AI Infrastructure2026-04-21
MIT Technology Review
Chinese Tech Workers Train AI Doubles, Face Job Concerns
A surreal and ethically fraught trend is emerging in China's tech sector: employees are being tasked with training the very AI systems designed to replace them. Management directives instructing workers to 'distill' their professional expertise into AI agents have sparked widespread anxiety and soul-searching among the workforce.
Projects with names like 'Colleague Skill,' hosted on platforms like GitHub, offer frameworks for codifying human knowledge and workflows into automated counterparts. While framed as efficiency tools, the underlying implication is clear—these AI doubles are being built to perform the same tasks without human intervention. This creates a profound professional dilemma for early-adopter tech workers who are simultaneously driving innovation and potentially engineering their own obsolescence.
The phenomenon highlights the accelerating pace of AI automation and its complex human cost. Beyond immediate job security concerns, it raises questions about the ownership of expertise and the moral responsibility of companies. As workers digitize their problem-solving instincts and institutional knowledge, they are essentially creating a legacy system that may not require their future presence. This trend is forcing a difficult conversation about reskilling, corporate responsibility, and the nature of work in an era where creating your own replacement could become a standard job requirement.
