Do LLMs obscure the historical processes behind their answers?

Horning reads LLMs through Lukács's concept of reification: the bourgeois tendency to treat "facts" as "petrified factuality," frozen into "a fixed magnitude" that "precludes any theory that could throw light on even this immediate reality." He argues LLMs exemplify this directly — they offer "a static map of facts" to users, standing in for the dynamic social processes and relations that actually produced those facts. Where Lukács held that the proletariat's alienated position under capitalism opens a privileged standpoint for seeing "reified forms as processes between men," Horning argues AI as a project works to keep that insight "maligned, stunted, eradicated, abandoned" by gratifying users with immediate "answers" instead.

The mechanism he gives is a substitution of calculation for understanding: LLMs show "how data are 'conjoined' — how words and concepts have historically fit together — while leaving them unchanged and unexplained." He treats this as deliberate obfuscation at the level of the technology's social function, not a side effect: "Capitalism has to produce ignorance and apathy to perpetuate itself; 'AI' is merely the latest means of production." As supporting evidence he cites a Wall Street Journal report that "lower artificial intelligence literacy predicts greater AI receptivity," plus a business professor's advocacy of "calibrated literacy" — teaching users just enough to find AI "delightful" without understanding it as "algorithmic pattern matching." Horning's prescription is explicitly dialectical: "Thinking rather than prompting; collaborating with other people and socializing rather than withdrawing into nonreciprocal machine chat" are what count as "de-reification."

This is a Marxist ideology-critique reading that sits closest to Why does AI discourse feel obscene in Baudrillard's sense?, which makes a structurally similar move from a different theorist: both argue that AI output presents a finished, "too-visible" surface detached from the process that would normally make it legible — Baudrillard's missing "scene" is Lukács's missing "historical totality." Horning's reading also stands in tension with Does Marxist alienation theory explain what AI does to cognitive work?, which argues that applying Marxist categories (there, alienation) to AI imports a presupposition — a prior authentic form being degraded — that doesn't survive scrutiny for cognitive labor. Horning's reification argument doesn't share that vulnerability in the same way, since his claim is about immediacy obscuring process rather than about degraded craft, but both notes show the vault holding competing views on how far classical Marxist categories travel to LLMs.

The excerpt does not establish that users who rely on LLM answers actually lose capacity for dialectical or process-oriented thinking — the WSJ-reported literacy/receptivity correlation is about willingness to use AI, not about effects of use on cognition, and Horning does not cite evidence for the causal suppression mechanism he describes. The argument is a theoretical application of Lukács's framework to a new object, offered with the rhetorical confidence of the framework rather than empirical demonstration specific to LLMs. If the application holds, the implication is that AI literacy campaigns framed as "delight" rather than mechanism are not incidental marketing choices but serve the same ideological function Lukács attributed to capitalist common sense.