What would humans do when work is no longer needed?

This series is adapted, with permission, from a Chinese-language series by our friend Marcus Ji, who looks at historical events through an economic lens. We have condensed and adapted the original for an international audience.

The previous instalment looked at the Black Death, when labour suddenly became scarce and wages doubled for a hundred and fifty years. The opposite is happening right now.

Instead of a plague removing workers, machines and algorithms are taking over the work itself. Human labour is becoming abundant rather than scarce. If the Black Death was a natural experiment in what happens when workers are suddenly rare, we are now living through the reverse experiment: what happens when they are increasingly unnecessary. The clearest place to see where it leads is in one of the oldest industries – farming.

In 1790, about nine in ten American workers farmed. Today, it is under two in a hundred. This did not happen because people eat less, or because no one wanted to farm. It happened because machines, fertiliser and better seeds took over the work that human hands used to do – America now grows more food than ever, using a tiny fraction of the number of workers. By one industry measure, a single American farmer now feeds 183 people a year.

Which raises the question this piece is about: when a society barely needs people’s labour, what do the people do?

Work is becoming unnecessary

The rise in how many people one person can feed can be illustrated in this table:

One fully mechanised worker growing nothing but corn could theoretically produce enough output to cover the yearly calories of around thirteen thousand people (though that worker is really the last pair of hands in a long chain of fertiliser plants, seed companies, machine makers and refineries).

The same split between output and jobs is now appearing in other industries too. Since 1979, American manufacturing output has grown about 80% while manufacturing employment fell from a peak of 19.3 million to about 12.8 million – output up, jobs down, the two lines parting around 1980 after an entire industrial age in which they had risen together. Digital industries push it further still: when WhatsApp was bought in 2014, 55 employees served about 450 million users. This is no longer a few people producing for the many. It is a very few producing for almost everyone.

The bargaining power reverses

The previous piece gave a formula: the price of labour is its scarcity multiplied by the bargaining power institutions allow. The inverse holds true as well. When labour becomes abundant, both parts of that formula fall: with many equally qualified people chasing each job, applicants lower the wage they will accept in order to be the one hired, and employers, now spoilt for choice, have far less reason to pay more. The evidence has been building for forty years. Labour’s share of national income has dropped about five percentage points since 1980; the economic pie is growing, but the slice going to labour is shrinking. Over the same period, the pay ratio between big American companies’ chief executives and their workers rose from about 21 to 1 in 1965 to 344 to 1 in 2022. This tilt has a structural cause: capital and software can move anywhere and copy themselves endlessly, while a worker cannot cross borders freely or clone himself – so when the two sides split what they produce together, the mobile side holds the stronger hand.

There is a hard paradox here. In the Black Death, workers gained leverage because their labour was both scarce and needed. But when there are more people than there is work, most people’s labour is no longer needed – and the less the economy needs you, the less power you have to bargain with. A strike in an empty factory means nothing. Yet the economy still needs those same people in another role: as customers. Production may stop needing workers, but consumption still needs people with money to spend. That gap – between what the economy needs to make and what it needs to sell – is why ideas like paying everyone a basic income are now discussed seriously rather than dismissed as charity. And the only lever the unneeded majority still holds to demand it is political rather than economic: automation does not shrink a vote.

What people actually do when you pay them not to work

So what do people do when they no longer have to work for money? This is not only speculation – it has been tested repeatedly, and each test was built around the same fear: that free money would make people stop working.

• The most rigorous is OpenResearch’s American experiment (2020–2023), which gave 1,000 low-income people US$1,000 a month, with no conditions, for three years. The outcome it was designed to measure was mass withdrawal from work. That is not what it found: recipients’ working hours fell by only about 1.3 a week. But it also found that the money did not move them into better jobs or change the direction of their lives – it eased hardship without redirecting it.
• Finland’s trial (2017–2018) gave 2,000 unemployed people €560 a month, and its government was hoping for the opposite of withdrawal: that removing the benefit rules which punish the unemployed for taking work would push more of them into jobs. That did not happen either – the effect on employment was roughly zero. Instead, wellbeing, trust, and mental health of the participants improved substantially.
• Kenya’s GiveDirectly programme tested whether unconditional cash in a very poor setting would sap the will to work. It did not reduce work; it changed the shape of it, as recipients moved from insecure odd jobs into their own small businesses.
• And a natural version of the experiment – decades of studies of lottery winners – asks what people do with a sudden windfall. Most keep working, trimming their hours only modestly.

Four separate bodies of evidence point the same way: unconditional money does not make people idle. The political philosopher Karl Widerquist put the deeper point plainly. What such schemes really give a worker is not money but the power to say no. When a decent option to walk away exists, taking a job becomes, in the strict sense, voluntary rather than necessary. What changes is not how much people work, but the starting point of every negotiation.

