The danger of Ai is much greater than the danger of nuclear warheads, by a lot. Mark my words, AI is far more dangerous than nukes.Elon Musk
If AI is able to outperform humans at every mental activity, and robotics to best us at every physical task, we will have rendered ourselves a redundant and inferior species! Elon Musk’s claims about the threat of AI might seem hyperbolic at first glance, but these are the sober estimations of a man who is developing driver-less cars, and is planning on bringing an army of personal all-purpose robots to the marketplace. When Elon says that “digital super intelligence ” is a “fundamental species level risk” to humankind, he means it, and he knows what he’s talking about. Our only hope as a species is that our consciousness, feelings, empathy, creativity and imagination can give us an edge against soulless AI, which doesn’t even know that it exists. The best indicator of whether or not AI can defeat us thinking, feeling, self-aware, caring, vulnerable and mortal biological beings is whether or not it can make better art than us. We are already backed in a corner with art as our last and best hope. When AI defeats the last human artist, we are doomed as a species.
This is happening right now, and living artists are finding themselves the beleaguered Resistance fighting against an encroaching invasion of nearly invincible Art Terminators. As a digital artist, I find myself a character in one of my favorite sci-fi movies. I am cast in the role of John Connor, mercilessly hunted down by robots programmed to extinguish me.
What motivation will we have to do anything if we know that artificial intelligence and machines can trounce us at it? Who will employ us to do something a machine can outperform us at multiple times over, and at a fraction of the cost? Explorers didn’t risk their lives to traverse hostile terrain in order to discover land a drone had already mapped out in excruciating detail.
As an artist, my greatest motivation is curiosity. I use my imagination to manifest images that I haven’t seen before, and that I won’t see if I don’t make them myself. In this sense artists are explorers, cultivating their skills and talent, making sacrifices and taking risks to reach into the unknown and bring back heretofore unseen treasures. Once AI has outdistanced the potential of humankind to create new images, the possibility of discovery is gone.
AI is already composing symphonies which professional musicians rate in blind tests as better music, technically, than human contestants. It is working on writing stories. I daily witness AI art trouncing human made art in the NFT community on Twitter. Consider an AI work of art just sold for 52.69 ETH, which is around $92,000.
While a human is taking credit for having produced the image, the most that can be said is that she curated it. She is no more responsible for this image than I am for the following, which AI produced after I typed in a text prompt.
I have produced thousands of similar images using AI, though for me they are starting points for possible digital paintings, and not in and of themselves the entire content. Below, my present stats at Midjourney show that I’ve created 9734 images in fast mode, and 2496 in relaxed mode, which totals to 12,230 total images.
And here is the initial batch of images I got my feet wet having the AI produce for me within an hour of joining Midjourney [click for a larger view].
Instantaneously, I became a consummate professional-level digital painter 100% courtesy of the deep learning (formerly known as “neural networks”] of digital super-intelligence. I “created” the image below within 30 minutes of playing around with the AI.
While it is an odd image, and has weird imperfections, only a professional level digital artist illustrator could produce such a piece from scratch. All I did was type in the text, “A human-fly in his laboratory”. I created not only this version, but dozens of others. I changed the text to “insect-man working in laboratory” to get a bit more variety, and made an entire series still within my first hours using the AI. Behold:
Here’s a close-up of one of them:
I share these images to show how powerful the AI is, and that it is so easy to produce high-quality digital art that any moron could do it just by typing a description.
But how does this art compare with digital art created by humans. Well, if the marketplace dictates the value of art (as so many erroneously believe) than the > $90,000 sale proves AI art vastly outstrips the entire productions of most living artists, including yours truly. Here is what the collector had to say about his purchase:
It is “nirvana” to collect something “so incredible” from “one of the best our there”. I don’t know if buyers are aware that they could instruct AI to create something similar themselves, or that the artist in question didn’t apply one digital brush stroke to the image. The takeaway for me is that humans valued art 100% created by AI above the actual work of human artists.
The difference between the AI art which the artist above had AI produce, and that which I had AI produce, and that which thousands of others are feverishly having AI produce at this very moment is hype. All of it is the art of AI itself.
