Karl Marx

In the [YOU] on AI Field Guide

The cycle that began with [YOU] on AI insists on looking clearly at what is actually happening in the AI transition, without the narcotic of hype or the paralysis of fear. Marx is the methodological model for this insistence. His central move—what he called ideology critique, illustrated by the camera obscura image in The German Ideology—was to refuse to take the smooth surface of an economic system at its word, to ask what it conceals about the labor and ownership relations that produce it. Applied to AI, the move is straightforward and uncomfortable: the smooth, friendly interface of a generative AI product conceals the labor of millions of data annotators, the appropriated work of millions of writers and artists, and the ownership structure that captures the value all of this produces.

The book’s central question—what does the orange pill see?—is answered differently by different thinkers in the cycle. Marx answers it at the level of political economy. To take the orange pill, on his account, is to see the machine not as a neutral tool but as a crystallization of social relations: of who invested, who labored, who owns, and who benefits. Primitive accumulation in the AI age is the enclosure of the cognitive commons—the conversion of humanity’s collective creative output into training data that is then returned to humanity as a proprietary service, for a fee.

The cycle’s most Marxian chapter is the one about dead labor. Marx’s gothic image of capital as “dead labour that, vampire-like, only lives by sucking living labour” names the structural condition of AI with a precision that borders on the prophetic. The model is an enormous crystallization of past human labor—writing, code, art, scientific papers—reanimated by the compute investment of a corporation and deployed to capture value from the living labor of current users. The living labor of the prompt engineer, the creative director, the programmer pairs with the dead labor of the model in proportions that shift continuously as the model improves.

He stands in the cycle alongside Shoshana Zuboff—who extends the surveillance capitalism analysis from Marx’s framework—and Andrew Feenberg, who draws on the critical theory tradition Marx inaugurated to argue that technology is never politically neutral. Marx’s specific contribution is the most basic and the hardest to look at: before asking what AI does for us, ask what it is made of, who made it, and who owns the making.

Origin

Marx was born in 1818 in Trier, in what was then Prussian Rhineland, to a family of Jewish lawyers recently converted to Lutheranism. He studied law and then philosophy at Bonn and Berlin, where he encountered Hegel’s dialectic and immediately subjected it to the inversion that would define his mature work: Hegel’s system moved from ideas to material reality; Marx proposed to move from material, economic relations to the ideas they produce. The method he developed—historical materialism—holds that the dominant ideas of any age are the ideas of the dominant class, and that the dominant class is defined by its ownership of the means of production.

His collaboration with Friedrich Engels, begun in 1844 and lasting until Marx’s death in 1883, produced the foundational texts: The German Ideology (1845-6, unpublished in his lifetime), which laid out the camera obscura image of ideology; The Communist Manifesto (1848); and the sprawling manuscript of the Grundrisse (1857-8), which contained the Fragment on Machines—the passage that has become the most cited text in any Marxist reading of AI. The three volumes of Capital, begun in the 1860s and continued by Engels after Marx’s death, provided the theoretical apparatus of surplus value, dead labor, commodity fetishism, and the labor theory of value that remains his central contribution to economic thought.

The Fragment on Machines in the Grundrisse is the text that makes Marx irreplaceable as a thinker for the AI age. Writing in 1858, before the invention of the telephone, Marx described a stage of production in which the general intellect—the accumulated scientific and social knowledge of the species—would become directly embodied in the machinery itself, so that the machine would not merely assist human labor but replace it as the primary productive force. The distribution of value in such a society, he argued, could no longer be based on labor time—and the contradiction between the social production of knowledge and its private appropriation as capital would become the central antagonism of the age.

Key Ideas

The general intellect. Marx’s term for the aggregate scientific, technical, and social knowledge of humanity—developed collectively across generations and, in the advanced stage of production, embodied directly in the machinery itself. AI training corpora are the general intellect in the precise sense Marx described: the crystallized product of collective human cognitive labor, now converted into a privately owned productive force. The question he presses is not whether this is useful but who owns it and who benefits from the rent it generates.

Dead labor and living labor. Marx’s distinction between the past labor crystallized in machinery and the current labor of the worker operating alongside it names the structural condition of human-AI collaboration with uncomfortable precision. A large language model is the largest crystallization of dead labor in human history—billions of hours of human writing, coding, and thinking, reanimated by capital. The practitioner who uses it is the living labor that makes the dead labor productive; the value generated flows disproportionately to the owner of the model.

Commodity fetishism. The way a product’s social origins in human labor are concealed behind the smooth surface of the commodity. The AI interface presents as a neutral, helpful service; the labor of the annotators, the writers, the programmers, the researchers whose work was scraped to train it is made invisible. Commodity fetishism in the AI age is the smoothness of the product concealing the exploitation of the process—and Marx’s method of ideology critique is the instrument for making it visible.

Alienation. Marx’s 1844 analysis of what wage labor does to the human being—estrangement from the product, from the process, from other workers, and from the species-being of humanity as a creative, purposive animal—describes the experience of working under AI systems with painful accuracy. The practitioner who produces excellent output through an AI that manages the cognition is estranged from their own labor in the fourth sense Marx identified: estranged from the specifically human capacity for deliberate, self-directed creation that distinguishes homo faber from the machine.

The enclosure of the cognitive commons. Marx’s analysis of primitive accumulation—the violent conversion of common land into private property that preceded the development of industrial capitalism—provides the framework for understanding the appropriation of AI training data. The enclosure of the intelligence commons converts humanity’s collectively produced knowledge into proprietary training data and returns it as a commercial service, extracting rent from the very people whose labor generated the underlying value.