dalle-2 - the crude attempt of apes to recreate the part of themselves that conceives, the way the neural nets are forced into desperate (mathematical) free-fall through millions of human-labelled images - there is in fact intrinsic substance behind Dalle. Pollock drips, cavemen place a hand on the wall, Pistoletto smashes mirrors, and AI researchers force mathematical constructs to reverse the diffusion of pictures. You see Dalle as devoid of humans, like a rock, but I see it as fiercely, farcically, tragically, human, like a cathedral.
All that said, Dalle isn't just hanging in MOMA. Perhaps you would have a different response if it were. Dalle is poised to cause the type of displacement you rightly lament. Like I said, I agree with much of what you say.
So on point, as usual--but the more I think about it, you gotta admit DALL-E is an utterly brilliant name.
First, this is a brilliant summary of a complex topic. I find myself unconvinced, though, that AI-graphics are going to put artists out of commission. The rise of search engine optimization (SEO) drove a massive increase in the amount of writing needed, delivering a deluge of low quality, spammy content that serves in stark relief to quality writing (both copywriting and blogging). While GPT-3 and DALL-E could lower friction there even more, the market for authentic, well-crafted arguments and art will remain.
In the 19th C people thought photography would be the end of art.
It wasn't, but it certainly had an effect. For one thing, it diminished the demand for portrait painting, a commercial art, although not to zero.
I think the effect of AI art will mainly be on commercial art.
Expressive art will continue as humans will still have the urge to create and marvel at the creations of others. Expressive art will live in the real world space, with tactile materials, there to see when people raise their eyes from their phones.
Thanks for a well thought out essay! While I do agree with your main theme that AI Art is, to put it simply, essentially soulless, I do have a quibble or two. For example, if art is about emotion, then we should define emotion, and I am not convinced emotion is outside the ken of computation. I hold the view that emotion is at heart the the experience or anticipation of reward or punishment. The positive emotions reflect rewards, and the negative emotions precede or follow punishment. If, as many of your references including Tolstoy do, we define art as the ability to communicate emotions as experienced to another -- which I believe to be a bit of a reductive definition -- then I don't see a serious obstacle to producing true "AI Art" by centering it on computational emotions that are designed to be human-like. Granted, that's not at all how language models or reinforcement learners today work, but that's a matter of implementation.
Secondly, although these AI systems are not at all conscious, what ARE is a learned probability distribution that replicates human judgements of the connections between images and text. And thus they cannot truly be said to be *unconsciously* generated; ultimately they are particular interpolations of true human experiences. In that sense, they are an emanation of human consciousness. We may find this AI art to be wooden or banal, but let's be honest: "art" produced by an average human isn't really all that great either. So the fact that certain "art" is boring does not say anything much about its relationship to consciousness.
I would bet that a good creative artist could use the descendant of a tool like this in an iterative process to create are that does convey emotion, by critiquing and augmenting each iteration with verbal commentary. In fact, what such toolsets could enable in the end is for verbal artists -- also known as poets and writers -- to create visual art inaccessible to them previously.
Despite the comments, I appreciate your point of view and wish more people could have a more reasonable view of these mathematical systems and their capabilities or lack thereof.
Thank for this piece. It obviously hit a nerve with a lot of people. It inspired my most recent issue on my newsletter, and I linked to you, of course.
Beautifully written and argued. My favourite line - “In such a world the art isn’t art, in the same way that a photograph of a hurricane doesn’t get anything wet.”
Not much of a commenter, but I was moved. Enjoyed the article. Thinking on it took me down a few different streams of attitudes and possibilities:
- Purist stream, the "immutables" who believe the lack of intentionality and consciousness, decision, curation by the creator is lost, and that taste is inseparable from art. I think I fall into that camp while also finding it a bit shaky. Humans work hard at art, and don't always do it well, but there is an intense pleasure in witnessing a human perform masterfully.
- The economical stream, the corporate opportunity presented by having a company or in-house AI generate all your artwork. Consider all those articles with weird or fancy accompanying art, now to be created to their specs and at, eventually, a fraction of the cost. Presumably a sufficiently powerful AI could complete many projects at once or sequentially. They could even generate a large stockpile for future use. Legalities around ownership may ensue, but for these circumstances the art is a dressing, and thus ensuring it has the immutable human touch might seem superfluous. All commerce seems to spawn a gutter and boutique version, so if we recall the internet is filled with low-effort, plagiarized, paraphrased, and otherwise lazy content, there's no reason to believe a similar environment won't spring up in at least one place for AI art. Keeping in mind those article are viewed by many as trash, their authors as hacks, and their websites as cesspools. Keeping in mind further that AI is already writing some of those articles.
- The over-performing stream, likely consists of taste-makers, future-thinking artists (who may be viewed by skeptics the same way digital music artists can be, as "pretend artists with laptops and pro tools"), but who will no doubt produce works that are indisputably beautiful (if only in their extrinsic properties, as mentioned in the article). They will have a nice PR angle, the misunderstood, the avant garde, the bold. Their opponent, the Luddite art snob.
-The long gamers, who believe as you do now, but also believe that the future is bright with possibility. DALL-E is the Wright Flyer, or maybe one of those ornithopters that shakes itself to pieces, or a lady with feathers on her back who rolls down a hill - looking ever forward to the day that AI steps out from under the human shadow and expresses itself, truly, in a way that may be either indistinguishable from human art, or perhaps more interestingly, be something entirely novel. Maybe AI art will cease to be for humans, and AIs will "hold hands" with AIs, even while disconnected from Skynet. This call to me too.
I freakin hate AI's forays into art. They seem grotesquely empty and lame to me -- and yet they are, you know, coherent and reasonable at first glance. I worry that people exposed to a lot of crappy plastic AI "art" will never know the real thing, never know what they're missing. Maybe they'll smoke a bunch of weed, or the 2060 equivalent, and that will enable them to experience some shivers of delight from AI prose and AI art. But they won't be profound and rich shivers.
I entered a favorite passage from Nabokov's autobiography, and had GPT-3 translate it to "standard English.' Here's what I got:
"Another part of the ritual was to ascend with closed eyes. 'Step, step, step,' came my mother's voice as she led me up - and sure enough, the surface of the next tread would receive the blind child's confident foot; all one had to do was lift it a little higher than usual, so as to avoid stubbing one's toe against the riser. This slow, somewhat somnambulistic ascension in self-engendered darkness held obvious delights. The keenest of them was not knowing when the last step would come. At the top of the stairs, one's foot would be automatically lifted to the deceptive call of 'Step,' and then, with a momentary sense of exquisite panic, with a wild contraction of muscles, would sink into the phantasm of a step, padded, as it were, with the infinitely elastic stuff of its own nonexistence."
"The child is blindfolded and led up a flight of stairs by their mother. They must lift their foot higher than usual to avoid stubbing their toe. They don't know when the last step will come, so they have to be careful. At the top of the stairs, their foot sinks into something that isn't really there."
"The infinitely elastic stuff of its own nonexistence" is just gorgeous. And teaching AI to recognize that is quite a task. I can't imagine how one would even begin to do it, much less to teach AI to produce similarly gorgeous bits of phenomenology.
Haha, fuck you GPT-3, you have Engineer's Disease -- i.e., mild Asperger's -- you're earnest and intelligent, but with fail to appreciate the other person's nuance, and have complete lack of insight into your own's interpersonal, esthetic and intellectual deficits.
You are a great writer, Erik, I am happy I discovered you. Powerful ideas.