China’s Professionals Turn to Gig Work Training AI Models

Summary

Chinese technology companies are recruiting lawyers, architects, engineers, teachers, journalists, and other specialists to create data for AI models. The work includes designing realistic tasks from their professions, uploading relevant documents, recording their reasoning, and evaluating model responses; trainers generally cannot use AI to complete the assignments. A task can take several hours and pay 100 to 500 yuan, but submissions may be rejected without payment if they fail quality reviews. Platforms including Alibaba’s Siriser, ByteDance’s Xpert, TalentsAI, and MeetChances are building this labor market, while China’s National Data Administration has encouraged experts and college graduates to participate in data annotation and called for higher-density training data. The jobs are emerging as layoffs, weak demand, and AI-driven disruption squeeze white-collar workers. ByteDance says Xpert has recruited more than 50,000 experts, and the article says China’s AI training-data market is expected to reach 7.8 billion yuan in 2026, up 25% from 2025, according to IDC. Experts are increasingly being recruited for productivity applications such as legal, financial, office, and engineering tools, where general-purpose model ability is insufficient. Workers interviewed by Rest of World described the income as unstable and often too low for the time involved, while some also worried about the use of uploaded work files. Even so, they view the gigs as supplementary income and a way to remain relevant as AI changes their professions.