Is Your Organization Paralyzed in the Age of AI?

The AI technology is sweeping in comprehensively, and its impact on the business world is as earth - shattering as a complete transformation. In the era of new technology, not only the business logic but also the organizational logic has been reshaped. Thus, the "agent organization" has become a hot topic. However, as various viewpoints pour in, this concept has become even more blurred. Beyond the surface concept, we are more eager to inquire: In an "AI - first" organization, what will happen to human employees? And what kind of organizational form will enterprises present?

Business World has specially invited Mu Sheng, a well - known domestic management expert who has long been deeply involved in organizational change and continuously concerned about the AI trend, the founder of Musheng Consulting, and a Doctor of Business Administration from the Guanghua School of Management at Peking University. Through 17 key questions, he systematically analyzes the organizational logic in the AI era. This exclusive interview is divided into two parts. The first part focuses on human resources, directly addressing the "feud" between employees and AI and dissecting the incentive challenges; the second part focuses on the organization, analyzing the real concerns of bosses and exploring how enterprises can transform into agent organizations.

The following is the first part, "Human Resources":

Business World Magazine (hereinafter referred to as "Business World"): In the past, when enterprises talked about automation, it was often "machines replacing humans", and the main ones being replaced were blue - collar workers. However, generative AI has started to target office white - collar workers. Is AI a good thing for humans? In some enterprises, there have been extreme behaviors where employees refuse to use AI on a large scale and even feed false data to AI to oppose the "distillation" of their personal skills.

Dr. Mu Sheng (hereinafter referred to as "Mu"): Any progress in productivity is marked by the replacement of certain human functions. If productivity keeps advancing, human activities will be forced into the fields where creativity can be most effectively exerted. Humans do what they are good at, and machines and AI do what they are good at. Then, productivity can develop explosively, and the ultimate beneficiaries will still be humans. In this sense, the popularization of AI is positive.

It's just that the progress of productivity is always more beneficial to the elite, who possess non - replaceable abilities. Those whose jobs are taken away will resent the new technology, but they can't stop technological progress, just like when workers angrily smashed the spinning jenny in the past, they couldn't stop the Industrial Revolution.

Business World: Are those who are sacrificed the price of productivity progress? If a large number of humans lose their jobs, will it undermine social stability? How can we balance productivity progress and social stability?

Mu: The social welfare system protects the weak. It levies taxes from the elite and transfers the funds to the weak to maintain their survival and dignity. This is a warm - hearted system. Every country has such a welfare system. However, in the AI era, this system may need to play a greater role.

The short - term unemployment caused by the huge productivity progress of AI is a bit like a "run on the bank", which the welfare system may not be able to fully handle. Enterprises first need to establish an internal welfare system to help employees replaced by AI receive retraining and return to work, or let them leave the enterprise in a more stable way to seek new careers. This requires enterprises to bear costs and tests the strength of enterprises and the social responsibility of entrepreneurs.

Business World: Liu Qiangdong insists not to lay off employees after AI replacement. Taking large enterprises like JD.com, which have a large number of employees and complex processes, as an example, if a large number of repetitive positions are taken over by AI, but the enterprise does not want to simply lay off employees, how should the organization arrange these people?

Mu: This is indeed a very tricky issue. JD.com's success largely lies in its self - built logistics system, which enables it to compete with platform - type e - commerce like Alibaba through its self - operated model. In this process, Liu Qiangdong has been very kind to blue - collar employees, including those in the logistics system, creating an image of being kind - hearted. Obviously, his hundreds of thousands of brothers cannot be easily abandoned. In this sense, they are the most difficult to transform. If they can get through this, they will set a benchmark for all large enterprises.

I have communicated with executive friends at JD.com, and they are investing a lot of energy in researching this topic. Recently, Liu Qiangdong also disclosed the internal "Nirvana Plan", hoping to send the existing 700,000 blue - collar workers, such as couriers, to schools for technical training to help them adapt to the AI era. His concept is that in the future, these people can maintain and repair robots, coexist with AI, rather than having their jobs taken away by AI.

Many people don't understand Liu Qiangdong's move and think that he is paying the price for his past image and is on the defensive. In fact, it's not the case. It's impossible for all these 700,000 people to maintain and repair JD.com's robots in the future. Otherwise, JD.com might as well not use robots because the labor cost won't decrease at all. Liu Qiangdong's move is an offensive in advance. The market for robot maintenance and repair in the future is limited. If JD.com's brothers occupy it first, other enterprises will have fewer opportunities.

