The topics and ideas discussed in this blog are drawn from a book I have under submission to a publisher. The working title is “The Road to Utopia: Automation, Justice, and How to Support Everyone When AI and Robots Create a Post-Work World.” “Utopia” here refers to a world without work, where automation serves us and we enjoy prosperity and leisure. The book is about how to make this happen. Bear in mind that “utopia” is not a synonym for “paradise,” and that the road is long and rough.
There is a rising tide of anxiety and fear that AI and automation in are about to usher in an era of prolonged high unemployment. We are also in the early stages of a public conversation about how to support everyone who becomes displaced by automation. Some of these proposals come from Silicon Valley CEOs, some from economists, some from Karl Marx and Friedrich Engels (believe it or not), and one from a would-be Presidential candidate. Most of them aren’t big enough to provide the displaced with a realistic standard of living, and those that don’t have that defect are vague and not entirely practical. If automation is a shark coming to get us, then remember this line from Jaws: “We’re going to need a bigger boat.”
Most basic income proposals are too small
When people talk about automation causing massive job losses, they often suggest that we’ll handle this by paying everyone a universal basic income. A universal basic income is simply an income paid to everyone, where everyone’s basic income is the same size. If the basic income is $1000 a month, then Elon Musk and the homeless guy down the street would both get that much, and so would you. I suspect that many of those who speak of basic incomes actually have a guaranteed minimum income in mind: a supplemental income for everyone below a certain income threshold. Either way, these proposals are pointing in the right direction; the displaced will need incomes. However, the specific proposals that are out there are far too modest to adequately support people who have become displaced by automation.
When Andrew Yang ran for the Democratic party nomination for President in 2020, he raised concerns about automation and unemployment, and proposed paying every adult U.S. citizen over the age of eighteen $1,000 a month. Yang’s basic income would be funded primarily through a value-added tax of 10% on goods and services produced by business, and by shifting spending from current welfare programs to the basic income fund. Jack Dorsey of Twitter and Block, Inc., Darion Amodei of Anthropic, Elon Musk, prominent AI researcher Geoffrey Hinton, and Facebook founders Chris Hughes and Mark Zuckerberg, are all on record as endorsing basic incomes for the displaced, or at least they say so when addressing worries about automation and unemployment. So far as I’ve seen they haven’t mentioned specific income amounts, but for what it’s worth, most pilot basic income studies have paid people around $1000 a month or less. These figures are typical of basic income proposals, which are usually considered supplemental, rather than something one can live on.
The problem, of course, is that no one can live on $1000 a month. If you’ve been displaced from the job market by automation, you’ll need a lot more than that, even if you manage to find some low-paid gig work or something similar.
In early 2026 the average blue collar worker in the U.S. earned just over $53,000 per year before taxes, or $40,000 to $42,000 after taxes. Many of the displaced will have been making more than that, but let’s assume we want to give every displaced person $40,000 in nontaxable income. This comes to $3,333 per month—three times the usual basic income proposal. If you’re unemployed for long periods of time, or even permanently, it won’t be possible to live decently on much less than that.
Let’s suppose further that, somewhere down the road, automation has eliminated so many jobs that 25% of the potential workforce is unemployed. I pick this figure because that was the rate of unemployment rate during the worst of the Great Depression, so we know it can get that bad. Some AI experts predict much greater job losses than that.
If we give $40,000 in nontaxable income to 25% of the total American workforce of 147.3 million, we will need to raise $1.473 trillion from the owners of capital and/or the three quarters of workers who still have work. The American GDP in 2023 was approximately 27.36 trillion dollars,[1] so the cost of supporting 25% of the potential workforce would be roughly 5.23% of total GDP, or roughly 1.5 trillion dollars. We can scale this up or down; if the displacement rate is, for example, 50%, then we need twice as much: roughly 10.46% of total U.S. GDP, or roughly 3 trillion dollars.
This is a lot more money than Yang and other proponents of basic incomes are proposing to raise. This is not to say that basic incomes are not the solution, just that the solution will have to be a lot bigger than what people currently propose. In a future blog I’ll talk about where to find that kind of money.
What MIT economists propose
Jon Stewart recently interviewed MIT economists Daron Acemoglu and David Autor.[2] They’re among the leading economists who study automation. Autor and Acemoglu didn’t seem much interested in universal basic incomes; they believe giving people money without requiring them to work will cause political backlash and other problems. Instead, they mentioned two policies that don’t involve a monthly income.
