The Formation and Defining Features of the Intelligent Society

We have already discussed the basic aims of education in the age of AI and the historical mission of AI general education. Those discussions rest on one fundamental judgment: artificial intelligence is taking a deep part in human cognition and social activity, and is moving human society into a new social form: the intelligent society.

We approach this important judgment from several angles. How does the intelligent society come into being? What is its historical connection to earlier societies? And what fundamental change has AI brought that gives us reason to regard it as a new social form?

This article argues that the formation of the intelligent society has deep historical drivers. The development of production and the social division of labour keeps raising the complexity of society and pushes some cognitive tasks onto external tools and systems. The arrival of AI has greatly accelerated this trend: it has made part of intelligent activity computational, so that cognitive capabilities can be stored, copied and called upon at scale; and as these capabilities enter the workflows and institutional arrangements of society, the way social activity is organized changes with them. The intelligent society is the inevitable outcome of this process.

1. Growing Social Complexity and Cognitive Externalization

As productive forces develop, social structures and processes grow ever more complex: production steps become finer, organizational ties wider, and institutional designs more varied. Existing tools and forms of cooperation gradually approach the limits of what they can do, which pushes people to improve their tools and reorganize. With these new tools and forms of organization, traditional modes of production are broken through and productive forces are set free.

The freed productive forces in turn raise the complexity of society further. Mechanized production and modern transport, for example, made larger factories, longer-distance trade and a finer industrial division of labour possible. The computer greatly increased the speed of information processing and transmission and gave rise to modern financial trading networks. Old difficulties were eased, and a more complex society took shape in the process.

Social development thus follows a cycle that keeps moving forward: growing complexity pushes existing tools and forms of organization toward their limits; new tools and systems break through those limits; and the widened scope of activity builds an even more complex society. The process is clearly cumulative and self-reinforcing. As long as society keeps developing, the need to push past existing limits of capability will keep arising, and increasing social complexity becomes a historical trend that is hard to reverse.

Alongside this process comes a redistribution of human cognitive activity. The knowledge, memory, computation and coordination that a complex society needs increasingly exceed what any individual can carry alone. People therefore rely on external media and systems, handing some cognitive tasks over to external tools and systems.

Writing and books preserve knowledge, lightening the burden on individual memory and allowing experience to outlast a single life; calculators take over arithmetic, letting people handle quantities far beyond mental calculation; navigation systems provide positioning and route planning, so no one has to remember every road. In these activities, specific parts of memory, calculation and judgment are allocated to different external tools.

We can summarize this process as "cognitive externalization": some cognitive tasks are carried by external media, tools and systems, and cognitive activity is accomplished through cooperation between people and external resources.

It is worth noting that cognitive externalization is itself part of social development. Written records support the accumulation of knowledge across time and place; computational tools support more complex engineering and science; scheduling systems support coordination among large numbers of people and facilities. Tools help people complete existing tasks and also create the conditions for activities that were once hard to undertake. Cognitive externalization therefore helps drive social complexity, and a more complex society in turn creates the need for it to develop further.

Artificial intelligence has developed within precisely this historical process. It continues humanity's long-standing direction of extending cognitive capability through external systems, and pushes the redistribution of cognitive tasks between people and machines into a new stage.

2. The Computationalization of Intelligent Activity

If AI is regarded as a tool like the steam engine, electricity or the computer, its arrival simply makes society more complex and cognitive externalization more pronounced. Yet AI's fundamental contribution is not higher capability or efficiency, but a "general simulation" of human cognitive capabilities.

The Redbook describes this process as "the computationalization of intelligent activity": turning human activities such as learning, recognition, reasoning, judgment, generation and planning into computational processes that can be represented, constructed, run and tested. This means that machines can acquire cognitive capabilities "similar" to those of humans, enter cognitive settings in which previously only humans could take part, share work with people, and replace them in some parts of it.

Crucially, by imitating human learning, machines can accumulate and update knowledge and skills; by imitating human reasoning, they can think in depth, plan tasks, and make and use tools. These "meta-capabilities" drive AI's "general simulation" of human cognition, rather than isolated capabilities such as memory or recognition.

Simulation, however, is only the first step. Once machines acquire cognitive capabilities comparable to those of humans, their comparative advantages over people quickly become apparent, such as continuous accumulation, transferability, and speed of computation.

Continuous accumulation: human capabilities are bound to the course of an individual life, and when a person leaves, experience and practised judgment that were never fully articulated may be lost with them. By contrast, some of the cognitive capabilities that machines form can be preserved and copied through models and related systems, outlasting any particular device.

