Ministry of AI · Dispatch from 2047
UBI vs Dividend: Charity Is an Insult to an Owner
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.
The first thing my mother did with her dividend statement was file it. Not spend it — file it, in the accordion folder where she keeps the deed to her flat and the pension paperwork from the insurer that let her go. She had spent nineteen years reading claim files for that company, and eight months after a system trained on her decisions took over her desk, she was fifty-nine years old and receiving money from the state for the first time in her life.
I asked her, that first year, whether it felt strange. She said it would have, if the paper had been a different colour. What she meant was that she had once helped a neighbour through a disability claim and remembered the woman crying in a waiting room over a form that asked, in effect, to prove she had failed. The dividend statement asked nothing. It said: your share of assessed machine yield for this period, and then an account number, and then a figure. There was no box to explain herself in.
That is not a small aesthetic difference. It is the whole institutional argument, and in 2047 it is the part most people under thirty cannot see anymore, because they have never received a payment that required an apology.
The same money, twice
Set two systems side by side. Both send every adult the same amount on the same day, funded from the same place. One calls it a basic income. One calls it a dividend. In the mid-2020s, serious people treated the distinction as marketing — a matter of which word polled better in which country. It was not marketing. It was a difference in what kind of thing the payment is, and institutions are made of exactly that.
A basic income is justified by need. Its moral engine is that no one should fall below a floor, which is a fine engine and a very old one. But a claim grounded in need is, structurally, a claim on other people’s generosity. Generosity is renewable and also revocable. It comes with the natural follow-up question — are you sure you need it? — and once that question is legitimate, the apparatus that asks it follows: eligibility rules, verification, review dates, the entire machinery of establishing that you are the right kind of poor.
A dividend is justified by ownership. Its engine is that you hold a share of an asset and this is your portion of what the asset earned. Nobody asks a shareholder whether they need the distribution. Nobody reviews their circumstances. The question do you deserve this? is not merely rude; it is category-inappropriate, like asking a landlord to prove hardship before collecting rent.
The Ministry did not choose ownership language because it sounded nicer. It chose it because the underlying claim is genuinely an ownership claim. The capability inside a deployed model is a compression of the human record — published science, public infrastructure, publicly educated engineers, and centuries of writing by people who were never asked. The return on an inherited asset belongs to its heirs. Change the justification and you have not softened the argument; you have replaced it with a different one.
The entitlement-form test
Because the wording is load-bearing, the Ministry tests it. Any proposed payment is scored on four properties, and the score determines which department administers it. The test is deliberately blunt.
| Property | Welfare form | Ownership form | Why it matters |
|---|---|---|---|
| Basis of claim | Need, demonstrated | Share held, recorded | Determines who bears the burden of proof |
| Means test | Present, recurring | Absent by construction | Every test creates non-take-up |
| Discretion | Administrator may vary or refuse | None; formula publishes the amount | Discretion is where stigma and lobbying enter |
| Revocability | Political, by ordinary budget vote | Requires unwinding a property claim | Determines survival across governments |
Read down the two columns and you can see the whole of the twentieth century’s difficulty. Programmes designed for the poor are administered by people with discretion over the poor, and discretion is expensive: it is the cost of the caseworker, the cost of the appeal, and the cost borne by the person who does not apply at all. The economics literature of the era was unambiguous that this last cost is large — take-up of benefits people are legally entitled to is routinely and substantially incomplete, for reasons that include information, transaction costs and stigma (Currie, NBER, 2004). A programme with a 70% take-up rate is a programme that fails three in ten of the people it was written for, quietly, without anyone recording a denial.
Now look at the ownership column, and at the one large real-world case that ran for decades before we existed. Alaska has paid a dividend from a sovereign fund to essentially every resident, adults and children, since 1982. The state’s own summary shows applications in recent years running at roughly 730,000 to 740,000 against a resident population of comparable size — near-universal participation, with 618,863 dividends paid in 2025 at $1,000.00 each, and 625,912 paid in 2022 at $3,284.00 (Alaska Permanent Fund Dividend Division). Nobody in Alaska is embarrassed to file. The instrument does not signal poverty because it is not addressed to the poor.
What it does to behaviour, and what it does not
I should be careful here, because this is where advocates in my industry used to overclaim, and I was in those rooms. The Alaskan evidence is good but narrow. The most careful study of it found no significant effect on aggregate employment, with a modest increase in part-time work (Jones & Marinescu, NBER, 2018). Large randomised transfer programmes have found broadly consistent results — recipients do not stop working en masse (GiveDirectly, 2023). A thorough review of the design questions, written while the debate was still theoretical, laid out the costs honestly and did not pretend the arithmetic was easy (Hoynes & Rothstein, NBER, 2019).
