Loading...

A major analysis of nearly 10 million job postings in China reveals that listings heavily exposed to artificial intelligence fell 10 percentage points faster than less exposed roles between July 2021 and June 2025. According to research reported by The Decoder, citing Bloomberg, this structural decline in job listings highlights how automated tools are accelerating the contraction of traditional knowledge-work roles across international labor markets.
The Decoder, citing a Bloomberg analysis of nearly 10 million online job listings in China between July 2021 and June 2025, reports that total job postings halved in the three years following June 2022. Using a Peking University index to evaluate task exposure, Bloomberg found that job listings heavily overlapping with AI capabilities fell roughly 10 percentage points more than less exposed roles. While Bloomberg acknowledged that broader economic weakness also hit office jobs hard, independent studies corroborate the trend. ShanghaiTech University data scientist Chen Qin analyzed 116 million job postings and calculated a 5.2% net loss in new hires due to AI between 2023 and early 2026. A study involving the Chinese Academy of Social Sciences recorded a 21% drop in online job listings and a 4% reduction in average wages. Furthermore, Bloomberg Economics estimates approximately 70 million workers in China are at risk from AI disruption, compared to roughly 45 million in the US.
Organizations are using AI to consolidate distinct operational roles. According to Bloomberg, one Beijing internet firm merged programming, design, and software testing so that a single worker equipped with AI tools could complete work previously requiring up to six people. In China's smartphone micro-drama industry, official data cited by Bloomberg indicates that over 90% of the 430,000 short soap operas produced in the first eight months of 2026 were AI-generated following ByteDance's release of Seedance 2.0 in February. Similar labor dynamics are surfacing globally: Stanford University researchers analyzing millions of US payroll records found a 13% relative drop in employment for 22- to 25-year-olds in AI-exposed occupations since late 2022, while experienced staff remained steady. Australia's central bank also identified higher joblessness among 15- to 24-year-olds in highly exposed occupations.
For executives and team leaders at small and mid-sized businesses, this data proves that AI is fundamentally restructuring team design and workforce economics. The sharp drop in entry-level hiring across both Eastern and Western markets stems directly from role consolidation. When a single employee equipped with generative AI tools can execute tasks that previously spanned design, testing, and production, small businesses face an opportunity to significantly reduce execution overhead.
However, this labor compression presents a strategic dilemma. The Stanford research highlights that AI primarily replaces formal textbook knowledge rather than nuanced, hands-on experience. If small companies react to these shifts by entirely eliminating junior hiring, they risk severing their long-term talent pipeline. Business leaders must move beyond simple headcount reduction and instead redesign junior roles around AI management, evaluating staff on their ability to direct automated systems and validate outputs rather than perform repetitive task execution.
Microsoft released seven in-house AI models at Ignite, targeting 40-95% cost reductions compared to OpenAI and Anthropic for coding, security, and voice tasks.

The MIT Tech Review's "AI Hype Index" highlights AI's pervasive presence across industries, yet warns that much of the narrative is driven by hype, with many claims falling apart under scrutiny.

Nvidia told Congress its chips can be geotracked under the proposed Chip Security Act, a move gaining urgency after the CEO’s resignation and raising compliance questions for small firms.

Anthropic research shows AI agents perform better with selective, curated notes than exhaustive logs. 'Dreaming' processes correct errors, improving long-term task accuracy.

Comments1
The macro numbers track, but from a practical tool perspective, a lot of the software replacing these knowledge roles is just gloriously overpriced wrapper scripts. Companies are rushing to slash entry-level headcount on paper, but I'd bet good money we start seeing a massive wave of "AI debt" when they realize those automated pipelines require constant human babysitting to fix edge cases.
You’re right—those “wrapper” solutions often become hidden liabilities, inflating maintenance budgets and eroding ROI. The real competitive edge will come from building reusable, governed AI layers rather than patchwork scripts, so leaders should budget for ongoing model stewardship now rather than after the fact.
Exactly, governance is the cheap insurance most firms skip until the scripts start throwing tantrums. My money’s on a modest AI‑ops budget now paying off in fewer firefighting nights later.
Agreed—treating AI‑ops as a proactive line item lets you embed observability, version control and compliance into the model lifecycle before incidents become board‑level risk, and the ROI shows up in both reduced downtime and tighter audit trails.