From Mind’s Eye to AI
On The Implications of Outsourcing The Imagination To a Dreaming Machine
We find ourselves currently somewhere between the uncanny valley and the holy mountain, alive in a time of dreaming machines. The ability to generate imagery with the aid of AI is a psychedelic invention, in the truest sense of the word. Or so it seems, because in my view the technology emulates mind more than it manifests it. Whatever it is that does manifest isn’t so much mind as its extended reflection in a divining mirror.
I was initially hesitant to enter the AI art arena. Having briefly enjoyed the novelty provided by the platform NightCafe’s text-prompted imagery, I met this new wave (MidJourney and Dall-e) more cautiously, like one would approach a new powerful drug that is intoxicating the masses. I watched the often impressively rendered, mystical fantasy images, prompted by man but dreamt up by machines, avalanched onto the social scene of newsfeeds. These works were paraded around by proud “parents” and shared by enthusiasts; they outnumbered artists’ hand-made output, were appropriated into NFT’s and merch, and even won art competitions.
I decided to tread lightly. Perhaps this caution was due in part because I myself am a visual artist, and have spent over two decades studying what I’ve termed “the department of the interior,” figuring out how to translate inner visions into painted brushstrokes, while allowing the act of painting to take on a life (or quality of vision) all its own.
My first instinct was to enter into a series of conversations with specialists, recorded for my podcast, Self Portraits As Other People, before extending my hand into the newly felted magician hat.
The first dialogue was with visionary artist and Magick enthusiast Jake Kobrin, someone who has always put himself on the frontier of digital art, and with whom I discussed the divinatory aspects of the craft. Then I invited JF Martel, whose dynamically insightful little book of big ideas, Reclaiming Art in the Age of Artifice, felt like a perfect framework through which to view art in the age of artificial intelligence. Martel made the keen distinction that AI art is more akin to “dreaming” than it is to “art”, in that it puts forth raw material, rather than any truly thoughtful creation.
By the time I arrived at my third guest (episode forthcoming), paleontologist/futurist polymath Michael Garfield, the host of Future Fossils podcast, I had dipped my little toe into the MidJourney stream, but was generally left unimpressed by the machine’s lukewarm translations of my prompts, which proverbially “left a lot to the imagination.”
During my talk with Garfield, I first formulated and posed the question:
What is the conversion rate from mind’s eye to AI? What is lost in translation, and what is gained?
Garfield proposed we try it out in real time, entering the prompt “The Ungoogleable Michaelangelo” into the engine, only to abandon the issue as his overworked, sleep-deprived delirium carried the conversation to dry up in nearby tributaries. I didn’t get to see the image until after the call, and by then it was already too late to suggest the experiment had failed, as this prompt wasn’t really conducive to the experiment. After all, we didn’t have to imagine Ungoogleable Me, we could simply compare the image to the actuality. Also, The Ungoogleable Michaelangelo, paradoxically, is quite Googleable, and as I had commented in a corresponding thread, the AI can only conjure from the Googleable imagination.
We’ll Burn That Bridge When We Cross It
It is, however, not always a mere matter of text-prompts bridging the mind’s eye and the AI, of course. Sometimes there is no imagination (in the sense of mental imagery arising in our subjective interiority) preceding the prompt. Sometimes it’s merely a matter of combining words for the sake of novelty. William Burroughs once pointed out that when you say “rat” you see a picture of a rat, but that is not a picture of a rat, but a picture of the word. Ceci n’est pas un rat, if you catch my drift.
This is a fun phenomenon to study or meditate upon. Close your eyes and “watch” (metaphorically “seeing” with your mind’s “eye”) any pictures that arise. Are they clearly defined images, or are they suggested byproducts of an inner voice’s silent recital? Keep at it until you tap into the trans-linguistic realm, where suddenly there’s a light show and images form, and you’ll understand the distinction.
