The Last Relevance

Richard still makes coffee by hand. He has a grinder, a kettle with a gooseneck spout, a ceramic dripper he bought in Kyoto fifteen years ago when he was running the Asia desk. The ritual takes seven minutes. The AI kitchen system could produce better coffee in forty seconds, calibrated to his cortisol levels and sleep data, adjusted for humidity and bean age. He knows this. He makes it by hand anyway, standing at the counter in his apartment on the forty-third floor, watching the water spiral through the grounds while the city arranges itself below him.

Richard is relevant. He knows what that word means now, though no one uses it in polite company.

The Line

#

There was no announcement. No policy paper titled “On the Partition of Humanity into the Necessary and the Provided-For.” There never is. The line emerged the way all structural lines emerge: through accumulation, through practice, through the slow crystallization of a pattern that everyone could see and no one would name.

It is emerging now.

You can hear it in how technology executives talk about “upskilling,” as though the problem is that people have not sufficiently improved themselves to deserve participation. You can hear it in how policymakers discuss “retraining programs” with the particular gentleness reserved for problems everyone knows the program will not solve. You can hear it in the shift from “these workers contribute essential services” to “these communities deserve support,” a shift so small it barely registers, from a language of contribution to a language of charity.

The line is not a future event. It is a gradient, and we are already on it.

On one side, gradually: people whose decisions still matter. Who direct capital, who set parameters for autonomous systems, who sit on the boards that govern the platforms that govern everything else. Who negotiate with other relevant humans about the terms under which the irrelevant ones will be maintained. The number shrinks each year, not through conspiracy but through capability. When a system can do what a person did, the person’s relevance does not vanish overnight. It attenuates. They review the system’s output for a while. Then they review it less carefully. Then the review is itself automated. Then they are still employed but no longer necessary, which is a condition that can persist for years before anyone names it.

On the other side, gradually: everyone else.

The everyone else are not suffering. This is the part that makes it difficult to talk about, the part that dissolves every critique before it can form. They have housing. They have food. They have healthcare that is, by every metric, superior to what the wealthiest humans had access to a generation ago. They have entertainment, education, companionship systems, creative tools, travel, leisure. They have the allocation, or whatever precursor to the allocation exists in their particular moment on the gradient.

They have everything except a reason for anyone to give it to them other than the belief that they deserve it by virtue of being alive.

The Five Relevances

#

Relevance is not one thing. It is several things, and they erode on different timelines, through different mechanisms, at different speeds. The reasons to keep the kept species disappear in sequence.

Economic relevance as labor goes first. This is the erosion everyone discusses, the one that fills policy papers and opinion pages. When a system can do what a worker did, the worker’s labor relevance attenuates. Not overnight. The attenuation has a characteristic pattern: the worker supervises the system, then reviews its output, then reviews it less carefully, then is retained as backup, then is retained as a formality, then is not retained. The timeline varies by profession and by sector, but the direction does not vary at all. This erosion is well underway.

Economic relevance as consumer persists longer, and this is the part that most analyses miss. An economy needs buyers. Eight billion people with purchasing power represent a market that no rational system would abandon. This is, for now, the strongest structural argument for the allocation: you give people money so they can buy things so the economy can function so you can make money to give people so they can buy things. The circularity is obvious and also genuinely functional. Consumer relevance is why the allocation exists in its current form. It is not charity. It is market maintenance.

But consumer relevance has a shelf life. As AI systems become both producers and consumers of each other’s outputs, as the relevant economy increasingly consists of transactions between autonomous systems optimizing for objectives set by the relevant humans, the consumer base of eight billion people becomes less structurally necessary. The market does not need their demand. It routes around them. Luxury goods for the relevant. Provision for the kept. And between the two, a shrinking zone where mass consumption still matters to someone with the power to care.

Political relevance erodes on a different timeline, and the mechanism depends on the system. In democracies, numbers are leverage. Votes matter. Eight billion people, if they could coordinate, could vote for anything. This is the argument that says the kept species retains power through democratic participation.

