Chinese tech workers are starting to train their AI doubles–and pushing back

The News

Chinese tech workers are increasingly being directed by their employers to train artificial intelligence agents designed to replicate their workflows and, ultimately, replace them [1]. This directive, coupled with the recent viral spread of a GitHub project called "Colleague Skill," has sparked a wave of introspection and resistance within the Chinese tech sector [1]. The "Colleague, Skill" project, while details remain sparse, reportedly allows users to "distill" the skills and personality traits of colleagues and replicate them within an AI agent [1]. While the precise methodology behind this distillation process isn't publicly available, its emergence coincides with a broader trend of Chinese companies seeking to leverage AI for workforce optimization and cost reduction [1]. The situation has moved beyond theoretical discussion, with reports indicating that workers are actively participating in the training of these AI doubles, albeit with growing unease and a degree of pushback [1]. This phenomenon represents a significant shift in the adoption of AI within Chinese workplaces, moving from enthusiastic experimentation to a more complex and potentially contentious dynamic [2].

The Context

The current situation is rooted in a confluence of factors, including the aggressive pursuit of AI-driven efficiency by Chinese corporations and the readily available, albeit nascent, tools to facilitate this transition [3]. The underlying technical architecture enabling this "AI double" creation likely involves a combination of large language models (LLMs), behavioral cloning, and potentially reinforcement learning techniques [1]. LLMs, trained on vast datasets of code, documentation, and internal communications, provide the foundational knowledge base for the AI agent [1]. Behavioral cloning, a supervised learning technique, allows the agent to mimic the actions and decision-making processes of the targeted employee by observing their workflows and interactions [1]. The "Colleague Skill" project likely leverages these techniques, although the specific algorithms and data used remain proprietary [1]. The project's viral spread suggests a degree of accessibility and ease of use, lowering the barrier to entry for companies wanting to implement this technology [1].

The push for AI-driven workforce optimization is not new, but its recent acceleration is tied to several economic and strategic imperatives [3]. Chinese tech CEOs, mirroring trends observed in Western companies like those led by Mark Zuckerberg and Jack Dorsey [3], are increasingly seeking ways to maximize productivity and control through AI [3]. Uber’s recent pivot towards “assetmaxxing,” as described by TechCrunch, exemplifies this broader trend of leveraging AI to optimize existing resources and reduce reliance on human labor [4]. This strategy involves shifting from a model of owning assets (drivers, vehicles) to a model of optimizing the use of existing assets through AI-powered management and automation [4]. The Chinese context is further complicated by the government’s strong support for AI development and its ambition to become a global leader in the field [1]. This support manifests in the form of funding, policy incentives, and a generally favorable regulatory environment for AI adoption [1]. The development of "mirror" bacteria, mentioned in The Download, highlights the broader, and sometimes unsettling, frontier of synthetic biology and AI convergence, reflecting a willingness to explore high-risk, advanced technologies [2]. While seemingly disparate, both developments point to a culture of pushing technological boundaries, even with potential ethical and societal ramifications [2].

Why It Matters

The implications of this trend extend far beyond the immediate impact on Chinese tech workers [1]. From a developer/engineer perspective, the forced participation in AI double training creates a significant technical friction point [1]. While some engineers may be enthusiastic about contributing to AI development, others are understandably resistant to being effectively replaced by their own digital replicas [1]. This resistance can manifest as decreased productivity, lower morale, and even attrition, potentially offsetting the cost savings initially anticipated by employers [1]. The adoption of this technology also introduces new technical challenges, including ensuring the accuracy and fairness of the AI models, mitigating bias in the training data, and addressing potential security vulnerabilities [1]. Details are not yet public regarding the specific performance metrics being used to evaluate these AI doubles, but it's likely that companies are focusing on efficiency gains, cost reduction, and improved output [1].