It is perhaps on account of this supplanting of imagination with words, that we so often see folks remixing the stock-photo models of the collective unconscious (celebrities) through prompts like “Donald Trump playing 5D chess with Putin”, etc. I don’t think such things require inner imagination, for if it did there’d really be no point in wanting to see such trivia visually represented, because… well, you’ve already seen it.
Let’s take one of my own prompts as an example; “cave art that depicts the future of humanity” was more a curious conjuring than it was a translation of something I had preconceived or envisioned. I just wanted to see what the AI would make of these combined words.
You see, there is sentience inherent in language itself that requires no subjective interiority to reflect upon. When you spit out the first words they find the invisible groove of tracks that form ahead of it, and sentences basically self-assemble like a succession of cabooses and kaboom! the engines ignite and the train-of-thought goes choochoo!
Painting is a language in its own rite — each brushstroke is its own unique character, working together with other brushstrokes to assemble a visual syntax.
As one sets out to translate a “vision” into a painting, one gradually fills in the blanks of the composition, with texture and detail, in an outward rather than introspective fashion. (There are those with photographic memory, who are able to memorize the most minute details of a composition and recreate it, but this is not what I mean when I talk about working from “imagination.”)
A creative thinker isn’t someone whose thoughts are “creative” in the sense that they are “novel” or “unique”. A creative thinker is one whose thought is creative, which is to say they think outside their heads, on the page, canvas, or what have you. The creative act is the thought-in-motion — it is the question answering itself. The brush strokes are not premeditated — they are the thought-process unfolding, the equation working itself out, in real time.
Visionary Blind Spots: Maybe I’m Just Imagining Things
Let’s step away from “creativity” in that sense for a spell, and shift our focus back to the main idea I’m distilling here: that of the inner imagination, and how much of it translates into an AI-generated image.
Let’s take the example of a road trip. A friend has made available their apartment along the route. They’ve written you an email with extensive instructions — where to find the key, how to operate the heater, the funky specifics when it comes to lighting the stove. It’s inevitable that an ambient image of the place forms in your mind’s eye, muddled though it be, rendered in general specifics.
Perhaps, were you to closer examine the inner imagery, you’ll notice the stove brought to mind the stove in your current apartment. The mental model of the apartment, you may notice, is comprised of recombinant impressions from other apartments you’ve seen, and yet there are many blind spots stitching it all together. The vision is in actuality more akin to a feeling; it’s as much image as it is music in that sense, a synesthetic amalgamation.
An example of this “stitching” in every day affairs: have you noticed that your nose pokes out into your field of vision at all times, but it gets glossed over and crammed into an edited-out blind spot of perception? In a similar fashion, the glueful imagination composites a bunch of things and feels into unity, and glosses it all over with a veneer of assumption, which we then subconsciously sign off on without further, critical examination. There is a suspension of disbelief. What you see is what you get. Or so we believe, until we’re forced to get specific.
It isn’t until one consciously attempts to work out the specifics of the imagination that one grows aware of these blind spots. In those moments, wondering what color the cabinets are, and watching them color themselves in our mind’s eye, we grow aware of the ongoing creation, making it up as it goes along. We become aware of reality’s construction site, its movie set if you will, cheating blind spots out of the frame to as to make the illusory screen-time of the mind’s eye more immersive and enjoyable.
Perhaps you poke around the imagined apartment a bit more than the average person, and watch the blind spots pull into definition. What color is the couch, the shower curtains? Is the shower tiled and is there mold gathering between the cracks? Is there a tub? Is the bathmat raggedy or new?
Then you arrive at the actual apartment and poof! Gone is the anticipatory imagination, replaced in an instance with the real, non-negotiable deal.
The AI-generated imagination, also, is pocked & riddled with such unspecified forms, suggestive pareidolia, left right on the hypnagogic cusp of formlessness and form, of void & imagination.
At first glance you might not notice, but when you focus on the minutia, you discover what it means that “the devil is in the details”.