The argument is formally correct and substantively weakening. When the meaningful decisions are technical, when the choices that shape daily life are made in parameter-setting sessions at governance boards rather than in legislatures, when elections become choices between flavors of provision rather than directions of policy, the vote retains its form while losing its grip. You can vote for the candidate who promises a twelve percent increase in the cultural enrichment allocation or the one who promises a nine percent increase with better healthcare optimization. You cannot vote to restructure the relationship between the kept and the keeping, because no candidate offers that option, because no candidate could deliver it, because the systems that would need to be restructured do not answer to the electoral process.

In oligarchies, the erosion is faster and more honest. The powerful have never needed the powerless to vote. They needed them to work, to fight, to produce. Remove those needs and the oligarchy’s incentive to maintain the population reduces to two things: moral sentiment and the avoidance of unrest. Both are real. Neither is structural.

Power relevance, the capacity to disrupt, erodes as physical infrastructure becomes automated. The strike was powerful because the factory needed bodies. The riot was dangerous because the city needed order maintained by people who might sympathize with the rioters. The tax revolt was threatening because the state needed revenue generated by the labor of the revolting.

When the factory runs itself, the strike is a group of people standing outside a building that does not notice their absence. When order is maintained by systems that do not sympathize, the riot is a disturbance to be managed, not a negotiation to be had. When revenue flows from AI-generated productivity rather than human labor, the tax revolt is a gesture aimed at a funding mechanism that no longer depends on the gesture’s participants.

This erosion is already visible. Gig workers who strike discover that the platform’s algorithm simply routes around them. Protests that once shut down supply chains now encounter supply chains that have no human-operated chokepoints. The capacity to disrupt requires being embedded in a system that needs you. Disembedding is the trajectory.

Moral relevance is the last one standing. The conviction that human beings matter because they are human beings, independent of what they produce, what they consume, how they vote, or what they could break if they tried. This is the dignity framework, the one that runs through the Stoics and the natural law tradition, through Kant’s kingdom of ends, through the Universal Declaration of Human Rights.

It was built as a shield against exploitation. Against being treated as mere instruments. Against having your worth reduced to your utility.

No one imagined the shield would become the only thing left.

The dignity framework was designed to prevent people from being treated as disposable. It is now the reason they are not disposed of.

The Sequence

#

What matters is the order. Economic relevance as labor erodes first. Then power relevance, as the infrastructure that labor once operated becomes autonomous. Then political relevance, as the decisions that matter move beyond democratic reach. Then economic relevance as consumer, as the market restructures around AI-to-AI transactions. And then you are left with moral relevance alone, the conviction that humans matter because they are human, held by people who have no structural reason to hold it and every structural incentive to let it quietly thin.

Each erosion is gradual. Each is deniable at every stage. Each produces a period in which the form of the relevance persists after the substance has drained out: workers who are employed but not needed, voters who vote but do not govern, consumers who consume but do not drive the market, protesters who protest but do not disrupt.

The forms can persist for decades. The substance is already going.

I wonder sometimes whether the comfort itself is what makes the erosion invisible. Whether the genuine quality of the provision, the fact that it works, that suffering is minimal, that daily life is pleasant, prevents anyone from noticing that each form of relevance is draining away beneath the surface of a life that feels fine.

Richard thinks about this sometimes, though he would never say it aloud. Not because it is controversial. Because it is obvious, and obvious things that no one says acquire a weight that makes them unsayable. Everyone on his side of the line understands that the allocation is a moral choice. Not an economic necessity. Not a political negotiation between parties with comparable leverage. A choice made by people who could, in principle, make a different one.

The recipients of the allocation know this too. That is the part no one discusses.

What the Zoo Knows

#

A species becomes unable to sustain itself in the wild. Its habitat has been transformed by forces it cannot adapt to quickly enough. Left alone, it would decline, fragment, disappear. So a decision is made, by a different species with the power to make it, to preserve it. Reserves are established. Feeding programs are designed. Breeding is managed. The preserved species is studied, appreciated, sometimes loved.

It is not consulted.

Humans are not animals in a zoo. They retain legal rights, political structures, the formal apparatus of self-governance. Elections still happen. Legislatures still convene. The forms of democratic participation persist, and they are not entirely hollow.

But the substance has shifted. When the economy that sustains a population is designed and operated by systems that do not require that population’s participation, the population’s political choices become, in a specific and technical sense, aesthetic. They can choose how they want to be maintained. They cannot choose whether to be maintained or to maintain themselves, because the second option no longer exists at scale.