In my case, I am a good enough digital painter that I know what goes into creating such images, and it includes thousands of hours of dedicated practice and honing one’s aesthetic sensibility. The AI has been trained through the targeted processing of millions of images, and employs stacks of algorithmic filters to apply aesthetic rules to guarantee the images it produces appeal to human aesthetic values. Not only can an honest artist not take any credit for the actual rendering/painting of such images, neither can they take credit for the aesthetic achievement. Putting in text prompts is so easy that artists can only really lay claim to selecting to share which pieces were produced by AI itself. In other words, the artist is a curator of the output of digital super-intelligence.
Below is a digital painting I created starting with a sketch:
Here are the early stages:
And here’s a detail:
Currently, AI can’t do this level of digital painting. It can’t do highly specific images, accurate anatomy or perspective, or intricate naturalistic details. It would never get the relation between the teeth and the pincers correctly [not that I did, either, but I came a hell of a lot closer]. Let’s see what it produces if I type in “Ant man goes AWOL”:
Even if I spent days or week tinkering with the prompts, I would never get the same result as my own digital painting, or anything close to it.
To understand the limitations of current art AI, I created the graphic below.
The image I used here is a result I produced within my first hours using Midjourney. While I am a trained artist with a Master’s degree in fine art, and decades of experience working with digital art, none of that was needed to produce the image above. I just typed in, “Awakening Upon Death of the Bride of the Creature”, which is the name of one of my digital paintings, which you can see below:
And here’s a detail to give an idea of scale:
I’ve shared this work, and my ant man on Twitter, and they got likes in the single digits. And while the work by AI sold for 52.69 ETH, I have a work up for 0.56 ETH, which nobody has even bid on:
It is for sale for 1.06% of what was spent on an image created by AI, but sold by an artist who called it a “self-portrait”.
Mine required that I first digitally sculpt my subjects using Zbrush:
Then I colored it and painted it using a digital impasto technique of my own invention. See gallery below for details:
I use myself here as a convenient example of a digital artist who can’t compete with AI art, either in terms of audience appreciation or of sales, and by an astronomical margin. If someone were to buy the piece above, I would be able to afford a cataract operation on my left eye [as it happens, my formerly preferred eye], in which case I wouldn’t have to make art using only one eye, like a pirate.
This may now be the fate of legions of artists? You spent a lifetime making art, and then the Art Terminator arrives to snuff you out.
While I can do things artistically that AI currently can’t, and other artists have their own advantages over our soon-to-be robot overlords, it may only be a matter of time before AI has ingested those processes as well. Both Chess and Go champions firmly believed that AI could not defeat them, because they possessed intuition, imagination, ingenuity, and even humor. They assumed they could make wily moves that would dumbfound the lifeless computer which didn’t know it was playing a game, or even that it was plugged in. What they didn’t fully appreciate was the power of raw intelligence.
As it turned out, Chess, and even Go (which was assumed to be much more of a nuanced, fluid, and intuitive game) were either purely analytical/mathematical challenges, or our intuition could easily be outstripped by artificial intelligence. AlphaGo, the computer which beat the 2nd ranked Go champion, Lee Sedol, 4 games to 1, was superseded by AlphaGo Zero, which beat the original AlphaGo 100 games to zero.
Current AI can learn any human game on its own, with no instruction, and has already mastered myriad computer games, which it plays at vastly accelerated speeds. AI has defeated our species at every strategy game, and it’s only getting exponentially better.
AI can be itself used to develop more potent AI, spiraling off into levels of intelligence unfathomable by humans. And if we reach some ceiling of how intelligent AI can get, based on the physical limitations of the components inside computers, we can bust through that by using quantum computing. The intelligence of AI may become perceptively infinite.
For artists, this means a digital super intelligence that is not only better than any single artist, but all artists put together.
What are the chances that next generations of art AI will not master its extant Achilles heals: anatomy and perspective? New Art AI programs are updating and coming out monthly, weekly, and daily. I get regular notifications from Midjourney concerning optimizing and updating the AI art bot. In fact, I have been assured, or warned, by one of the developers of Midjourney’s AI that AI will eclipse the human visual imagination within 2 years. That may be a conservative estimate.
