But once ten workers divided the labor among themselves, they could produce more than 48,000 pins a day. Even if we use 48,000 as the figure, each worker produced at least 240 times what one person could have made alone.
This is one of the most famous stories in economics. It first shows how division of labor raises output; the exchange argument that follows explains how markets allow strangers to cooperate.
Behind the pin factory is a simpler judgment: in a productive society, more people mean finer specialization, and finer specialization means higher output. Ten people do not merely multiply one person's output by ten. Through cooperation, they push output into another order of magnitude. Cities, markets, and global supply chains are all larger versions of the same logic.
But what happens next? If machines begin to take over reproducible tasks, the chain of "more people bring more production, and more production brings more social positions" no longer holds naturally. The question shifts from "how do we organize more people into production?" to "when production no longer automatically distributes social positions, how are people still needed?"
Imagine an extreme future scene: inside a windowless factory, from design to quality inspection, humans are left only in distant supervisory seats. Robots maintain robots. AI schedules AI. Once the shipping order is generated, driverless trucks move toward their destinations.
Consumers live on basic income, public transfers, and cheap goods produced by machines. They no longer participate directly in most production.
This is not a news report. It is a thought experiment that pushes automation, autonomous driving, basic income, and early deployments to their extreme. It does not require a fully automated factory to already exist. It only asks us to consider what happens when the recognition function of some production roles is weakened.
Roughly 250 years have passed since Smith wrote about the pin factory. If we push the automation trend to its extreme, this is the question:
If production no longer automatically makes us need one another, why else do we need one another?
Smith Saw Another Layer
Most people remember Smith for saying that "division of labor raises efficiency." That is the layer directly displayed in Book I, Chapter 1 of The Wealth of Nations.
But another sentence from Chapter 2, quoted even more often yet often misunderstood, pushes the issue toward exchange and cooperation among strangers:
"It is not from the benevolence of the butcher, the brewer, or the baker, that we expect our dinner, but from their regard to their own interest."
The point is not that "selfishness is good." It is something else: strangers can form stable networks of cooperation through mutually beneficial exchange.
In many premodern societies, people depended more directly on kin, neighbors, and hierarchical ties. You needed the people around you because you could not survive without them.
After the Industrial Revolution, you began to depend on people you would never meet. The people who grind coffee beans, sew clothes, or write the operating system on your phone may be in Colombia, Bangladesh, California, or somewhere else. You do not know them, but you need them every day. What they need is not you as a specific person, but the demand, market, and institutions formed by you and countless others.
This means the economy is not only about material production. It is also about how people are connected to one another: economic relations are social relations.
In the public economic narrative centered on production, employment, exchange, and growth, this question has rarely been the main question over the past two centuries, because during the age in which that narrative was most powerful, the answer seemed to require no separate explanation:
Why do we need each other?
An Assumption That Was Never Spelled Out
This essay calls that public narrative, centered on productivity, employment, and market exchange, the productivist economic narrative. Inside it is a social fact that has long been taken for granted:
Market production has long required large amounts of paid labor, so people could become visible to institutions through production roles.
World Bank data show that the global labor-force participation rate was about 65% in 1991 and still about 61% in 2024. For more than three decades, paid work has remained an important entry point through which most adults are counted and recognized by institutions.
You need the baker not because only he can bake bread, but because specialization lets him do it for you more steadily and efficiently. The baker needs you because your money can be exchanged for what he wants. You do not know each other, yet through production and exchange you enter each other's lives. This mutual need is so common that it is rarely argued for on its own.
Division of labor itself is not exploitation, but it naturally creates the conditions for exploitation. When ten people cooperate, the additional output is not the independent product of any one person, and each person's contribution is hard to quantify precisely. How the surplus is distributed is therefore not merely a technical issue. It becomes an issue of power, rules, and narrative.
If the people who organize production, own capital, or control platforms mainly credit the cooperative gain to themselves, while reducing participants to wages, performance metrics, and replaceable positions, division of labor can shift from a mechanism of "needing one another" into a source of inequality.
"Being needed" has several layers: survival interdependence, market demand, institutional recognition, subjective meaning, and irreplaceability in intimate relationships. This essay is especially concerned with institutional recognition: whether public institutions see, name, evaluate, and accept a form of contribution. Meaning is the echo of that recognition inside an individual. What employment, taxation, consumption, and statistical systems capture most easily are still workers, consumers, or taxpayers.
Marx saw the other side of this point. Commodity fetishism shows how social relations among producers can appear, through commodities and exchange, as relations among things. Alienated labor reminds us that this kind of connection is not the same as full recognition.
This essay adds another layer: even when this connection carries alienation, it has still given modern people a visible social position at the institutional level. You may not be recognized as a whole person, but at least you are recognized as a worker, a professional role, or a link in an exchange chain.
Keynes vaguely saw the next step. In 1930, he published "Economic Possibilities for our Grandchildren," imagining that, assuming no major wars or large population increases, developed societies might come close to solving the problem of scarcity in the survival sense by around 2030. Remaining work might be shared out in three-hour shifts or fifteen-hour workweeks, delaying the adjustment problem caused by leisure.
