AI is causing Chinese office workers to do the work of three people.

TEXT START: Wu Qiong, a data analyst in China , was once tasked with automating repetitive reports using artificial intelligence (AI).

THE DISSECTION

The article documents labor compression in real time. AI lets one worker absorb the output of several, eliminates departments, bundles formerly separate occupations, shortens deadlines, and weakens pay. It also performs a containment maneuver: it frames structural displacement as an adaptation problem caused partly by economic slowdown and solvable through retraining.

THE CORE FALLACY

The central error is treating a shift toward AI skills as evidence that mass employment can be preserved. Rising demand for AI engineers and product managers does not demonstrate that enough jobs are being created to replace the editorial, service, design, and analyst roles being destroyed. The article measures changing skill demand, not net productive participation.

Under the Discontinuity Thesis, this is the crucial distinction. AI does not merely move workers into better jobs; it increases the output expected from each remaining worker while reducing the number required. “Do the work of three people” is not a productivity dividend when ownership remains concentrated. It is a headcount deletion followed by a wage reset.

HIDDEN ASSUMPTIONS

• Retraining will occur faster than AI eliminates existing roles.
• New AI jobs will be numerous enough to absorb displaced workers.
• Workers can acquire scarce technical skills rather than merely add more commoditized tasks to their workload.
• Productivity gains will be shared through wages instead of captured by owners.
• Employers will preserve employment rather than use AI to remove bargaining power.
• Policy assessment, subsidies, and retraining can restore productive participation rather than merely manage displaced consumption.
• The recruitment figures represent the whole labor market and measure job quality, permanence, and total volume. The article supplies no such proof.

SOCIAL FUNCTION

Partial truth wrapped in transition management and ideological anesthetic. The article is too grounded to deny displacement, but its conclusion redirects attention toward adaptability, multi-skilling, and policy support. That makes a structural collapse appear to be a difficult but manageable transition. The “skills advantage” it advertises is itself unstable: once AI makes a capability widely accessible, that capability becomes another employer expectation and another reason to reduce headcount.

THE VERDICT

This is evidence for Cognitive Automation Dominance and Productive Participation Collapse, though not by itself proof of Coordination Impossibility. Its own facts destroy its soft conclusion: workers produce more, earn less, lose departments, and compete for fewer roles while a narrow AI-control class expands. The system is not replacing every worker with an AI job. It is converting many workers into cheaper, broader servitors and discarding the surplus outright. Retraining may create individual survivors; it does not repair the mass employment–wage–consumption circuit.