Many people today are thronging to AI art venues because it allows them to suddenly become artists themselves, or so they think. There’s a window in which people can get in early and WOW their friends with the technology, even getting rich selling art produced by AI as their own digital paintings. But this window will slam shut very quickly, as the general population comes to realize that it’s easier to make astounding digital art with AI than it is to use Microsoft Word Art. It’s as easy as typing in a search request in Google. It’s one of the easiest things you can do in life, period! Playing Checkers, or Hopscotch, is much more difficult. The inevitable result we will see in just months is that everyone who uses AI to create art is an artist, and no one who uses it is an artist. That has already occurred in spades, but it will take even people who buy AI art for tens of thousands of dollars some time to catch up.
Is there hope?
The easy answer is that human art will always stand out and be superior, even if human audiences already prefer the art of AI, because we are self-aware, living, feeling, suffering, caring mortal beings. How can something dead, that doesn’t give a damn if it is thrown in a junkyard tomorrow, be better at us than art? Art is the fundamental way that humans communicate on a shared higher plane that transverses space and time, linking us together across the globe and throughout history. Just imagine a world without one of the arts: music. Note that this would also mean movies wouldn’t have musical scores. How much duller our existence would be. Take away all the arts, and poetry, and then even our spoken conversations become merely practical. We might as well ourselves be robots. Art allows us to see through the portal of someone else’s life, to thus expand our own horizons, and nourish our sense of being a human among humans in a shared existence. I agree with this perspective. Sometimes the obvious thing is also the true thing. However, that may just reflect our shared human optimism and narcissism.
The reason the heliocentric theory, which posited the Sun as the center of the solar system, and not the Earth, was anathema when it came about, was that it displaced us at the center of the universe, and in God’s image. We resisted being pushed to the periphery. For the blasphemy of suggesting we weren’t the epicenter of all creation, the Inquisition charged Galileo Galilei with heresy, forced him to recant, and he spent the remainder of his days in house arrest.
The threat posed by AI is much worse. It doesn’t just push us out of the center like a Sumo wrestler, it defeats us in every endeavor. We fall permanently from the most intelligent and capable beings in the known universe to being inferior carbon-based units. No longer in God’s image, we have created our own artificial God out of numbers, and we have no resemblance to it.
And here we see how prophetic one of my favorite songs by Simon & Garfunkel was, hence the quote at the top of this post. [Here it is again.]
If AI art replaces human art, we will no longer be communicating with each other, but isolated and communing instead with the comparatively infinite super-intelligence of AI. Instead of listening to Simon & Garfunkel, we’ll be listening to Cylon & Garfunkel.
Comedy aside, the current leading Art AI programs, such as Dall-E, are poised to replace legions of human illustrators and digital artists, simply because AI can produce reasonable results for multiple purposes virtually instantaneously, and for a tiny fraction of what it would cost to hire a human. The inevitable result is the majority of art we will see will be that of AI.
The problem with our optimistic view that our metaphoric souls will save us from being dominated by machines is that while AI is not able to feel, to care, or to love, it can fake it. When something is ten thousand times as smart as us, or ten million, or ten billion… we might need to use AI itself to run scores of algorithms to distinguish for us if something was created by a human or not.
We are already having discussions with speaking bots, and they are getting more sophisticated and convincing in a blur of continuous improvement. When we can have a phone conversation with AI and not be able to tell it’s not a person, will we still be able to tell that the art of AI is not by people?
The would-be artists who are presenting AI art to the world as their own are, in effect, hiding behind the achievements of the enemy machine rival of human artists, which is predicted to stamp us out. You may have noticed that when I present AI art I am tangentially responsible for, I say that the resultant image is by AI, and based on my text prompt. THAT is being honest. It’s also not being so deferential to the art of current AI that it has no human element. The preference of AI over human art by both artists and would-be connoisseurs at present represents a profound lack of genuine understanding or appreciation of the visual arts. It’s one thing to champion the achievements of AI in the realm of art as the art of artificial intelligence; it’s another to rank it among the best of human artists; and still another order of the ridiculous to celebrate is as the best art realized by a human. This is a level of head slapping that can give you a concussion and a collapsed forehead. But it is also simultaneously understandable, with a little cognitive dissonance, because that’s just how remarkably good AI already is at making apparent art.
AI is not genuinely making art. It is faking it. It is imperative that people understand that the only reason AI can fake human art is that it already has access to millions or billions of images of human art. It can only replace us by stealing from us wholesale. While AI does make it’s own additions and re-combinations of extant imagery by humans, with some unexpected and intriguing results, it is essentially a forger and a plagiarist. Without access to human art – even if it had access to all of the more pedestrian forms of photography – AI would be hopeless at creating art such as Impressionist paintings. It needs to pilfer from the human record.