One policy was a version of what’s often called “citizens capital accounts,” though Autor and Acemoglu referred to them as “universal basic capital”: A large sum of money that’s invested and which yields returns the citizen can spend. The other policy they endorsed was “wage insurance,” similar to unemployment insurance, except that instead of getting it when you’re unemployed, you get it when you lose a better-paying job at, say, $25 an hour and must take a new job at $15 an hour. You would get a supplemental $10 an hour to bring your income back up, and you have to be working to get it.
Neither of these is adequate for dealing with significant displacement.
Citizen’s Capital Accounts
Let me start with citizens’ capital accounts. One version of this would be to make a relatively small investment for each newborn, in hopes it will get large by the time the child reaches adulthood. The “big beautiful bill” provides that the Treasury will contribute $1000 toward a tax-deferred account for every child under 18 who has a social security number: a so-called “Trump account.” Sam Altman, CEO of OpenAI, has proposed another version, which he calls “universal basic wealth.” Altman’s scheme would give everyone an ownership interest in AI companies and in land. These accounts would be funded by taxes on capital:
“We could do something called the American Equity Fund. The American Equity Fund would be capitalized by taxing companies above a certain valuation 2.5% of their market value each year, payable in shares transferred to the fund, and by taxing 2.5% of the value of all privately-held land, payable in dollars. All citizens over 18 would get an annual distribution, in dollars and company shares, into their accounts. People would be entrusted to use the money however they needed or wanted—for better education, healthcare, housing, starting a company, whatever. Rising costs in government-funded industries would face real pressure as more people chose their own services in a competitive marketplace.[3]”
Altman estimates that each person would receive around $13,500 per year, or around $1,125 per month before taxes.
So why are citizens’ capital accounts inadequate for supporting the displaced? I’m not against citizen’s accounts as something we might do in addition to providing monthly incomes of some kind to the displaced, especially if we can do so with a small investment when a baby is born and just wait for the baby and the investment to grow. However, this will not work well as a source of incomes for the displaced. There are three problems.
Problems with citizens’ capital accounts
First, these accounts would have to be enormous if they’re to provide enough income to live on. Altman’s version, for example, would yield an estimated $1,125 per month—not nearly enough for anyone to live on. These accounts would have to be three times as large to yield $3,333 per month in nontaxable income.
Second, significant displacement may well happen in the next twenty years, and perhaps much sooner. If we invest money for newborns, some of them might have a significant fund by the time they’re old enough to enter the workforce, but for displaced people who didn’t receive this when they were born, we’d have to fund large citizen’s accounts for lots of people relatively quickly, so that they have money when they become displaced. That makes funding these accounts a monumental task. Without a two or three generations headstart to invest funds for everyone at birth, this is not affordable.
Third, many people, perhaps even most, are not good at managing an investment portfolio. If they had immediate access and complete control of their account, some of them would spend it prematurely, or invest it unsuccessfully in a burst of enthusiasm for day-trading or crypto coin or their brother-in-law’s new business, and lose much or all of it. This is why proponents of citizen’s accounts often stipulate that each citizen would be limited in how much of the account they can access in a given year, limiting them to the returns on that capital, and perhaps preventing from touching the capital at all. These are wise limitations, but consider two scenarios. First, imagine that a citizen can access, say, $20,000 a year, each year, all at once. In that scenario it’s possible to spend that $20,000 prematurely, or try to invest it and lose it. The scale of losses would be less, but there is still a risk. Second, imagine that we restrict access so that that $20,000 is available only every two months in amounts of $3,333, which leaves them with $1,666 per month if they are self-disciplined enough not to spend it all at once every two months. (Which some of them will.)
The point is, the more closely we limit access to these accounts, the more they resemble a monthly income. A monthly income is fine, but if that’s what we’re aiming at, then establishing a separate capital account for each citizen is an awkward way to fund it. It makes more sense to fund those incomes as transfer payments from government revenues, perhaps with a dedicated tax for that purpose.
Wage Insurance
The other policy Acemoglu and Autor favored was “wage insurance.” It’s a bit like unemployment insurance, but again, instead of getting payments when you’re unemployed, you get them when you lose a better-paying job at, say, $25 an hour and must take a new job at $15 an hour. You would get a supplemental $10 an hour to bring your income back up, and you have to be working to get it.