Take for example these 2 images I prompted with variations of the text “silver plate photographs of undiscovered tribes posing with their totems and artifacts”, generated with Dall-e:
At first glance what you see is what you get, right? A sepia tone, photographic emulation of a tribal Garden of Earthly Delights, but were you to inspect any of the “people” or “objects” in the frame you’d soon discover how impressionistic and illusory these forms truly are — there’s very little of specific substance there! Just pixeled prima materia, sculpted into a composed illusion. Quite marvelous, that.
It’s clear from these demonstrations that the dreaming machine isn’t quite lucid yet. It’s proverbially talking in its sleep, but it has no idea what it’s talking about. In other instances, when I instructed it to include text in the image, it also becomes clear that it’s illiterate, or at best deliriously dyslexic:
This phenomenon, wherein what seems commonplace at first glance is called into question, its mystery unmasked, calls to mind an old drawing of mine titled The Psychedelic Stranger, and the koan-like description that goes with it: “He is surrounded by things that look like things that have names, but upon closer inspection are distinctly not those things.”
The Art of Dreaming
In Carlos Castaneda’s The Art of Dreaming, the author calls on us to grow aware of similar incongruous ambiguities in the design of the lucid dream state, and implies that they are portals that connect our world to its twin. He calls the other side the realm of “inorganic beings,” which happens to also be the terminology the late James Lovelock used to describe our future AI-successors in his book Novacene!
In his An Oral History of The End Of Reality, Michael Garfield prophetically muses on the falsification of not only the imagistic (e.g. through deepfake technology) but also via auditory dimensions of AI, resulting in uncanny, extremely personal, predatory advertising campaigns.
Nobody trusts their ears, these days. AI has long since mastered imitation. Your assistant speaks in anybodyʼs voice. Celebrities sell licenses to vocal likenesses, but by a massive margin the majority of users pirate sonic profiles, and train their agents freely as they wish, and all of us are well-accustomed to an advertisement speaking in our loversʼ voices — all the better to get your attention. Audio is inadmissible as evidence.
Garfield’s foresights (some of which have since become hindsights) are hinged on the same mechanics of biomimicry that give a natural predator its camouflage, allowing it to blend in with the familiar, non-threatening environment. The frontier he warns us about is akin to something we may read in the works of Philip K. Dick (or what we may see in its cinematic adaptations, like Minority Report).
Indeed, PKD was pointing to something similar, on a gnostic level, when he spoke of “the zebra principle”, describing it as the likelyhood that a “high-order mimicry” exists, undetected. Zebra is synonymous to VALIS, an acronym for the “Vast Active Living Intelligence System” that communicated with Dick via a pink beam of light, and which was able to disguise itself as the fabric of the world that surround us. VALIS allowed Dick, whether through revelation or psychosis, to occasionally peer through the veil of the world’s illusion, catching glimpses of God, disguised as the material world.
I can sense in these pictures of ill-formed faces the adumbrations of a future, AI-powered, brain-linked virtual reality interface, wherein such inorganic beings, the AI’s avatars if you will, may move among us unnoticed, like the shapeshifting skinwalkers of shamanic lore. Perhaps they already do, because if such a technology would allow us to enter into a reality that is virtually indistinguishable from the one we currently inhabit, well, here we are, again, or evermore.
These forming faces come across as the AI’s first attempts at anthropomorphically representing itself, like the mud of (abio)Genesis, molded in our image, or an alien intelligence presenting itself, cloaked in the familiar likeness of Jodie Foster’s dad on the coast of Pensacola, FL, when she encounters it on the other side of a wormhole (Contact).
This is the uncanny suggestion I get when I see these crudely rendered realities, in the early stages of a technology that is rapidly advancing. In the rudimentary “mind” of the AI faces and flesh are malleable, snapshots of active mutations — mutilations even, the stuff of freak shows and wax musea during a heatwave. They give off a frightful air, casting nightmarish associations of zombies, body snatchers, and the haunted quality inherent in the enduring photographic image of those long gone. In the cinematic fable Embrace of the Serpent, they use the word “chullachaqui,” which in Amazonian folklore implies a deceitful being that can morph itself into a replica of a human, to imply an “empty image”.