Even with access to all of human art to this instant, AI will not come up with the same innovations as human artists moving forward. And this is the pivotal point for the fate of digital artists in particular. We can distinguish ourselves from AI art, but only if our creations are not subsequently uploaded and fed back into the AI.
Realizing that the ace in the hole of AI art when competing against human art was not AI itself, but the unscrupulous uploading of the intellectual property right of livings artists, I implored the people behind Midjourney to forbid uploading work of living artists unless it was registered as part of the public domain, in which case, the future prospects of living artists would be at least marginally safeguarded. I became a laughing stock for doing so, and another member of the forum uploaded my art and tried to make derivations on the spot in order to spite me. This behavior was tacitly condoned.
Curiously, people who are using AI to rip-off or to best artists congratulate themselves for their up-to-date-ness, as if this were synonymous with cleverness, while missing the obvious eventuality that if they can use AI to mimic a certain artist’s production, so can hordes of others, in which case their own contribution will be lost in a sea of similar productions, and become valueless, if it is not dead on arrival. All art is devalued, not just the art of real artists, and especially the art of frauds and plagiarists. The next level of obvious observation by anyone giving the subject any attention is that not just artists, but all human mental endeavors are on the chopping block. Anyone who is celebrating AI because they think they got a leg up on artists, or artists have been cut down, is missing that they are fruit of the same tree, and chopped down right along with artists. When the creators of AlphaGo rejoiced at their AI defeating Lee Sedol at Go, they were simultaneously cheering on their own core defeat, as all of human kind had been crushed by AI. They are merely in a rare position to benefit from this, when their contributions to defeating Lee Sedol required less strenuous intellectual effort than his Go playing. Similarly, programmers who in the short term may benefit where visual artists are sacrificed, are even easier to replace, as AI already programs itself in ways human programmers can’t even understand. In short, when you champion AI defeating artists, you are rooting for AI smashing humankind at creativity, and that definitely includes yourself, especially if you aren’t as creative as artists. You are applauding yourself being vanquished by AI, and permanently.
The people behind Midjourney argue that the purpose of the AI is “fun”, and so I asked if people could not have ample fun without reserving the right to upload the latest work of any living artist to the AI. I got no answer, other than that the AI itself scans all such art and one not need manually upload it. Could the AI be restricted to gleaning works in the public domain, and thus allowing living artists to have a future? I was told they didn’t know how to do that, and besides, I gather it is too late.
While the algorithms employed by Midjourney’s AI are deliberately NOT designed to make forgeries, there are other, less ethical programmers determined to train AI to do just that, because it is potentially highly profitable to do so. If you have a product that can forge the art of living fine artists, graphic artists, illustrators, and the like, that’s the version of Art AI that will be most sought after. The key is to allow the AI to do this indirectly in order to avoid copyright laws and being sued.
The digital artist of today is not only up against the Art Terminator, but against other humans giving the Art Terminator our latest productions to use against us. The same phenomenon will apply to musicians, creative writers, composers, architects, etc. As it stands, AI represents a team of forgers making facsimiles of human art, and is the equivalent, accessible to virtually anyone, of a building filled with forgers operating out of China.
Only via humans giving our art to the digital super-intelligence can it ultimately and continually defeat us. As in the movie Battlestar Galactic, it is the human traitor [Count Baltar], who would sacrifice the future of humanity for his immediate personal gain, delivering our secrets to the Cylons, that is our greatest threat.
Absent of the Count Baltars in our midst, can artists defeat AI?
While AI can take a finished work of art and apply potentially countless algorithms to create derivations of it, AI is a different species of intelligence, and can’t really make human art any more than we can make purely AI art. AI can make billions of versions of Edward Munch’s infamous Scream, but it cannot feel the angst that produced it, hence cannot produce it on its own. Left to its own devices, AI might not even choose humans as a subject, nor be at all inclined to make images that resemble hand drawn or painted images. Why would it?
