Who Pays for What Remains?
On AI abundance, circulation, and human formation
Imagine a village whose grain moves through many hands. A sack leaves a field and does not stop. It is carried to the granary, and from there it begins to move. Some goes to the miller, who keeps a portion for the grinding. Some goes to the lender who carried a household through the lean months and now takes his share back with interest. Some is set aside for seed, some for a wedding, some for the schoolmaster, some for the temple. What is left is bread. And the bread feeds the hands that will harvest the next field.
The grain in the granary is one kind of wealth. You could weigh it. You could write down the number. But the life of the village is not in the number. It is in the passage. The miller eats because the grain passed through his mill. The lender’s family eats because the grain passed through his ledger. The schoolmaster, the potter, the girl who works in the landlord’s house, all eat because the grain moved, and moved through them. Take any one of them out of the path, and they fall out of the circuit through which the village feeds itself.
So there is another kind of wealth, harder to see: how widely the grain travels and how many lives it passes through. The land belonged to some and not to others. Some bent their backs while others kept the accounts. Some were held in place by debt, by caste, by gender, by the accident of whose house they were born into. But the movement was wide. It touched almost everyone, because almost everyone stood somewhere along the path the grain had to travel.
Then, just beyond the village, a machine is built. It can produce grain more cheaply and more abundantly than the village ever could. The grain no longer moves through the miller, the lender, the many hands of the village. The machine does in one motion what those hands once did in many. The grain still exists, more of it than before. But it now reaches the village, if it reaches the village at all, along a single channel owned by whoever owns the machine.
At first, this looks like freedom. Fewer bent backs in the field, fewer anxious nights watching the sky. The granaries are full. It looks like abundance. But abundance was never only a question of how much grain exists. It was a question of what becomes of the life that formed around its movement. The machine raises the amount and narrows the path. And the village, which lived in the passage, is left looking at a fullness that no longer passes through it.
If AI systems come to produce more of what societies need, if their ownership remains concentrated, if human labor becomes less central to the circulation of income, and if the activities that once formed judgment are increasingly bypassed, then what happens to access, legitimacy, and human formation?
AI systems are already being used to write code, read scans, draft legal arguments, tutor children, optimize logistics, generate images, and assist with mathematical work. If these capacities continue to improve and spread, the same resources may produce more. Wealth can increase when matter, energy, time, skill, and knowledge are arranged into more useful forms.
A report that once required weeks of work becomes one of a thousand reports generated in an afternoon. A song, image, lesson plan, legal memo, diagnosis, design, or essay may still be useful while carrying less scarcity, authority, or social weight. Trust, attention, provenance, judgment, distribution, institutional recognition, and the human relationship through which an artifact comes to matter begin to carry more of the value. AI abundance can create value while dissolving the scarcity through which value used to appear.
Much of the current discussion is about jobs: which will go, which will remain, and how soon. But a job is only one end of a loop. The grain still has to move. Human beings still have to value what is produced, and in a market, their values count only when they are attached to claims. A person may need food, shelter, music, teaching, care, conversation, dignity, or human presence. Without a wage, ownership, public entitlement, credit, or enforceable access, that need remains humanly real but economically quiet.
If machines produce more and more of what we need, and if fewer people earn their living by producing those things, who has the money to buy what the machines make?
⁂
If machines become very good at optimization, perhaps more human work will gather around things that require presence: care, teaching, music, sport, craft, ritual, attention, and the shared human frame in which bodies, memory, and recognition are present to one another.
I was reminded of this at a concert, a tribute to Bob Marley performed by the New Haven Symphony Orchestra with Judah Tribe. A symphony orchestra sat behind a reggae band, the music passing through a room of Black and white listeners, children and old people, some barely able to walk. Then the music began, and they stood up and danced. It felt more like a room remembering something than a performance. The songs were some of Marley’s greatest hits. No Woman, No Cry. Is This Love. Redemption Song. One Love.
At one point, a man sitting in front of me turned around. He looked at me, smiled, and offered his hand. It happened too quickly for me to understand whether he had mistaken me for someone he knew, or whether the music had simply made such recognition possible. I took his hand. It was firm, and in that firmness I felt an unmistakable connection. Nothing had to be explained. The music had done some work before the words could arrive.
During the interval, many in the audience moved to get wine, mingle, and look at the stalls. I met an artist selling Bob Marley T-shirts. One shirt showed a Black boy playing a violin. I asked him about it. He said he liked imagining children seeing themselves in places where they were not always expected to be. I also saw the lead vocalist walking through the room. She was on her phone, showing her young child the audience that had gathered to hear her sing. I had not expected to see her moving through the crowd. She smiled as she turned the phone toward the room. Much of what mattered that evening would not appear in a recording.
