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

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

The Dissection

This article documents the first visible phase of the Discontinuity Thesis while packaging it as an adaptation story. Its evidence shows rapid AI diffusion, direct layoffs, wage collapse, declining demand for training, and automation entering physical service work. It then softens the mechanism with aging demographics, human judgment, AI errors, and independent ventures.

The underlying process is simple: AI expands output, firms cut headcount, and remaining workers are forced to use the same tools to defend declining pay. “Join them” is not a solution. It is a labor-market arms race in which universal access destroys the scarcity that once supported wages.

The Core Fallacy

The article confuses demographic labor scarcity with economically necessary human labor. A shrinking workforce may reduce the number of available workers while leaving millions whose tasks have been automated redundant or economically peripheral. That is not a contradiction; it is the transition.

It also confuses productivity with mass prosperity. Higher output benefits the owners and controllers of the systems unless ownership and purchasing power are redistributed. If every competitor has AI, it becomes a baseline requirement, not a moat. Human selection, editing, teaching, and supervision may survive, but the number of people paid to perform them can still collapse.

The supplied evidence sketches P1 and the opening of P3: coding, translation, scripts, industrial work, and delivery are being compressed; firms are shedding staff; and translation pay has reportedly fallen by more than half. China’s aggressive “AI Plus” diffusion accelerates P2 by making human-only economic enclaves harder to preserve.

Hidden Assumptions

• Displaced workers can create enough independent businesses to replace lost wages.
• AI-generated work will still require roughly the same number of human decision-makers.
• Demographic decline will absorb displaced labor rather than reduce the consuming population.
• Productivity gains will flow to workers instead of owners.
• Vlogging and small studios can scale beyond a few visible survivors.
• Unemployment rates capture underemployment, discouraged workers, wage collapse, and involuntary exit.
• AI errors will remain frequent enough to preserve human labor demand.
• Government promotion of adoption will create compensating jobs rather than accelerate substitution.

The text does not establish these assumptions. Several are contradicted by its own evidence: AI use among Chinese industrial enterprises rose from 9.6% to 47.5%; translation pay fell by more than half; and the number of live-action short and vertical video series reportedly fell about 75% year over year.

Social Function

Partial truth disguised as transition management, with a strong ideological-anesthetic component.

The article accurately records the mechanism’s early symptoms. Its framing then converts systemic displacement into personal adaptation: learn the tool, launch a venture, remain the human decision-maker, or wait for demographics to repair the labor market. The individual stories are survivorship exhibits. They show who is experimenting, not whether the system can provide economically necessary roles for the displaced majority.

The aging-population argument is a lullaby. It treats fewer workers per retiree as evidence that automation may be helpful while ignoring who owns the automated capacity and who retains purchasing power. A society can have labor shortages in selected essential functions and mass productive exclusion elsewhere.

The Verdict

This is an early-collapse report wearing the costume of an adaptation article. It confirms the Discontinuity Thesis more than it refutes it: cognitive automation is spreading, wages are being compressed, firms are reducing headcount, and government policy is increasing diffusion speed. Human ingenuity and demographic decline may create temporary niches, but they do not preserve the post-WWII employment–wage–consumption circuit.

The article’s “if you can’t beat them, join them” prescription is surrender with a productivity app attached. The decisive question is not whether people can use AI. It is whether they own or control AI capital, or remain indispensable to those who do. The supplied cases mostly describe servitors trying to avoid becoming surplus. That is transition management, not system survival.