At a time when AI technology is developing rapidly, we are standing at a watershed moment in an era. Just as the Internet has revolutionized production and lifestyle, the rise of AI has also brought great disruption and opportunity. Starting from the core feature of “intelligent low-cost centralized supply”, this paper delves into how AI can subvert traditional production models and organizational structures, and how it will give rise to new production relations.
When a big technology rises, some existing production models are destined to be destroyed, and at the same time, some new models will begin to be established.
In the Internet age, this combination of destruction and establishment is department store VS e-commerce, SMS VS WeChat, etc.
In today’s rapid rise of AI, it is obvious that it is also necessary to understand what is the role of department stores in 2000 and what is the role of e-commerce in 2000.
Finding the combination of destruction and establishment that belongs to AI is to identify risks and opportunities in the intelligent era.
And these have little to do with people’s preferences to return to the fundamental characteristics of big technology.
This technical feature is simply summarized:
The Internet is characterized by connectivity, while AI is characterized by intelligent low-cost centralized supply.
All predictions for the future actually have to go back to this starting point.
Intelligent centralized supply
In the past, intelligence presented a state of decentralized supply in most scenarios, such as programming required N programmers, N doctors to see a doctor, and N lawyers to fight lawsuits, and everyone to deal with their own affairs.
The way of organizing production is actually adapted to this state of intelligent decentralized supply.
In reality, it is manifested as: either managing each other, or defining roles and processes to cooperate with each other.
To achieve these three challenges, product managers will only continue to appreciate
Good product managers are very scarce, and product managers who understand users, business, and data are still in demand when they go out of the Internet. On the contrary, if you only do simple communication, inefficient execution, and shallow thinking, I am afraid that you will not be able to go through the torrent of the next 3-5 years.
View details >
A large number of measures in the company (processes and tools, mission, vision, values, etc.) can be seen as an aid to the intelligence of coordinating decentralized supply and achieving the set goals.
This is clearly not the case with AI.
The direction of general intelligence is that it is one or ten thousand, less or more.
No coordination is required, only the conditions for it to be effective.
Obviously, large models cannot only be used for a single thing, such as programming.
What does this lead to?
This will lead to many existing production relationship designs becoming obstacles to productivity improvement and need to be overturned.
(The organizational structure can be reinterpreted from the perspective of intelligent centralized supply)
What is the new production relationship like?
New production relations after centralized supply
The principle of the new production relations is simple, you need to give full play to all the capabilities of centralized intelligence.
So what is the obstacle to its ability?
The lowest level is obviously data.
It does not know how the situation will work.
The higher level is obviously feedback and knowledge, and in a fluid world, a rigid system is useless no matter how many agents are used.
Teacher Hou Hong has drawn such a picture, which is very important:
In the past, the organizational system was used to coordinate the intelligence of decentralized supply, and the intelligence of centralized supply required not coordination but global data, feedback and dynamic knowledge system.
This is a completely different paradigm, and obviously the focus is also different.
Because they are two different sequences, there is a high probability that they cannot be changed from the existing system. Or the cost of improvement may be much greater than that of new construction.
E-commerce is not changed by department stores, and Didi is not changed by taxi companies, so this will basically be disruptive innovation.
From this point of view, “Unmanned Company” may really be the only book at present that tells the story of applied intelligence from the perspective of a new model.
Versatility
The biggest similarity between this change and the Internet is its versatility.
The Internet does not care what it is connected to, so the model of the Internet penetrates into N multiple fields.
The intelligence of AI large models is not so clear at the moment, so it is destined to penetrate into every intelligence-related field.
Suppose lawyers can change from N lawyers doing 1V1 services to providing services centrally around a large model of lawyers, while others are scattered around the system to solve obstacles to the effectiveness of the model. So what areas cannot reuse this model? (In “Unmanned Company”, this model is summarized as: intelligence first, everything is counted, real-time feedback, and central decision-making.) ）
So this period is actually a time when electronic octopuses are generally born.
The octopuses are still very small, but they have basically begun to grow and develop.
The octopus’s brain is a large model, and the octopus’s tentacles have domain attributes.
From this point of view, the depth of this destruction far exceeds that of the Internet (I don’t know if there are still people who think that the impact of AI is not as good as the Internet). ）
Core Challenge
The core challenge is actually that the model of intelligent centralized supply is contrary to common sense and anti-Internet.
The Internet has accumulated a series of methodologies, such as single-point ultimates, such as rapid iteration, but these successful experiences are all obstacles to the model of intelligent centralized supply.
For example, some students may say:
Today, I finally solved a customer’s problem, which made me feel at ease.
This is wrong because it results in infinite improvements to taxi companies without leading to jumping to Didi ride-hailing.
It will also lead to better department stores, but not e-commerce.
Infinitely close to a certain limit value does not mean that it can be broken.
The so-called rapid iteration in the field can only lead to infinitely close to a certain extreme.
The single point ultimate is similar, under the premise of intelligent centralized supply and rapid pull-up, there is no tool-type single point. Because general intelligence itself is the biggest tool. In the past, decentralized tools were the result of intelligent decentralized supply.
All products and methods produced by intelligent decentralized supply are destined to disappear because of intelligent centralized supply.
Now every time I encounter this point of view, I have a new saying: I can read the unmanned company again……
Hypothesis of intelligent jumping
The growth cycle of electronic octopuses will also be more contrary to common sense.
Its core nutrient is actually intelligence, and the level of intelligence is a nonlinear jump progression.
I made a deduction like this:
If we put the progress of AI since 1956 on the timeline, we will find two distinct characteristics:
First, the progress of AI level is leapfrog. Pattern recognition to AI large models are deep learning, but this ability is difficult to describe as a multiplier increase, and it is more like jumping over a monkey and becoming a human.
Second, the jumping cycle is getting shorter.
This is not accurate, but if it holds, then:
The last jump took about ten years from 2010~2022, so the next jump will be 5~7 years after 2023, and then there may be another jump in 2~3 years.
So before 2033, there is a high probability that we will usher in two intelligent jumps like chatGPT.
Every time it is a jump from pattern recognition to chatGPT, which is very exaggerated.
And every big jump will obviously make the electronic octopus expand rapidly.
Intelligent centralized supply coupled with the assumption of intelligent jumping, you will eventually find that the typical application model of the future is actually exactly what is described in “Unmanned Company”.
brief summary
If the new words for the future in 2000 are e-commerce and social networks, each of these words implies a mode of connection; One of the new words for the future today may be an unmanned company, which contains a new production model spawned by intelligent centralized supply…
It happens that some people will ask about the difference between a one-person company and an unmanned company, and here is an answer:
The limit of enhancement is a one-person company, and a one-person company can be imagined as a person who is good at 18 weapons, and then he does the job himself. In fact, a one-person company is likely not to be able to do anything complicated, and can only do simple work. Because once the business expands, the necessary information flow will lead to people being the biggest obstacle, and people cannot process a few meters of data in seconds.
Unmanned companies are the next generation of automation, and people are not in the basic business process, but the coordinator of the business.
