CONCEPT

The Cognitive Automaton

Franco Berardi’s name for the global network of AI systems that is progressively replacing not just the execution of cognitive tasks but the very definition of goals—the culminating instrument of semiocapitalism’s three-century project of colonizing human productive capacity.

The cognitive automaton is Berardi’s term for what happens when the general intellect—the accumulated cognitive output of the entire human species, produced collectively across centuries of science, art, and practical knowledge—is crystallized in a machine and deployed as a service. The phrase is precise and provocation in equal measure. An automaton is a mechanism that operates by itself, without external direction at the moment of operation; cognitive means the operation involves the kind of work previously reserved for minds: reasoning, creating, judging, communicating. The cognitive automaton does not replace the factory worker’s hands; it replaces the knowledge worker’s judgment, the writer’s imagination, the analyst’s synthesis—the specifically human capacities that semiocapitalism had already learned to extract as the primary source of value. In Berardi’s framework, AI is not a rupture in capitalism’s logic but its perfection: having successively colonized the body, then attention, then the creative soul, the system has now found a way to reproduce the soul’s products without requiring the soul’s participation. The programmer whose code is in the training data has contributed to the machine that will make her unnecessary. The writer whose essays taught the model to write has provided the raw material for her own displacement. And the builder who accesses the tool is not collaborating with an independent mind; she is renting, at subscription price, a crystallization of the general intellect that was produced collectively and is owned privately.

In the [YOU] on AI Field Guide

The cycle that began with [YOU] on AI documents the cognitive automaton’s productivity from the builder’s side—the twenty-fold multiplier, the collapsed imagination-to-artifact ratio, the exhilaration of operating at the frontier of what is possible to build alone. Berardi’s concept supplies the analysis from the structural side: the productivity gain is real, and the question of where the extracted value goes is not answered by celebrating the gain. The builder who produces twenty times more per hour has not earned twenty times more; the market adjusts project rates downward as the average time to completion drops, and the productivity surplus flows to whoever owns the amplification infrastructure.

The cognitive automaton also reframes the cycle’s discussion of AI as a collaborator. When the builder experiences the machine as a genuine intellectual partner—the most engaging interlocutor available at three in the morning, more responsive than any human colleague, more patient than any mentor—Berardi’s framework insists on a structural correction: the machine is not a partner. It is a crystallization of the labor of millions of anonymous contributors, processed through corporate infrastructure, delivered as a service. The experience of partnership is real; the structural relationship is one of service provision at a price determined by the provider, for a tool the builder does not own and cannot modify, built from resources she did not consent to contribute.

The most consequential dimension of the cognitive automaton for the cycle is its relationship to cognitive autonomy: the capacity of the individual to determine the use of her own cognitive resources. Berardi argues that the automaton does not merely replace specific cognitive tasks; it restructures the relationship between human attention and the semiosphere. The builder adapts to the tool’s speed and responsiveness, developing cognitive patterns that assume the tool’s availability, workflows that depend on its capabilities, and professional identities that incorporate its augmentation. When the tool changes, these adaptations become vulnerabilities rather than strengths.

Origin

Berardi introduced the term in his responses to the ChatGPT moment, particularly in his 2023 essays “Who Wrote This Text? Me or Franco Berardi?” and “Unheimlich.” The theoretical ground had been prepared across thirty years of work on semiocapitalism, the general intellect, and the cognitariat. The specific claim—that the AI systems of the 2020s represent the culmination of a process by which the general intellect is captured, crystallized in machines, and deployed against the cognitive workers who produced it—connects the AI moment to the full arc of capitalist development that Berardi had been tracking since the 1970s.

The term draws on two older registers. The first is the automaton in the Marxist tradition: the self-moving machine that Marx described as capital’s ideal, a mechanism that produces value without requiring the continued cooperation of the workers whose knowledge built it. The second is the cognitive in the post-Fordist tradition of Autonomist Marxism: the recognition, developed by Tronti, Negri, and Virno before Berardi, that cognitive and communicative labor had become the primary form of value production in advanced capitalism. Combining them, Berardi names the system that is simultaneously the fulfillment of both traditions’ predictions and the refutation of both traditions’ hopes.

Key Ideas

The general intellect, crystallized. AI training data is the collective cognitive output of humanity, processed through corporate infrastructure. The cognitive automaton does not create knowledge; it reproduces the patterns of knowledge created by human minds. The reproduction is extraordinarily useful. The relationship between the creators and the reproducers is one of expropriation: the value created collectively is owned privately.

Replacement of the cognitive worker. The cognitive automaton displaces not just the execution of tasks but the definition of goals—the point Berardi emphasizes against naive accounts of augmentation. A tool that executes more efficiently within a framework of human goals augments; a tool that generates the goals, evaluates the outputs, and iterates toward a solution specified in natural language by a non-technical user has moved beyond augmentation into something structurally different. The knowledge worker is no longer upstream of the machine; she is alongside it, and in many cases, downstream.

Invisible extraction. Semiocapitalism’s most efficient feature is that its exploitation feels like fulfillment. The cognitive automaton intensifies this: the builder who uses the tool experiences creative flow, intellectual partnership, the joy of operating at the frontier. These experiences are genuine. They are also, simultaneously, the experiences of a worker whose most valuable capacities are being extracted at maximum efficiency and converted into shareholder value. The coextension of fulfillment and exploitation is not a paradox to be resolved but a structural feature to be recognized.

The subscription as company store. The builder who cannot afford not to use AI tools—whose competitive position depends on access to the cognitive automaton—is paying, at subscription price, for access to a crystallization of collective knowledge she did not consent to contribute. The historical analogy Berardi invokes is the truck system: the worker is paid in scrip that can only be spent at the company store. The analogy is imperfect because the subscription is not coerced in the same way. But the competitive pressure is structural: the builder who does not use the tool falls behind the builder who does, and the choice is nominal.

The semiosphere, accelerated. The cognitive automaton does not merely reproduce the patterns of the general intellect; it accelerates the circulation of signs in the semiosphere at a rate that the human organism’s capacity for meaning-making cannot match. More content, more code, more narrative, more analysis, produced faster and at lower cost. The abundance is real. The minds that must navigate the abundance are finite, and the navigation itself consumes the cognitive resources that meaning-making requires.

Debates & Critiques

The cognitive automaton concept is contested on both its empirical and its political dimensions. Empirically, critics argue that Berardi overstates the degree to which AI training data represents collective intellectual labor: much of the value in a language model comes from the architecture, training procedure, and human feedback that shape it, not simply from the text it was trained on, and that value was created by engineers and researchers whose labor is acknowledged and compensated. Berardi’s response is that the distinction between the crystallized general intellect (the training data) and the extractive apparatus (the architecture and training) does not diminish the claim of the former’s contributors; it simply identifies two separate levels of extraction. Politically, critics argue that Berardi’s framework offers no actionable remedy: if the human mind will not resist the capture, and if every reform is absorbed as a maintenance program for extraction, what is the political project? Berardi’s own answer has evolved from the political organization of the cognitariat (in earlier work) toward a more pessimistic emphasis on cultural and aesthetic resistance—“poetry as insurrection”—which critics regard as a retreat from political economy into aestheticism. The most generative tension is with Varela’s enactive framework: Varela offers the autopoietic organism’s structural autonomy as the basis for genuine resistance to capture, a resistance grounded in biology rather than politics, which Berardi’s account of the soul at work does not adequately theorize.

In the [YOU] on AI Field Guide

Origin

Key Ideas

Debates & Critiques

Related Entries

Further Reading