The machines may eventually be able to make almost everything – except customers.
That’s the paradox at the heart of the AI boom, and it’s the one Wall Street, Silicon Valley, and most of the political class are working hard not to notice.
From the point of view of a single company, replacing workers with AI looks like simple arithmetic. Labor is expensive while software scales. If a department of thirty people can be reduced to five people supervising a collection of AI agents, the savings flow straight to the bottom line. If your competitor does it and you don’t, your costs are higher and your survival may be at risk. If one country permits it and another restricts it, the first gains a cost advantage over the second. Once the technology works well enough, refusal becomes expensive.
That’s the microeconomic logic, but the macroeconomic logic is nastier.
The worker whose wage is an expense for one company is the customer of another. The paycheck that looks like waste on one income statement is the purchasing power that keeps someone else’s revenue stream flowing.
Slash enough paychecks and the problem ceases to be unemployment in the ordinary sense. It becomes a demand crisis: the machines can make the product. They cannot make the paychecks that buy it.
That’s why the debate over whether AI will “take jobs” is too limited. It is the obvious question, so naturally it is the one everyone asks first. But work was never just a way of producing goods and services. In modern capitalism, work became the load-bearing wall of our entire social structure. We use paychecks to distribute income, finance government, support credit, build retirement savings, organize cities, train young adults, and give people proof that they have worth.
If AI weakens work badly enough, it does not merely threaten jobs; it threatens the whole architecture built around jobs.
I worried about this thirty years ago, in a book called Facing the Future. At the time, people thought it was pure speculation, almost science-fictional: what happens if machines eventually do so much of the work that the demand for human labor isn’t enough to fuel purchasing power?
That question has now returned with a vengeance, wearing an AI mask.
The standard, comforting reply is that people have always feared new machines. The spinning jenny, power loom, tractor, assembly line, and computer each destroyed some jobs yet created others. That’s historically true. But it’s no longer sufficient, because the game is different this time.
Earlier waves of automation usually replaced specific forms of physical labor while raising demand for other forms of human effort. AI is different not because it is conscious, but because it attacks the cognitive middle: the reports, analyses, translations, memos, customer interactions, marketing plans, compliance forms, and administrative judgments that filled the desks of the educated class.
The old advice was: get educated.
The new requirement may be: own the machine.
Paul Krugman recently framed the larger economic shift as the “new inequality.” The old inequality story focused on wages: the premium earned by college graduates, the split between high-skill and low-skill labor, and the idea that education would protect people from economic decline. But that story has been fading for years. The more important story now is the shift from labor to capital: a rising share of income now flows to profits and ownership, with ownership itself concentrated in fewer and fewer hands.
AI fits that pattern almost perfectly. It requires a lot of expensive hardware and systems. If it succeeds, almost all the gains go to the owners of the systems that replace the workers.
There will still be jobs. There may even be many jobs.
But work may survive while the wage system fails.
The current AI bubble will probably collapse at some point. I’ve written about that repeatedly. Hundreds of billions may be wasted. Today’s indispensable chip may become tomorrow’s back-of-the-rack junk, and become stranded assets before the debt incurred to build them is paid off. Some of the companies now selling the future won’t have one.
But bubbles are financial events. Technological substitution will be a tectonic shift.
Railroad bubbles burst, yet the railroads remain. The fiber-optic bubble burst, yet the networks remain. If the AI bubble bursts, the crash will kill companies, but it will not un-invent the machine.
Nor will it remove the competitive pressure. “Humanity won’t let this happen” sounds reassuring only if humanity is imagined as a single actor, thinking carefully and weighing consequences, then choosing wisely.
But “humanity” is companies competing with companies, countries competing with countries, investors competing with investors, militaries competing with militaries, and politicians trying not to be blamed for decline.
A democracy may regulate AI. It may restrict its use in law, medicine, education, journalism, entertainment, or public administration. Some of those restrictions may be wise and necessary. But another company may operate with fewer scruples. Another state may decide that lower costs, military advantage, cyber capability, or strategic autonomy matter more than preserving careers.
AI does not need permission from humanity. It only needs adoption by enough competitors to make turning away from it expensive.
The cracks show first where governments meet finance.
Modern government is financed by the wage economy for a simple reason: paychecks are visible, regular, and remarkably difficult for an ordinary employee to hide.
Capital possesses no such loyalty. It is slippery: it migrates across borders, morphs into intellectual property structures, and vanishes behind unrealized capital gains. If AI systematically replaces taxable payrolls with elusive corporate profits, the state loses its easiest tax base at the precise moment the social safety net requires reinforcement.