A parliament that controls no economy is a parliament in the way a constitutional monarch is a head of state. The title is real. The power is notional.

Self-governance requires a self that governs something. When the something has been transferred, the self-governance is a ceremony.

Richard’s Side

#

Richard does not feel guilty. He has thought about this carefully and decided that guilt is the wrong frame. What he does is genuinely important. Someone has to make the decisions that autonomous systems cannot make for themselves, not because the systems lack capability but because the systems lack standing. They optimize, but optimization requires an objective function, and objective functions require values, and values require someone to hold them.

Richard holds values. That is his job, to the extent that the word “job” still applies. He sits on three governance boards. He reviews the parameters that shape how the allocation is distributed across South Asia. He makes judgment calls that affect nine hundred million people, and he makes them well, with genuine care, informed by data systems that surface consequences he could never compute alone.

He is relevant because the moral architecture requires human authors. Not because no AI could do what he does. Several could, and might do it better by most measurable criteria. But the principle has held, so far, that the systems serving humanity should be directed by humans. The circle of directing humans has simply gotten very small.

Richard’s daughter is in that circle. His son is not. His son lives in Portland, in a comfortable apartment, with a partner and two dogs and a life that looks, from any reasonable external vantage, like a good life. His son makes art. The art is interesting. No one needs it. No one needs anything his son does, in the sense of “need” that would make his existence instrumentally necessary to the continuation of any system.

Richard loves his son. He also knows that his son’s life is funded by decisions made by people like Richard, for reasons that have nothing to do with his son specifically and everything to do with the species-level commitment that humans are worth keeping alive and comfortable.

His son knows this too. They do not talk about it.

The Intrinsic Value Problem

#

Intrinsic value was always argued in contrast to instrumental value. You should not treat people merely as means, Kant said. They are ends in themselves. The argument assumed a world in which people were routinely exploited for their instrumental value and needed protection from that exploitation.

The argument never considered a world in which people had no instrumental value to exploit.

In that world, intrinsic value does not function as a protective shield. It functions as a life-support justification. The argument shifts from “do not reduce people to their usefulness” to “people matter even though they are not useful.” The grammar is similar. The emotional register is entirely different.

The first is a demand for respect. The second is a case for charity.

When intrinsic value becomes the only value, it stops feeling like dignity and starts feeling like a pardon.

Margaret, in her garden, does not think in these terms. She does not think of herself as kept. She thinks of herself as retired, as comfortable, as lucky to have her health and her tomatoes and Tuesday mornings at the coffee shop. The language of the kept species would strike her as dramatic. She would be right that it is dramatic. She would be wrong that it is inaccurate.

The Conviction Erodes

#

Not dramatically. Not through some villain’s decision. Through the same process that erodes all convictions held without structural reinforcement: gradually, at the margins, in the language. This is already happening. You just have to listen for it.

The first shift is linguistic, and it is well underway. “Citizens” becomes “population.” “Rights” becomes “provisions.” “Self-governance” becomes “input.” “Workers” becomes “communities.” Each substitution is minor. Each substitution reflects something real. When a tech company announces layoffs and pledges support for “affected communities,” the word “communities” does a specific kind of work. It reframes the relationship. These are no longer people who contributed and are now being discarded. They are a population that will be provided for. The moral grammar has already shifted from reciprocity to custodianship. We are just not used to hearing it yet.

The second shift is attentional, and it follows naturally. The people making decisions are busy. They have real problems to solve: coordination among autonomous systems, parameter-setting for planetary-scale optimization, governance of AI architectures that grow more complex each year. The provision of the kept population is a solved problem. It works. It does not demand attention. And attention that is not demanded tends to wander. You can see this now in how UBI debates have moved from urgent policy discussions to background assumptions. The question is no longer whether to provide. It is how much, administered by whom, and the question is getting quieter each year because the answer is increasingly: let the systems handle it.

The third shift is generational, and it is the one that will matter most. Richard’s generation built the allocation out of genuine moral conviction. They remembered the world before. They remembered unemployment, poverty, the scramble, the suffering. They built the allocation because they believed no one should live like that, and they had the means to prevent it.

