Summary of Key Points
This discussion revolves around the impact of AI on corporate organizations. Dr. Mu Sheng, a management expert, argues that it is a misconception for companies to view AI as a panacea. Improving individual productivity does not necessarily equate to enhanced organizational efficiency. The future organization will consist of a dynamic network of a small number of human elites and a large number of AI employees. Middle managers, especially those who merely pass information up and down, will be the first to be replaced by AI. Human creativity, complex decision-making abilities, and empathy are areas that AI cannot yet penetrate—what he calls the “forbidden zones of God.” The true pillars of corporate resilience are the organization itself, its people, data, and models. Even large companies still need to make significant changes in their organizational structures.
1. Using AI Without Greater Strength? The Problem Lies with the “Organizational Drag”
Many companies force employees to use AI and measure their performance based on the amount of tokens used, but efficiency does not improve. The reason is similar to the 19th-century transition from steam power to electricity: the technology has changed, but the organizational model remains the same. For example, in the steam engine era, factories relied on a central axis to operate all machines; switching to generators did not increase efficiency until each machine was independently powered and assembly lines were implemented. The same principle applies to AI today. Programmers may write code faster with AI, but if code review processes remain inefficient, it can lead to code accumulation and slower deployment. Similarly, although employees can use AI to streamline resume screening, outdated approval procedures and meetings can hinder overall progress.
Mu Sheng emphasizes that technology cannot overcome organizational barriers; without changing the organization, even the most advanced AI will be ineffective.
2. Who Will AI Replace First? Middle Managers (MOMs) Are the First to Go
It is often assumed that AI will replace ordinary employees first, but in reality, middle managers who merely coordinate information flow will be the first to be eliminated. Their roles—collecting data, conveying instructions, and monitoring outcomes—are perfectly capable of being automated by AI. Silicon Valley is already phasing out such middle managers, as they view their departments as their own “territories” and resist changes (similar to how workers resisted the introduction of the spinning jenny in the past). However, skilled middle managers (such as those who train AI models) will not be replaced; they will simply transition from managing people to managing AI systems.
3. Human “Forbidden Zones of God”: Three Abilities That AI Cannot Take Away
Mu Sheng insists that there are three human abilities that AI cannot replace:
1. Creativity: The ability to think outside the box, such as when Steve Jobs envisioned the smartphone; no algorithm can teach AI how to innovate in new areas.
2. Complex Decision-Making: The ability to discern key insights from vast amounts of information, like experienced investors who can quickly evaluate projects.
3. Empathy: The ability to understand and respond to human emotions, such as customer service representatives who provide comfort during difficult times; AI cannot simulate this level of emotional connection.
These abilities are fundamental to human capabilities and cannot be replicated by algorithms.
4. What Will the Future Companies Look Like? Three Characteristics of “Intelligent Organizations”
Mu Sheng describes the future organization as an “intelligent entity,” which has the following characteristics:
1. Fewer Humans, More AI: Standardized tasks will be handled by AI, leaving humans for strategic roles (creators) and frontline execution.
2. Decentralized Networks: AI systems will collaborate through APIs, forming a dynamic network without fixed hierarchies.
3. Customer-Centric Structure: The backend will consist of AI clusters and infrastructure, while the front end will be customer-focused units that can leverage company resources as needed.
Such organizations are not small-scale workshops; they rely on AI to increase productivity and functionality, making them more powerful than traditional companies.
5. The Pillars of Corporate Resilience: Organization > People > Data > Models
Many believe that models are the most important in the AI era, but Mu Sheng argues that the correct order is:
1. Organization: A well-designed organization can maximize the value of its talents; for example, a pyramid structure wastes elites, while an intelligent entity allows them to focus on core tasks.
2. People: Human abilities are the foundation of corporate resilience.
3. Data: Corporate data can enhance AI accuracy, but data lakes now make it easier to store and analyze various types of information.
4. Models: While public models (like GPT) are widely available, it is not difficult for companies to train their own models.
Although large companies like Tencent, Alibaba, and ByteDance claim to be platform-based and decentralized, their organizational structures are still pyramid-shaped. Changing the organization is much more challenging than changing technology.
Conclusion
AI is not just about replacing tools; it represents an opportunity to transform the entire organizational structure. The future of business competition will depend on which companies can first become intelligent entities. Humans should not fear being replaced by AI but should focus on developing abilities that cannot be automated. Companies should also shift their focus from merely using AI tools to reimagining their organizational foundations.