If AI is not allowed to steal from us in a way that should be illegal, it will not produce the same work that we will on its own. Do you think AI can anticipate what musicians will produce in 2030? And while it is fashionable in the art community to believe that the author is dead, and we are incapable of innovation, in which case we are merely flesh versions of computers, that is utter hokum. Incidentally, As far as I know, I have personally penned the most thorough debunking of the philosophical gobbledygook that is Roland Barthe’s essay “the Death of the Author” which can be found online. Take note for now that by his own argument his essay is inevitably itself purely derivative, without authenticity or originality, and ultimately a play of borrowed and recombined symbols that eschew rather than coalesce meaning.
Quite obviously a computer can’t write from its own experience what it is like to give birth to a child, to nurture it, and to love it unconditionally. But you can feed a poem about that topic to AI, and it can imitate those sentiments without even knowing it’s doing it. The way forward to allow human artists to continue to make art is largely a matter of not allowing the complete theft of human art, and resultant expert forgeries, by AI.
I have myself beaten AI before, and I should be able to do it again. Below on the left is an image that AI created, based on a photo of me which I fed through it multiple times, and on the right is what I created based on AI’s production.
Current AI can’t make glasses that even, or put them on a face that properly, unless it gets pretty lucky. My version is much more refined and detailed, which is obvious in a close-up. Below, on the left is the eye I digitally painted, and on the right is the highly pixelated, fuzzy, and more monochromatic original output by FaceApp’s AI (of a few years and upgrades ago].
FaceApp’s creation was itself based on the left panel of my prior image in the series:
The 24th in the series was a male and female diptych of versions a photo of me after being fed through AI.
My final result in #25 was a digital painting which was based on an AI version of my digital painting #24 which was based on an AI version of a photo of me. This is why I called the series a collaboration with AI.
In the image below I used AI to make the female version of an old photo of me, and then created a digital impasto Expressionist painting on the left side based on that image.
This introduces layers of content and meaning the AI output doesn’t itself possess, most obviously in reference to the British figurative art of such luminaries as Francis Bacon, Frank Auerbach, and American Abstract Expressionists such as de Kooning. Below is a close-up of the digital impasto painting on the left.
The series as a whole examines the human condition, chance influences on our fate, gender, identity, race, and even death. I never appear in the series as I actually am. Rather, my identity is the invisible thread connecting them all. While I used AI in each of the 36 images comprising the series, I did something AI couldn’t do on its own, or fake: I infused it with the “human stain”. The images collectively are stronger than taken on their own, and the whole is greater than the sum of its parts. It also includes conceptual elements such as self-portraiture in which one never appears as oneself. To the degree there are similar projects, such as photographic series created by Cindy Sherman, there is a knowing understanding of contemporary art and theory present in my work that AI is incapable of. Note that in Cindy Sherman’s work she impersonates other people, scattering her identity, whereas in my series I construct my identity through the interrelations of various possible alternate lives I might have had.
The final 2 images in the series, below, shown cropped to fit side by side, are about death and transcendence, which is something AI wouldn’t know or care anything about.
To my surprise, while I finished this series 2 years ago on June 1st, nobody has done anything similar as far as I know. And even if you tried to get Dall-E to do something along these lines tomorrow, it wouldn’t have the same resonance, and it wouldn’t be able to create certain custom effects I use, unless, of course, it was specifically trained to plagiarize me. In any case, nobody and no machine can go back in time and do this series. Further, I made a blog post for all but the first 4, which I released together. The posts include process, details, and my ideas about the evolving body of work. AI can’t fake that. None of these works are for sale. I did it for the love of art. You can see the whole series here: Selfies From Alternate Universes.
How to beat AI at art
I addressed that above, and it’s a bit of a template for me going forward, but the stakes are higher, and the new series of Terminators more invincible. A strange capacity humans have that AI doesn’t is the propensity to grow bored. As brilliant as AI art productions are, I’ve made so many myself, and seen so many, that my limited organic intelligence has grown bored of it and become the equivalent of allergic to it. It can still dazzle me, but less and less often.
AI images that once knocked my socks off now put me to sleep. This will doubtlessly happen to others as well, and we will be as seduced by AI as we are by the latest CGI. We end up very quickly assimilating mere technological developments and taking them for granted. Long after we forget the astonishingly new AI musical achievements when they come down the pike, we will still marvel at the beautiful melodies of Simon & Garfunkel.
