How would organizations / enterprises change/evolve in the AI era?

If we continue to develop AI at the current speed, what will our organizational structure look like in a few years?

First, the scale of enterprises may be further reduced in the future, because the human resources required to complete the same business may be reduced. This has already appeared in many small startups with AI native as the core concept. Especially in Silicon Valley, some investors prefer to support efficient small teams, believing that such a model can adapt to market demand more quickly. From the overall trend, the future organizational form may develop towards a "people + AI" collaborative model. The popularity of AI tools has made many tasks that traditionally require a lot of manpower easier. This does not reduce the importance of people, but frees people from repetitive tasks and focuses more on creative and decision-making work. This also simplifies the organizational structure. For example, large projects that used to require dozens of people to complete may now be efficiently promoted by a small team of three or five people with AI tools; medium-sized teams may also be reduced to two or three people, or even one person in charge of certain departments, and complete collaboration through AI technology. This not only reduces communication costs, but also makes team operations more flexible and efficient. At the same time, AI's auxiliary role in transactional decision-making, information transmission, resource coordination, etc. is becoming more and more obvious. This change not only changes the organizational operation mode, but also allows enterprises to find a new balance between cost and efficiency.

Enterprises that have not completed digital transformation will directly become intelligent in the process of AIization.

The core of digitalization is to upload data online, such as converting voice into digital assets, and then manage through these data. Digitalization is very helpful to the development of AI because it provides a large amount of data accumulation for AI. Tools such as Slack, Microsoft Teams, and Google Workspace have accumulated a lot of corporate data, and AI can effectively analyze and manage based on these data. However, for many small and medium-sized enterprises, they have not yet completed digital transformation, and AI may be more helpful to these enterprises. In the past, small and medium-sized enterprises did not lack data, but lacked systematic data management tools. In the past, processing unstructured or non-standardized data required a lot of manpower and capital investment, and the cost of IT department intervention was also high. But now, with the development of AI technology, these thresholds have been greatly lowered. In other words, some large enterprises have digitalization and then naturally turn to intelligence, but many small enterprises can skip digitalization and directly enter the stage of intelligence.

In the past few years, some enterprises have performed well in digitalization, usually relying on strong R&D teams and technical capabilities, and have opened up a gap with small and medium-sized enterprises. However, with the popularization of AI, the threshold is being lowered, and tools are becoming more universal and intelligent. So, where will the competition dimension between a good small and medium-sized enterprise and an ordinary enterprise be in the future?

From a technical perspective, the development of AI has indeed narrowed the gap between enterprises with resources and those without resources. On the contrary, large enterprises are often reluctant to make major adjustments due to the path dependence and sunk costs of existing systems. Even if better AI systems are available, they may hesitate because they involve department adjustments and resource reorganization. In contrast, small enterprises are more flexible. In the past, a large IT team may have been needed, but now only two or three people can complete the operation of the entire system with AI tools. For example, a large property company has only a dozen people in its IT department, but it has realized many functions through AI. In the past, it required the support of algorithm engineers, and some even required a doctoral degree to be competent. Now, with the help of the capabilities of big models, such as natural language processing and data processing, employees may only need to know how to adjust the API and how to describe the scenario clearly to complete what an algorithm team could do before. The key point of competition lies in "people". Whether small and medium-sized enterprises can attract "super individuals" who understand both AI and business determines whether they can seize the initiative in the era of intelligence. In fact, many popular AI products come from small companies, not big companies. Even those launched by big companies are usually completed by small teams. This shows that resource requirements are declining sharply and scale advantages are weakening. Nowadays, as long as you access the API of the big model, you can use ordinary computers to achieve many functions, which has brought unprecedented opportunities to small businesses.

In the next 2 years, what kind of changes will AI drive in the organizational structure of enterprises?

I think from the perspective of existing technical capabilities and the practices of some enterprises, the trend of smaller organizational scale is clear. However, many enterprises currently do not fully understand the difference between generative AI and traditional AI. Generative AI brings a new logic, and this cognitive gap may directly affect the ability of enterprises to seize opportunities. In theory, transformation can happen quickly, but in reality it is not easy, especially for large enterprises. They already have mature customers, businesses and organizational models, and it is difficult to transform quickly while maintaining existing businesses. In contrast, many small enterprises are more flexible. From the beginning, they may think about "Can I become an AI native enterprise?" when planning their business. I have seen some entrepreneurs who first ask when doing business: Can AI do this? Even though the efficiency of AI may not be as good as manual work at present, they are willing to incorporate AI into the workflow. In this way, when the technology is upgraded, the results will be immediately apparent. For example, out of ten things, AI may already be able to do 65% well. Although there are still some things that are not ideal, they are still working in this direction. In the process of building AI-native enterprises, they have begun to achieve results. For enterprises and individuals, this idea is both a driving force and a pressure. Individuals will also go through a similar process when using AI. At the beginning, they may feel that AI is not easy to use, but the problem is often not the technology, but the lack of cognition. Once we improve our own cognitive level, even if the technology has not been upgraded, we will find that the effect of AI will exceed expectations.

