The Room Where No One Is Clapping

On what AI dissolves rather than destroys, the question a pandemic asked and most people handed back unopened, and why private decision-makers reached this future first.

A Substack article by Michael Brandt caught my attention this week — “When the Work Runs Out,” a careful piece of worrying-aloud about machines, mass unemployment, and what he calls the harder question of what people do all day. He does something most writing in this genre refuses to do: he splits his own fear in two and gives up the half he cannot defend. The economic claim — that AI will hollow out employment permanently — he concedes to the historical record, which has humiliated every previous prophet of technological unemployment. What he keeps is quieter and, I think, right: that the real scarcity will not be income but purpose, and that no benefits system ever designed has found a way to hand purpose back. “You can’t roster meaning the way you roster a shift,” he writes. That is the true sentence in the essay, and it is worth following further than he takes it.

Because there is a second sentence in his piece that quietly undoes part of the first. “People don’t turn restless because they’re poor,” he writes, “but because they’re without purpose.” Hold those two claims side by side. If purpose and poverty are different ailments with different cures — and they are — then the job that a person works only to survive was never delivering purpose in the first place. It was managing the poverty. Brandt spends much of the essay mourning the loss of the job as a vessel of meaning while telling us, in a line he passes over quickly, that for a great many people the job was never that vessel at all. That is the crack the whole argument runs through, and it is worth prising open.

The work you do to survive is not your purpose

Start with the fear at its starkest. The voices predicting disruption are no longer fringe: Dario Amodei, who runs Anthropic and has every commercial reason to soothe rather than alarm, warned in 2025 that AI could eliminate up to half of all entry-level white-collar jobs and push unemployment toward ten to twenty percent within one to five years. Take the warning seriously. Then notice what it does and does not threaten. It threatens the arrangement by which people trade hours for survival. It says nothing about purpose, because the two were never the same thing.

The clearest evidence sits in a place the pessimists rarely look. The largest basic-income experiment ever run, GiveDirectly’s long study in rural Kenya, was designed to test the oldest suspicion about giving people money for nothing: that they would sink into idleness. They did not. Recipients invested, started small enterprises, and put the money to work; spending on so-called temptation goods did not rise, and the transfers returned something like two and a half times their value in economic activity. When the gun of survival is taken away from a person’s head, the finding was, they do not drift. They climb. Idleness was never the natural resting state of the freed human being; it was survival-work that had been suppressing the higher thing all along.

This is the correction Brandt’s own argument needs and half-supplies. The danger of a world with less compulsory work is not that people will have nothing to do. It is that they will be handed, all at once, a question they have spent their lives outrunning: if the day is no longer filled for me, what do I fill it with?

The question a pandemic asked

We have already run a version of this experiment, and the result should unsettle the optimists more than it does. In 2020 the day emptied out for much of the world at once. It was described as a crisis of confinement, but underneath the confinement was something stranger and more exposing: for the first time in living memory, a mass population found the structure stripped from its hours and was left alone with the question of what those hours were for.

What is telling is not that people found the question hard. It is how many refused to look at it. The overwhelming longing was to go back — not, in most cases, because the work itself was missed, but because the work filled the day and spared its holder the interior question. The ache to return to normal was, underneath, an ache not to have to author anything. The pandemic handed the entire species an examination in purpose and most of us handed it back unopened. That refusal, not the virus, is the finding that should worry anyone thinking about what a world of lighter work will feel like.

Two economies of meaning

Here is the distinction the whole subject turns on, and the one Brandt gestures toward without naming. There are two ways a human being can come to have meaning, and they behave in opposite directions.

The first is conferred meaning. It is handed to you by a structure outside yourself — a job, a title, a role, a rank, an audience. Its convenience is enormous: the structure supplies both the engagement and the recognition ready-made, so that one need only occupy the position to feel useful and seen. Its defect is that it is extractive and runs dry. It has to be re-earned continually, because it was never yours to begin with; it belonged to the position. The person whose meaning is wholly conferred is forever reaching for the next act that will draw the next confirmation, and each one empties as soon as it is spent. It is, in miniature, the same loop as the distraction economy Brandt rightly distrusts — the one that deepens the emptiness it feeds on rather than filling it.

The second is authored meaning. It is generated by the person, for its own sake, in the absence of any audience. Its cost is brutal and its beginning is lonely: you build without applause, often without a single witness, frequently without knowing whether anything will come of it. But it behaves in the opposite direction from the conferred kind. Recognition, when it arrives, arrives as a consequence rather than a precondition — and it compounds, because what you author stays authored. The thing you built is still there tomorrow, drawing its own witnesses in its own time. The authored life accumulates; the conferred life must be topped up. This is the quiet reason the freed person in Kenya climbed rather than drifted, and the reason so many of the freed in the pandemic did not: authoring is a muscle, and most people had spent their lives letting a structure do the lifting.