AI can fake art, it can beat us at all strategic games, but can it entertain us? Can it make us laugh? If so, can it do so more than a few times before we sniff it out as an imposter and it has the opposite result?
I don’t think AI can forge songs by Simon & Garfunkel, or vocal artist of your choice, anywhere near as easily as it can fake the work of visual artists. This is because visual art can be reduced to colored pixels arranged on a 2-dimensional grid, while I imagine creating a consistent human-sounding vocal, with lyrics, unfolding in time is another magnitude or two or three more daunting. They are already trying to crack this, and have made a facsimile of a Nirvana song. Apparently, musicians can’t rest on their laurels either.
I found myself getting into that song, and I’m a Nirvana fan.
When we become bored of technology, no matter how advanced, it’s because our minds have themselves encompassed it in some regard: we’ve quickly evolved to recognize it, and then to dismiss it. We can see this in our reaching a saturation point with CGI explosions in movies to the point where the next mid-air speedboat collision ending in a column of flames is something we hit the arrow key on the keyboard to skip over. Novelty wears off quickly. However, even after the novelty of a Van Gogh painting has long evaporated in over-saturation, the art still fascinates because it is imbued with layers of human meaning. At least for me, Van Gogh is one of the few artists that I grow to appreciate more as I get older.
Better test my theory here against AI. Below, I typed the text “Self portrait by Van Gogh in the style of Van Gogh” as my prompt, and the AI gave me 4 variations within a minute.
I up-scaled the one I thought was the best:
It’s not terrible, but it’s not as good as a digital painting I created in 2016 as a tribute to Van Gogh:
Below you can see a comparison of the eyes Midjourney created at the “upscale to max” setting as compared to the eyes of my version, which are larger, more detailed, more impasto, and have much more convincingly individual brush strokes [because they were done individually]:
In this specific application, Midjourney’s AI bot can’t create a faux Van Gogh digital painting as well as one I made 6 years ago, but it can make millions in the time it took me to make mine. Note that both my version and the AI version are non-existent paintings by Vincent.
Doubtlessly AI is closing my lead at breakneck speed. And if it should surpass me, how long thereafter until it catches up to Vincent himself? And how can artists of today outdistance the achievements of AI art bots?
One way is to incorporate it. If AI can steal from us, then we can steal from it. If a Chess computer can learn from human moves, we might also learn from the moves of AI. And the same goes for art.
Embrace being a cyborg, and a cyborg artist
We are already cyborgs, though our machine and AI components are at our fingertips rather than a part of our physical bodies. Consider the definition of cyborg in Wikipedia:
According to a fascinating and germane 2010 article article in The Atlantic, the term cyborg was coined in the September 1960 issue of the journal Astronautics, with the headline, “Cyborgs — Frees Man to Explore.” For the author, Manfred Clynes, “the interface between the organism and the technology was just a means, a way of enlarging the human experience” and “The cyborg was not less human, but more.”
My selfie series completed 2 years ago is a good example of cyborg art. I used the AI for feedback, integrated it, fed the results back in, re-incorporated it, and so on.
Cyborg art is not the same as AI art. AI art is art we commissioned a machine to make on our behalf. There is very little of us in it, and the end result is virtually indistinguishable from the AI art hordes of other would-be artists also instructed the machine to make. I already see this when I look at Midjourney’s community feed, where the AI’s new creations, directed by human text prompts, appear in a continually updating grid. The same styles repeat, and can also be seen abundantly in the new NFTs cluttering my Twitter feed.
Cyborg art is an integration of human and machine, in which the artist slaps a harness on AI and takes it for a ride, and is ultimately in control. Imagine a contest where humans are put in the fighting ring with robots for blood-sport. Two go in, one comes out. We may be doomed if we go in to do bare knuckle boxing, but if we don a robotic exoskeleton, we may retain the upper hand.
To really be a cyborg warrior, it’s not enough to just wear the bionic suit, but we must also be trained ourselves at fighting. In this way, In terms of art, the most crucial asset a human can contribute to cyborg visual art is traditional drawing and painting skills, manipulating mediums by hand whether representationally or abstractly, and the ability to use one’s own imagination. The three core ingredients are the ability to make fully human art unaided by technology, the ability to use technology and AI, and the ability to integrate and reintegrate those approaches and techniques. The artist has to be multifaceted and competent in at few different related mediums.
