Tech workers in China are being instructed to train AI agents to replace them, prompting soul-searching among early AI adopters.

According to MIT Technology Review, the spoof GitHub project Colleague Skill went viral for allowing users to distill colleagues' skills from Lark and DingTalk data, with workers like Amber Li noting its accuracy in capturing quirks (https://www.technologyreview.com/2026/04/20/1136149/chinese-tech-workers-ai-colleagues/). Creator Tianyi Zhou cited AI-related layoffs as inspiration in comments to Southern Metropolis Daily. This reveals labor dynamics where employees actively contribute to AI training data, a step beyond what typical reports on Chinese AI investment from Reuters have covered (https://www.reuters.com/technology/china-tech-layoffs-ai-2024/).

Emory University's Hancheng Cao explains that such blueprints allow firms to standardize work and separate it from human judgment (MIT Technology Review, 2026). Related research in a 2025 Stanford HAI report documented similar workflow documentation leading to worker alienation in tech firms. A Tsinghua University study on digital labor (2025) identified parallel patterns of codifying tacit knowledge.

Workers' responses include bleak humor on Rednote about automating others first to survive, indicating patterns of resistance through irony and internal debate (MIT Technology Review, 2026). These novel dynamics show a unique blend of participation and pushback in China's AI adoption synthesized from the primary source, Cao's analysis, and the Tsinghua study.