The Bosses' Oldest Dream - Jacobin

TEXT START: The first computers were designed with labor discipline in mind.

The Dissection

The article performs a historical continuity argument: computing began as a technology for measuring, disciplining, and replacing labor, and AI is the culmination of that project. It strips away the productivity euphemism and identifies the real managerial attraction—lower costs, weaker bargaining power, surveillance, and eventually labor elimination.

It also performs a political maneuver. After presenting evidence of employment damage and extreme wealth concentration, it offers collective action—especially the WGA strike—as the available counterforce. The article is strongest as an indictment of AI-enabled labor discipline and weakest when it converts that indictment into a strategy for preserving labor’s economic centrality.

The Core Fallacy

The article treats automation as a political choice that organized workers might sufficiently constrain. Under the Discontinuity Thesis, it is a competitive system outcome.

If AI can perform economically useful cognitive work at lower cost, firms that refuse to deploy it will eventually be punished by competitors that do. Collective bargaining can delay deployment, regulate specific employers, and extract concessions. It cannot permanently preserve mass human employment against superior machine labor—especially when the AI companies themselves are outside the workers’ bargaining relationship.

The article also mistakes weak present evidence for a serious challenge to the terminal thesis. Hiring freezes, avoidance of entry-level workers, and productivity gains without equivalent job creation are lag indicators. They are the early architecture of displacement, not proof that displacement has failed. The absence of mass layoffs can mean firms are choosing not to hire replacements, not that labor remains secure.

Most importantly, the article identifies the possibility that machines may become labor, then continues reasoning inside the old automation framework. Earlier machines replaced tasks while leaving humans with new economically necessary tasks. General AI attacks the supply of economically necessary human labor itself. That is not another productivity cycle. It is the severing of the wage-to-consumption circuit.

Hidden Assumptions

• Human workers can sustain durable coordination against globally mobile capital and AI deployment. This conflicts with the Coordination Impossibility condition.
• Regulation and collective bargaining can bind AI firms that are not the workers’ employers and prevent capital from arbitraging around restrictions.
• Tacit judgment, social interaction, and hard-to-verify outcomes are permanent human moats rather than temporary lag defenses.
• Rising GDP will remain socially useful to workers even when wages and employment collapse. Output abundance does not create bargaining power for people who no longer control productive assets.
• The WGA’s success generalizes to less organized, less scarce, and more globally substitutable workers. It does not.
• Workers can turn AI into a tool they use rather than a system that replaces them without owning or controlling the AI capital. Without ownership, productivity gains accrue primarily to the Sovereign class.
• Political institutions can redistribute AI-generated wealth at sufficient scale and permanence to preserve productive participation. Transfers may preserve consumption; they do not restore economic necessity.

Social Function

This is a partial truth wrapped in transition-management rhetoric and labor-mobilization propaganda.

It is not simple copium. The historical account of automation as labor discipline is materially useful, and the article correctly exposes the class function of surveillance and substitution. But its closing prescription risks becoming a labor-movement lullaby: collective action is presented as if it can preserve the old order, when it can mainly purchase delay, compensation, and protected niches.

The WGA example is a bargaining success, not a refutation of obsolescence. It shows that organized workers can regulate an employer during a transition. It does not show that workers can retain sovereignty once the productive system no longer needs them.

The Verdict

The article correctly diagnoses the weapon and misidentifies the battlefield. AI is not merely another boss’s tool for squeezing labor; it is the mechanism by which labor can cease to be economically necessary. Collective bargaining can slow the execution and improve terms for selected groups. It cannot reverse P1, overcome P2, or prevent P3.

The article sees the guillotine, documents who built it, and then recommends negotiating over the blade’s speed. Under the Discontinuity Thesis, survival depends on ownership or control of AI capital—or indispensable access to energy, logistics, maintenance, verification, and transition bottlenecks. Everything else is managed decline.