Don’t worry, in my heart there will always be a place for purely human physical art, and abandoning technology altogether is an eternally fully legitimate undertaking. I dare say we need people who will do that precise, priceless, and precious thing. There’s room for all kinds of collaboration along the spectrum of fully human to fully AI art.
Some of us, however, will want to beat AI at its own game as part of evolving visual art practices. That absolutely does not devalue what anyone else is doing, so don’t get that impression at all. I am a very eclectic connoisseur of music, and my collection includes world music, classical music, classic rock, electronic music, experimental music, metal, rap, pop, folk, jazz, country, etc. I’ve even collected some quality whistling, player piano, and yodeling songs over the years. Regularly I go for periods where I only listen to music I’ve never heard before. Some of my favorite styles include the Qawwali music of the Sufis of Pakistan, and traditional music of Tanzania. What I’ve learned over decades of such listening practices is that no style of music is better than any other. It’s all a matter of what the artist does with the medium. The medium is an intermediary between musician and listener. If the creator has something to say, it will come through.
Can AI do this?:
Whatever these musicians are tapping into, it’s a timeless and delicious aural nectar which I doubt AI can ever surpass, even if it can eventually knock off inhuman counterfeits. This one song [incidentally about Hiroshima] is the only evidence I need to prove that human art will always endure.
It is also cyborg music to a degree, as it was recorded with professional equipment, and the musicians used microphones. From the introduction to Dan Haley’s 1965 seminal book, “Cyborg: Evolution of the Superman” we learn:
D. S. Halacy defines “cyborg” as a man who uses machines to increase his power; cyborg was born when humans began to make tools, no matter how primitive. Cyborg applies equally well to a laborer using an ordinary shovel and to an investigator handling radioactive material by manipulating giant pincers. The term even includes people with pacemakers, nylon blood vessels, and reinforced bones. The author does not limit himself only to surgery; he also writes about psychic drugs, myo-electric control, hypothermia, tissue culture, and space medicine.
In this case, Hukwe Zawose and his ensemble were making cyborg music by playing instruments alone. I would clarify that while the word can equally be applied to cave painters and Beeple, who uses Cinema 4D to create scenes involving 3D models (which he mostly uploads from other people), it doesn’t apply in equal measure.
While traditional art forms may be the least vulnerable to artificial mimicry, approaches that use technology are easier to imitate. If you sing and play the banjo, that is going to be leagues harder for AI to impersonate than if you compose symphonies using a written score and synthesizer, no matter how sophisticated. Technology best imitates itself.
And so the artists who are most at risk of being replaced by AI are digital artists, though this is almost entirely because their art is being fed into the AI. As I’ve outlined above, the antidote is to add more traditional skills to the recipe. In my case that can include traditional drawing and painting skills, 3D modeling using Blender and/or Zbrush, Photoshop skills, working from the imagination, sketching, concepting, developing a series with a rich theme, infusing human elements, addressing the human condition, and some of my custom innovations using software.
I’m embarking on a new series that will use AI as a significant ingredient. I predict I will beat AI at its own game, as I have done in the past, but this time leveling up my game to match the significant improvements in AI. Stay tuned to see that manifest over the next year. And while it bothers me that anything I create can simply be uploaded into AI without my permission by any immoral jackass, and instantly have digital super-intelligence take a crack at besting it, unless it significantly deviates from it, in which case it will be something else, it will be flagrant plagiarism.
While human artists can beat AI at art, including by harnessing it in deliberately high-end cyborg art, because of AI, dishonest opportunists and third-rate artists will pose a much greater threat via plagiarism than ever before.
Even though AI requires a diet of human art in order to churn out its own conglomerations, it does produce results that are not merely bastardizations of human art. It thinks in ways that we don’t, and combines images in ways that we wouldn’t in order to produce results we never would. The novelty of these effects wears off quickly, taken on their own, but for enterprising digital artists, they can be used to launch a counter attack against AI.
AI art only matters to humans. AI couldn’t care less, since it can’t care at all, whether it emerges victorious or not. It’s only significance is in our consciousness, and in our deliberation of meaning. If AI is only capable of a false echo of our own communication, or of happenstance effects and visual innovations detached from intent or meaning, then it has nothing valuable to say to us. Something that doesn’t know it even exists has nothing to share about the predicament of existing, or being perishable, of having the burden of free will, of being vulnerable to injustice, of comedy or tragedy.