The paralysis extends instantly into the credit markets because financial adulthood is an underwriting game. Banks do not lend money to people based on their survival needs: they underwrite career trajectories.
I see this friction in my own life. As a retiree with assets and a clean credit history, I still watch traditional financial institutions hesitate when considering a credit card application. They’re engineered to read a payroll template. They don’t know how to handle an anomaly like me.
Now scale that institutional confusion to millions of people living on a Universal Basic Income (UBI) stipend. Will a bank extend a thirty-year mortgage to an applicant whose income is capped by a public voucher? Will a landlord lease an apartment to a tenant with zero career upside? A stipend can buy groceries: it cannot buy leverage, and without leverage, you can’t accumulate assets. UBI quickly reveals its true architecture: a floor with a ceiling bolted above it.
Worse: if the state does create a predictable, guaranteed stream of public money flowing into millions of bank accounts, financial vultures will immediately appear to harvest it. Payday lenders, gambling platforms, and digital rent-extractors will circle the UBI stipend like digital parasites. A guaranteed income stream is simply a money pool waiting to be swindled from the vulnerable long before it can build generational stability.
The politics will be uglier still.
If UBI is too low, it humiliates and traps people. If it’s high enough to provide a decent life, it will be attacked as rewarding idleness. If it permits modest pleasures – a birthday cake, a streaming subscription, a rotisserie chicken, a popsicle for a child – the usual monetary preachers will thunder about the decadence of it all. That happens loudly enough today. In a post-work future, the volume will rise to a roar.
A society that resents poor children getting popsicles is not ready for a serious conversation about UBI and post-work dignity.
That’s why the status of the payment matters. Is it charity? Welfare? Compensation? A citizen’s dividend? A share of national productivity? Hush money to keep the plebs quiet?
Or something colder and more practical, for UBI may actually be billionaire life insurance.
That’s not the way it’s usually sold. It’s usually presented as compassion, justice, or liberation from drudgery. Perhaps it can be some of those things. But if the owners of the machines capture the income while millions are told their work is no longer needed, redistribution becomes more than benevolence. It becomes social fire insurance for the people at the top.
Because capitalism has three fundamental stages.
In the first, strivers claw and fight to make enough to own a yacht.
In the second, they kick-ass and fight to see who can own the biggest yacht in the sewer.
In the third, they discover the tumbrils have arrived to take them to the guillotine.
That second-stage sewer is not merely decadence. It’s the world despoiled by the winners while they continue competing ruthlessly for status. The yacht isn’t just luxury. It’s insulation from the sewage of the society they despoiled.
The guillotine enters the story when that insulation burns through.
This isn’t a Marxist call for violence. It’s a warning about legitimacy because systems can delegitimize themselves. A ruling class that hollows out the wage system, concentrates ownership, resents redistribution, and mocks the displaced as freeloaders should not be surprised if the question stops being about economics and starts being about revolution.
But income is still not the deepest problem. Work also helps turn people into adults. We forget that the workplace has long been one of society’s primary maturing institutions. It teaches accountability, restraint, diplomacy, and the awkward art of cooperating with strangers who are under no obligation to love you.
By automating entry-level cognitive tasks, AI eliminates the apprenticeship tier entirely. The junior lawyer, rookie accountant, assistant designer, and so on: these roles were never valued solely for their immediate output. They were the training grounds for judgment. If AI eats away the lower rungs of the ladder, it destroys the mechanism of becoming.
The corporations executing this shift are transforming from social organisms into legal wrappers around leverage: a small group of founders, a pile of capital, and an army of software APIs. The result is not a utopian playground of universal leisure, but something that looks older, uglier, and deeply familiar: a feudal society.
Machine feudalism chimes with medieval Europe: a tiny aristocracy owning the digital engines of production, a dependent majority stripped of bargaining power, and a vast servant class doing the messy, low-status work that machines cannot yet execute cheaply enough. Elder care, plumbing, cleaning, emergency repairs in the middle of the night: the tasks least replaceable by computer code remain the very jobs capitalism least wishes to reward.
For half a century, the cultural elite told young people that safety lay in cognitive complexity. Learn to code. Go to university. Physical work was branded as the destination for those who failed the knowledge economy. AI flips that hierarchy on its head. The digital strategist is exposed; the residential plumber is secure. This inversion will be culturally explosive: a generation trained to believe credentials guaranteed status is about to discover that the work most resistant to automation is the exact work they were taught to look down upon.
