US vs China: Which Economy Faces More AI Job Risk?

The same peak score, very different scale

When the WorldJobsData scoring framework assigns AI exposure scores to ISCO-08 major occupation groups, the US and China land at identical peaks: 8.5/10 for clerical support workers. This is not a coincidence. Both countries' clerical categories cover the same type of work - data entry, correspondence, scheduling, classification, record management - which sits squarely within the capability of large language models deployed in 2026. The task profile is the same because the ISCO-08 category definition is the same.

What differs dramatically is scale. US clerical support workers number 16.5 million (BLS OEWS, 2024 data year, published May 2025 via ILO ILOSTAT CC BY 4.0). China's equivalent group numbers 33.6 million (National Bureau of Statistics of China via ILO ILOSTAT CC BY 4.0, 2025 data year). China's single highest-risk occupation group is larger than the entire US clerical workforce at the same exposure level. And clerical workers make up only 9.3% of China's 362.2 million tracked workers - meaning the workforce is actually more buffered than the US despite the larger absolute numbers, because the manufacturing and trades sectors dominate.

The US average AI exposure of 5.07/10 sits meaningfully above China's 4.48/10 (WorldJobsData weighted averages, from occupation employment counts and group-level AI scores). The gap reflects a structural difference: China has 93.6 million craft and trades workers (25.8% of its workforce, 2.5/10 exposure) and 50 million plant and machine operators (13.8%, 3.0/10 exposure). Those 143 million workers in manufacturing-related roles pull China's average down considerably. The US has 11.2 million trades workers and 18.8 million plant operators - a combined 30 million in equivalent low-exposure roles out of 143.1 million total.

Side-by-side comparison: major occupation groups

The table below compares both countries across the major ISCO-08 occupation categories. China has no wage data available from ILO ILOSTAT for this dataset; employment counts are from the National Bureau of Statistics of China (NBS) 2025 via ILO ILOSTAT (CC BY 4.0). US wages are from BLS OEWS 2024.

Note: China's ILO ILOSTAT dataset does not separately enumerate ISCO-08 groups 6 (skilled agricultural workers) and 9 (elementary occupations) in the 2025 release. The major reported groups cover 362.2 million workers total across groups 1-5, 7, and 8.

Why the US faces disruption sooner despite a smaller workforce

The WorldJobsData risk velocity rating for the US is imminent (1-3 years) - a 10.0/10 on the velocity scale. China rates at disruption arriving (3-7 years) at 5.2/10. This gap reflects deployment speed rather than exposure level. Both countries have the same task automation potential for clerical workers; what differs is how fast employers are converting that potential into actual headcount changes.

The US private sector has led AI deployment in enterprise software. Tools built on large language models for document processing, customer service automation, and back-office data work have been commercially available and adopted at scale since 2023-2024 (source: McKinsey Global Institute "The Economic Potential of Generative AI," June 2023, estimates 60-70% of work activities automatable in US financial services and insurance). US clerical employment in the BLS OEWS May 2025 data already shows a projected growth rate of -4.15% for clerical support workers - the only major group with a negative outlook. That negative projection is the risk velocity playing out in official labour market data before broad AI displacement has even fully arrived.

China's AI deployment timeline is longer for a different reason: state-direction shapes adoption speed. China's government has published AI development plans (the 2017 "New Generation Artificial Intelligence Development Plan" and subsequent 5-year plans) targeting AI leadership by 2030, but enterprise automation in clerical functions is proceeding more unevenly than in the US private sector. Additionally, China's manufacturing-heavy workforce means the proportion of workers in the highest-risk clerical and professional categories - though large in absolute numbers - is smaller as a share of total employment. This creates both a longer runway and a larger buffer.

The AI race context: chips, data, and state capacity

The US-China AI comparison cannot be separated from the geopolitical AI race. US export controls on advanced semiconductor chips (NVIDIA A100 and H100 restrictions enacted October 2022, extended October 2023) have constrained China's access to the highest-performance training hardware. This affects AI research and frontier model development more than it affects deployment of existing models into enterprise clerical workflows - which require inference rather than large-scale training compute. The practical effect on workforce displacement timelines is smaller than the chip-war framing suggests.

China's advantage is state-directed scale and data access. The Chinese government can mandate AI adoption across state-owned enterprises, which employ a substantial share of formal-sector workers. The NBS China data tracked by ILO ILOSTAT does not fully capture the informal economy (estimated at 30-40% of employment, World Bank 2024), meaning the 362.2 million workers in this comparison represent only the formal workforce. China's actual AI displacement exposure - when informal sector workers are included - would look different, with a higher share of agricultural and informal service workers who score lower on AI exposure.

