The reified mind

A note on generative AI as commodified language

Many worthy critiques of “AI” pursue technical analyses of what the systems are capable of, sorting through system cards and benchmarks and all the hype that mystifies the models’ performance. But one can also gain perspective by revisiting canonical critiques of capitalism that assess the role that technology in general, any kind of technology, plays in systemically extending capital’s grasp. Regardless of whatever a particular technology can supposedly do, capital is always compelled to use it for ultimately the same purposes, reshaping the production process and the consciousness of those caught within it.

Consider Georg Lukács’s “Reification and the Consciousness of the Proletariat.” It begins with a straightforward reminder of Marx’s definition of “commodity”:

The essence of commodity-structure has often been pointed out. Its basis is that a relation between people takes on the character of a thing and thus acquires a ‘phantom objectivity’, an autonomy that seems so strictly rational and all-embracing as to conceal every trace of its fundamental nature: the relation between people.

Maybe it seems too obvious to belabor, but this kind of “phantom objectivity” applies equally to generative models, which attempt the reification of language on an unfathomably large, industrialized scale, absorbing vast amounts of capital to the degree that it’s now not uncommon to hear it claimed that investment in “AI” is propping up the entire U.S economy.

If language itself is a kind of reification — supplanting live things with dead words — then generative models raise this exponentially, further obfuscating the relations between people that first animated language and imbued its systematized matrix of differences with socially established meanings. Models don’t “generate” any new meanings but enclose and commodify those that already exist, working to freeze the embedded social relations in whatever turns out to be the most profitable forms for the interests that own and promote “AI.” (Right now, as it happens, businesses are finding it hard to put the technology to economic use, with 95% failing to find any profit boost from it, according to recent MIT research.)

LLMs make language available in a commodified form whose value rests in “phantom objectivity” — its apparent autonomy from any determining contexts or subjective intentions. It’s as if language has been made fixed, static, concrete, in defiance of its actual fluidity in social practice. LLMs thus offer users the apparent ability to acquire neutralized, finished language to accomplish some economic purpose instantly without any need for speaking subjects; it makes available language that seemingly can’t be undermined by any idiosyncratic personality or usage, by any ambiguity or requirement for interpretation.

The “hope” (for capital) implicit in LLMs seems to be that if the language deployed is sufficiently frozen, it can have a freezing effect on the people subjected to it — that it will help kill the language in their brain that still is alive, in the sense that the person regards it as a tool for expressing as yet unformulated things — as making possible a kind of craft production with language — rather than a reified product, a commodity that must necessarily be purchased on a market. LLMs perform a kind of taxidermy with words, presenting them in a lifelike fashion that mainly serves to remind you that they are dead. Being compelled to consume LLM-generated content is like being stuffed with that sawdust.

Describing the world as dominated by the commodity form, Lukács writes:

The commodity can only be understood in its undistorted essence when it becomes the universal category of society as a whole. Only in this context does the reificiation produced by commodity relations assume decisive importance both for the objective evolution of society and for the stance adopted by men toward it. Only then does the commodity become crucial for the subjugation of men’s consciousness to the forms in which this reification finds expression and for their attempts to comprehend the process or to rebel against its disastrous effects and liberate themselves from servitude to the ‘second nature’ so created.

LLMs could be understood as one of the forms in which reification — the commodified world — finds “expression,” manifests as expression, and extends the “second nature” of the world as made of so many commodities into the realm of language, disseminated as inert tokens to be shuffled and exchanged through processes that are ultimately closed off to creative use by human subjects.

Instead, we will be brought to “buy” the language we want for whatever reason from machines; that the desire for language itself may even regress into prelinguistic or nonlinguistic forms, since we may be systematically deskilled from language use to the point where their is no longer desire conceivable in words. Why bother to think your desires as expressive thoughts when, say, a machine implanted in your brain can translate your brainwaves into words for you? When words are reified as things, the desire for them becomes a kind of depthless desire for simple accumulation.

One might think of language before LLMs as akin to the “organic unities of pre-capitalist society” that Lukács invokes as a point of contrast to the world determined by the universalized commodity form, in which “all the social and economic conditions necessary for the emergence of modern capitalism tend to replace ‘natural’ relations which exhibit human relations more plainly by rationally reified relations.” Lukács traces these conditions in the production process: “If we follow the path taken by labour in its development from the handicrafts via cooperation and manufacture to machine industry,” he argues, “we can see a continuous trend towards greater rationalization, the progressive elimination of the qualitative, human and individual attributes of the worker.”

LLMs reflect a similar process of by which language is rationalized and the qualitative components of human language use are progressively eliminated; communication becomes another aspect of abstract labor that can be carried out by interchangeable agents, conveying nothing idiosyncratic or individual about them.

This rational mechanization extends right into the worker’s ‘soul’: even his psychological attributes are separated from his total personality and placed in opposition to it so as to facilitate their integration into specialized rational systems and their reduction to statistically viable concepts.

If tech companies and their sponsors have their way, language and communication would be similarly reduced so that it would never occur to anyone to use such tools for anything other than alienated production. “The mechanical disintegration of the process of production into its components also destroys those bonds that had bound individuals to a community in the days when production was still ‘organic,’” Lukács writes. “In this respect, too, mechanization makes of them isolated abstract atoms whose work no longer brings them together directly and organically; it becomes mediated to an increasing extent exclusively by the abstract laws of the mechanism which imprisons them.”

Even if you are skeptical of this kind of talk about “organic” social organization, it is not hard to conceive of its opposite in mechanized language being unleashed as a kind of imprisoning force, binding isolated chatbot users to their mechanical mediators. Reified language works to reinforce the development of the “reified mind,” chatting with LLMs makes this process vividly interactive. One can engage with the ideological training module directly and derive a kind of direct enjoyment from it, whether as a sense of mastery, or control, or flattery, or entertainment. This process casts human conversations as a kind of “organic” relation of production that has become moribund, outmoded, no longer efficient enough to keep up with the demands of capital.

The human use of human language threatens to decommodify it, destabilize its reified form. So the ideological campaign against conversation will likely intensify, transitioning from apologetics for “relationships with chatbots” and sympathy with their sycophancy, to castigating human relationships as first inefficient and inconvenient, and then dangerously chaotic, and then a kind of menace to the established order.