AI will transform capitalism – but how? | AI (artificial intelligence) - The Guardian

TEXT START: The idea that autonomous, thinking machines may one day destroy property, hierarchy and inequality is as old as western political thought.

The Dissection

This is a left-reformist attempt to seize the meaning of an AI-induced crisis before capital does. It correctly recognizes that AI may destroy not only jobs but also entrepreneurial and innovative functions. It proposes public provisioning, cooperative ownership, AI-directed allocation, and a free “un-ownable” model as the escape route.

The central weakness is that it treats technological direction as a choice available to “humanity” rather than a prize contested by whoever controls compute, energy, chips, data, networks, finance, and states. The text identifies the approaching corpse accurately, then assumes the mourners can redesign the organism.

The Core Fallacy

The article confuses a technically available configuration with an economically selectable equilibrium. Open source, free access, and non-profit ownership do not abolish the material ownership stack. Someone still controls compute, electricity, fabrication, cloud distribution, model updates, security, logistics, and enforcement. “Un-ownable AI” can be owned at every layer that makes it usable.

It also confuses consumption preservation with human economic agency. Basic incomes and services can keep demand alive, but under DT they do not restore productive participation. They replace the wage-consumption circuit with administered distribution.

The article sees P1 and much of P3: AI may outperform human cognitive labor and eliminate economically necessary roles. It largely evades P2. Existing institutions cannot coordinate a stable, globally humanist AI commons while states and firms compete for strategic dominance. The article’s “we” has no command center, enforcement mechanism, or answer to defection.

Hidden Assumptions

• AI development is a neutral path that can be redirected after frontier capabilities are captured.
• Open source diffuses power rather than merely diffusing software while compute and infrastructure remain concentrated.
• A free model can outcompete subscription and cloud monopolies without explaining financing, energy, maintenance, security, and distribution.
• “Universal values” can be encoded without deciding who defines them, updates them, and enforces them.
• States and firms will surrender rents and strategic advantage because a cooperative alternative is more rational.
• Weakening patents severs the innovation-profit link rather than shifting rents to infrastructure owners, deployment monopolies, and sovereign power.
• Replacing markets with AI planning solves political allocation. It merely moves the choke point to whoever controls the planner, its data, and its enforcement apparatus.
• “Humanity” is a coherent actor. It is not. The transition is a struggle among owners, states, servitors, displaced workers, and opportunists.
• Redistribution can compensate for the loss of work’s necessity, status, bargaining power, and control.

Social Function

Primary classification: partial truth wrapped in ideological anesthetic and transition management.

The article provides a useful diagnosis for the political left: it admits AI may destroy jobs, entrepreneurs, and the wage-to-consumption circuit instead of repeating the dead promise of universal reskilling. That is the partial truth.

Its anesthetic is the word “we.” The article turns a power struggle into a design workshop, allowing readers to imagine that a benevolent collective can choose AI’s social purpose before infrastructure owners and states lock in the outcome. Its UBI, universal services, cooperatives, and planning proposals are transition-management tools: mechanisms for preventing unrest and maintaining demand after productive participation collapses. They may stabilize the carcass, but they do not create a classless system.

It is also prestige signaling. Marx, Aristotle, “general intellect,” and universalist language provide historical grandeur for a proposal whose enforcement mechanics are absent.

The Verdict

The article correctly identifies the approaching crisis: AI can sever the labor, innovation, and profit links that allowed post-WWII capitalism to reproduce itself. It fails at the decisive step by assuming that recognizing the crisis creates collective agency to shape the replacement.

Under DT logic, the likely result is not an automatically emancipatory commons but a struggle over AI’s physical and institutional bottlenecks. Open models may weaken software monopolies while leaving energy, compute, logistics, maintenance, capital, and state power intact. UBI and public services can preserve consumption while making the population less economically necessary. That is replacement, not liberation.

The essay is a lucid premonition wrapped in a power fantasy: it sees the old machine losing its workers, but imagines that the abandoned passengers can redesign the engine while its owners are still driving.