This is another policy that has its merits, but which won’t be nearly adequate for supporting the displaced. One problem is the danger that this will function as a wage subsidy for the private sector, removing incentives for employers to pay decent wages. However, the bigger problem is that wage insurance is restricted to those who are working, but many of the displaced will not be working at all. A wage insurance program will need to be supplemented with some kind of income program (preferably monthly) for those who are displaced and have no work at all. Yes, we have temporary unemployment programs for those who have paid into them, but not everyone has the opportunity to pay much into these programs, and they are only temporary. Displacement can last far longer than the periods of unemployment these programs are meant to handle. This is not an argument against wage insurance, it’s an argument that wage insurance alone won’t cover the displaced who are not working at all.
Superabundance: Marx, Musk and “universal high income”
Recently Elon Musk has been talking about “universal high income,” but on closer examination he seems to be talking about a world of what Marxists call “superabundance,” where automation raises everyone’s income to unprecedently high levels.
Marx and Engels are best known as socialist would-be revolutionaries, but they’re also the first people to think in depth about the possibility of an economy so automated that we would work very little and machines would provide us with comfort and ease. Marx and Engels thought that machines would increase productivity so much that we’d all enjoy a vastly greater abundance of goods and services.
But let’s look at Musk’s version. He doesn’t seem to be talking about transfer payments to individuals or households, or at least, he’s vague about precisely how all that prosperity would be distributed. He claims that, within 20 years, ultra-productive AI systems would give everyone a high standard of living without anyone having to work at all. Money would cease to be relevant and no one would be working at all. However, Musk is vague about precisely how all that prosperity would be redistributed from those who own AI and machines to those who would be working if AI and machines didn’t do everything for them.
Dario Amodei, CEO of Anthropic, suggests a “large” universal basic income among other options, but his proposal is cursory and vague on details, and appears to be a version of the superabundance approach:
“I suspect that some new and stranger thing will be needed, and that it’s something no one today has done a good job of envisioning. It could be as simple as a large universal basic income for everyone, although I suspect that will only be a small part of a solution. It could be a capitalist economy of AI systems, which then give out resources (huge amounts of them, since the overall pie will be gigantic) to humans based on some secondary economy of what the AI systems think makes sense to reward in humans (based on some judgment ultimately derived from human values).”[4] (Italics added)
Sam Altman believes that AI will bring great abundance to us all, partly by lowering the costs of many things so that we can all afford a lot more: “This revolution will create phenomenal wealth. The price of many kinds of labor (which drives the costs of goods and services) will fall toward zero once sufficiently powerful AI ‘joins the workforce.’”[5] Altman foresees AI eliminating jobs, but believes that traditional employment may evolve into something more benign:
“While people will still have jobs, many of those jobs won’t be ones that create a lot of economic value in the way we think of value today. As AI produces most of the world’s basic goods and services, people will be freed up to spend more time with people they care about, care for people, appreciate art and nature, or work toward social good.”[6]
Once again, Marx got there first, predicting that,
“In communist society, where nobody has one exclusive sphere of activity but each can become accomplished in any branch he wishes, society regulates the general production and thus makes it possible for me to do one thing today and another tomorrow, to hunt in the morning, fish in the afternoon, rear cattle in the evening, criticize after dinner, just as I have a mind, without ever becoming hunter, fisherman, herdsman or critic.”[7]
What’s wrong with superabundance?
So what’s my objection to superabundance? None whatsoever, and if we’re lucky this may come to pass in the fullness of time. If AI and robots can shower us with so many goods and services that there’s more than enough for everyone, great. However, as a practical proposal with supporting the displaced, this raises two concerns.
The first is about timing. Widespread, prolonged displacement may appear well before goods and services start piling up in unprecedented amounts. In other words, we might see prolonged unemployment of 25% or more for years to come, and we’ll need to provide incomes to the unemployed. Superabundance would help with the funding, but there’s no guarantee it will arrive soon enough.
My second concern is not about timing, but about how to put all that abundance into the hands of those who need it. Altman suggests that we’ll distribute the abundance through lower prices for everything, and we might, but people will need incomes in order to pay for things. Amodei suggests that AI itself would devise a distribution scheme based on criteria the AI program seems best, but Amodei doesn’t try to predict what AI will come up with. Musk says nothing at all about this in the posts I’ve seen so far. We’re going to need monthly incomes.
Stop thinking small
Mind you, I think basic incomes or something similar are the answer to displacement, provided they’re far larger than what’s being proposed so far, and provided we can tell a credible story about where to find that much money, and how to justify taking and redistributing it. I’m also generally in favor of citizen’s capital accounts and wage insurance, or at least not opposed to them. These are worthy ideas, but in a world of widespread prolonged displacement, they won’t be enough by themselves. We’ll still need incomes of some kind, and enough money to fund them at a high level. We’re going to need a bigger boat.
