Living in Uncanny Valley: On the Forces and Relations of Production of AI

 

One common refrain one hears about AI is that it is inevitable. It is nothing other than the progressive development of the possibilities of technology. Such an assertion could be considered a version of technological determinism. It is technology, what Marx called the forces of production, that drive history. On this view history proceeds from the engineers workshop to the factory and into society.

Some have attributed this position on technology to Marx himself, drawing on his remark in The Poverty of Philosophy, "Social relations are closely bound up with productive forces. In acquiring new productive forces men change their mode of production; and in changing their mode of production, in changing the way of earning their living, they change all their social relations. The hand-mill gives you society with the feudal lord; the steam-mill society with the industrial capitalist." Without rehearsing the entire argument on Marx and technology, one that often draws from a series of quotes and passages cited from various texts and thus, as is often case, a changing position, it is possible, at least for this post, to shift to a different position, one predicated not on the primacy of the forces of production, of technology, but the relations of production, of the social and economic relations. 

As Louis Althusser gives this formulation his characteristic precise formulation when he writes the following, "Within the specific unity of the productive forces and relations of production constituting a mode of production, the relations of production play the determining role, on the basis of, and within the the objective limits set by, the existing productive forces." Althusser formulation poses a problem in the form of a thesis. How is it possible grasp the determining instance of the relations of production in what appears to be the determination of the forces? This question is complicated by the fact that forces are not simply posited as an effect of the relations, but have their own role to play in setting limits. 

To return to the question of AI we can see that one of the primary uses of this technology is to reduce the costs of production, to not so much eliminate the role of workers in writing text and creating images but to reduce the role of such work to a negligible amount, to that of proofreading and editing what a machine produces. In this way the worker is reduced to a "conscious organ" of the machine, as Marx wrote. AI is adopted not out of an inevitable drive to make things faster or better;  the primary role that AI provides is to make this process cheaper, standardized, and easier to control. Cheap comes at the price of quality, and thus we see "slop" as the defining word of the year, as poor quality images, videos, and texts permeate the online space. Beyond this general imperative to make labor cheaper and more disposable, there are two examples where it is possible to see the hype around AI being more about the relations than the forces of production, Hollywood and the university. 

To take the first example Hollywood. Hollywood, and film and television production in general, has been one of the most visible lines of conflict around AI use. It featured prominently in the 2023 Writers strike and continues to be a contentious issue, with directors and filmmakers taking visible stands for and against the use of this technology. So far there have been stories of it being used in movie posters and other peripheral aspects of film production, and I imagine more than a few screenplays have passed through ChatGPT, but it is not these ideas of cost cutting that dominants the image of AI in Hollywood. One often hears about the idea that with AI it will be possible to train a machine on various existing screenplays, television shows, or novels and get a new screenplay. I do not know if this technologically possible, but I do know that this image of the future is in some sense already here.

What I have referred to as the "Intellectual Property Film," the sequels, reboots, and prequels of some existing movie franchise, comic book, or television show are often written in exactly this way. A writer, or more often, a group of writers comb through existing versions to produce something new, yet not quite new. One of the defining characteristics of the "comic book movie" is that with a few exceptions, such as the first Watchmen film and the Sandman series, the movies are not adaptations of existing stories or narratives. They are instead adaptations of general tendencies or directions of existing plots and storylines in the original versions. Thus we get several films that are modifications of the gritty and dark Batman from Frank Miller's comics and related stories. What is produced at the film is both like and unlike what has already been produced in other medium, an uncanny image of what has come before. Déjà vu is our historical situation and dominant form of cultural production. 

I plan to have more to say about this book when I finish it. 

In their forthcoming book, Imagination Artificielle, Intelligences Aliénées, Grégory Chatonsky and Yves Citton have argued that one of the defining characteristic of AI image production is a kind of déjà vu in which what is new appears to be at the same time old, everything new is at once familiar. This has to do with the way in which AI images draw from the stockpile of existing images in order to produce something which is a slight variation of what we have already seen. As they write, "The generated images resemble other images without being those images: influence inscribes these similarities in generalized family resemblances." One can thus see in AI image generation a fundamental dynamic of capitalism, there must be something new, new stories new images, but these stories and images must resemble as much as possible what has already proven profitable. In the technology of AI, in the forces of production, we see what already existed in the relations of production, the idea that some existing intellectual property can be a source of nearly infinite, but slightly different, reproductions. It is the fantasy of dead labor becoming a productive force in and of itself with labor reduced to a minimal role.

The idea of AI produced art or music is in some sense of culmination of the dead weight of existing intellectual production over the living. One of the striking things of the streaming model, of movies, television, and music, is that its ability to cheaply deliver content is almost matched by its inability to produce anything new. Spotify can make the entire discography of dead artists available at the click of a button, but it cannot pay or support living artists; Netflix can buy up Warner Brothers, Disney can buy everything else, but the cost of this is an undermining the entire film industry by shifting distribution to streaming. In some sense AI solves a problem that the streaming services has created. Spotify may turn out to be better at creating the basis for AI generated songs than promoting living artists.

A similar pattern of the primacy of relations over forces can be found in the second example, the university. Here I am not referring to the problem of cheating, and writing prompts replacing writing, but to the idea that AI can replace writing centers, language instructors, and professors. The fantasy of replacing (relatively) expensive and labor intensive instruction with some new technology has been an ongoing story of the university as long as I can remember, remember MOOCs, and the dream that lectures would come from a handful of "world class lecturers," reducing everyone else to TAs? What we are seeing with AI is the same old dream reborn with new machinery, we are told that AI can grade essays, teach, and instruct better than humans. Technology is the fantasy of capital extricating itself from its dependence on workers. The dream of an AI university is coupled with a fantasy, or nightmare, of a faculty that is radicalizing students, turning them against their own society. AI offers the dream of a cheaper and more docile professoriate. As Marx also writes about technology, "It would be possible to write quite a history of the inventions, made since 1830, for the sole purpose of supplying capital with weapons against the revolts of the working class."

These are just sketches and not analyses, my point with both is to draw a different picture of AI than the inevitable march of technology, of the forces of production, and instead to see the way in which the demand for such technology, its uses, and the dreams associated with it, are products of the relations of production, of the existing relations of capitalism.