This article was translated automatically by AI.
Here is part 1 of the six-part series. You can find the opening article here: The Cello in the Engine Room
In 1930, John Maynard Keynes promised his grandchildren a 15-hour week. He predicted the prosperity correctly, but not the freedom. Now, for the first time, AI is hitting the well-paid office work of the middle class – and Hannah Arendt's warning about a society of laborers without labor is drawing ominously close.
Why John Maynard Keynes was wrong about his 15-hour week, what the Austrian Court of Audit's income reports reveal about our true situation – and why Hannah Arendt's most famous warning is suddenly becoming reality in the age of AI.In June 1930, amid the upheavals of the Great Depression, the British economist John Maynard Keynes traveled to Madrid to deliver a lecture entitled "Economic Possibilities for our Grandchildren." Instead of joining in the cultural pessimism that was customary at the time, Keynes ventured a thought experiment about the world a hundred years on – in other words, almost exactly about our present day in 2026.
Keynes predicted two things. First, technological progress and the accumulation of capital would raise the standard of living in the Western industrial nations four- to eightfold within a century. Second, this enormous productivity would essentially solve humanity's economic problem – the struggle for material survival. For us, his symbolic grandchildren, Keynes therefore foresaw a new problem looming: how do we meaningfully fill our newfound freedom when no one has to work more than 15 hours a week?
Almost a hundred years later, the verdict on this famous prophecy is as astonishing as it is sobering. With his first prediction, Keynes was spot on: per capita economic output in Western Europe and the US has in fact more than quintupled since 1930. As a society, we are more productive and materially prosperous than any generation before us. But with his second prediction, he was spectacularly wrong: a relaxed 15-hour week for everyone is out of the question. Instead, we are experiencing a grotesque paradox of overwork on the one hand and structural exclusion on the other. Why is that – and what happens when artificial intelligence now slams into this lopsided system with full force?
What the numbers reveal: the decoupling of productivity and wages
To understand why technological progress has so far failed to bring us the freedom Keynes hoped for, a sober look at the statistics is enough – for instance at the general income reports of Statistik Austria or the long-term data from WIFO and the OECD.
For decades in the postwar era, an unwritten law held: when productivity rises thanks to better machines and smarter organization, workers' real wages grow at the same pace – or working hours fall with full wage compensation. But since the 1980s and 1990s, this bond has been broken. While labor productivity kept climbing steadily through digitalization and automation, median earned incomes, adjusted for purchasing power, remained virtually flat for decades.
The data from Statistik Austria and WIFO clearly show this imbalance: since the early 1980s, the share of labor income in value added – the wage share – has declined for decades, while an ever-larger portion flows into corporate profits and returns on capital. At the same time, the labor market has split. We maintain an army of several hundred thousand people in unemployment, training schemes or involuntary underemployment – and in public discourse we often go so far as to denounce these people as "lazy," even though the number of job vacancies in no Western country would ever be sufficient to create genuine full employment for everyone able to work.
That we nevertheless do not work 15 hours, as Keynes hoped, has two reasons. On the one hand, consumer capitalism has immediately filled every window of time that opened up with new status needs and rising housing and living costs, so that an average income is often barely enough to support a family securely. On the other hand, as the anthropologist David Graeber aptly analyzed, the economy has created millions of administrative "bullshit jobs" in bureaucracy, reporting, compliance and self-administration, merely to uphold the dogma that every person must look busy forty hours a week.
The AI shock: when automation changes workplaces
It is precisely into this bloated architecture of knowledge and administrative work that generative artificial intelligence is now striking. And it differs fundamentally from all previous waves of automation.
When the power loom, the steam engine and later the industrial robot entered the factories in the 19th and 20th centuries, it was above all physical and manual routine work that was hit. The middle class's way out back then was education and the move up into the office. Anyone who studied, programmed, wrote texts, audited balance sheets, drafted contracts or analyzed markets was considered immune to the machine.
Current labor market studies – from the International Labour Organization (ILO) to Goldman Sachs to analyses by the IMF – paint a completely reversed picture for the developed economies, however: between 40 and 60 percent of all current jobs in Europe and North America show high "exposure" to AI. And for the first time in economic history, it is not construction sites, the skilled trades or care work that are most easily replaced, but precisely those well-paid white-collar professions of the middle class: clerical work, financial analysis, legal groundwork, marketing, translation, customer service, software development and administration.