Transferability: once a system has learned, its model can be copied and deployed to other machines with the necessary operating conditions, giving them the same capability without each having to go through the same learning process from scratch.

Speed of computation: compared with people, computers already have an advantage in speed, and parallel computing lets them trade resources for even more speed. This provides the basis for supplying cognitive capability at scale.

3. The Computationalization of Social Activity: The Formation of the Intelligent Society

"The computationalization of intelligent activity" allows some cognitive functions once performed mainly by people to be realized as computational systems and, through copying and deployment, supplied at scale. These computational cognitive capabilities can enter social activity in bulk and affect how society runs. The Redbook describes change at this level as "the computationalization of social activity".

The computationalization of social activity involves a chain of linked transformations: real-world objects are turned into data, historical experience is organized into models, social goals are expressed as metrics and constraints, model outputs enter workflows, and organizational rules then give those outputs corresponding force. Computation thereby takes part in allocating opportunities, scheduling resources and shaping social action.

Take recruitment as an example. AI can help staff read résumés and organize materials. If an organization writes AI scores into its screening rules, the model's output acquires real power over people's job opportunities. In this example, the model takes over a screening judgment that people used to make, and the organization goes on to build the model's output into its hiring rules. The computational realization of a cognitive function thus enters the structure of how society operates.

When AI capabilities spread to large numbers of organizations and are linked through networks, the computability and deployability of cognitive capabilities move from the individual scale to the social scale. How tasks are divided, through which channels information flows, at which step decisions are made, and who bears responsibility may all shift as a result. At the same time, services and forms of cooperation once held back by processing capacity and organizational cost become feasible.

This is the key step from technical progress to a change in social form. The computationalization of intelligent activity provides the capability base for machines to take part in cognitive labour; the computationalization of social activity embeds that capability in organizations, platforms and institutions, making it a normal condition of how society runs. The two processes of computationalization interlock and act together, ultimately producing a new social form, which the Redbook calls "the intelligent society".

As the Redbook defines it, the intelligent society is "a social form in which human knowledge production, information dissemination, judgment formation and social action are continuously mediated jointly by AI systems and institutions". This definition emphasizes the structural link between AI and the workings of society. People increasingly obtain materials, understand problems, form judgments and act through intelligent systems, and society increasingly allocates resources, organizes cooperation and assigns responsibility around these systems.

In this sense, the formation of the intelligent society rests on solid historical drivers. Growing social complexity keeps generating new processing needs, the computationalization of cognitive capability opens new possibilities, and the copying of capabilities and their embedding in institutions drive expansion in scale. Together these forces give the intelligentization of society a sustained momentum.

4. Six Defining Features of the Intelligent Society

First, AI becomes infrastructure.

Once cognitive capabilities can be copied and called upon at scale, they gradually become a basic condition on which many different activities depend. AI can be embedded in the workflows of offices, education, research, production and public services, continuously carrying out recognition, screening, generation, prediction and scheduling in the background. When people use a service, they need not deliberately launch an AI tool; they may already be relying on its capabilities.

Becoming infrastructure means that AI's influence reaches beyond its active users. Even a person who never uses AI directly may find that the information they encounter, the services they receive and the evaluations they are subject to have passed through intelligent systems. The reliability, accessibility and operating rules of AI thus become matters of public significance.

Second, social processes become data and models.

For social activity to enter computational systems, it has to be represented as something those systems can process. People's experiences, the operation of organizations and the results of activities are recorded as data; relationships in historical material are used to build models; and complex social goals are turned into computable metrics and constraints. Society thereby gains a much larger capacity for observation, analysis and prediction.

These transformations also involve choices. Which experiences are recorded, which differences are ignored, and what counts as success all affect what a system can recognize and what it optimizes. When model outputs go on to influence the allocation of resources, people and organizations also adjust their behaviour around the relevant metrics. Data and models thus help shape the very social processes they describe.

Third, cognition and action become hybrids of human and machine.

The computationalization of intelligent activity allows posing questions, selecting materials, generating options, forming judgments and carrying out actions to be redistributed between people and machines. A piece of work may have its direction set by a person, its materials processed and suggestions offered by a system, and then be reviewed by a person and executed with the system's help. In the process, people gain extended capabilities and also develop an ongoing dependence on external intelligence.