Business World: Previous technological progress has also replaced some positions, but the arrival of AI technology is not only replacing positions but also reconstructing organizational division of labor. How will this change the organizational design of large enterprises?

Mu: Any revolutionary progress in productivity will change the organizational division of labor and be reflected in the organizational model.

Looking back at the industrial development history: In the Industrial 1.0 era, it was mechanization, marked by the steam engine, and the organizational model was the line system, a simple structure of "head factory - branch factory"; in the Industrial 2.0 era, it was electrification, marked by electricity and assembly lines, and the organizational model was the functional system, including the later improved line - functional system, which established a functional system acting as the brain on the basis of the line system; in the Industrial 3.0 era, it was informatization, marked by computers and the Internet, and the organizational model was the divisional system, which allocated some functions to business BUs (Business Units) to give them more autonomy; in the Industrial 4.0 era, it is intelligentization, marked by the Internet of Things, big data, and AI, and the platform - type or network - like organizational model has become the mainstream, which is also the focus of our current attention.

I call this organizational model a "platform - type organization". When this organizational model is integrated with AI technology, it becomes the currently highly - concerned "agent organization".

The front - end will become a number of extremely small but fully functional business units that can achieve "from idea to currency". The middle - and back - end will become an intelligent platform, providing resources or services through various functional agents. Small business units focus on a specific track with their target customers at the center. They are composed of several functional BPs (Business Partners) and are connected to the APIs (Application Programming Interfaces) of various functional agents, linking the entire company's resources. When each business unit is connected to several functional agents and each functional agent is connected to several business units, the enterprise becomes a criss - cross network.

Business World: If all functions become agents, will the actual workload and decision - making power of human employees be compressed by AI? How should enterprises redefine the job value, performance indicators, and promotion standards of human employees to prevent employees from becoming "AI operators" or "process watchmen"?

Mu: If all functions become agents, the scale of human employees will definitely be compressed. The work of human elites will be to design and train large models, make regulatory judgments and emergency take - overs at critical moments, or do hands - on work on the front line. The compression of workload and decision - making power is inevitable, but it's not a bad thing. Humans can focus on more meaningful things that AI cannot replace.

Actually, people with work experience can more or less feel that there is "fluff" in the work content of many positions. Bosses also know that half of the labor cost is wasted, but they don't know which half. Is the enterprise's performance management system useful? Most Chinese enterprises haven't made good use of it. Musheng Consulting has an original indicator called the "real incentive index", which is used to measure the proportion of the actual variable part in the total performance score. Large - sample surveys show that since 2020, this indicator has been around 5% every year. In other words, when employees' performance is scored on a 100 - point scale, the results are basically between 60 and 70 points.

In this "equal sharing system", real hard - workers cannot be identified. What's the meaning of defining job value and performance indicators? And what are the promotion standards? However, if all functions become agents, the value of employees working around these agents will be more easily manifested. Their incentive rewards and career advancement will be clearer because the process of their efforts and the results of their contributions are all fed back by precise numbers.

Business World: Large enterprises provide employees with Tokens and AI tools, but employees only regard them as efficiency tools and are reluctant to let AI change their job boundaries. Enterprises hope that employees use AI to reshape the organization, but employees are worried about being replaced. How should enterprises design incentive mechanisms to make employees willing to actively expose processes that can be optimized by AI instead of hiding their real work?

Mu: First, we need to clarify a concept. Using AI to reshape the organization, that is, building an agent organization, is itself an organizational change and a re - distribution of responsibilities, rights, and interests, which is bound to be impossible to be accepted by all employees. The question of "how to make employees accept" doesn't make much sense because "employees" are not a unified whole.

To carry out organizational change well, on the one hand, more opportunities should be given to top elites, providing them with a stage to display their talents and earn incentives, making them the vanguards of change; on the other hand, humanistic care should be given to the bottom employees, enabling them to land smoothly without becoming a resistance to change. Once these two ends are set, the middle - level employees will automatically divide. Some will strive to become vanguards, and some will withdraw in time, each getting what they want.

The key to organizational change is to let top elites stand out, which requires the injection of "market - based incentives". The market can best create heroes. In the design of the platform - type organization, those with market sensitivity study customers, design products, and organize deliveries in the front - end business units. After making money, they share the profits with the enterprise in proportion and then distribute them internally according to virtual equity. This is a "partnership"; those with professional functional abilities provide professional services in the middle - and back - end and charge fees according to the service content and effects, perhaps in the form of energy contracts. This is a "transactional relationship". Whether it's a partnership or a transactional relationship, they are all market - based relationships. Capable people can earn far more than their salaries. This is their greatest motivation to participate in the change.

