跪拜 Guibai
← All articles
AIGC

The Automation-Creates-Jobs Law Is True, Except for the People It Should Save

By 狼爷 ·
Read original on juejin.cn ↗ Google Translate ↗ Alt translation

The aggregate unemployment numbers mask a generational squeeze: junior roles are the ones disappearing, seniors survive, and compensation arrives too late for anyone with a monthly mortgage. Developers making career bets need to plan for the individual timeline, not just the trend line.

Summary

The historical rule that automation boosts total employment is sound, but it operates on factories, industries, and countries, not on any single person. Its blind spot shows when demand for a product hits a ceiling, as with radiology reports; the rule then runs backward. Machine translation at 60–80 quality turns human work into editing, not a new profession, and the much-promised “new jobs” are often the old job with a different process.

Entry-level workers are paying the price. Stanford researchers found that in high-exposure occupations, employment for 22-25 year olds dropped 13% relative to other ages, while seniors kept growing. A recruiter said a team that once hired 100 juniors now files for 50 and must justify it. The missing juniors become a missing middle in five years, which MIT calls an expensive long-term mistake. Meanwhile, the US labor income share fell to 53.8% in Q3 2025, the lowest since 1947, so productivity gains aren't reaching workers.

The author's own past example of “prompt engineer” as a new role evaporated within two years as models ate the task. The real issue is that compensation is a promise on a twenty-year clock, while mortgages are monthly. Fear isn't a cognitive failure; it's arithmetic. The practical answer is to shift accountability from process to outcome and leave a record an AI can't reproduce.

Takeaways
— For every 1 percentage point of factory automation, employment rises 0.25% after two years and 0.4% after ten years.
— US labor income share dropped to 53.8% in Q3 2025, the lowest since records began in 1947.
— In high-exposure occupations, employment for 22-25 year olds fell 13% relative to other ages.
— A tech recruiter reported a team cutting junior hires from 100 to 50 and needing to justify the smaller number.
— Machine translation now scores 60–80 on a hundred-point scale, shifting human work to polishing rather than creation.
— The previously highlighted new role “prompt engineer” stopped being a distinct job within two years as models absorbed the task.
— MIT researchers described the expected shortage of middle management in five years as a costly long-term mistake.
— The Nobel-winning economist who formulated the automation rule also warned that without proper institutions, technology revolution can become a barrier to growth.
Conclusions

Fear of AI is better framed as an individual accounting problem than a macroeconomic one; the distinction between process and outcome responsibility gives developers a practical handle on AI-era job security.

The labor income share data undercuts the “rising tide” argument: productivity gains are not automatically distributed to labor, so total employment growth tells you little about your own paycheck.

The junior-role squeeze creates a pipeline risk for the whole industry, which companies ignore because the cost is postponed and falls on someone else's headcount.

Concepts & terms
Price elasticity of demand
An economics measure of how much quantity demanded changes when price changes. In this context it determines whether cheaper products from automation actually lead to more sales and hiring, or whether demand is finite, as with radiology reports.
Labor income share
The portion of national income paid as wages and salaries rather than to capital owners. The article notes it fell to 53.8% in the US in Q3 2025, the lowest since 1947, indicating that productivity gains are not flowing to workers.
High-exposure occupations
Jobs with high potential for AI automation. The Stanford study tracked employment changes for 22-25 year olds in these roles and found a relative decline of 13%, while senior workers in the same occupations were unaffected.
From the discussion
Featured comments
NARUTOne

This article is also AI-written.

狼爷

AI subsidy.

See top comments, translated →
Source: juejin.cn ↗ Google Translate ↗ Backup ↗