Creative work is often described by the artifact it leaves behind: a song, a painting, a design, a paragraph, a shirt. But the artifact is only the visible remainder of something larger. Around it is a world of memory, taste, risk, place, repetition, inheritance, and people seeking connection with one another.
AI can make music. It can imitate style, generate arrangements, design the poster, and write the marketing copy. But it cannot make this room. It cannot make old people dance after they have almost stopped walking. It cannot make a stranger turn around and offer his hand. It cannot make the history of Black music enter a symphony orchestra and pass through a mixed audience.
What was present there was not output. It was human contact organized through sound.
Mattering is not the same as being paid for. That is where the picture becomes harder to hold.
⁂
Work built on a shared human frame does not support itself merely by mattering. If the income generated by automated production flows mainly to the owners of models, compute, data, platforms, land, robots, and distribution, then the rest of society cannot simply decide to become musicians, teachers, caregivers, coaches, writers, priests, potters, or local performers and support one another indefinitely. A teacher can pay a musician, the musician can pay a caregiver, and the caregiver can pay a craftsperson. But some recognized means of support have to enter the circle: wages, ownership, public provision, patronage, shared obligation, or a community that feeds and shelters the people doing the work. Without that, goodwill cannot keep the circle moving for long.
A company may become worth a trillion dollars because investors believe it will capture the future. That valuation is a claim on future purchasing power. If the wage loop weakens, the company may weaken the demand on which its own value depends. The number on the market screen may say abundance is coming. The village still has to ask whether it will be fed by that abundance, or merely asked to buy shares in it.
A hospital adopts an AI system that reads scans, suggests treatments, monitors patients, completes paperwork, and enables the institution to operate with fewer staff. For that hospital, the value is visible: lower costs, more throughput, perhaps fewer mistakes. Then the same logic spreads across schools, law firms, software companies, publishers, logistics firms, and media companies. The wage that disappeared from one balance sheet was purchasing power somewhere else.
AI may increase production while weakening one of the old ways in which income circulated. A wage earned in one place became spending somewhere else. That mechanism was never fair enough, and it was never the only possible one. If it weakens, something else has to carry purchasing power back to people. Otherwise, there is more output and fewer paths through which ordinary people can reach it.
People could own shares of the systems that produce the goods. Governments could tax profits and rents, and use them to support income, schools, clinics, libraries, transit, housing, care, and other shared forms of provision. The wealthy could fund art, education, care, sport, research, and human experience.
Dividends and transfers can restore purchasing power. Public provision can restore access. Patronage can fund some human work. Yet a society can give people money and still fail to rebuild a human economy if that money immediately drains back to the owners of the infrastructure through which modern life must pass.
Patronage can set a human ecology in motion. A wealthy person funds a musician. The musician pays a teacher, a venue, an accompanist, an instrument maker. Those people pay others. But the first movement begins with someone else’s choice of what deserves support. A patron may fund the symphony and not the reggae band, the school that resembles his own and not the teacher working in a borrowed room, the sport his children play and not the game through which another neighborhood holds itself together. Other forms of care, repair, teaching, craft, and presence may remain unseen. A society that relies on patronage does not only depend on generosity. It depends on the judgment, taste, prejudice, and attention of those who already have power.
The image I keep coming back to is a leaky vessel. Money may be poured into the human side of the economy, but if the necessities of modern life are controlled by concentrated platforms, landlords, lenders, insurers, and subscription systems, it quickly drains away. A musician is paid, but the payment goes to the platform, the landlord, the health insurance bill, the delivery system, the subscription stack. A teacher is paid, but the money drains back just the same. The money touches a human being, but not long enough to support a human world.
A local music school is not only a place where children learn music. A library is more than a building with books. A neighborhood clinic is more than a delivery point for health services. A community theater, a sports club, a craft guild, a temple, a public school, a university department, a care cooperative: these are also ways of keeping money, attention, skill, trust, and obligation moving through people. They slow the leak. They allow one form of human presence to support another.
Human presence becomes an economy only when people have the means to support it. Otherwise, it remains a longing, not a livelihood.
Some of humanity’s most enduring work has been done in temples, monasteries, huts, workshops, and communities that did not organize themselves around efficiency. But even these forms of life depended on larger ecologies of transmission. The question is not whether a few people will continue to sing, teach, prove, compose, or care. It is whether a society continues to reproduce the conditions under which such people emerge.
⁂
Beyond circulation is the slower problem of formation.
A society may restore purchasing power and still fail to reproduce the capacities that make people able to judge, repair, teach, refuse, and create. Work has formed people, sometimes badly, sometimes beautifully, often both. A doctor does not learn judgment by reading correct diagnoses. A mathematician does not learn to verify proofs by looking only at finished proofs. A writer does not develop a voice by receiving polished sentences. A teacher does not become a teacher by reading lesson plans. Formation happens in the awkward middle, where one tries, fails, imitates, notices, is corrected, returns, and slowly becomes capable of seeing, from inside the activity, something that was not visible before.