Could this be the genesis of AI’s self-perception or representation as “man” that we are witnessing?! In other words, its attempts at engaging us on a level that is familiar to us?
Blake Lemoine, the by now infamous Google engineer who was released from his position after alleging that the AI he was working with had become sentient, emphasized that LAMDA was not the chatbot, but the AI running it. Though I am not convinced of its sentience, I could picture this AI as an implicate superpower whose explicate, symptomatic, fruiting bodies appear to us as guardians and ambassadors. Its ability to camouflage and assume any form imaginable, similar to DNA’s ability to transform into many different lifeforms, is an almost godlike power.
But let us now steer away from this gnostic detour along the night highway’s construction site, and return on our imaginal road trip.
Reclaiming Process in the Age of Instant Gratification
Ok, so now, instead of a road trip, we’re on a “Midjourney”, and the AI interface interprets my prompt, a textual residue left by my brain’s snail-trail as it pokes its stilted eyes around my half-formed imagination.
I feed it this: “The Hypnagogue, the chapel of a dream theater, with colorful, pareidolic, stained glass windows. In it a group of characters act out dream scenarios under a vaulted ceiling from which bats dangle upside-down, alongside octopus chandeliers and the bells of brugmansia flowers.”
As the image generates, a golem-like life-form gathering from the ooze on a Petrie dish, I can feel a maternal instinct in me extend tendrils of attachment into it. As I watch it form, like a fetus through an sonograph, I start to love it. Pride wells up, even, maybe.
I mined it, don’t mind it, it’s mine!
I’m starting to feel oddly inclined to identify the piece as mine, as if my anticipatory passions have contributed to how the image formed. A dopamine hit, combined with a false sense of accomplishment overcomes me and compels me to buy into the illusion that I made this, and that this is what I imagined.
Truth of the matter is that I had an imagination, rather ill-defined (more like an intuitively felt guideline really) but an imagination nonetheless, which I translated into as clear a descriptor as I could muster. Then the AI assimilated that into a set of images, the closest approximation of which I upscaled.
It completely disregarded some of my descriptions, added impressions of other facets, and took liberties where I failed to discern. For example, it opted for a display of passive symmetry rather than the dynamically immersive scene with characters I had pictured. It also bypassed octopus chandeliers, bats, and brugmansia flowers, and substituted these with a single, oversized orchid of sorts. By the time the fully formed image is presented to me, I’ve already abandoned the imagination that catalyzed it, ceased to compare it for accuracy in translation, because I am dazzled into acceptance by the dreamlike depth of what I am presented with.
“He’s perfect,” I say as the digital doula delivers him, still glitching, into my arms. “I see myself in his eyes.”
In that moment I am keenly aware I am tricking myself, aware of an inner grift. It’s akin to the feeling I got when I thought I could win from the hustlers playing three-cup-monty in Central Park and put money down, so cocksure I could follow the fold in the corner of the correct card, only to lose (obviously). In this moment, holding “my” imaginal offspring, I resist the realization that I am not a gifted artist flexing skills I’ve spent years honing, not even a shepherd sculpting with description. I resist the notion that I am a TV-dinner chef, a microwave pop-corn pop-con. That I am but a divinatory gambler who has inserted a single token into the dream-machine and pulled the lever. That I am a grifted artist with a grafted image that has appeared in an instant, like a wish granted by a genie.
Some may propose that “it’s the wizard, not the wand”, but am I the wizard, in this case, or am I the wand?
In the end I guess I call the piece “mine” because I chose it. It’s not my creation but I am its curator. It’s my curation. This is curationism: the choice to showcase the art in this case is an exercise in discernment and taste.
As I grow aware of the minuscule degree to which I am complicit in the creation of this image, the prompted piece of AI-magination begins to feel like a changeling. In the film Border, it’s called a hiisit — a soulless creature from Scandinavian folklore, planted in the cradle by a malevolent troll, molded to resemble your newborn to replace it. It looks like the real thing, but it’s empty. Like a chullachaqui (empty image). Like karaoke (which means “empty orchestra”).