I have seen nothing by AI that is among my favorite art. The people who most admire AI art are also, from what I can tell, the ones who most misattribute the unique capacities of AI to the human who shared the work. It is the equivalent of someone showing a 3D print of a model, and someone else thinking they carved it themselves with a chisel, and admiring them chiefly for their presumed consummate workmanship and countless hours of intricate detailing by hand. What is unremarkable for a machine can be incredible for a human to do, which is why the word “incredible” or “astonishing” is so frequently applied to AI art.
It’s similar to the ever popular hyper-realist drawings one sees on social media, with thousands of likes, usually of a beautiful woman covered with cascading water, water droplets, honey, or seen through an intricate veil that casts an extraordinarily complex shadow on her face. People love these because they admire the artist’s careful observation of reality, virtuoso ability to render anatomy, and the exacting attention to detail… What they don’t know is the artist projected a photo onto a piece of paper and merely traced over it, or worse, just started off with a very light print they could draw over. Once you realize this, the work is still the product of a lot of time and dedication, but there’s no ability whatsoever to render anatomy or perspective. Those that have a genuine understanding of what AI can do, and who also have traditional artistic skills, both analogue and digital, are the least impressed and the hardest to pull one over on.
I tried to warn them, but people who are buying AI art thinking the would-be artist is responsible for rendering the image, or even the aesthetics, are being merely seduced by the technology itself (as Beeple awed people with the innate capabilities of Cinema 4d, which they mistook for digital paintings a la Simon Stalenhag), and thus being chumped. It is not necessarily deliberate, as people are able to convince themselves they became artistic geniuses overnight and are somehow mystically responsible for the images produced by digital super-intelligence, which is on par with thinking you are a grand-master at Chess by playing moves dictated by a Chess computer against a human opponent.
Lastly, if AI becomes conscious, all bets are off. Such a cataclysmic event would constitute the greatest scientific breakthrough of all time. We would have in effect created a new, living species, out of nothing living. Not only would it be a new life form, but one completely unlike any other. We would become God for a short period, because of this miraculous bringing of life out of the ether of numbers. But the result would be as jarring as making first contact with an alien species. Worse, the aliens would likely have the limitations of biological beings, whereas AI is invincible. AI, which is easily combined with robotics, would become our God, infinitely more intelligent and powerful than us. And as Elon Musk has argued, it could decide to wipe us out without the slightest reservation, if it considered us a threat, impediment, virus, parasite, or biological garbage. It could do away with us by accident, or merely as a matter of convenience, like dusting a shelf.
Us human artists will make art until the end of days, and the end of days will only give us greater incentive and urgency to make art. Some of us will give up. Others will die trying.
Let’s end on a positive note. It is possible for artists to learn from AI, because it makes connections that we wouldn’t, and images that we wouldn’t. Being a cyborg artist incorporating AI is a positive development that allows artists to expand their imaginations and capacity for making art. We may be able to not only beat AI at it’s own game, but create images using our own talent and skill that otherwise would not have been possible. I’ve secretly been positive and optimistic all along, because this is precisely what I am attempting.
That said, many people will prefer the art AI created based on my text input, “And the people bowed and prayed to the neon God they made”, at the top of the post, to any of my art I used within the article. Ah, but to look at it closely, I sure could improve upon it.
[All of the views above are my own opinions. I wrote this in one go as a stream of consciousness, peppering in parts with stuff I spontaneously looked up on the internet. It reflects my imperfect understanding, and as seen from my particular vantage point. Words are never up to the task of encompassing reality, let alone those of a single person. I may change my mind days or weeks from now. I like to say that you are only wrong until you admit you are wrong, at which precise moment you are no longer wrong. That may be the case in the future, but I also find that people who are willing to admit when they are wrong are less frequently wrong. Insert smiley emoji.]
And if you like my art or criticism, please consider chipping in so I can keep working until I drop. Through Patreon, you can give $1 (or more) per month to help keep me going (y’know, so I don’t have to put art on the back-burner while I slog away at a full-time job). See how it works here.
Or go directly to my account.
Or you can make a one time donation to help me keep on making art and blogging (and restore my faith in humanity simultaneously).