Business World: Many bosses, executives of large Internet companies, or well - known scholars don't emphasize the so - called "market - based incentives" you mentioned. Instead, they think that elites should be self - governing and self - driven, and the highest incentive should be to challenge themselves and achieve great careers. How do you view this view?

Mu: This kind of view is a bit like asking "Why don't they eat meat?", and I'm not going to argue about its right or wrong because those who say these words may not even believe them themselves. I want to talk about why such views exist.

For enterprise bosses, they want to achieve "free acquisition" on the basis of avoiding the pain and risks of organizational change, using a few slogans to arouse employees' enthusiasm. For enterprise executives, they neither want to think about solutions nor hope that market - based incentives will weaken their power. So the best incentive is to "divide the money in sacks".

Some scholars don't study organizational and human resource management and don't want to cross - discipline to challenge. But they need to form a closed - loop for business model and strategic theories. Then this simple view of "working for the love of it" is the most suitable. They won't discuss the details of organizational structure and incentives but simply mention "networking" and "self - driving". They will also give extreme cases of employees "working for the love of it" and "creating miracles" to support their views.

Since scholars are willing to say, bosses and executives are willing to listen, and the media is willing to promote concepts and stories, such views have become very popular.

Business World: Let's get back to a more practical incentive mechanism. If employees use AI to compress their original 8 - hour work into 3 hours, should the enterprise increase tasks, reduce the number of employees, increase salaries, or let employees take on more complex work? If the distribution mechanism is not well - designed, will it lead to employees' resistance to AI?

Mu: Most enterprises will arrange other work for the saved 5 hours, which leads to a phenomenon that employees secretly use AI tools and "slack off" with the time saved by AI. When bosses find out, they are unhappy and start laying off employees. A good thing turns into a bad thing.

This situation is the boss "dealing with" employees. However, after the enterprise transforms into an agent organization, this problem will be easily solved.

AI tools only improve individual productivity, not organizational productivity - it is not reflected in the enterprise's final business results. The productivity burst in a certain position is still consumed in meaningless division of labor, processes, and approvals. Even worse, the burst of productivity may even block the enterprise's operation, just like a weak person vomiting and having diarrhea after drinking a tonic soup.

The significance of the agent organization is that it enables all AI agents to aim at creating value for end - customers. Those that cannot create value will be squeezed out of this network. At this time, why would the boss care about your working - time problem and assign you specific work tasks? At this time, the boss and employees work together to "deal with" the market, make money, and share the profits together.

Business World: When applying agents or AI, there will be problems of cognitive boundaries and work experience. Experienced employees are much better than new employees in terms of usage efficiency and results. Will this make it more difficult for enterprises to recruit new employees and lead to the "aging" of enterprises?

Mu: There is no such thing as aging. Those with young minds are the "young people".

What kind of thinking is young? Based on understanding the status quo, being unconventional and innovating ingeniously. Experienced employees are more efficient when using AI. On the one hand, they understand the status quo, have accumulated experience, and have a "feeling" for certain fields; on the other hand, AI creates infinite possibilities, allowing them to break through boundaries. In the past, they didn't dare to put forward ideas because no one would implement them.

Let's take a specific scenario. An employee proposes: "Why not do this in this way?" The leader waves his hand and says: "Good idea. Please take the trouble to write a plan!" After going through this a few times, who would still dare to put forward creative ideas? Naturally, they become "old timers". But now it's different. With AI, as long as you have a creative idea and can provide prompts, there will be outputs.

There have always been many absurd views on the issue of age. Those enterprises that make a one - size - fits - all decision based on age either have too many talents and think it doesn't matter to "misjudge", or are too incompetent and have no other way to identify talents, so they can only use age, an obvious label.

I've also seen many enterprise bosses blindly worship the young, especially in the era when the "Internet thinking" was in full swing. They adored young heroes and humbled themselves to the dust. Coupled with the fueling of some media, they shouted "Learn from the young" and "Surrender to the young". This phenomenon even became a fashion at that time. However, when facing such "submission", the young may not respect the "oldies" but are more high - spirited and determined to trample on them. In fact, this is completely unnecessary. As I said before, those with young minds are the "young people". Youth itself is a natural phenomenon and is not worthy of admiration. What we should admire is the "young thinking".