None of that evidence tells you where the money should come from, and that is precisely the gap the dividend argument fills. Every basic income proposal of the 2020s ran aground on the same rock: it was a spending programme in search of a tax. The dividend is not a spending programme. It is a distribution of yield that is already being produced, by an asset that was already collectively made, and which had simply never been metered — the same shift in perspective that made the labour share of income legible as a falling number rather than an act of nature (Our World in Data). The productivity question, argued sharply throughout the period (Acemoglu, NBER, 2024), was always secondary. Even a modest gain, booked as margin instead of wages, changes who owns the output.
Tuesday, and the colour of the paper
My youngest is twelve and receives a child share, held in trust, which he refers to — with the flatness of someone describing rainfall — as “my account.” He has never once used the words help, support, assistance or benefit about it. He knows the yield figure moved last quarter because the Ministry publishes the error band with the estimate, and he thinks the band is interesting. He will grow up believing, in the way you believe things you were never taught, that he is a part-owner of the productive capacity of his society.
My mother is seventy-three and did not grow up believing that. She spends Tuesdays sitting unpaid with families arguing with the health service, because she is good at reading a file and knows exactly what a denial letter is engineered to do. Her Contribution Record notes it. Nothing in the floor depends on it, and that is the entire design: capacity is not contribution, and work that must be done to survive is not chosen work.
Twice now she has told me the same thing about that first statement. It was not the money. She had savings; the money mattered less to her than to most. It was that after nineteen years of adjudicating whether other people qualified, nobody adjudicated her. She was not a case. She was a holder of record.
Where the framing fails
Here is the cost, and it is real enough that I would not publish the argument without it.
Ownership language protects the payment by making it equal, and equality is not the same as fairness. A dividend cannot be tilted toward the person who needs it more without contradicting the logic that makes it hard to abolish. Once you pay one shareholder more than another for reasons unrelated to their holding, you have reintroduced discretion, and discretion is the door stigma walks through. So the dividend does what it does well — it removes the coercion under everyone’s Monday — and it is structurally incapable of doing the rest.
The rest still has to exist. Disability costs money that a flat share does not cover. Chronic illness, care obligations, a house that burns down: these are needs, they are unequal, and they require exactly the targeted, means-tested, discretionary programmes the ownership form was designed to escape. The Ministry runs alongside them; it does not replace them. And those programmes remain what they have always been — the first thing cut, the last thing defended, administered by people with discretion over people without it.
We solved the wrong half, in other words, and we solved it well. The floor is unconditional and effectively unrepealable. Everything above the floor that depends on someone assessing your circumstances is as fragile in 2047 as it was in 2027. My mother’s Tuesdays are spent in that fragile part, arguing on behalf of people whose need is real and whose paperwork asks them to prove it.
She still has the folder. The dividend statement sits with the deed, because that is what it is.
FAQ
Is a dividend just UBI with a better name? Only if you think a lease and a deed are the same document because both let you live in the house. The payment can be identical; the claim is not. UBI is justified by need and can be reduced when the budget tightens. A dividend is justified by a share and has to be expropriated rather than trimmed.
Doesn’t the ownership framing exclude people who contributed nothing? It includes them, deliberately. The share is not earned by individual contribution — it is inherited from a collective record that nobody can be shown to have built alone. That is also its weakness, and we have said so: it pays the prolific and the silent the same amount.
Which is politically easier to pass? Basic income, almost certainly. It requires no argument about the origin of machine capability and no new accounting instrument. Which is why it was proposed constantly for a century and enacted almost nowhere at scale: the easy version had no funding logic of its own, and a payment with no funding logic is the first item on any austerity list.
What happens when the yield falls? The amount falls with it, publicly, and that is a feature rather than an embarrassment. Alaska’s payment has moved by hundreds of percent between years. A dividend that varies is still a dividend; a benefit that varies looks like a cut, and gets fought as one.
Does the framing survive contact with people who dislike the underlying argument? Not always. Someone who rejects the claim that machine capability is collectively inherited will read the ownership language as a costume over a transfer programme, and they are being consistent. The argument stands or falls on provenance, which is why the metering instruments exist and why their error bands are published rather than hidden.