Chinese tech giants are hiring skilled professionals as specialized AI trainers to build high-quality datasets, mirroring efforts by US platforms like Mercor
Squeezed by a stagnant economy and state directives, China's underemployed lawyers, architects, and engineers are taking cheap gig work …
Context & Ripple Effects
Chinese AI development has already relied on large-scale labeling work: a 2023 investigation found data-labeling companies serving Baidu, Alibaba, and JD using vocational-school internships, raising a contrast with the turn to professionals supplying domain-specific training material.
The hiring shift also fits a cost-conscious model-building environment. Chinese developers including DeepSeek and 01.ai had been pursuing smaller training datasets to reduce model costs amid export controls, while workforce cuts at major platforms had increased workers’ concerns about AI-driven displacement.
First-order effects
- Chinese tech giants gain access to lawyers’, architects’, and engineers’ job-specific tasks, documents, and reasoning for datasets intended to make workplace AI more useful.
- Underemployed professionals gain a gig-work outlet, but at low pay that converts specialized occupational knowledge into training inputs.
Second-order effects
- Data-labeling vendors that previously supplied mass annotation for Baidu, Alibaba, and JD face pressure to recruit and manage credentialed contributors rather than rely chiefly on student labor.
- For Chinese model developers pursuing smaller datasets, expert-generated examples raise the value of each training item and make data quality a more important competitive input than raw volume.
Third-order effects
- If platforms continue pairing workforce reductions with expert-training gigs, professional knowledge work may split more sharply between a smaller set of retained roles and lower-paid work that transfers expertise into AI systems.
- The pattern places AI labor policy at the intersection of state-backed industrial priorities and employment pressure, making the terms of training work as consequential as access to compute.
The trend: Chinese AI development is moving toward higher-value, domain-specific training data as platforms seek workplace usefulness while managing compute and labor constraints.