This hybridity changes the basic unit of analysis. We need to look at the whole process: who poses the question, who chooses the grounds, who sets the goal, who retains the power to decide, and who can find and correct errors. The final result is produced jointly by people and machines, and the actual role of the human has to be understood through these specific steps.

Fourth, social connection becomes mediated by platforms.

In large-scale social connection, platforms organize the flow of information, services and attention through interfaces, rules, ranking and recommendation. AI further takes part in matching needs, filtering content and arranging interactions, so that vast numbers of relationships can be formed and adjusted instantly.

Platforms thus influence what people can encounter, whom they connect with, and which things gain attention more easily. The visibility of information and the availability of opportunities are increasingly shaped by these arrangements. Connections between people therefore also carry the effects of platform goals, technical rules and organizational interests.

Fifth, systems evolve through feedback.

The outputs of intelligent systems influence human behaviour, and that behaviour in turn produces new data that becomes the basis for further analysis and system updates. Recommendations, for example, shape what users see, while users' clicks and dwell time feed into later recommendations; evaluation metrics shape how organizations work, while organizations' adjusted performance becomes material for the next round of evaluation.

Systems and society thus form an ongoing feedback relationship. This relationship can help systems find problems and improve their services, but it can also reinforce certain biases in a loop. Evaluating a system requires looking at how, after sustained use, it changes people's behaviour, organizations' incentives and subsequent data. The initial design and one-off testing can hardly capture this long-term evolution.

Sixth, technology, institutions and values become deeply coupled.

Computational systems always run toward certain goals, and setting those goals involves value choices. Demands such as improving efficiency, extending coverage, ensuring fairness and protecting privacy have to be arranged in concrete settings. Institutions determine what information a system may process, what tasks it may take on, and how much force its outputs carry.

The same scoring model can be used to offer a reference, or it can be granted the authority to screen automatically, and its social consequences differ accordingly. How a technology works is always tied to how organizations use it and how institutions constrain it. Building an intelligent society therefore requires discussing model capabilities, rules of use and public values together, and establishing corresponding mechanisms for review, correction and accountability.

These six features are interconnected. Infrastructure supplies capability continuously; data and models bring social objects into computation; human–machine hybridity redistributes cognition and action; platforms organize social connection; feedback drives the ongoing evolution of systems; and institutions and values run through all of them, jointly determining how these capabilities act on people's lives. Together they present the operating structure of the intelligent society.

5. The Cognitive Predicament and Cognitive Autonomy

Once we understand how the intelligent society forms and what features it has, the cognitive predicament within it also comes into view.

AI expands people's ability to acquire knowledge, understand problems and complete tasks, while also distributing more steps of the cognitive process into external systems. People can reach conclusions more easily, but it is not equally easy for them to understand where those conclusions come from, what evidence supports them, and where they apply; they can complete tasks more smoothly, but may not take full part in setting goals and forming judgments. How information enters our field of view, how metrics influence choices, and how system suggestions turn into action all need to be actively examined.

The Redbook analyses this new cognitive environment in terms of information overload, mixed sources, generated content, algorithmic mediation, boundary limits, governance by metrics and automated decision-making. When individuals struggle to understand and regulate their own cognitive behaviour, their capacity for judgment and their acceptance of responsibility may fall out of balance; when such imbalance becomes widespread, the basis on which society jointly discusses facts, tests what is true and corrects errors is also affected.

The Redbook therefore proposes "cognitive autonomy": the process by which individuals, in complex cognitive environments, actively regulate their own belief formation, use of evidence, allocation of trust, boundaries of judgment and responsibility for action. People need to make active use of external intelligence while retaining the ability to pose questions, probe the grounds for claims, calibrate trust, revise judgments and take responsibility. Education should help people develop these capacities, and institutions should provide the preconditions for verification, questioning and correction.

6. Conclusion

Human society is entering the intelligent society, a social form in which AI and the mechanisms by which society runs are deeply fused. In such a society, preserving, obtaining and sharing knowledge becomes enormously easier, individual capabilities are greatly extended, and people have more choices. At the same time, cognitive tasks that belong to people are outsourced on a large scale, bringing a risk of cognitive offloading; the information people encounter is mediated and filtered by intelligent systems, and the risks in decision-making rise accordingly. How to face this new social form while preserving individual agency and the capacity to take part in public life is a key question that urgently needs an answer.

The mission of AI general education in our time is precisely to help everyone understand the intelligent society that is arriving, take an active part in it, and keep their own cognitive agency.