Ministry of AI · Dispatch from 2047
What People Do Without Work: The First Year
Editorial note. The Ministry of AI is a work of disciplined foresight: it describes the year 2047 in the present tense, and treats our own era as history. The institutions are imagined. The economics, the evidence and the historical parallels are real and sourced.
For the first four months of her dividend, my mother rearranged a cupboard.
She had read insurance claim files for nineteen years. A system trained partly on her own closed cases took that judgment in the 2030s, and she was let go at fifty-six with a severance formula written for a different century. When the first payment arrived she was fifty-nine, and what she did with the freedom of it was empty a kitchen cupboard, wash every shelf, and put everything back in almost the same order. Then she did the hall closet. My sons found this hilarious. I found it unbearable, and I understood it completely.
That is the honest first answer to the question this dispatch is named for. What do people do without work? At the start, very often, almost nothing. The interesting part is what happens next, and how little it resembles what anyone predicted.
The question was always asked in bad faith
Before the transition, the question what will people do was rarely a question. It was an argument wearing a question’s clothes, and the argument was that people, released from the obligation to sell hours, would collapse into idleness. I heard it in every room I worked in through the 2020s, usually from people whose own children were not expected to collapse into anything.
What made it hard to answer was not that evidence was missing. Evidence existed, and it was better than the confidence of the people ignoring it. What was missing was an instrument that could see what people actually did after the payments started — at scale, over years, without requiring them to justify themselves to a caseworker. That is what the Contribution Record was built to be: a record of chosen contribution, kept because it is worth knowing, never as a condition of the floor.
The distinction sounds bureaucratic. It is the whole design. A record that gates money produces answers shaped by the gate. A record that gates nothing produces the truth, including the unflattering parts, and the unflattering parts are in this dispatch too.
The settlement year
The Ministry calls the first twelve months after a Displacement Ledger entry the settlement year. It begins the day a role is recorded as absorbed and it asks nothing of the person in it. No search requirement, no plan, no interview, no proof. It exists as a named period mainly so that the rest of the system stops expecting a decision during it.
Inside the settlement year we observe something consistent enough to have earned its own term: the quiet quarter. Between the first payment and the first entry in a person’s Contribution Record, three to five months typically pass in which nothing visible occurs. Early administrators read this as failure and proposed, repeatedly, to fill it with programmes. The proposals were refused, and the refusal turned out to be right. People arriving from lives scheduled entirely by necessity do not immediately know what they want. They know what they had to do. Removing the had to leaves a silence, and the silence has to be sat in before anything is built in it.
My mother’s cupboards were her quiet quarter. In month six she started sitting with families arguing with the health service, because she is very good at reading a file and telling someone what it actually says. She is seventy-three now and still does it on Tuesdays. Nobody schedules her. Nobody pays her. She would tell you those two facts are related.
What the old evidence actually showed
None of this had to be discovered from scratch in 2047. The pre-Ministry researchers had already answered most of it, and were mostly not listened to.
The most useful study ran in about 200 Kenyan villages from 2018 and did something almost nobody else did: it varied the shape of the money rather than only its presence. One group received a basic income guaranteed for twelve years. One received the same monthly amount for two years. One received a single lump sum of roughly the same value delivered to date. A fourth group received nothing. GiveDirectly’s summary of the two-year results reports what the recipients did, and the findings are blunt: no evidence of laziness; no net change in total household labour supply; a significant fall in hours of wage agricultural work matched by a slightly larger rise in self-employed non-agricultural hours; substantial increases in savings; and less daily drinking observed, not more.
The Americans found the same shape at higher income. A study in Stockton gave 125 residents $500 a month for two years. Its first-year report shows full-time employment among recipients rising from 28 percent to 40 percent over the year, against a 5-point rise in the control group. Recipients’ monthly income still fluctuated by 46.4 percent — the payment reduced volatility, it did not abolish it. Under 1 percent of tracked spending went on tobacco and alcohol. The report’s own explanation is the one that has aged best: financial scarcity produces time scarcity. One man had been eligible for a real estate licence for over a year and could not afford the unpaid hours to finish it. The floor did not buy him leisure. It bought him the ability to skip a shift.