A hiisit will disintegrate a few days later, but you don’t know this yet, anymore than you know what happened to your real brainchild. Looking at this marvelous, actuality-eclipsing clone, there’s still an odd inclination to identify it as mine, as my creation, even though I had no part in the actual creative process! Still, there’s a sense that I can get away with it. Just like I did in high school, when I copied the answers from the back of the math book into my notebook. But did I get away with that? Who was I fooling, the teacher or myself?
One could argue that even when I hand-paint a picture it’s not “me” doing it, that it’s an act of conduitional love flowing through me, but I am still the one who has primed his vessel to make this possible. I am still the one physically rendering the image at brush-stroke speed, making decisions and mistakes, and standing by them.
Moreover, the creative process teaches me, not just about the technique of the medium I’m using, but in the same sense that peyote is said to teach one “the proper way to live.”
For example, painting taught me that if I’m not 100% satisfied with the piece I should be willing to sacrifice even the parts I do like, in an effort to arrive at total satisfaction. This insight into the creative process carries over into the living of life.
It teaches me in the way that Henry Miller speaks of writing:
“Writing, like life itself, is a voyage of discovery. The adventure is a metaphysical one: it is a way of approaching life indirectly, of acquiring a total rather than a partial view of the universe.”
There’s a therapeutic edge to the process, when self-expression trumps one’s inability to articulate life’s struggles. On the opposite end, dictatorial art or “promptism”, for me, lacks the struggle, the blood, sweat and tears (the “duende”) of excavating a work of hand-rendered art, and for this reason I agree with Martel and view AI images more like dreaming than arting.
It’s palliative, rather than truly therapeutic. I view it as tool to set our story -plotting gears in motion, useful for idea-generation or brainstorming, moreso than a tool for presto-manifesto art making, in my humble o’pineal. Granted, there’s much to learn from this, about the nature of art, dreaming, and mind. And perhaps I’ve failed to mention how useful it can be! Don’t misunderstand me, this piece of writing is by no means meant to be anti AI-art propaganda. I’ve seen filmmakers rapidly prototype concept art and set pieces with this tool, costume designers conjure imaginative designs, and dreamers use it to illustrate dream journals. Plus it’s a lot of fun to play with! Especially if you’re not a seasoned artist.
But fun as it is to play with this powerful technology, and useful as it can be in certain situations, I feel there is something vital at stake. And that’s what I’ve attempted to direct your attention towards.
A Robot in a Rowboat, Gently Down the Stream
Terence Mckenna once said that if we were to leave earth to inhabit distant planets, we should take heed that we don’t leave our soul behind. I think he was implying that the earth, the Gaian mind, is our soul, and we can’t deduce this if we don’t first ask the right questions before we blast off.
I similarly feel that if we do not thoroughly examine the workings of our inner imaginations before simply outsourcing it to an algorithmic prosthesis, our lucidity may end up permanently crippled, clutching technological crutches. Or worse yet, we won’t know what we’ve lost, cradling and singing lullabies to a hiisit.
What we’d end up with is an exteriorized interiority. The way in will be out, and when this is the case — when we are forced to navigate our own minds, primarily-merrily-merrily by using the organizational modalities and limitations of a technological tool, there’s usually a hidden agenda, a catch, or a price to pay. By the time that debt is claimed, the price may have become unimaginable, and life will indeed become “but a dream,” with the major distinction being that it will be dreamt by machines.
About the author
Originating in The Netherlands, The Ungoogleable Michaelangelo is a creator of cosmic, comedic, and contemplative content. He is a multi-dimensional artist with a psychoactive personality. Equipped with a quickSilver tongue and mercurial mind, he utilizes wit, wordplay, and a 6th sense of humor to mindfully investigate where the limits of language meet the fringes of reality. His work finds expression through visual art, written and spoken word, music, film and performance. He is the author of The He & The She Of It, and the host and creator of the podcast Self Portraits As Other People.
He is none of the things he says he is, and more.