From the Solow paradox to a structural dam break
Many companies and politicians are still lulled into a false sense of security because the really big clear-cut has not yet shown up in the statistics. But that is due to a phenomenon economists have known since 1987 as the "Solow paradox": when the first personal computers moved into offices back then, Nobel laureate Robert Solow wondered why you could see the computer age everywhere – except in the productivity statistics. The reason was simple: the new computers had just been placed on the old desks, without changing the workflows themselves.
This is exactly the point on the so-called productivity J-curve where we find ourselves today with artificial intelligence. In many companies, employees already use chatbots and copilots and complete individual subtasks five times faster – but as the management theory of bottlenecks (Theory of Constraints) teaches, any acceleration fizzles out into nothing as long as the organization around it stays the same: people then simply send even more emails, reports and presentations back and forth in less time.
But this is precisely what is now changing radically. With the transition from mere text chatbots to autonomous AI agents, companies and financial investors are no longer just speeding up individual tasks but are beginning to rebuild entire value chains from the ground up (zero-based design). Where a team of five junior analysts, copywriters, graphic designers or programmers once spent a week preparing data and producing drafts, today a single experienced professional steers a fleet of digital agents and does the same work in a morning. For the company's balance sheet, this is the longed-for breakthrough. For the labor market, however, it means that the classic entry-level ladder for young university graduates and skilled workers is being sawn off rung by rung.
Hannah Arendt's warning: when a society of laborers runs out of labor
We are thus standing at precisely the historical abyss that the philosopher Hannah Arendt described with almost uncanny foresight back in 1958 in the prologue to her magnum opus "The Human Condition." Writing at the time of the first debates about automation, Arendt observed:
"What we are confronted with is the prospect of a society of laborers without labor, that is, without the only activity left to them. Surely, nothing could be worse."
To grasp the full sharpness of Arendt's diagnosis for the age of AI, one has to know her famous threefold division of human activity. Arendt drew a strict distinction between three forms:
Labor: the endless, constantly repeated toil that serves biological survival, consumption and the maintenance of everyday life. No sooner is the result consumed than one must begin all over again.
Work: the craftsmanlike, artistic or intellectual creation of lasting things – a table, a book, a building, a reliable institution – that lend our world permanence and stability.
Action: the highest and most human form of activity: speaking and acting directly between people in the public and political sphere, without the detour via things or mere usefulness. This is where human beings reveal themselves in their uniqueness.
Arendt's devastating finding about modernity was this: in the course of industrialization and the economy of utility, we have almost completely devalued action and genuine work, reducing human beings to a mere animal laborans – a laboring and consuming animal. We have turned even our highest intellectual activities into "jobs" that serve only to earn money and keep the economic process running.
And that is exactly why the AI revolution catches us so unprepared at our most vulnerable point: because for two hundred years we have trained people to define themselves exclusively through paid employment, we experience the disappearance of this work not as the liberation for work and action that Keynes dreamed of, but as an existential devaluation.
Overcoming the false alternative
"Endless work or the end of work?" – as long as we remain stuck in the logic of the past thirty years, both lead to disaster. In the age of AI, "endless work" means a desperate race of human against algorithm, in which employees have to deliver ever more output at ever higher speed, to the point of complete exhaustion. The "end of work," in turn, means under today's system nothing less than the loss of income, social participation and dignity for millions of people.
In the 21st century, we are not running out of meaningful activity – there is more than enough to do in care, in education, in the ecological transition, in culture, in the skilled trades and in human interaction. What we are running out of is the kind of paid work that, by the standards of the 19th and 20th centuries, can be sold profitably on a purely private-sector market.
If we want to prevent the productivity gains of AI from merely filling the accounts of a few tech monopolists while the middle class erodes, we have to pose the questions of distribution and organization anew. The good news is that we do not have to reinvent the wheel entirely. Austria and Central Europe in particular have a unique historical experience of taming technological upheavals through smart institutions of solidarity.
"The Bridge Builders": Austria has managed this once before. When the machines of the 19th century shattered the old order, citizens, farmers and workers invented savings banks, cooperatives and chambers. Next week: why, of all things, the much-maligned chambers could be our best answer to AI – and what has to change about them for that to happen.
Published in European Cultural News (ISSN 2308-085X) on 16 November 2026
https://european-cultural-news.com/en/endless-work-or-the-end-of-work-ai-and-the-labor-market/59660/