The truly difficult question would become: after being freed from economic pressure, how should people use freedom, arrange leisure, and live wisely and well? Keynes even worried that humanity might face a broad crisis of adaptation.
He was broadly right about the direction of rising living standards in developed countries. On the fifteen-hour workweek, he was plainly too optimistic. But his question about leisure, freedom, and meaning can be translated here into the question of social recognition in the age of AI.
AI Does Not Merely Replace Labor
When people discuss AI, they tend to place it inside the narrative frame of the previous round: machines replace workers. That frame was inherited from the Industrial Revolution, as a continuation of the story in which looms replaced textile workers.
But AI is something else.
The steam engine replaced muscle, but it still needed people to operate, design, maintain, and decide. Computers replaced memory and calculation, but they still needed people to program, judge, and assign meaning. Every technological revolution has taken away part of human capability while rearranging the division of labor between humans and machines.
Generative AI significantly expands the range of judgment, creation, decision-making, and interactive tasks that can be automated or assisted - tasks that were once often considered closer to what makes humans human. The ILO's 2023 report and its 2025 ILO-NASK update both warn that generative AI exposure indicators are more like early warnings than unemployment forecasts; most jobs are more likely to be transformed than to disappear entirely.
The IMF's 2024 analysis estimated that about 40% of global employment is exposed to the influence of AI: about 60% in advanced economies, about 40% in emerging economies, and about 26% in low-income economies. This includes both complementarity and substitution.
These are estimates of task and occupation exposure. They do not equal actual job disappearance.
In "The Endgame of Programming: From Human-Machine Compromise to Silicon-Based Awakening," I discussed a similar direction: programming is shifting from "machine-centered logical construction" toward "intent-centered probabilistic generation." The relationship between AI and the steam engine or the computer is not merely quantitatively stronger. It is qualitatively different.
Every technological revolution redraws the boundary between humans and machines. This round is different: the boundary begins to pass through "meaning."
This means the revolutionary character of AI is not only about "which occupation disappears." The more important question is whether, once AI is combined with industrial robots, logistics robots, algorithmic management systems, and enterprise deployment, more and more standardized production processes will reduce their human density.
Data from the International Federation of Robotics in 2025 already show that in 2024 the global operational stock of industrial robots was about 4.66 million units, with annual installations exceeding 540,000. This is not a conclusion about employment, but it is a factual anchor showing that production processes continue to automate.
These data do not form one single causal chain. They only show that cognitive task exposure and physical production automation are advancing at the same time. What this essay cares about is not a reality that has already fully arrived, but a trend that may expand: some processable production links may increasingly rely less on direct human labor participation.
The Crisis Is Not Only Unemployment
When discussing AI unemployment, one common answer in thought experiments is UBI - universal basic income. If machines create abundance, distribute that abundance to people. This sounds reasonable, and it may create room for care, learning, and public participation, but it does not touch the whole problem.
UBI can provide an income floor and relieve part of the distribution problem. Income security is necessary, but it is not sufficient. If it is only an individual cash transfer, it mainly solves purchasing power. It does not automatically generate roles, responsibilities, mutual dependence, or public recognition.
The crisis this essay worries about is not only unemployment. It is also not being needed.
In The Human Condition in 1958, Hannah Arendt divided human activities into labor, work, and action: labor sustains life, work creates a relatively durable world, and action takes place among plural human beings, involving public speech, the appearance of identity, and shared responsibility.
This is not Arendt's technological prophecy, nor is it an attempt to rewrite her threefold distinction as a history of technology. I borrow the distinction only to make one point: AI's impact is not only on productive labor. It also touches some outward forms of response, coordination, companionship, and recognition. It is not true action, but it forces out a question: if these outward forms can also be simulated by machines, how are people still needed in a common world?
This is not only an income problem. Even if some form of basic income reduces material pressure, a person may still lack the feeling of being needed - the feeling that "if I were not here, something would be different."
Above subsistence, what people fear most is not having nothing. It is having no place.
Over the past two centuries, that feeling has largely been granted through occupation and economic relations. You are a worker, a doctor, a teacher, an engineer - your role and labor matter to certain people and certain things. Personal irreplaceability comes more from concrete relationships, but the systems of markets, employment, performance evaluation, and statistics most easily recognize occupations and production roles.
What AI may weaken is precisely this form of necessity that society finds easiest to see.
Recognition Beyond Production
This points to a question long simplified by modern market society: what makes a person recognized within public institutions and feel that their existence has meaning?
The Christian tradition of imago Dei often gives one answer: as the image of God, or as a created being, the human person has a dignity prior to achievement and does not need to prove the right to exist through productive capacity.
Modern legal and human-rights traditions also recognize inherent human dignity, and do not equate a person's standing with productive capacity. But in the everyday institutions of many modern market societies, a different and more operational answer often prevails: because you can produce. You can provide labor, services, professional capacity, or purchasing power, so you are easier to count, evaluate, hire, and include. This answer had great explanatory power during the Industrial Revolution and the information revolution, because production capacity was still scarce, and production roles were easiest for institutions to see.