Work was never only about money

That people stay busy without a wage still leaves the harder question, because a job was never only about income. In 1930, the economist John Maynard Keynes predicted that within a century productivity would rise fourfold to eightfold and people would work fifteen-hour weeks. He got the productivity right and the hours wrong: the working week in rich countries fell from about 48 hours to 34–38, and then stopped falling. Some of the surplus went into competition for status, some into an uneven split of the gains, and some into the rising cost of the things machines have not made cheaper – housing, healthcare, education.

The deepest reason, though, is that a job supplies far more than income, and we know this from a natural experiment. In the 1930s an Austrian mill town called Marienthal lost its factory, and almost the whole town was thrown out of work at once, with relief payments keeping people fed. Sociologists moved in to see what people did with sudden, total free time. The result was the opposite of flourishing: men walked more slowly across the village, put their watches away as their sense of time dissolved, and borrowed fewer library books than before, despite having unlimited hours. From that study, the sociologist Marie Jahoda drew up five things a job quietly provides beyond pay:

• A structure to the day – a job decides when you wake, where you go, and how the hours are divided; without it, as in Marienthal, the sense of time itself dissolves.
• Social contact – for many people, colleagues are their main source of company outside the family.
• A shared purpose – working towards goals larger than yourself, alongside others doing the same.
• A sense of status – a recognised place and identity in society: “I am a nurse,” “I am an engineer.”
• A reason to be active – regular demands that pull you out of the house and into effort.

Pay is only the visible half of a job. Any life after work has to find a replacement for all five of these; whichever one it fails to replace is the one that collapses.

History offers both endings, and both are real. On one side, a great deal of science was done by people who never had to earn a living. Charles Darwin never held a job; he spent his time and his family’s fortune travelling, observing and studying, and gave us the theory of evolution. Today, millions of people build Wikipedia and open-source software for no pay at all. On the other side stands Rome. For centuries the Roman state handed out free grain to a few hundred thousand of its citizens – the corn dole – and staged public games to keep a large idle population fed and quiet. What that produced is remembered not as a flowering of philosophy but as “bread and circuses”, a byword ever since for a people kept passive by handouts and entertainment. Removing the pressure to work does not release a single hidden “true self”. It widens the range – more Darwins, and more drift. Which end a society lands on depends on whether it can grow new structures of purpose, and that has never been designed in advance. Every previous time, it grew on its own.

Change sides, don’t double down

That is the collective picture. For any one person living through the change, the same framework points to a few durable rules.

The first is that you cannot control the wave, only where you stand in it. In the sixty years after 1764, nothing slowed the mechanisation of spinning – not petitions, not protests, not machine-breaking. The hand-loom weavers from an earlier instalment are the model of the wrong position: as power looms crushed their wages, they wove longer and longer hours to survive, so their own hard work drove prices down further and ruined them faster. The rule that follows is simple – when your kind of work gets its own machine, move to a different kind of work rather than trying to out-work the machine.

The second is that guessing which jobs will be automated has a poor record, so bet on tasks instead. In 2016, a leading AI researcher advised against training new radiologists, sure the software would replace them; a decade later radiologists are in shorter supply than ever. What is more predictable is the kind of task AI takes over: repetitive cognitive work whose results can be checked quickly. And it tends to amplify the person using it – in trials it helped novices far more than experts, narrowing the gap between workers. The farmer did not try to out-plough the tractor; he climbed into the driver’s seat. The move now is to use the tool, not race it.

The third is that what has fallen is not only the cost of labour but the cost of trying. Many worthwhile projects were never started because the checking, research and repetitive work were too expensive to justify – and much of that is now cheap. Things that once needed a team and a year can be attempted by one person in a few weeks. The new openings sit in the freshly cleared ground just behind a bottleneck that has recently broken.

The question economics now has to answer

For three thousand years, economics answered one question: what do you do when there is not enough to go around? The next hundred years will force a different one: what do you do when people are no longer needed?

There is no answer yet. But twice before, labour looked like it was vanishing and instead changed shape. After the Black Death, the surviving peasants did not sink into idleness – they won higher wages, better terms and the freedom to move, and their descendants left the land for crafts and trades. In 1900, no one could say what nine in ten American farmers would do once machines took over the fields; the answer turned out to be whole categories of work that did not yet exist. The MIT economist David Autor found that about 60% of the jobs Americans held in 2018 belonged to occupations that did not exist in 1940. Whether the pattern holds a third time, no one can prove – what makes AI different from every earlier machine is that it aims at cognitive work itself, the very thing displaced workers moved into last time. But an unanswered question is not new. Twice before, societies faced one this large, and worked out an answer they could not have written in advance.

Previously in this series: Why did the Black Death lead to very different economic realities in Eastern and Western Europe?

Next in this series: A brief history of finance