ILO Publishes Study of AI Gains and Worker Risks in Chinese Firms
Reported benefits are concentrated in routine, data-heavy work, while weak measurement and income concerns complicate the productivity story.
Listen to this story
The audio brief
Story brief
3 key pointsThe ILO’s China study points to a measured, uneven AI payoff: firms report faster processing and higher throughput, but most deployments still keep humans in the loop. Evidence comes from 21 interviewed enterprises and 1,591 surveyed professionals, not a representative or independently audited sample. The business risk is distributional: 39% of workers surveyed expect AI to cut their income, while firms cite skills,...
- 01
An insurer reported customer-service capacity rising from 6,000 to 15,000 daily issues across 300 employees.
- 02
A smart-manufacturing facility reported 30% higher production efficiency; figures came from firms and were not independently verified.
- 03
One large insurance group cut recruiting time 57%, from 30 days to 13.
The International Labour Organization has published research on AI adoption in Chinese enterprises that finds reported productivity gains alongside a workforce still organized mainly around people working with AI, rather than full automation. The study also flags skills gaps, income anxiety and weak measurement beyond output.
The research brief draws on in-depth interviews with 21 enterprises and a survey of 1,591 professionals in China. The firms ranged from an eight-person startup to a conglomerate with 270,000 employees, across manufacturing, finance, business services, construction, education, media and travel.
Faster routines were the first gains
The reported improvements were concentrated in repetitive, data-heavy work: processing documents, responding to customer queries, screening résumés and collecting data. One insurance company said 300 customer-service employees raised daily issue handling from 6,000 to 15,000 after adopting AI. A smart manufacturing facility reported a 30% production-efficiency increase.
Those results are not a representative scorecard for Chinese business. The enterprises were purposively selected, and the productivity figures were supplied by firms rather than independently verified. Most also lacked systematic frameworks to assess AI’s effects, especially beyond conventional productivity measures.
Automation has not displaced oversight
The study finds human–AI collaboration and hybrid workflows more common than full automation. It identifies three ways companies organize adoption: centralized specialist teams, AI embedded in business units, and bottom-up use that spreads organically through smaller firms.
That does not remove pressure on workers. Among surveyed professionals, 56% viewed AI adoption as inevitable and 47% thought it would create more jobs than it displaces. Yet 39% expected AI adoption to reduce their income. Firms cited skills gaps, employee resistance, output quality, data security, regulation and integration with existing systems as barriers to deeper use.
The next move is to spread the gains
The ILO’s proposed response focuses on helping workers and smaller firms participate in adoption, while giving employers a clearer way to judge its effects on job quality and working conditions.
The ILO’s proposed next steps
- Build AI skills and lifelong learning, with attention to mid-career and older workers.
- Support workers moving toward higher-value tasks.
- Measure job quality and working conditions alongside productivity.
- Give smaller firms access through shared platforms, training and affordable services.
Better measurement would help firms assess workplace quality alongside throughput, rather than treating reported efficiency as the whole result of AI adoption.
Reader comments
Newest comments first. Replies stay oldest first.