Chinese Workers Are Training Their Own Replacements — And Fighting Back
There's a particular cruelty to being asked to train your own replacement. It's happened in corporate America for decades during offshore outsourcing waves, where departing employees had to document their processes and train their cheaper successors before receiving severance packages. But the latest iteration of this dynamic, now unfolding in China's tech sector, has a dystopian twist: the replacement isn't a person in another country. It's an AI agent built from your own digital footprint.
According to recent reports, Chinese tech workers are being asked by employers to actively participate in creating AI agents that replicate their workflows, communication styles, and professional personas. A GitHub project called 'Colleague Skill' has gone viral for its ability to automatically reconstruct coworkers as AI agents by analyzing workplace communications. The backlash has been swift and visceral, sparking debates about worker dignity, job security, and the ethics of forced participation in one's own obsolescence.
This isn't just about automation anxiety. What makes this situation distinct is the psychological dimension of active collaboration. Previous waves of automation typically happened to workers — factories closed, software replaced manual processes, algorithms took over tasks. But asking employees to consciously build detailed replicas of themselves crosses a different line. It transforms workers from victims of technological displacement into unwitting architects of it.
The timing is significant. We're seeing similar patterns emerge globally, though usually with more corporate euphemism attached. Hyatt's recent deployment of ChatGPT Enterprise across its workforce, for instance, is framed as 'enhancing employee productivity.' Perplexity's new Personal Computer assistant promises to 'manage workflows on behalf of users.' The language is always about augmentation and assistance, never about replacement. But workers aren't stupid. They understand what productivity enhancement often means in practice: doing more with fewer people.
What's remarkable about the Chinese workers' response is the directness of their resistance. By making the 'Colleague Skill' project a topic of public debate rather than quiet resignation, they're forcing a conversation that Silicon Valley has carefully avoided. The Western tech industry has become masterful at obscuring the labor implications of AI through aspirational language about human-machine collaboration. Chinese workers, it seems, are less willing to pretend.
The deeper issue here isn't really about AI capability — it's about power and consent. An AI agent that learns from your work patterns without your meaningful agreement isn't a productivity tool. It's a surveillance and extraction mechanism. The fact that it might eventually replace you is almost secondary to the violation of being digitally replicated without genuine choice.
This matters because China's tech sector often serves as a preview of trends that later arrive in Western markets. The bluntness of what's happening there may simply be an unvarnished version of dynamics already present everywhere AI is deployed in workplaces. The main difference is that Chinese workers are calling it what it is.
The question facing workers globally isn't whether AI will automate knowledge work — that's already happening. It's whether they'll have any say in how that automation unfolds, whether their expertise will be extracted or collaborated with, and whether they'll be compensated for the intellectual labor of making themselves redundant. Right now, the answer appears to be no. But the fact that workers are starting to organize resistance around these questions suggests the conversation is just beginning.