Share some impressive cases that show the application of AI in small and medium-sized enterprises

For example, in multimodal technology, some companies use AI for production safety management. Through video surveillance, AI can automatically analyze potential safety hazards. To put it bluntly, it is to let AI have a basic understanding of common hazards, and then slightly adjust the rules according to specific needs. This solution is low-cost, but the effect is critical. In the past, it may be necessary to arrange several shifts of people to patrol every day, but now AI can do it, and the efficiency improvement is very obvious.

For example, the application of AI in the field of customer service is also very representative. In the past, customer service recordings were mainly archived, and at most some keywords were extracted, but these keywords were still quite different from the real expression, and basically could not be used. Now, with AI, when encountering customer complaints, AI can directly analyze where the problem lies. For example, customer complaints are sometimes due to inadequate service by employees, and sometimes they may be unreasonable by customers. In the past, such problems were often handled opaquely. The intervention of AI makes decision-making clearer and fairer, and reduces many unnecessary conflicts.

In addition, AI's performance in terms of rules and flexibility is also worth mentioning. We have done a study before. After introducing the company's values into the AI system, AI can be closer to the company's orientation when making decisions. For example, the headquarters of a chain restaurant company may emphasize "customer first", but in some stores, the implementation may be different. However, if AI is used to describe the specific scenario clearly and then set the handling principles, the suggestions given by AI will be more consistent, and the execution efficiency will be higher, and the public opinion risk will be greatly reduced.

Another point is AI's learning ability. As long as a few typical cases are accumulated, AI can gradually optimize the decision-making logic. For example, when encountering emergencies, AI can comprehensively consider factors such as weather and traffic and give reasonable handling methods for late arrival and early departure. Compared with traditional rigid rules, this handling method is more flexible and more humane. These examples show that AI can indeed help companies find a balance between rules and efficiency, while improving overall management methods.

AI technology is now widely used, and it can greatly improve efficiency. However, this efficiency improvement often seems to be more beneficial to senior management or the company as a whole. For ordinary employees, how can they really enjoy the technological dividends brought by AI?

For example, after the introduction of AI, can the company develop new growth points? This is an expansion demand. This expansion does not happen naturally, and requires the joint efforts of the company and employees. We have seen that some traditional industries have begun to get involved in AI-related businesses after the introduction of AI, which is the most ideal state. The company leaders have ideas, and employees will also put forward different ideas and ideas when using AI tools, so that employees have the opportunity to represent the company in various innovation and entrepreneurial activities. But in reality, another common problem is "involution". The current competitive environment is indeed very volley, but with AI tools, at least everyone does not need to be so volley. For example, a task that used to take 10 hours of work may now be completed in 6 hours. The problem is that many bosses still want employees to "volley", with fewer people and more things to do, and the working hours are still very long. However, if AI technology can be popularized and become the new normal, employees will be more efficient and work less, and the rhythm of the whole society will also be adjusted.

Transparent

Another major advantage of AI is that it makes employees' achievements more transparent. In the past, bosses often measured employees' efforts based on overtime hours, but with AI, the output of knowledge assets has become more important. If an employee can complete a task that used to take 10 hours in 2 hours, the boss will obviously pay more attention to efficiency rather than continue to rely on overtime hours to measure contribution.

Yes, this is very beneficial to efficient employees. If my work efficiency is more than ten times higher than before, and the company still wants to evaluate it in the old way, then such employees may not continue to stay. They have a strong choice space and can become "super individuals" and survive anywhere.

In the future, everyone will have their own AI Agents, and each department will have its own AI. Can these AIs communicate with each other? For example, can my AI and my colleague's AI work together?

