Chinese Engineers Train AI Doppelgangers, Resist Self-Replacement

Introduction: Engineers Training Their Own AI Doppelgangers

On April 20, 2026, MIT Technology Review reported an intriguing development: workers at Chinese technology companies are, under orders from their superiors, training AI agents of themselves. This is not merely a technological implementation but a deeply conflicted endeavor involving entrusting one’s professional future to AI. The GitHub project “Colleague Skill” symbolizes this phenomenon, spreading a wave of introspection across the tech industry.

Background: China’s Tech Industry and the AI Adoption Trend

China leads the world in the development and application of AI technology. Under the government’s digitalization strategy, tech companies are aggressively adopting AI to boost productivity. However, this rapid automation is bringing significant changes to the labor market. Particularly in fields like software development and data analysis, AI is not just complementing human work but increasingly has the potential to replace it.

Against this backdrop, companies are advancing the introduction of AI agents, citing cost reduction and efficiency. One of the most effective methods is having actual workers train the AI. This is said to allow the AI to acquire more human-like judgment and skills. However, this approach places a psychological burden on workers, resulting in prompting self-reflection.

Specific Example: The GitHub Project “Colleague Skill”

The “Colleague Skill” mentioned in the news is an open-source project hosted on GitHub. This tool offers a function to “distill” the skills and personality traits of colleagues. In other words, by analyzing conversation records and work outputs, it can train an AI model to create a digital doppelganger of that person.

For instance, by having a veteran engineer’s years of accumulated coding tips and project management know-how learned by AI, junior or new employees can instantly utilize that knowledge. However, this technology also carries the potential to threaten the role of that engineer themselves. Among workers forced to use this tool under their superior’s orders, many harbor anxiety about being dismissed in the future as their skills are “absorbed” by AI.

Impact: Worker Psychology and Industry Changes

This phenomenon is not merely a technical issue; it is profoundly affecting workers’ psychology. Even tech workers who have actively embraced AI are increasingly experiencing a sense of loss of self-worth by serving as AI trainers. Previously welcoming AI as a “convenient tool,” they are now beginning to perceive it as a “competitor replacing themselves.”

Across the industry, changes in employment structures may accelerate. By introducing AI agents, companies can reduce labor costs and build systems that operate 24/7. On the other hand, workers face the need to update their skills and develop creative capabilities that cannot be replaced by AI—