Smart Native Organization Theory: A New Framework for Survival and Growth in the AI Era

In the field of management studies, a theoretical counterargument has been presented against the notion that advancements in AI technology threaten the very foundation of organizations. A paper by Hou Hong and Li Zhiyong, published in the Chinese management journal Tsinghua Management Review, critiques the narrative that “AI will completely replace humans” amid the rise of large language models, agents, and multi-agent systems. It redefines the survival and growth of organizations as meso-level units of analysis and proposes the concept of a “smart native organization.”

The core of this article lies in viewing organizations not merely as aggregates of actions, but as complex systems embedded in uncertain environments. While AI has indeed raised the level of technical rationality, it has neither eliminated environmental uncertainty nor automatically answered fundamental questions such as “Why do organizations exist?” and “How do companies grow?” Rather, the authors argue that a smart native organization is a dialectical unity of AI and humans, possessing “bounded infinite possibilities.”

Response to the Crisis Narrative

The widespread narrative that “AI will completely replace humans” and “management studies are facing twilight” is rooted in Herbert A. Simon’s bounded rationality framework. The argument goes: if AI removes cognitive bottlenecks and significantly reduces coordination costs, organizations lose their raison d’être as “compensation mechanisms for human rationality.”

In response, some emphasize human value rationality, while others stress the importance of human-machine collaboration. However, the authors point out that the former is too grand and the latter too micro, lacking a systematic discussion focused on the meso-level unit of the organization.

Thus, this paper returns to two classic meso-level frameworks: Thompson’s open rational system theory and Penrose’s theory of firm growth. Based on the core objectives of organizational survival and growth, it attempts to reconfigure the relationship between AI and humans.

Conditions for Survival:

Uncertainty and Uneven AI Penetration

The essence of organizational survival lies in maintaining the capacity for action in an uncertain environment. The authors classify uncertainty into three types: general uncertainty, contingent uncertainty, and technical uncertainty. To address these, organizations have evolved a functionally layered architecture comprising technical, managerial, and institutional layers.

AI has a significant efficiency advantage in dealing with technical uncertainty and contingent uncertainty. However, it has inherent limitations in addressing general uncertainty—scenarios where reliable causal knowledge is lacking. Consensus-building among stakeholders is a political negotiation process, and AI cannot provide politically symbolic actions. Entirely novel variables lie beyond AI’s existing experiential framework, and AI cannot recognize their significance.

In a smart native organization, AI exhibits an uneven penetration structure. In the technical layer, the potential for AI substitution is highest; it can deeply replace human labor to produce efficient and stable outcomes. In the institutional layer, the core lies in meaning-making and consensus maintenance, where AI’s substitution potential is fundamentally limited. The managerial layer evolves into a complex human-machine coordination structure, but its core function of “delineating the boundary of uncertainty processing” cannot be technologized.

Conditions for Growth: Beyond the Penrose Effect

The essence of firm growth lies in endogenous growth through discovering the versatility of unused resources. The driving force of growth comes from endogenous resources; the direction depends on the entrepreneur’s imagination and beliefs about new uses of resources. The speed of growth is constrained by the supply of managerial services—this is the “Penrose effect.”

AI can improve information processing capacity, alleviate bottlenecks in the supply of managerial services, and solve constraints on growth speed. However, it cannot replace entrepreneurial services. AI excels at single-loop learning within a given framework but cannot complete double-loop learning that transcends existing frameworks. AI is data-driven backward thinking; it lacks the initiative and creativity of entrepreneurs who, based on beliefs, envision the future.

Smart native organizations favor growth through vertical integration rather than horizontal diversification. AI requires upstream and downstream data optimization, and technical capabilities tend to spill over upstream and downstream. In vertical expansion, computing resources are often used on demand in a cloud model, with less resource redundancy. Moreover, trained intelligent agents are difficult to generalize across businesses, resulting in insufficient endogenous driving force for horizontal expansion.

Bounded Infinite Possibilities

AI, as a super lever, breaks through the traditional quadruple rigid constraints of organizations. It ultimately compensates for individual bounded rationality, shifting decision-making bottlenecks to highly elastic data and computing networks. Through fluidization, it restructures fixed processes and organizational walls, allowing organizations to build dynamic collaborative networks on demand as needed. It achieves continuous 24/7 evolution, breaking the cyclical constraints of traditional organizational learning. It breaks through the Penrose effect, transforming growth from a linear model dependent on human capital to an exponential model dependent on intelligence density.

However, AI has three clear functional boundaries. First, it cannot handle incalculable value judgments or problems involving entirely new variables. Second, it lacks generative rationality and entrepreneurial spirit, unable to complete creative framework reconstruction. Third, the meta-management function of “defining the nature of problems and delineating the boundaries of technological application” cannot be substituted by AI.

The core of a smart native organization lies in the crucial division of labor between humans and machines. In the technical layer, AI deeply substitutes; in the managerial layer, humans and machines coordinate; in the strategic and institutional layers, core functions remain in human hands. Ultimately, it channels the infinite possibilities of technical rationality into human value pursuits, achieving bounded infinite development.

Editorial Perspective

This paper is valuable in that it provides a realistic analytical framework for the “organization unnecessary” narrative spawned by the AI bubble, through a reinterpretation of classic organizational theory. In the short term, it will prompt corporate executives advancing AI adoption to reconsider the allocation between technological investment and human capital. Notably, its warning about the risk of excessive AI investment in the technical layer overlooking the importance of the managerial and institutional layers offers practical insights.

In the long term, the framework outlines the direction of organizational design in the AI era. Rather than merely pursuing cost reduction through full automation, the idea of clarifying areas where human value judgment and entrepreneurship are indispensable and expanding them with AI can serve as a foundation for sustainable corporate growth. However, it remains to be verified whether the Penrose theory and Thompson theory on which the authors rely are equally applicable to cloud-native or AI-native startups.

From an editorial standpoint, the concept of “boundaries” presented in this paper is likely to become a focal point for future discussion. The question of where to draw the line for AI’s scope is influenced not only by technical limitations but also by managers’ values and social acceptance. It is hoped that the dialogue between management studies and practice will deepen to a more profound dimension.

Reference

• Huxiu.com: “Survival and Growth of Smart Native Organizations: Bounded Infinite Possibilities” (https://www.huxiu.com/article/4866234.html?f=rss) — Published June 10, 2026
• Tsinghua Management Review (WeChat official account) — Authors: Hou Hong, Li Zhiyong