Under the pressure of AI-driven labor rearrangement, this answer may begin to loosen.
If a person's labor role is no longer scarce, it becomes harder for that person to find a place inside the logic of "production as public recognition." We can give that person money through UBI, entertainment through VR and AI-generated content, or more free time. But if the recognition given to care, learning, public participation, and shared governance remains weaker than the recognition given to paid production, this logic will struggle to explain that person's social position.
Some theistic traditions can sustain some people, but they are unlikely to become a shared underlying narrative for highly secularized populations. Philosophical and religious traditions have of course offered other answers, but they differ from one another and have not become the shared default of modern market society.
More people may discover that the default narrative of "I am needed because I produce" is no longer so stable.
Having income, leisure, and entertainment, but no one who truly needs you - this is not poverty. It is another kind of scarcity.
The deeper issue this essay worries about is not only unemployment, distribution, and inequality. It is the fragmentation of shared meaning narratives and the failure of default recognition mechanisms.
From Production Recognition to Relationship Recognition
If productive capacity is no longer the scarcest thing, what is?
The answer is probably not relationship in the abstract, but the scarce inputs that carry relationships: time, attention, trust, and irreplaceable presence.
People need each other not only because each person is a production node. They also need each other because recognition, care, responsibility, shared history, and irreplaceable presence depend on real relationships, and are hard to replace fully with standardized services or AI interaction.
AI can create the experience of "being answered," but it lacks the structure of mutual consequence-bearing. It does not share history with us, bear responsibility, become answerable, or change in the way another vulnerable human being can.
In 2023, U.S. Surgeon General Vivek Murthy issued an advisory on the loneliness epidemic, saying that the mortality risk associated with social disconnection was comparable to smoking up to 15 cigarettes a day. In the same year, the AI companion app Replika publicly said it had more than 2 million active users and about 500,000 paid subscribers. These two figures do not prove each other, but together they point to the same problem: the scarcity of real relationships is being captured by substitute products.
Loneliness is evidence of demand. AI companionship is the projection of that demand - but a projection cannot bear consequences for you.
In the future, two kinds of things may be harder for markets to price fully.
One is irreplaceable presence in private relationships: someone waiting for your reply, someone who needs you because they know you, a shared experience that cannot be replayed. The other is non-market contribution in public life: care, education, public service, community cooperation, and shared governance. Their value is not just that a process is completed, but that someone is willing to bear consequences and help maintain a common world.
Economists can of course study these scarcities. Gary Becker brought time allocation and household production into economic analysis in 1965. Elinor Ostrom won the Nobel Prize in Economics in 2009 for her work on economic governance, especially the commons. Household production, matching, trust, attention, and non-market value have long been inside economics. But that is precisely the point: the boundary is not the object of study, but the authority to interpret. Without making its ethical and political premises explicit, economics should not replace public judgment and decide on its own which relationships deserve protection.
This economic narrative has gained enormous explanatory power over the past two centuries because the problems it was good at answering - production, distribution, and growth - happened to be the most urgent problems of that era. When AI rewrites those problems, the center of the economic narrative may be forced to shift.
A public narrative centered on growth, employment, and market exchange also needs to absorb the work of welfare economics, labor economics, household production, and institutional economics, and answer concrete questions together with philosophy, ethics, political science, and psychology: which relationships cannot be outsourced? Which responsibilities should be institutionally recognized? Which public contexts must preserve human judgment and consequence-bearing?
This is not a failure of economics. It is a reminder that the so-called "final question" is not the endpoint of economics, but a question that must be answered by many forms of knowledge together when the productivist narrative reaches its boundary.
Back to Smith
Adam Smith's first major work, published in 1759, was not The Wealth of Nations. It was The Theory of Moral Sentiments. At the beginning of that book, he discussed sympathy: humans are not only self-interested. They also have the ability to enter, through imagination, into another person's situation and feel another person's fate.
Smith continued revising The Theory of Moral Sentiments late in life, and the sixth edition in 1790 included major additions. At the very least, we can say that moral philosophy never left the center of his vision.
For the later public economic narrative centered on productivity, employment, and growth, "why do people need each other?" long had a simple answer that did not need to be asked again and again: because production requires cooperation.
AI will not make economics invalid. It merely pushes the productivist narrative back to its boundary: output, income, and efficiency can answer how we produce, but they cannot answer why we live together.
As machines become increasingly capable of production, humanity is not facing a more complicated production problem. It is facing an older relational problem: if a person is no longer needed because of a job, how can that person still leave an irreplaceable difference in someone else's life?
What is truly scarce in the next era may not be productive capacity, but relationships in which people are willing to answer one another, care for one another, and bear consequences with one another.
People need each other because human meaning has never been only about what one can produce. It is also about who takes one seriously, whom one is willing to bear responsibility for, and with whom one helps maintain a common world.
