An A.I. Utopia Is No Place For Humans
The A.I. future we’re being offered has little use for human labor—or human beings at all
by Christopher Pearce
Over the last year, an unlikely path to automated utopia has come into view. Built on an endless amount of stolen artistic labor, generative A.I. tools like DALL-E and ChatGPT can now generate a seemingly infinite amount of ready-made images and text—from illustrations to comics to short stories to essays and marketing material—all with just a simple written prompt. Automation has historically been focused primarily on physical labor like transportation, manufacturing, and logistics; and so this represents something of a reverse in the conventional wisdom. Sam Altman, the CEO of Open.Ai, the company that created both DALL-E and ChatGPT, has gone on record suggesting that the job automation these tools will bring about will follow this same counterintuitive order: beginning with creative rather than physical work.
This is the future laid out before us: the end of all artistic and intellectual professions, and the mass proletarianization of everyone. A.I. advocates might balk at this phrasing, but it’s the logical conclusion of their arguments: if A.I. really does outperform humans, what need do we have for human labor? A.I does not need to be better than humans to displace them, after all: it simply needs to be functional and cheap. Automated checkouts are slower and more susceptible to theft than human cashiers, but they are considerably cheaper than human labor in the long term and (presumably) less prone to unionization. Absent massive and systemic change in the socioeconomic order, the end result of algorithmic automation will be a mass shift of wealth and power into the hands of those who own these new engines of creation—the bourgeoisie—and, more troublingly, the end or severe curtailment of the ability of art to effect change in the world.
The law regarding A.I. generated art and copyright is currently ambiguous. Although long-standing doctrine from the USPTO suggests that works that are wholly created by an algorithm or some other form of machine cannot be copyrighted (since only the human creator of a work can hold its initial copyright), future court rulings are likely to overturn this standard. The tech companies that own these algorithms are enormously wealthy and powerful; artists are not. Whether these algorithms are simply a sophisticated form of photobashing, unable to actually create anything original, is irrelevant. Capital’s continued ability to extract profit from labor seems to depend on the use of A.I., and courts will inevitably reach an outcome that favors or at least facilitates this.
This might even be written into future contracts: since the company employing you (or the company publishing your work) owns the content, they will have the right to feed it to an A.I., and they will have the rights to whatever the A.I. produces. Given the sheer volume of material it takes to train an AI, the claim will likely be that the “art” produced isn’t derived from your work at all; after all, there’s a thousand other artists who had their art uploaded into the electronic hivemind, and that’s why no one except the corporation itself can really claim ownership over what the A.I. produces.
As these technologies and their inexpensive, on-demand output proliferate, they will crowd out human artists—and human writers, doctors, scientists, and lawyers will all follow, as the capabilities of A.I. tools grow more sophisticated and become more accepted throughout other industries. For the arts especially, this will lead to a rapid decline in the number of students pursuing those fields—even beyond the current pivot away from the arts. Even students who could, given time and effort, surpass the work of A.I. artists and writers, won’t bother attempting to do so if there is no market for that work. Despite its often stilted writing and uncanny images, the baseline output of A.I. is so vastly superior to all but a few savants that the effort to learn and improve will be worth it to a vanishingly small number of people.
Students who could, given time and effort, surpass the work of A.I. won’t bother trying if there is no market for that work.
But where will the displaced intelligentsia go? Past collapses of areas of the arts, such as journalism, has seen an exodus from those disappearing fields into adjacent fields like marketing or public relations—but with even those fields increasingly facing automation, there will be no such escape valve for the targets of the next automation wave. With the path to middle-class economic security via education closed, manual labor and service jobs will be flooded with people seeking employment, allowing those jobs to be paid pauper’s wages. What possibility of escape will there be from poverty and alienation, once corralled into easily replaceable, increasingly disenfranchised work? There will be only the assembly line, the gig economy, and servility.
Of course, creative work will not cease to exist overnight. Many jobs will still exist in fields like animation and video games—but even these once-dreamed-of jobs will face their own sort of A.I. impingement. The dream of automation was a dream of liberation: that mundane, menial, and dangerous tasks would be given to unthinking machines, freeing humans to perform more creative and fulfilling work. Even the creation of art has work that leaves its laborers alienated: frame-by-frame animation is grueling work, and conducted on what is essentially an assembly line. The current generation of A.I. struggles with this sort of “oversight” labor. It can create images, concepts, and ideas with ease—but combining these into a finished project may well remain work fit only for humans… but only this work. Illustrators will not be commissioned to create finished pieces, but they might have the “luxury” of eking out a living fixing the strangeness that the algorithm produces, unmelting hands out of faces one after the other. Writers, too, are already being tasked with hammering together endless reams of A.I. text into coherent novels, blogs posts, and essays. Soon enough, this sort of “clean up” work might be the only way for a human to make a living off of artistic work at all.