And that brings us to the psychological heart of the issue.
The techno-utopian answer to mass unemployment is leisure. AI, we are told, will free humanity from drudgery! People will write poetry, make music, become cabinetmakers, study philosophy, tend gardens, volunteer, care for one another, and finally become fully themselves.
Some will.
In a small way, I live that. I’m retired. I have enough to live, yet I find deep satisfaction in writing, in working on a book tracing the invention of marine geophysics, in thinking, in writing stories, and in trying to make sense of a difficult world. I know that a life without wage work can be rich and purposeful.
But I am not sure I’m typical.
A society cannot simply hand people a stipend and say, “Go become interesting.”
Many people are held together by external structure: somewhere to be, and someone expecting them to be there. Like the refrain from A Chorus Line goes: “Give me a job and the rest of the crap will get solved.”
Remove that structure and some people might well flourish, but many others may drift.
Leisure is wonderful when it’s a relief from toil. Infinite leisure, though, imposed on people who have lost purpose, status, and proof of worth, can curdle into something uglier: ennui, resentment, addiction, conspiracy, cruelty, or entering into destructive actions out of spite.
H.G. Wells painted a world like this in The Time Machine. The Eloi were not simply a leisured class. They were comfortable, dependent, childlike, provided for – and less than fully human. Wells’ picture isn’t a policy forecast, but it is a warning. Comfort without challenge does not necessarily elevate. It may shrivel instead.
This is where work may matter most, and where a paycheck matters in a way economists often miss.
Robert Heinlein once wrote that real applause is the kind you can fold and put in your wallet. Crass, but true.
A paycheck is folded applause. It is a signal from the world that someone wanted what you did, that your time had value, that your presence mattered enough that someone exchanged money for it. That money bought food and shelter, but it also carried recognition.
Work tells people they belong. More than that, it tells them they have worth.
UBI may answer hunger. It probably won’t answer humiliation. It may keep the body alive while quietly gnawing away at the spirit.
Viktor Frankl, the Viennese psychiatrist and Holocaust survivor who wrote Man’s Search for Meaning after surviving Nazi concentration camps, argued that human beings are not sustained by comfort. They need meaning.
That matters because a post-work society cannot be judged only by whether people have calories, shelter, healthcare, and entertainment. A life can be materially fulfilled and still be vacant because maintenance is not meaning.
So where does that leave us?
Wage insurance is humane, but it is a transition policy – but a transition to what? Retraining helps if there is another durable occupation to train for, but a bridge is not the answer if it ends halfway across the canyon.
It’s easy to mistake my concerns for cynicism, but the difference between a cynic and a skeptic is that the skeptic has hope.
The cynic says the oligarchs have won, the machines will take the work, the people will get scraps, and nothing can be done. The skeptic says the danger is real, and that’s why we must name it while there is still time to shape an answer.
We do not examine a crumbling wall because we enjoy collapse. We examine it because we would prefer not to be under the roof when it falls.
The end of work is not about jobs. It’s about the social arrangement that used jobs to distribute income, maturity, and personal worth. If AI breaks that arrangement, the question cannot be answered by a clever chatbot, or a billionaire’s TED-talk patter about universal leisure.
If the future no longer needs most people’s work, then the question we had better ask before the owners of the machines answer it for us is:
What are human beings for?
Summary: This essay argues that the real danger of AI is not simply mass unemployment, but the collapse of the wage-based social architecture that modern capitalism depends on. AI may allow companies to replace large amounts of cognitive labour, shifting income from workers to the owners of machines, data centres, software systems, and capital. But workers are also customers, taxpayers, borrowers, apprentices, and citizens; if paycheques shrink, the effects ripple through demand, public finance, credit, adulthood, family formation, and political legitimacy. Universal Basic Income may become necessary, but it risks functioning as “billionaire life insurance” unless it provides dignity, protection from predation, and a real share in productive wealth. Beyond economics, work gives people structure, discipline, social recognition, and proof of worth. A post-work society cannot be judged merely by whether people are fed and housed, because maintenance is not meaning. The central question is not whether AI will replace jobs, but what can replace the income, responsibility, status, and human purpose that work once supplied.



This skeptic has flash scenes of Star Trek: The Next Generation. THE BORG.
Replacing human labor may be the last straw that tears the fabric of society. Is it any wonder the super rich buy islands with bunkers?
A very thoughtful and terrifying look into a future that may be inevitable. One of the sci-fi writers, many years ago envisioned the “machine wars” of the future.