This may change the mode of organizational communication. For example, there is a management concept called "employee advice", and many companies will encourage employees to put forward ideas and suggestions. Employees may have good ideas when they are in contact with the front line. Some companies have achieved good results by designing suggestion systems and evaluation committees. However, most employees may not take the initiative to make suggestions, thinking that they are useless or afraid of saying the wrong thing. In the future, everyone will have their own Agent. For example, an employee's AI can collect his ideas and pass them to the CEO's AI, and the CEO's AI will filter this information and match it with the company's strategic goals. If a suggestion is particularly in line with the strategic direction, the CEO's AI will notify the boss and call the employee to learn more about it. This model can greatly improve the efficiency of information flow, and at the same time, through privacy protection design, it can prevent sensitive information from being over-exposed. To some extent, it makes the information of the organization more transparent. The current process is that the leader holds a meeting to convey the information to the middle level, and the middle level then conveys it to the lower level. The information in the middle is easily distorted. But if the leader tells AI directly, AI will convey the task according to the clear goal, and employees can communicate directly with AI if they don’t understand, avoiding misunderstandings.

Nowadays, many young people no longer simply focus on salary when choosing a company, but pay more attention to the company's organizational culture and values. What they care about is whether they can feel happy at work and whether they can realize their personal values. Under this trend, will AI bring some changes to corporate culture, especially the cultural atmosphere of small and medium-sized enterprises?

The introduction of AI can more effectively promote the implementation of organizational culture. Every company has an organizational culture, but many times these cultures are difficult to truly implement. For example, the culture promoted by some companies during recruitment is very attractive, but employees find that the actual situation is not the case after joining the company. With the help of AI, the culture declared by the company can be more consistent with the actual culture. By analyzing a large amount of text data within the company, we can intuitively understand its cultural attributes. For example, a company claims to focus on cross-departmental collaboration, but if the meeting minutes show a large number of blame-shifting behaviors, it means that there are problems with the culture. The formation of culture requires continuous accumulation and iteration. If bad behaviors are not corrected and good behaviors are not encouraged, culture will be difficult to implement. AI can make behaviors transparent, such as using reasoning to remind employees which behaviors do not meet cultural requirements, thereby gradually guiding the shaping of culture. But AI cannot determine whether a culture is good or bad. If the company's orientation is to oppress employees or suppress different voices, AI may strengthen this bad culture; on the contrary, a positive orientation will make AI a tool to promote a good culture. For example, Google collects employee data not only to improve productivity, but also to balance employee experience. Amazon focuses more on efficiency, and the collected data is completely used to serve productivity goals. If data collection is used to measure efficiency, employees may feel oppressed; but if it is for employee health, such as the system finds that someone needs a vacation and automatically arranges it, employees will feel cared for.

Therefore, the premise is that the objective function must include the dimension of employee experience. You can add indicators such as "employees' weekly overtime hours cannot exceed 15 hours" to your objective function. In this way, employees will feel that their work situation is being paid attention to, and management will be more caring.

Some people who have studied AI in depth, doing AI innovation within the company or starting their own business independently, which path is more common?

At present, whether it is internal entrepreneurship or independent entrepreneurship, in fact, there are not many, this is more of a direction. Some supporting mechanisms are needed here. The popularity of AI tools has indeed reduced the demand for resources, so many employees will think, why not just jump out and do it themselves? After all, in terms of revenue, the gap between internal entrepreneurship and independent entrepreneurship is still quite large. For enterprises, the key is to design good systems and incentives, give employees a reasonable choice space, so that they are willing to stay in the company to start a business, and can see their own development prospects. They are "super individuals" rather than "super employees", because these people do not necessarily have to be affiliated with a certain company. They can be independent, have their own ideas and goals, and are not limited to the platform within the company. Now many excellent AI products are actually made by independent entrepreneurs, rather than from projects within large companies. The reason is simple: the resource threshold has been lowered. In the past, if I was doing hardware, I had to pull a lot of people to start a business, and none of the marketing, product, and development teams could be missing. But now it is different. For example, I know hardware, but I need some software support. If I know a little bit, I can directly use AI tools to get it done, use AI to generate code, design UI, and even make a prototype. In the past, these required a complete team, but now two or three people can use AI to get it done. For example, I pull a friend and use AI to make a demo within a few days, and then I can test the market and even find investment. After this method is implemented, small teams can do it on their own without relying on large companies to provide resources.

Therefore, for companies, how to give employees enough incentives so that they are willing to keep these achievements within the company instead of going out to start a business directly becomes particularly important. In the traditional model, the company provides a lot of resources, and the achievements naturally belong to the company. But now it is different. In fact, half or even more of many achievements are achieved by employees' personal abilities and creativity. Companies must redesign their mechanisms to retain these "super individuals" and make their creativity truly serve the company instead of being snatched away by the market in the end.