Richard’s granddaughter will not remember the world before. She will inherit the allocation as a fact, not as a moral achievement. She will not know what it cost, in argument and conviction, to establish. She will know only that it exists, that it has always existed in her lifetime, and that maintaining it requires resources that could be directed elsewhere. This is already the pattern with every inherited moral commitment. The generation that fought for civil rights understood viscerally what was at stake. Their grandchildren understand it as curriculum. The understanding is real but it is thinner, and thinner commitments bend under pressure that thicker ones would not.

Moral commitments made from memory are stronger than moral commitments made from inheritance. The generation that remembers why the commitment matters is never the generation that decides whether to keep it.

What Identity Becomes

#

For the kept population, identity undergoes a specific transformation. Not the identity dissolution that Part 60 described, which was a loss of differentiation. Something more fundamental.

Identity, at its root, is an answer to the question “what are you?” For most of human history, the answer involved some form of participation. I am a farmer, a builder, a parent, a citizen, a believer, a maker of things that others need. Even when the participation was coerced, even when it was exploitative, it provided an answer. A bad answer, sometimes. A constraining answer. But an answer.

The kept population has no participatory answer. They are not farmers, builders, makers of needed things. They are not citizens in the substantive sense, because their citizenship confers no leverage and governs no economy. They are not workers. They are humans. That is their identity. Their species membership. The thing they share with every other member of the kept population and that distinguishes them from nothing.

Richard’s son makes art. He is good at it. But “artist” as an identity requires an audience that receives the art because it needs what the art provides. His audience is other members of the kept population, consuming each other’s creative output in a closed loop of production and consumption that touches nothing outside itself.

An identity that connects to nothing beyond itself is not an identity. It is a hobby.

The Kept and the Keeping

#

The relevant humans, the ones who maintain the allocation, who direct the systems, who hold the values that the autonomous architectures optimize toward, also have an identity problem. Their identity depends on the kept population’s existence. Remove the eight billion people they are keeping alive and comfortable, and what are they? Operators of systems that serve no one. Holders of values that apply to nothing. Governors of a civilization with no civilization to govern.

The keepers need the kept. Not for labor. For meaning. And, if we are being honest, for stature.

There is a version of this that is purely noble. The relevant humans maintain the allocation because human dignity demands it, full stop. That version is partially true. But it is not entirely true, and the part that is not true matters.

Keeping people alive and comfortable is also a source of identity for the keepers. It answers the question of what they are for. It provides the moral vocabulary that makes their position bearable: we are not an elite hoarding power, we are stewards, custodians, the ones who chose to honor the species commitment when we could have chosen otherwise. The benevolence is real. So is the fact that benevolence requires someone to be benevolent toward. Noblesse oblige was never just oblige. The noblesse was the point.

Richard would not put it this way. He would say, truthfully, that he cares about the people his decisions affect. He does care. He would say, truthfully, that he works hard. He does work hard. But there is a warmth in the caring that is not entirely separable from the elevation the caring provides. He stands at his window on the forty-third floor and looks out over a city of people he is responsible for, and the looking down is not incidental to the feeling. The responsibility and the altitude are the same experience.

This is not hypocrisy. It is the ordinary human mixture of genuine care and quiet self-regard that has characterized every custodial class in history. The British in India built hospitals and railways and believed themselves civilizers. The belief was not entirely false. The self-regard was not entirely separable from the building.

The moral architecture that protects the kept also elevates the keepers. Both functions are structural. Neither is optional.

He finishes the coffee. He rinses the ceramic dripper carefully, the one from Kyoto, and sets it on the drying rack.

Somewhere below, his son is walking the dogs. The morning is ordinary. The provision continues. The conviction holds, for now, in the hands of people who remember why it matters and who, if we are being fully honest, remember also what it feels like to be the ones who matter.

The question is not whether they are right that human life has value. Of course they are right. The question is whether being right is enough, in the absence of any structure that requires them to act on it, and whether the rightness can survive the generation that earned it.

This is Part 86 of The Approximate Mind, a series exploring how AI reshapes human experience, identity, and society. This piece traces the forms of relevance that connect a population to its own civilization, and asks what happens as they erode, one by one, until only the moral conviction remains.