The General Intellect

In the AI Story

The general intellect does not exist in any individual mind. It lives in languages, cultural traditions, educational institutions, communicative practices, and the accumulated output of generations. Every individual draws on it when she thinks, creates, communicates, or solves problems. No one invents a language alone. No one creates a cultural tradition from nothing. The general intellect is, by its nature, common — a shared resource produced by collective human activity across centuries.

Semiocapitalism's distinctive achievement is the enclosure of this commons. The mechanisms of enclosure are various: intellectual property law converting shared cultural production into ownable assets; platform monopolies capturing the communicative commons and charging rent; educational debt forcing individuals to pay for access to knowledge their own society produced. But the most complete capture in history is the large language model — the training corpus of which comprises a substantial fraction of every text ever published on the internet, extracted without consent, embedded in proprietary systems, rented back to users.

The double capture matters. First, the creative output of billions of individuals is harvested as free raw material. The forum poster, the novelist, the programmer — none were asked, paid, or credited. Second, the model that embeds the captured general intellect is used to amplify individual builders' creative capacity, but only through a corporate product that charges for access. The builder who uses AI draws on the collective creative capacity of the species, mediated by a tool whose owner has converted the commons into a revenue stream.

The irony Berardi identifies is precise. The builder experiences AI amplification as enhancement of her own creativity. The experience is real — her individual contribution (vision, judgment, taste) remains hers. But the amplification is not coming from the machine in any meaningful sense. It is coming from the general intellect, from billions of minds whose output trained the model. The machine is a mediator. The corporation is a gatekeeper. The builder rents access to the collective creative capacity of her own species.

Origin

Marx's Grundrisse (1857–58), specifically the section known as the Fragment on Machines, first published in full in 1939 and translated to wider audiences in the 1970s. The concept was reclaimed by Italian Autonomist theorists — Paolo Virno, Antonio Negri, Maurizio Lazzarato — in the 1980s and 1990s to theorize post-Fordist cognitive labor.

Berardi's contribution is the application of the framework to the specific dynamics of semiocapitalism and, in his later work, to the unprecedented capture of collective cognitive capacity represented by large language models.

Key Ideas

Collective, not individual. The general intellect exists in the networks and traditions of the species, not in any single mind.

Commons by nature. Produced by all, belonging to no individual — the structural condition that enclosure must overcome.

Smoothing through capture. When embedded in machines, the general intellect is filtered through training objectives that optimize coherence and helpfulness while eroding disruption and contradiction.

Double extraction. Training data extracted without consent; amplification rented back to the populations that produced it.

The question of ownership. Who owns the aggregate creative output of the species? Current legal frameworks are not equipped to address the question.

Debates & Critiques

The question of whether AI training constitutes legitimate use of publicly available data or theft of the commons is being litigated (the Andersen v. Stability AI case, the Authors Guild letter) even as the commercial deployment of the captured intellect continues to accelerate. Berardi's framework provides an analytical vocabulary for these disputes but does not resolve them.