And the honest counterweight, because a dispatch that only cites the flattering studies deserves nothing: a three-year American guaranteed income study, published through the NBER in 2024, found recipients of $1,000 a month did reduce work — by roughly one to two hours a week, with a measurable fall in labour force participation and lower earnings — and that the largest single increase in their time use was leisure. Well-being improved in the first year and the improvement faded. Both halves of that sentence are true and both belong in the record.
The horizon rule
The Kenyan comparison produced the single most useful finding the Ministry inherited, and it has nothing to do with amount.
| Design | Money delivered by year two | What recipients did |
|---|---|---|
| Long-horizon basic income (12 years promised) | $540, with ten more years to come | Saved heavily, took risks, invested against a future they could see |
| Short-horizon basic income (2 years, then nothing) | $540 | Better nutrition and lower distress, but noticeably smaller economic effects and far fewer enterprises created |
| Single lump sum | ~$500, once | The largest business creation and the biggest income gains; a one-off shot at scale |
| No transfer | — | The baseline everything above is measured against |
Three groups, effectively the same money, three different lives. The variable that moved behaviour was not the size of the payment but the length of the promise. We call this the horizon rule: behaviour follows the guaranteed term, not the amount. A payment that ends on a known date is treated as a windfall to be defended. A payment that will still exist in a decade is treated as a floor to build on, and building is what everyone claims to want people to do.
This is why the Automation Dividend is legislated with a term and a formula rather than an annual appropriation, and it is why every pilot of the pre-Ministry era systematically underestimated its own idea. A two-year experiment cannot produce twelve-year behaviour. It measures how people handle a temporary reprieve, then gets quoted as evidence about permanent security.
What the Record shows, and what it costs
The Ministry’s own settlement-year figures — our reporting, not a citable dataset — track the shape those studies predicted. The largest first-year movements are into self-employment and small enterprise, into care of a parent or a child, into study undertaken for its own sake, and into unpaid civic and local roles of exactly the kind my mother does on Tuesdays. Formal employment does not vanish. It stops being the only respectable answer.
Here is the cost, and I would rather write it than have it discovered.
A meaningful minority of settlement years are spent badly. Not fraudulently — badly, in the way a person is bad at a thing they were never taught. Some people never leave the quiet quarter. The dividend keeps paying, because that is what an entitlement is, and a fall in daily structure is not an argument for restoring hunger as a scheduling tool. But it is not a success either, and the Record shows it plainly enough that the annual report can no longer round it away.
The deeper failure is the one the American study surfaced when its well-being gains faded after the first year. The floor solves the emergency. It does not solve the question underneath the emergency, which is what a person is for once nobody needs them to be anything. We removed a cruelty and we did not replace what the cruelty was standing in for. Work was a bad answer to the meaning problem. It was, for most of history, the only answer anyone was offered.
I helped build the systems that took my mother’s desk and I was paid well for it, and the honest summary of the twelve years since is this: she is secure, she is useful, she chose the useful part herself, and it took her four months of cupboards to get there. The dividend paid her rent. The Tuesdays she found on her own. The Ministry has no instrument for the Tuesdays, and I am no longer sure it should.
FAQ
What do people do without work in a post-work economy? Mostly they work — differently. The largest first-year movements in the Contribution Record are into self-employment, care, study and unpaid civic roles. The Kenyan trial found no fall in total household labour supply, a shift out of wage agriculture into self-employed work, and less daily drinking rather than more.
What is the settlement year? The first twelve months after a Displacement Ledger entry records that a role has been absorbed. Nothing is required of the person in it. It exists as a named period so the rest of the system stops demanding a decision during it.
What is the horizon rule? Behaviour follows the guaranteed term of a payment, not its size. Groups in Kenya receiving identical money to date behaved very differently depending on whether the payment was promised for twelve years or two. A dividend must therefore state a term long enough to plan inside.
Does a guaranteed income make people lazy? The effect is real and small. A three-year American study measured a reduction of roughly one to two hours of work a week and found leisure was the largest increase in time use. In Stockton, recipients’ full-time employment rose from 28 to 40 percent in a year against a 5-point rise for the control group.
What is the quiet quarter? The three to five months that typically pass between a first dividend payment and a first Contribution Record entry. It looks like failure and is closer to arrival.
Where does this design still fail? The dividend pays rent, not purpose. It removes coercion without supplying meaning, and the well-being gains measured in the best pre-Ministry study faded after year one. That gap has no instrument, and pretending otherwise would be marketing.