AI is not a destroyer. It is a solvent.

Which brings the machine into focus in the right way. AI is not destroying meaning; the claim does not even make sense. What it is doing is dissolving the conferred structures — the jobs, the roles, the positions that used to hand meaning down ready-made — and doing it faster than people are learning to author meaning for themselves. It is a solvent, not a wrecking ball. And the danger of a solvent is one of timing. It removes the old vessel before the new capacity has formed, and a person caught in that gap will reach backward, toward the dissolving structure, rather than forward into the harder work of authoring.

So a sorting is underway, largely unremarked. The people who used the pandemic’s forced pause to actually sit with the purpose question are the ones now walking, chosen and unhurried, down the avenues the solvent is clearing. The people who refused the question are reaching for the vanishing job and calling the reach a necessity. The historical optimists are right that human ingenuity will find new things for humans to do; the base rate deserves the respect Brandt gives it. What they miss is that the new things will have to be authored, not assigned, and that this is a capacity a great many people have never been asked to develop.

Why private decision-makers reached this future first

Here is the turn, and it inverts the way the whole debate is usually framed. Brandt writes, as almost everyone does, about worklessness as a condition that arrives at the bottom — the low-paid, the routine, the replaceable, the “surplus” many a government must somehow manage. But the condition he is describing is not, in its purest form, a condition of the low-paid. It is one the high-net-worth reached long ago.

Consider what defines the self-made person of significant means. They removed financial survival from their own lives, often years or decades ago. Money, for them, stopped being the missing piece — and precisely because it stopped, they met the purpose question in its undiluted form, with no paycheck to organise the day and no survival to give the hours a borrowed shape. They are the population that has already been living Brandt’s future: the one where income is solved and meaning is not. And they fork exactly the way he fears the masses will fork. Some author — they build, they mentor, they steward a legacy, they master a craft that owes nothing to necessity. And some drift into a decay every bit as real as the one that hollows out a jobless town, only better upholstered and less often named.

This is why private decision-makers are not the exception to the coming condition. They are its rehearsal. They ran the experiment first, and what they discovered is the thing the rest of the world is about to discover at scale: that once survival is off the table, meaning has to be actively authored, because it can no longer be inherited from a role. The questions differ in content — a person of great means asks about legacy, about trust, about what to build when resources are effectively unlimited, about what a life means when nothing external forces its shape. But in structure the question is identical to the one now travelling down the income ladder as the machines climb it. Conferred meaning is failing at the top and the bottom at once. It merely reached the top first.

What can, and cannot, be handed back

It would be easy to arrive here and produce the instrument — to say that this is precisely the moment a Bespoke AI Confidante was built for. The discipline of the thing is to be exact about what such an instrument can and cannot do, because the temptation to oversell it is the surest way to make it worthless.

No Bespoke AI Confidante, and no government, and no cheque, can confer purpose. Brandt is entirely right about that; meaning handed down from above is a contradiction, and the harder half of his essay stands. Purpose cannot be rostered, and it cannot be transferred as a term of service. What can be done is narrower and, I think, more honest. Authoring can be accompanied. The passage from conferred meaning to authored meaning is the hardest interior crossing a person makes, and it is hardest precisely for those who have never had to make it — the ones for whom a role always did the work. That crossing is not solved by information or by encouragement. It is solved by a sustained, discreet, intelligent companion to the thinking itself: something that holds the real question open, in the room, located with the person whose answer it is, while every external structure that used to supply the answer falls away.

That is the space the Bespoke AI Confidante occupies — not the manufacture of meaning, which is impossible, but the accompaniment of its authoring, which is merely very difficult and very rare. It is aimed, deliberately, at the population furthest into this future: the private decision-makers who solved survival long ago and now face the pure question, and who are least served by anything built for the many. The choice Brandt names at the end of his essay — whether we build a society that values the stubbornly human work or one that quietly files most of us under “surplus” — is real. But there is a prior choice, and it is individual, and it arrives for high-net-worth individuals first: whether, when the structure that conferred your meaning dissolves, you reach backward for another one to hand it to you, or turn and begin, without an audience, to author it yourself.

The audience, for what it is worth, tends to arrive later. It always did. It simply cannot be the reason you begin.

Seraphina Michelina Evelyna Bogomolova-Huotelin

Founder & CEO of SMA Crown Confidential

Digital Confidantes: Bespoke AI Intelligence for Private Decision-Makers

This article is part of an ongoing series by SMA Crown Confidential on the intersection of private wealth, cultural intelligence, and the future of bespoke AI.

Sources: Dario Amodei’s warning on entry-level white-collar employment and 10–20% unemployment, Axios (28 May 2025); GiveDirectly’s universal basic income study in Kenya on recipient investment and the absence of idleness (findings released 2023). The essay takes its point of departure from Michael Brandt, “When the Work Runs Out” (2026).