The US advantage is private sector capital velocity. US firms - including the hyperscalers (Google, Microsoft, Amazon, Meta) that build the underlying AI infrastructure - are deploying at a speed that no government-directed program matches. HDI data from the UNDP Human Development Report 2025 (2023 data year) shows the US at 0.938 (rank 17) versus China at 0.797 (rank 78). The HDI gap reflects differences in healthcare access, educational attainment, and income levels that shape both who builds AI tools and who is most vulnerable when those tools displace jobs.

China's 93.6 million trades workers: the biggest buffer in the dataset

The single most important structural difference between the US and China AI exposure profiles is China's manufacturing workforce. Craft and related trades workers - electricians, welders, fitters, construction workers, and related trades - number 93.6 million in the 2025 NBS China data via ILO ILOSTAT. They score 2.5/10 on AI exposure. This group alone is larger than the entire US workforce tracked by BLS OEWS.

These workers are not safe from automation indefinitely - robotics risk for China's trades workers is rated 4.5/10, reflecting the ongoing industrialisation of Chinese manufacturing. But AI specifically, as distinct from robotics and process automation, is not the primary threat to this group in the 2026-2030 timeframe. The physical, site-specific, judgment-intensive nature of skilled trades work remains beyond what currently deployed AI systems can replicate in real-world conditions. A welder in Shenzhen or a construction carpenter in Chongqing faces a robotics threat (industrial welding robots, rebar-tying machines) long before they face an LLM-based AI threat.

In contrast, the US craft and trades workforce of 11.2 million workers at 2.5/10 (BLS OEWS 2024) is both far smaller in absolute terms and in shorter supply. The US construction and trades shortage (National Association of Home Builders, 2024: 400,000 unfilled trades positions) means these workers have labour market power that their Chinese counterparts, in a much larger supply pool, do not. The economic dynamics of displacement differ even when the AI exposure score is identical.

What this means for workers in both countries

For US workers in clerical and administrative roles, the data points to a clear and near-term risk. BLS OEWS already projects negative growth (-4.15%) for the clerical category, and that projection precedes the full rollout of enterprise AI tools in 2025-2026. US clerical workers score 8.5/10 on AI exposure at a median annual wage of $45,432 (BLS OEWS 2024). The combination of high exposure, negative growth outlook, and low median wage means these workers have limited financial buffer for a job transition and face a skills gap that retraining programs have not yet closed at scale.

For Chinese clerical workers, the same exposure score applies but the timeline is longer. The 33.6 million workers in China's clerical group face the same task automation potential, but enterprise AI adoption in Chinese workplaces - particularly in state-owned enterprises and smaller private firms outside the tech sector - is proceeding more slowly. The practical result is a 3-5 year window rather than 1-3 years. That window matters: it means Chinese clerical workers have somewhat more time to prepare, but it does not change the destination. The 3.7% annual growth rate in China's professional class (NBS data 2025, ISCO-08 group 2, growing from 79.5M in 2023 to 81.8M in 2025) suggests the Chinese economy is moving up the value chain, which increases the share of AI-exposed knowledge work over time.

Economy context helps frame who absorbs the cost of displacement. World Bank data (2025) shows China's GDP per capita at $13,862 versus the US at $90,027. China's Gini coefficient of 36.0 (World Bank 2022) is lower than the US Gini of 41.8 (World Bank 2024), but China's lower absolute income levels mean that a displaced clerical worker has less personal savings to draw on. Both countries lack robust universal retraining safety nets - the US through structural welfare gaps, China through an urban-rural social insurance divide that leaves migrant workers in clerical roles with minimal protection.

Economy context: US vs China side by side

Both countries differ substantially on income levels, inequality, and the state's role in labour market outcomes. These factors shape how AI displacement plays out beyond what the exposure scores alone capture.

China's lower Gini (36.0) reflects a less unequal income distribution than the US, but the absolute income floor in China is far lower. A displaced Chinese clerical worker earning the national average urban wage (National Bureau of Statistics of China, 2024: approximately $14,000 USD equivalent per year) has far less margin than a US clerical worker at the $45,432 BLS OEWS 2024 median. Both face inadequate safety nets for AI-driven displacement; China faces the same problem at a larger scale with fewer personal financial resources per affected worker.

For a broader framing of how both countries fit within the global AI risk picture, see the US vs World comparison and the India vs China analysis, which puts China's 362.2 million workforce in the context of the other major emerging economy competing for the same manufacturing and services base.