As the “high concept” artistic labor—brainstorming, ideation, the raw stuff of artistic creation—passes into the hands of A.I., it will also be passing into the hands of the bourgeoisie. This transition will also gradually bring about an end of the ability of art to meaningfully critique the ruling class. Even in the most focus-grouped of Marvel movies, one can occasionally see the artist straining to critique the power structure that produced their work: Captain Marvel was both a feature length Air Force propaganda film and a critique (however limited) of the military complex that created it. But with A.I. in the director’s chair, the ruling class will find it even easier to filter out troublesome political themes. Rather than having to threaten a human director with firing or blacklisting, all while wondering what they might be sneaking in under your nose, an artistic process with no artist in the driver’s seat is one that be changed or rewritten at a whim with nothing greater than a slight change to the prompt you’ve fed to an algorithm, or a few adjustments to the code.
The ease with which art can be created via automation might seem like an upside; but alas, it is an illusion. Automated art is already effortlessly inoffensive, incapable of challenging the consumer or directly restricted from depicting subject matter deemed inappropriate—the apotheosis of Theodor Adorno’s “culture industry, wherein culture is homogenized to hide the alienation caused by capitalism. And while modern pop culture already aims for the overly broad, attempting to be palpable to all audiences while not being bespoke enough to be truly satisfying, automated art opens up the possibility of being both highly specific to the individual for whom it was created and highly limited in what themes that it can impart. (Imagine if one of those bizarrely specific, randomly generated t-shirts produced by bots could be a TV show: “This movie is for South Carolinians who love chocolate, went to the University of Texas, have two older brothers, are blonde, and intensely fear roller coasters!”)
If human dexterity can be replicated by machine learning, the perceived utility of the working classes will drop even further.
Of course, given that A.I. must be fed human labor for it to produce its content, its destruction of art as a field may create a hard limit on its ability to create. DALL-E, for instance, struggles with compositionality—ask it to put specific things in specific places, and it is often unable to produce the results you request. While tools like Chat GPT or A.I. Dungeon can produce impressively creative short descriptions, asking them to implement a complete story typically leads them to immediately return to the most generic of worn-out tropes and cliches. But even if such a limit exists, it won’t matter by the time it’s reached: capital has no interest in the quality of the art, merely its ability to produce wealth for owners.
All of this, of course, assumes that the problems of automated manual labor are insurmountable, which may not be true; as DALL-E’s advancement from vaguely nightmarish monstrosities to art that was nightmarish in an entirely different way demonstrates, things can change in an instant. If human dexterity can, ultimately, be replicated by machine learning, then, the perceived utility of the working classes will drop even further.
And if an educated worker is no longer a prerequisite for the functioning of capitalist society? Education of the masses could be greatly reduced or entirely eliminated. Education, as it currently exists, is a form of Marx’s bourgeoisie necromancy: necessary to sustain the ideological superstructure of capitalism, but at the cost of the risk of making workers more aware of their oppression and how to overthrow it. And if education is eliminated, the seeds of discontent it can sow will also be eliminated. Social efficiency, with its emphasis on schools as job training centers above all else, is already the order of the day in the American educational system. As more jobs are automated, more students will be shifted to only receiving the most basic of education—if even that. Another, more human system might see different value in the education of humans; in ours, it is necessarily the labor market that drives its perceived value, and as it moves away from humans, so too will education.
An uneducated populace is one that is easy to control, especially as the ability to process and analyze information shifts more and more into the hands of the ruling class. The ability of the average person to spot government and corporate malfeasance will slowly diminish as no one but the algorithm itself will have ever been taught how. Perhaps the facade of democracy will remain, but that won’t stop more and more people from being pushed into the most degrading and dangerous of jobs, placated by art entirely subordinated to capitalist ideology, unable to fully express their dissatisfaction or organize meaningful resistance.
The planet already strains to support the bourgeoisie’s lifestyle, and as the need for labor (and therefore the value of humans) for capitalism diminishes, what is to stop depopulation from becoming the order of the day? It was not so long ago that eugenics was openly practiced by capitalist countries. The outcome need not be a overnight return to genocide. Why not a return to “suggested” sterilization? Why not voluntary suicide?
And from where will the resistance to such measures come? Certainly not from the carefully controlled, automated entertainment that is the dream of A.I. boosters and which could well become the dominant and preferred form of art left in the world.
Absent the troubling need for human labor, the bourgeoisie could simultaneously preserve its lifestyle and “save” the planet. And when the mass of humanity is finally extinct… all that will be left is “utopia”. Just not one that includes us.
Christopher Pearce is a writer living in the southern United States.
