Subscribe
AI and jobs
3 min read

AI-Exposed Jobs Are Losing China's Hiring Race, 10-Million-Ad Analysis Shows

Bloomberg scored nearly 10 million Chinese job postings against a Peking University AI-exposure index. Roles deemed vulnerable fell about 10 percentage points faster than the rest between 2021 and 2025.

Mid-century poster, headline JOBS LOST TO AI over an isometric rooftop scene with a floating staircase
Mid-century poster, headline JOBS LOST TO AI over an isometric rooftop scene with a floating staircase

China's labor market has never produced clean data, so Bloomberg built its own. In an analysis published Friday, the outlet scored nearly 10 million online job postings from July 2021 to June 2025 against an AI-exposure index developed by academics at Peking University. The finding: advertisements for roles with greater AI exposure fell about 10 percentage points more than postings for less-exposed positions.

A worker dwarfed by an automation console

The scale is what makes the numbers matter. Bloomberg Economics estimates roughly 70 million Chinese workers face disruption from the technology, against about 45 million in the US, a country with a population less than half the size. Beijing publishes only limited official jobs statistics, which is precisely why an ad-level dataset drew attention.

Three studies, one direction

The Bloomberg analysis lines up with the small body of academic work on the question:

  • Chen Qin, a data scientist and visiting scholar at ShanghaiTech University who led a review of 116 million online job ads, estimated AI caused a 5.2% net loss in new hiring between 2023 and early 2026. The damage concentrated in roles requiring one to three years of experience, the classic entry-level band. Nearly a fifth of postings showed a significant change in listed responsibilities over the period.
  • A separate study involving researchers from institutions including the Chinese Academy of Social Sciences, which reports into the State Council, estimated AI led to a 21% decline in the average number of online job postings and a 4% drop in average salary.

The analysis has honest caveats, which Bloomberg states plainly. Parts of the divergence in hiring rates predate widespread AI adoption, so a decelerating macroeconomy may simply be hitting white-collar sectors harder. And an online posting pool overweights office work while excluding most informal and blue-collar roles.

An economy already absorbing a shock

The AI signal arrives into a labor market that is already strained. China created just 21 million net new jobs in the five years to 2024, according to World Bank figures, less than half the total of the prior five-year stretch. Covid-era expansions of college admissions produced graduating cohorts that Gavekal Dragonomics, a research consultancy, has called a "lost generation" of more than 70 million recent graduates. A Mycos survey found nearly 11% of last year's graduates were still jobless six months out, almost three times the share a decade earlier. Youth unemployment reached close to 19% this summer, and the gig economy now holds more than 200 million workers, roughly 40% of the urban workforce, in insecure positions.

Layoff stories add color, though attribution stays contested. Xiaomi, Meituan and Alibaba's travel unit Fliggy all denied local media reports of mass layoffs this year. Several workers at top tech firms described rounds of cuts beginning in spring that reached as much as 30% of employees in some departments. A Beijing programmer, fired in June along with half his eight-person team, said his employer merged programming, design and testing into a single role on the premise that AI lets one person do the work of as many as six. "The layoff this time is different because it's a reorganization driven by AI," he said.

What Beijing is doing

The topic is politically sensitive, and companies have largely refused to link job cuts to AI. Behind the scenes, though, officials are moving. In June, China's cabinet announced an unprecedented nationwide survey to track AI's impact on jobs over five years. Last month it called for "prudent management of the pace of AI adoption in labor intensive industries" to soften the blow to workers. Most concrete measures so far target training: roughly a third of university degree programs were overhauled toward AI, robotics and advanced computing, and local governments are subsidizing certificates in areas of labor shortage, including AI model training.

Cai Fang, a former adviser to China's central bank, called the current lull "a temporary period of calm" and said policymakers have a window to prepare.

The most visible disruption so far is in content production. More than 90% of the 430,000 microdramas launched in China in the first eight months of 2026 were AI-generated, according to official data, after tools like ByteDance's Seedance 2.0 collapsed the cost of producing the 90-second vertical episodes. At one Hangzhou filming venue that opened in late 2023 with government backing and ran hundreds of workers at peak, crews are now nearly gone.

Demand for retraining, meanwhile, is booming. The Zhejiang Digital Technology Talent Employment Training Base in Hangzhou expects 8,000 students this year, up from 5,000 last year, and moved into a new building in March to fit them.

Sources

AI and jobsChinaResearch