Workers in China worry over being replaced as they adapt to the growing impact of AI on jobs

TEXT START: HONG KONG (AP) — Computer programmer Fei Zhaojun's boss asked him if artificial intelligence could soon replace humans in coding jobs.

The Dissection

This is an early-warning report on the first stages of the Discontinuity Thesis: wage compression, role elimination, and growing fear of employment loss. It records AI moving through cognitive work, media production, translation, logistics, and physical service tasks while China’s government accelerates diffusion.

The article also packages the rupture as an adaptation problem. Workers are told to experiment with AI, launch independent ventures, or treat AI as a tool like Word. The demographic argument—that a shrinking workforce may absorb displaced labor—serves as the article’s principal escape hatch. The evidence is stronger than the framing: AI is already reducing the amount of human labor required to produce economically usable output, while the replacement jobs are not appearing at comparable scale.

The Core Fallacy

The central error is confusing productivity gains with continued human economic necessity.

A shrinking workforce does not preserve mass employment if AI eliminates the tasks those workers would have performed. It merely reduces the number of people competing for an economy that requires fewer human inputs. Retirements can delay visible unemployment, but they do not restore the wage-to-consumption circuit. The gains accrue first to firms and owners controlling models, data, robotics, energy, and distribution.

The “AI is only a tool” defense is equally weak. If one worker using AI can produce the output of ten, the tool has already displaced nine labor inputs. Human selection, supervision, and judgment may remain valuable, but that creates a thinner layer of high-leverage operators—not a preserved mass labor market. AI’s occasional errors are not a durable defense when employers can accept cheaper, imperfect output or use a small number of humans to verify it.

Hidden Assumptions

• AI-enabled businesses will create enough jobs to replace the roles they destroy.
• Productivity gains will flow into wages and household consumption rather than ownership income.
• Human oversight will remain necessary in roughly the same quantities as the automated work.
• AI’s current quality limitations will persist and matter enough to protect human labor.
• Individual experimentation and reskilling can turn displaced workers into owners or indispensable specialists.
• An aging population will generate sufficient new work instead of merely shrinking the consumer and tax base.
• Official unemployment rates capture wage cuts, underemployment, labor-force exit, and discouraged workers.
• Government can diffuse AI rapidly while maintaining a stable human employment floor.
• Human-made creative work will retain a price premium even when automated alternatives are adequate.

Social Function

Classification: partial truth functioning as transition management and ideological anesthetic.

The article accurately documents the mechanism’s symptoms: mid-level coding becoming replaceable, translation pay falling by more than half, entire creative departments being cut, and consumers spending less because employment feels insecure. But it relocates responsibility from the system to the individual. “Adapt” becomes the ritual word used while the productive role of the majority is being withdrawn.

Its demographic framing gives policymakers a socially acceptable delay narrative: automation is presented as a solution to labor scarcity rather than as an accelerator of labor devaluation. The result is a polished account of systemic replacement that still pretends the labor market is the institution doing the replacing and reabsorbing.

The Verdict

China is not disproving the Discontinuity Thesis; it is demonstrating it early and at scale. The article shows P1 already operating, P2 advancing through state-backed diffusion, and P3 emerging through wage collapse, layoffs, and employment anxiety. Aging may postpone the unemployment statistics, but it cannot resurrect the mass employment–wage–consumption circuit once AI makes human labor economically optional.