AI Isn't Coming for Our Jobs. It's Coming for Our Wages. - Jacobin

TEXT START: Over the short term, AI may create more jobs than it destroys.

The Dissection

This essay performs a controlled retreat from the “AI jobs apocalypse.” It correctly identifies AI as a tool for deskilling, wage suppression, surveillance, and the expropriation of worker autonomy. Then it relocates the crisis from employment to wages and proposes workplace struggle plus public ownership as the remedy.

Its historical comparison to mechanization, Marx, and Braverman is useful but incomplete. The essay treats AI as another industrial technology whose effects can be redirected politically. Its real function is to convert a terminal-system question into a manageable labor-reform program: fight over who controls AI instead of confronting whether human labor remains economically necessary at all.

The Core Fallacy

It mistakes the first symptom for the kill mechanism. Wage compression and deskilling are not alternatives to job destruction; they are the opening phase of it. Firms can retain workers temporarily at lower pay while AI strips away skill, autonomy, and bargaining power. Once AI reliably performs more cognitive work at lower cost, labor becomes an expense rather than a negotiating partner.

The infrastructure boom—data centers, utilities, transport—represents expanded capital reproduction, not durable compensation for displaced labor. Those jobs are a lag defense. The essay itself notes that machinery cannot reabsorb all the labor it replaces, yet still treats current expansion as evidence that compensation may prevail.

It also treats hallucinations and conventionality as fundamental limits. Under the Discontinuity Thesis, current defects are engineering obstacles, not proof of permanent human indispensability. P1 produces cognitive automation dominance; P2 prevents institutions from preserving large human-only economic domains; P3 follows as productive participation collapses.

Hidden Assumptions

• AI-created infrastructure and downstream jobs will be permanent, numerous, and resistant to further automation.
• Current LLM limitations will remain structural rather than being reduced through better systems, tools, data, and verification.
• Worker organization can outrun capital concentration and preserve human control across competitive markets.
• Public ownership will restore productive agency rather than merely redistribute the output of automated systems.
• Job counts measure meaningful economic necessity, even when the work is degraded, subsidized, or redundant.
• Public planning can prevent elite capture and competitive pressure from forcing further automation.
• More equal access to AI will preserve the wage system, although ownership and consumption can survive without mass labor participation.

Social Function

Classification: partial truth, transition management, and ideological anesthetic.

This is not empty copium. It accurately identifies the immediate attack on wages, skill, and autonomy. But it offers labor a battlefield it can plausibly win—ownership, regulation, workplace control—while avoiding the battlefield where labor loses necessity. “AI as a public utility” could improve distribution and cushion the transition. It does not resurrect the mass employment circuit. It manages the carcass and calls that survival.

The Verdict

The headline is right about chronology and wrong about ontology. AI may hit wages before it hits jobs; that is the first wound, not the final diagnosis. Class struggle and public control can delay, redistribute, or soften the collapse, but they do not defeat P1, P2, or P3. The post-WWII order dies when labor is no longer required to produce what society consumes—even if transfers preserve consumption. This article sees the blade entering wages and mistakes that for the whole autopsy.