For enterprises and startups, the widespread adoption of AI doubles has the potential to disrupt traditional business models [1]. Companies that successfully implement this technology could gain a significant competitive advantage by reducing labor costs and increasing productivity [1]. However, those that fail to manage the transition effectively risk alienating their workforce and damaging their reputation [1]. The rise of "assetmaxxing" strategies, as seen with Uber [4], demonstrates how AI can fundamentally reshape business operations, shifting the focus from human capital to optimized asset utilization [4]. The winners in this ecosystem will likely be companies that can balance the benefits of AI automation with the need to maintain a motivated and engaged workforce [1]. The losers will be those who prioritize short-term cost savings over long-term employee well-being and innovation [1]. The emergence of the "Colleague Skill" project also highlights the potential for open-source initiatives to democratize AI technology, empowering smaller companies and individuals to compete with larger corporations [1].

The Bigger Picture

The Chinese experience with AI doubles is not entirely unique, but it represents a particularly acute manifestation of a global trend [3]. Western tech companies have also been exploring ways to leverage AI for management purposes, albeit often with less overt coercion [3]. The focus on "being everywhere at once" by CEOs like Zuckerberg and Dorsey [3] suggests a desire to exert greater control over operations and decision-making, potentially through AI-powered systems [3]. However, the Chinese approach appears to be more aggressive and centralized, reflecting the country’s unique political and economic context [1]. This contrasts with the more decentralized and market-driven approach often seen in the West [1].

Looking ahead, the next 12-18 months will likely see increased scrutiny of the ethical and societal implications of AI-driven workforce automation [1]. The resistance from Chinese tech workers could serve as a catalyst for broader discussions about the role of AI in the workplace and the need for policies to protect workers’ rights [1]. Competitors are likely to observe the Chinese experience closely, adapting their own AI strategies based on the lessons learned [1]. The development of more sophisticated AI agents capable of truly replicating human skills and personality traits is also likely to continue, blurring the lines between human and artificial labor [1]. The emergence of "mirror" bacteria [2], while seemingly unrelated, underscores the broader trend of pushing the boundaries of synthetic biology and AI, potentially leading to unforeseen consequences [2]. The long-term impact on the global AI talent pool remains uncertain, but it is likely that the demand for AI specialists will continue to outstrip supply [1].

Daily Neural Digest Analysis

The mainstream media narrative often frames AI adoption as a purely positive development, emphasizing the potential for increased productivity and economic growth [1]. However, the situation unfolding in China reveals a more complex and potentially troubling reality [1]. The forced training of AI doubles, while presented as a means of efficiency, risks creating a climate of fear and resentment among workers [1]. The "Colleague Skill" project, while technically impressive, raises serious questions about the ethical implications of replicating human skills and personality traits without consent [1]. The focus on "assetmaxxing" [4] highlights a broader shift towards treating human labor as a commodity, rather than a valuable asset [4]. The sources do not specify the long-term psychological effects on workers who are essentially being replaced by AI versions of themselves [1]. This situation underscores the need for a more nuanced and critical examination of AI’s impact on society, one that considers not only the potential benefits but also the potential risks to human dignity and well-being [1]. What safeguards are necessary to ensure that AI serves humanity, rather than simply replacing it?

---

References

[1] Editorial_board — Original article — https://www.technologyreview.com/2026/04/20/1136149/chinese-tech-workers-ai-colleagues/

[2] MIT Tech Review — The Download: murderous ‘mirror’ bacteria, and Chinese workers fighting AI doubles — https://www.technologyreview.com/2026/04/20/1136154/the-download-murderous-mirror-bacteria-chinese-workers-fight-ai-agents/

[3] Wired — Tech CEOs Think AI Will Let Them Be Everywhere at Once — https://www.wired.com/story/tech-ceos-using-ai-to-be-everywhere-at-once/

[4] TechCrunch — TechCrunch Mobility: Uber enters its assetmaxxing era — https://techcrunch.com/2026/04/19/techcrunch-mobility-uber-enters-its-assetmaxxing-era/