China's White-Collar Workers Turn to AI Training Gigs Amid Downturn

Professionals become data workers in China's AI push

Chinese technology companies are recruiting architects, lawyers, engineers, and other skilled professionals to work as specialized AI trainers, creating high-quality datasets to improve model performance in professional domains. The trend reflects both the country's ambitious AI development goals and the economic pressures facing its middle class.

A Shenzhen-based architect identified as Cuicui told Rest of World that declining government infrastructure spending cut her income in half over five years. To cover mortgage payments and her child's expenses, she began logging onto TalentsAI after work to teach AI models how to write building proposals and conduct profitability analyses. "It is just like teaching a child," she said. "Every task has to come from my actual work."

Platforms operated by Alibaba, ByteDance, Tencent, and independent companies are recruiting professionals across dozens of specialties. ByteDance's Xpert platform, launched in 2025, claims to have recruited more than 50,000 experts ranging from writers to therapists. These workers create realistic tasks from their actual jobs, upload work documents, and document their reasoning processes—all without using AI assistance themselves.

Why it matters

China's shift toward recruiting domain experts for AI training signals a strategic move from consumer applications toward productivity tools, with the country's AI training data market expected to reach $1.1 billion in 2026. This development mirrors similar efforts by U.S. platforms like Mercor and Surge AI, but occurs against a backdrop of 17.9% youth unemployment and widespread economic anxiety among China's professional class. The trend reveals how AI development is simultaneously creating new forms of precarious work while threatening the very expertise it seeks to capture.

Economic pressure drives participation

The gig work pays 100 to 500 yuan ($15 to $74) per task, which typically requires several hours to complete. Trainers report that assignments are frequently rejected by quality control teams, resulting in unpaid work. Platform terms explicitly state that workers receive no payment for rejected submissions, though some allow revisions or appeals.

A Shanghai-based software engineer with two decades of experience, identified as Jessica, began taking AI training work in July despite minimal pay. "You never know how long you will get to keep your job," she said. "You have to keep evolving yourself to avoid getting completely replaced."

Another trainer, Liu, holds a master's degree in law but avoided joining a law firm due to concerns about long hours and workplace culture. She works at a legal AI company during the day and takes training gigs with multiple platforms in the evening, asking models to provide legal advice or draft judgments. "If I don't pass my knowledge to AI, others will do that," she said.

Government backing and market growth

In June, China's National Data Administration endorsed building "high quality data sets" and called for bringing more industry experts into data annotation to raise the "knowledge density" of training data. The agency encouraged college graduates to view annotation work as flexible employment.

Julie Yujie Chen, an associate professor at the University of Toronto who researches data labor in China, said government endorsement has lent legitimacy to the profession. "The middle class in China with ordinary white-collar jobs are living in a state where they face this urgency to find a supplementary income," she told Rest of World.

Tiezhen Wang, an AI researcher and former head of the Asia-Pacific ecosystem at Hugging Face, said Chinese AI labs are ramping up specialized data collection as they expect AI work tools to become the next competitive battleground. "Making models perform better at work will be a very strong differentiator in the future AI competition," he said.

These details were first reported by Rest of World.