In the 1960s and 1970s, Raymond Williams advanced an argument that orthodox Marxism found puzzling and liberal economics found irrelevant: that the means of communication are themselves means of production. Not metaphorically—Williams insisted on the material reality. Communication technologies require physical infrastructure (presses, transmitters, fiber-optic cables, server farms), organized labor (writers, editors, engineers, content moderators), and capital investment. They produce commodities that circulate in markets. And, most importantly, they shape the conditions under which all other production occurs—determining what can be known, coordinated, collectively decided, and imagined as possible. The argument rejected both the Marxist tendency to treat communication as superstructure (mere reflection of the economic base) and the liberal tendency to treat it as neutral infrastructure. Communication technologies are productive forces with their own material requirements and political consequences. The large language model makes Williams's claim undeniable: a system that produces working code from natural-language description is simultaneously a means of communication (it processes language) and a means of production (it manufactures artifacts). The boundary has collapsed.
Williams developed the framework through sustained engagement with broadcasting history. In Communications (1962) and Television: Technology and Cultural Form (1974), he traced how radio and television—technologies with genuine democratic potential—were progressively captured by commercial interests and organized according to market logic rather than public need. The pattern was structural: new communication technologies arrive carrying democratic possibilities (broader access, distributed production, weakened gatekeeping), but the technologies are developed and deployed within capitalist social relations, and those relations shape how the technologies' possibilities are realized. The democratic potential was real. The commercial realization was not inevitable—it was the product of specific political decisions about ownership, funding, and regulation. Williams's point was that different decisions would have produced different outcomes.
The framework's relevance to AI is direct. Claude Code, GPT, and their competitors are means of communication—they process natural language, engage in dialogue, produce text that carries meaning. They are also means of production—they generate functional code, working prototypes, deployable systems. The inseparability is unprecedented. When a builder describes a problem in natural language and receives working software in response, the act of communication is the act of production. Williams's argument ceases to be theoretical and becomes literal description of the technology's operation. The political consequences follow: if the means of communication are means of production, then questions about who owns the models, who controls access, and who captures the value are identical to questions about ownership of productive forces—the questions that have driven two centuries of political contestation.
Williams identified three modes of organizing communication technologies: authoritarian (state-controlled, serving state power), commercial (market-controlled, serving profit), and democratic (publicly governed, accountable to affected communities). The AI transition is being organized according to the commercial mode. The frontier models are developed by private corporations (Anthropic, OpenAI, Google), funded by private capital, distributed through subscription models. Democratic alternatives—open-source AI, publicly funded research, community governance—exist but operate at resource disadvantages. Williams argued that this commercial dominance is neither natural nor inevitable; it is the product of political decisions (or political failures), and different decisions remain possible.
The framework also clarifies why access is not sufficient. Even if every person on Earth could afford a Claude subscription, the fundamental question of control would remain unresolved. The technology's capabilities, its limitations, its embedded values, its governance—all determined by the corporations that own the models. Democratic access to a commercially controlled technology is not democratic control of the technology. Williams distinguished these rigorously in his broadcasting analysis, and the distinction applies with equal force to AI. The developer in Lagos may access capabilities similar to the Google engineer's, but she does not participate in determining what those capabilities are, how they are governed, or whether they serve her community's needs. The access is real. The power asymmetry persists.
First articulated in Communications (1962) and developed systematically in Television: Technology and Cultural Form (1974) and Marxism and Literature (1977). The argument built on Marx's analysis of productive forces but extended it into a domain Marx had treated as secondary—demonstrating that in modern societies, the production of meaning is as materially consequential as the production of goods.
Communication technologies are productive forces. They require material infrastructure, organized labor, and capital—they produce, not merely transmit.
They shape all other production. By determining what can be known, coordinated, and collectively decided, communication technologies structure the conditions under which economic and political life unfolds.
Ownership is a political question. Who controls communication technologies determines whose interests they serve—and the question is as politically consequential as ownership of factories.
Three modes: authoritarian, commercial, democratic. Every communication technology can be organized according to state control, market logic, or public accountability—the mode is chosen, not determined.
Democratic alternatives require political struggle. Commercial capture is the default; democratic organization requires continuous effort to build and maintain.