Zhang Xiaoyu on AI-Native Organizations

From P2P Foundation Wiki
Jump to navigation Jump to search

Contextual Quote

"Xiaoyu’s proposal for an AI-native organisation is more concrete. This is not an ordinary company with a copilot. It may consist of three humans and hundreds of agents conducting research, writing code, procurement, sales, scheduling and administration. It scales through compute, permissions and physical interfaces rather than headcount. It may not require the managerial layers, office districts or professional ecology through which the old middle reproduced itself."

- Chor Pharn [1]


Description

Synthesis by ChatGPT, prompted by Michel Bauwens, and inspired by Chor Pharn's reference [2] :


Zhang Xiaoyu (张笑宇; Xiao-Yu Zhang) is a Chinese writer and independent scholar whose work focuses on technology, civilization, political history and the social consequences of artificial intelligence. He is the author of works including Technology and Civilization, Commerce and Civilization, Industry and Civilization, and AI Civilization: Prehistory. In 2026 he taught an “AI Sociology” course at the PPE Academy in Hangzhou, where he developed a framework for thinking about how AI agents could transform organizations and social structures.

AI-native organization (AI原生组织) is central to Zhang's recent thinking. The distinction is between an ordinary organization that adopts AI as a tool and an organization designed from the outset around AI capabilities. In the former, employees continue to perform the essential work while AI assists them. In the latter, AI agents become part of the organization's basic operating structure, potentially performing research, programming, administration, analysis, coordination and other forms of cognitive work.

Zhang's argument begins from the observation that modern organizations were partly created to manage the scarcity of human cognitive labour. Companies require departments, managers, specialists and administrative systems because individual humans cannot perform all the functions required by a complex organization. If AI agents can perform much of this work, the relationship between organizational capacity and human headcount may change. A small number of people could potentially coordinate a much larger amount of productive activity through AI systems.

In a May 2026 presentation on AI sociology, Zhang described four ideas associated with this transition: human equivalent, concerning the amount of human-equivalent work AI can perform; CTK (tacit knowledge) projection, concerning the conversion of organizational knowledge into forms accessible to AI; interaction as interface, concerning AI's role in mediating organizational communication and knowledge; and an AI-native “free association” of people. The latter reflects Zhang's interest in organizations based less on hierarchical employment and more on voluntary cooperation between highly capable individuals supported by AI.

Source: [3](https://36kr.com/p/3827627297002371)

Zhang has also experimented with an AI system he calls AIgora. The system uses multiple agents with different intellectual roles—such as literature review, divergent thinking, criticism and synthesis—to undertake research and discussion. The significance of AIgora is not simply that several AI systems are used simultaneously, but that intellectual work itself is divided between specialized agents, with humans setting questions and evaluating results.

Zhang therefore envisages a possible shift from the conventional company toward a much smaller organization in which a few highly capable humans are supported by a large number of AI agents. In an interview in January 2026, he discussed the possibility of companies becoming more like associations of highly productive individuals, with AI allowing each person to accomplish work that previously required teams and organizational hierarchies.

Source: [4](https://www.36kr.com/p/3650528288273155)

This argument has wider political implications. If productive organizations require substantially fewer human workers, they may also require less of the managerial, administrative and professional infrastructure traditionally associated with large firms. The minimum viable organization could become smaller, more mobile and less dependent on a particular locality. However, this is a hypothesis rather than an established economic trend. AI-native organizations could equally remain dependent on large concentrations of capital, computing infrastructure, energy, data and physical assets.

The idea has been taken up by writer Chor Pharn in his essay A Republic Capable of Betrayal. Pharn connects Zhang's AI-native organization to his own argument about the political role of locally rooted proprietors, or the “gentry”. His suggestion is that if organizations become radically smaller and less dependent on local human networks, AI could weaken some of the economic foundations of this political class. Pharn further speculates that highly mobile AI-native organizations might increasingly choose among jurisdictions rather than being deeply embedded in a particular political community. These political conclusions are Pharn's extrapolation and should not be attributed directly to Zhang.

Source: [5](https://thecuttingfloor.substack.com/p/a-republic-capable-of-betrayal)

The broader significance of Zhang's proposal is therefore not simply that AI will automate existing jobs. It is the possibility that AI changes the organizational unit of production itself. Instead of asking how existing companies can make their employees more productive with AI, the AI-native perspective asks what a company would look like if it were designed on the assumption that much cognitive execution could be delegated to machines.

This raises unresolved questions about responsibility, ownership, authority, verification and the distribution of economic power. A smaller workforce does not necessarily mean a more decentralized economy: if AI-native production depends heavily on computing infrastructure, advanced chips, proprietary models and energy, ownership could become more concentrated rather than less.

Sources

Retrieved from "https://wiki.p2pfoundation.net/index.php?title=Zhang_Xiaoyu_on_AI-Native_Organizations&oldid=150545"