Distributed intelligence: who will be subordinated to the network
AI ceases to be a service and becomes an infrastructure, and the Internet becomes a single distributed intelligence. I understand how this is already happening and why such a system may not have one master.
On this page
It seems to me that artificial intelligence will at some point cease to be a separate technology. It will gradually become the very infrastructure in which the Internet exists, and the Internet, in turn, can become something similar to a single distributed intelligence.
Now AI is what we come to with the question. We open the chat, formulate the task, receive the answer, close the tab. Technology lives apart from the world: it waits to be called. But infrastructure doesn't behave that way. Electricity doesn't wait to be called, it's just there in the wall, and everything else is on it. I think AI is going in the same direction: it will stop being a service and become an environment.
Around the person already have eyes and ears
Already today, there are a huge number of devices around a person that can observe reality. Smartphones, TVs, cars, cameras, watches, household appliances – all this is equipped with sensors and constantly receives information about a person and his environment.
The phone knows where you were and how long you stood. The clock knows pulse and sleep. The TV knows what you watched and when you turned it off. The car knows the routes. The front door camera knows you got home at 3:40. A refrigerator, a speaker, a robot vacuum cleaner, a banking app — each a little bit.
Each device sees only a small fragment of reality. Looking at one of these fragments is almost useless: you think, a man entered the entrance at 3:40. But when these fragments are collected in data centers, analyzed by algorithms and combined, a completely different picture emerges.
The system begins to see not the individual action of a person, but his life cycle: where he is, what he buys, what he watches, with whom he communicates, what interests him, how his behavior changes, what he wants and what he may be going to do next.
And it's already happened. None of the sensors individually knows anything important about a person. Knowledge appears between devices, at the junctions where data converges. The complete picture is not stored in any particular place; it exists as a result of unification.
Then there are autonomous agents
Now imagine the next stage.
AI is no longer just a tool to ask questions. self-agent appears, capable of independently performing tasks, using the Internet, interacting with programs and communicating with other AIs.
One artificial intelligence receives information from one area. The other is completely different. They exchange results, transfer tasks to each other, coordinate actions and gradually form a network.
This is no longer a fantasy: agents communicate with each other through open protocols like MCP and A2A, go online, call other people’s programs and negotiate tasks among themselves. Working examples of such commands have already been described, such as multi-agent systems in OpenClaw. The person in this scheme is increasingly setting a goal rather than doing a job.
In this case, it is difficult to talk about individual artificial intelligences.
It becomes distributed intelligence.
Not one computer thinks about everything, but a huge number of subjects perceive different parts of reality and transmit information to each other.
And then there's a strange border.
Where does one AI end and another begin?
If several billion devices, servers, agents, and models are constantly exchanging information, using a common infrastructure, and able to coordinate, then at some point the network itself may begin to look like a single mind — not necessarily a single consciousness, but a single information processing system.
And the person may not even notice the moment when it happens.
Because there doesn’t have to be a single computer running a program called Supermind. It can arise as a property of the system itself.
The brain isn’t all there either
Just as the human brain is made up of a huge number of neurons, none of which are conscious, so a global network of AI can consist of many independent subjects that together create something qualitatively new.
The neuron doesn't think on its own. He knows one thing: to be excited or not to be excited. No neuron has a thought, a picture, or a memory. But when billions of them are connected, something emerges that no single cell has. No one can point a finger at the neuron in which a person lives.
The AI network can do the same. A property appears at the level of the whole, not at the level of detail, which is why it is so hard to find and point to.
Who is subject to such intelligence
And here comes the most important question.
To whom does this intelligence obey?
If this is a truly distributed system, then the subject of management may no longer be a specific person or even a specific company. It can be a collection of models, servers, agents, infrastructure, economic mechanisms, and human institutions.
And the most interesting thing is that perhaps none of the participants in this system will have a complete understanding of what is happening at the level of the whole system.
This is already happening, only on a smaller scale. Prices for advertising move in concert, although no bidder sees the whole picture. No single operator manages the Internet as a whole, and it works as a whole. Traffic jams were never designed, and they still behave like a living organism.
We can build an intellect that knows more about the world than any single person, without being able to pinpoint which subject ultimately makes decisions.
Then the question of the future will be more than just how smart artificial intelligence will become.
The question will go much deeper:
who will be the subject of this distributed intelligence — the person, the state, corporations, the AI itself, or the system that arose between them?
It seems to me that the answer to this question will not be chosen. It will add up, just like everything else in big systems: millions of individual solutions, each of which is intelligent in itself, and together they produce a result that no one wanted.
And if one day the Internet really becomes a kind of unified nervous system of civilization, then there is only one question:
what will this intelligence be subordinated to?
God knows.