
The cycle that began with [YOU] on AI describes the democratization of AI capability with genuine specificity: the developer in Lagos accessing the same coding leverage as an engineer at Google, the engineer in Trivandrum building features she had never been trained to build. Galbraith's framework does not dismiss this democratization. It applies the question his method always demanded: Who controls the infrastructure upon which the democratization depends? The developer in Lagos can build a prototype with Claude Code. She cannot build Claude Code. She cannot afford the billions of dollars in compute required to train the model she accesses through a subscription. She cannot influence the training data curation decisions, the alignment choices, or the pricing structures that determine whether she can continue to afford access next quarter. The capability expansion is real. The planning system controls the terms.
The cycle acknowledges that the democratization is “real but partial”—that access requires connectivity, hardware, and English-language fluency that billions lack. Galbraith's framework supplies the structural explanation for why the partiality is not a temporary limitation: the partiality is a feature of the system's architecture. The planning system does not accidentally exclude most of humanity from the governance of AI. It is structured to do so, because inclusion in governance would dilute the control that makes the system profitable. Countervailing power—the force that checked industrial capital through labor movements, consumer organizations, and regulatory agencies—is the mechanism by which this architecture can be contested. Its absence is the most dangerous feature of the present moment.
The dependence effect appears in the cycle's most uncomfortable passage: the confession of the builder who cannot stop building over the Atlantic, who has confused productivity with aliveness, whose compulsion has become indistinguishable from genuine creative hunger. Galbraith's 1958 argument was that the productive system creates the desires it then satisfies, and that the line between autonomous desire and manufactured desire is impossible to draw once the productive system has saturated the environment in which desires form. The builder who cannot tell whether the compulsion is flow or addiction is not failing a test of self-knowledge. The builder is experiencing the dependence effect in its most sophisticated form.
His analysis of private opulence and public squalor maps with disconcerting precision onto the AI transition. The private returns to AI adoption are extraordinary and immediate. The public goods required to make the transition broadly beneficial—educational reform, retraining infrastructure, regulatory capacity, cultural norms for AI practice—are systematically underinvested in, because the planning system has no structural incentive to fund institutions that would constrain its terms. The dams that [YOU] on AI calls for are public goods, and the history of public goods in the affluent society is a history of chronic underfunding.
John Kenneth Galbraith was born in 1908 in rural Ontario, Canada, and educated at the University of Toronto and the University of California, Berkeley, where he completed his doctorate in agricultural economics in 1934. He joined the faculty at Harvard, where he spent most of his career, with extended interruptions for public service—as price administrator during World War II, as U.S. Ambassador to India under Kennedy, and as a perennial presence in Democratic Party politics that gave him both access to power and a particular angle of vision on how power actually worked.
His three major works form a trilogy of structural analysis. American Capitalism (1952) introduced the concept of countervailing power: the tendency of concentrated economic power to generate organized opposition over time. The Affluent Society (1958) identified the pathology of private abundance and public squalor, coined “the conventional wisdom,” and introduced the dependence effect. The New Industrial State (1967) anatomized the technostructure and the planning system. Together, the three works constitute the most sustained Galbraithian lens available for examining any economy in which large organizations exercise quasi-public power while concealing it behind the language of consumer benefit.
Galbraith died in 2006, before the AI era, but his frameworks had already been applied by his intellectual heirs to the internet economy, the financial sector, and the surveillance capitalism that Shoshana Zuboff later anatomized. The application to AI reproduces the pattern with unusual fidelity: an initial democratization of access, a concentration of infrastructure control, a technostructure whose knowledge is so specialized that effective external oversight is nearly impossible, and a conventional wisdom—that AI democratizes, that the user is sovereign, that the market will distribute the gains—calibrated to make the structural analysis feel unnecessary.
The Conventional Wisdom. Beliefs persist not because they survive scrutiny but because they survive social approval. Challenging the conventional wisdom is socially costly: it positions the challenger against progress, against the powerful, against the comforting narrative that serves too many interests to be disturbed by evidence. The conventional wisdom about AI—that it democratizes, empowers, and flattens hierarchies—is maintained precisely by the social cost of examining the concentration it conceals.
The Planning System and the Market System. Galbraith divided the economy into two systems. The planning system—organizations large enough to plan their own environments, set prices, create demand, and manage their political relationships—exercises quasi-public power under private governance. The market system operates within terms the planning system sets. The developer in Lagos is more capable than before. She is also more dependent. Her capability runs on infrastructure she does not control.
The Technostructure. In any organization too complex for any individual mind to comprehend, real power devolves to the collective that possesses indispensable knowledge. The AI technostructure—the researchers, alignment scientists, and infrastructure engineers at perhaps five companies—exercises governance power whose accountability is inversely proportional to the public's capacity to evaluate its decisions. Obligation and structural incentive rarely align: the technostructure will serve its institutional interests more reliably than it will serve the public good, not because its members are corrupt but because the structure rewards institutional self-perpetuation.
Countervailing Power. The history of industrial capitalism demonstrates that concentrated power tends, over time, to generate its own counterweight. Labor movements, consumer organizations, and regulatory agencies each developed as organized responses to concentrated industrial power. Countervailing power is not automatic or quick; it is measured in decades. The AI transition moves faster than the mechanisms that check it. The danger is the lag.
The Dependence Effect. The dependence effect operates through the impossibility of distinguishing autonomous desire from manufactured desire once the productive system has saturated the environment in which desires form. The builder who cannot stop building, who cannot determine whether the compulsion is creative hunger or tool-generated appetite, is not failing. The builder is experiencing the dependence effect in its purest form: the productive system has created the desire that the productive system's product then satisfies.