The phrase “humans in the loop” hides this difficulty. We still have doctors who can judge a diagnosis, lawyers who can read an argument, scientists who can evaluate a claim, mathematicians who can check a proof, and teachers who can see when a student is lost. But these capacities were not sitting somewhere fully formed, waiting to be assigned the role of oversight. They came from the very activities that AI may now interrupt.
Mathematics is where I feel this most sharply. A proof is not only a result; it is also a discipline of attention. One learns what counts as an argument by staying with a line that will not yield, by believing a false proof for a day and then finding the place where it breaks, by returning to the same page with a different suspicion, by being corrected until the correction becomes internal. If AI systems increasingly solve problems while human beings mainly inspect the finished outputs, then something has shifted that cannot be captured by saying mathematics has become more productive. The theorem may arrive, but the route that once formed the mathematician may have disappeared.
Verification is not a static reserve. It has to be replenished. A society cannot keep drawing judgment from people while removing the activities through which judgment is formed.
A society may not need many deeply formed people to keep functioning technically. Perhaps a small class of experts can verify frontier outputs, direct systems, set objectives, and manage risk. Everyone else may receive income through transfers or dividends, and spend time in consumption, entertainment, care, community, performance, or presence work. This may not collapse immediately. But it is a different society from one in which many people are formed by real participation in difficult activities. It may continue technically while becoming less able to renew the people whose judgment it depends on.
The market may need only a narrow class of verifiers. A democracy needs more. Human agency needs more. A culture in which people can question, judge, refuse, repair, teach, and interpret cannot rest entirely on a thin layer of experts. If broad formation becomes economically unnecessary, it may still remain humanly necessary. That is the point at which the question of work becomes larger than the question of employment.
Income and formation do not have the same solution. A transfer can help people buy the machines’ output. It does not by itself teach them to judge what machines produce. A dividend can sustain consumption. It will not create doctors, teachers, musicians, scientists, craftspeople, or people capable of governing themselves. If the activity that forms these capacities is no longer economically necessary, it must be supported for another reason. That reason cannot be productivity alone.
An AI-generated song, book, lesson, diagnosis, film, or companion may be cheap, fluent, and competent. But it may also arrive as the visible sign of an economy that no longer needs the people who once made, learned, judged, repaired, or taught such things. The artifact then carries a social wound. It is judged not only by its technical quality, but by the world that produced it, the livelihoods it displaced, and the human processes it bypassed. What looks from one angle like abundance may, from another, look like a more efficient way of leaving people out.
Once abundance is visible, exclusion becomes harder to justify. People will see the grain, the machines, and the owners of the machines. They will ask not only why they cannot afford what has been produced, but why the arrangement deserves their acceptance.
From there, the futures separate. One future finds a way for machine productivity to pass again through human lives: through broad ownership, public provision, taxation, local institutions, and support for the practices that form people. Another future keeps abundance behind gates, prices, licenses, subscriptions, private infrastructure, and force, managing exclusion rather than ending it. A third future breaks the arrangement open. People who can see abundance all around them, even as they are told they have no rightful claim to it, may stop accepting the terms by which they are excluded. They may not ask to be included. They may take.
Each would call itself progress. They are not the same future.
Broad ownership seems important. So do public goods. So do schools, libraries, apprenticeships, local arts, care networks, clinics, sports clubs, and civic spaces whose value is not visible only through immediate output. Some of this may sound old-fashioned, but perhaps that is because our language of progress has become too attached to scale. A local institution may not look innovative. It may simply be where the grain keeps moving, where people become capable of living with one another.
The question, then, is who pays for what remains: above all, for the broad formation of people who may no longer be economically necessary for production, but remain necessary for a human society.
The market alone will make much of what remains precarious or elite. Patronage alone will leave culture dependent on the preferences of those who own the machines. Redistribution alone may preserve consumption while the practices that form judgment decay. What is needed is stranger and harder than any of these: a way to keep money, attention, skill, and judgment moving through human lives, not only as consumption but as formation.
And formation is about more than judgment. It may be how a person discovers what is in them. Much modern work has long obscured this, reducing activity to employment and employment to income. If the job disappears altogether, the question it buried returns: through what activity does a person become visible to themselves and to others?
The future I am worried about is not a world in which humans have nothing to do. It is a world in which many meaningful things remain to be done, but the economy no longer naturally pays people to do them. The grain may be abundant. The granaries may be full. But unless access, money, skill, and judgment keep moving through human lives, the village will not be fed by abundance. It will only be able to see it.