AI Can End Labor. It Cannot End the Need to Be Needed.

November 17, 2026

In 1930 the textile factory in Marienthal, a village south of Vienna, closed. Almost the whole village lost its work at once. A team around Marie Jahoda, Paul Lazarsfeld and Hans Zeisel went there to study what happens to people when the work goes away.

One of the things they looked at was the village library. Before the closure, people paid a small fee to borrow books. After the closure the fee was dropped. People had more free time than ever, and the books were now free.

Loans fell by almost half between 1929 and 1931.

Lazarsfeld wrote that the people did not know what to do with the time on their hands, and that they even stopped the things they had done before. The study team expected the opposite. So would most of us.

I keep coming back to that library when I read the optimistic case for a world after work.

The debate

I recently read an exchange on X between two people who think hard about this, one asking and one answering. It is the best version of the argument I have seen.

The question side goes like this. Smith and Montesquieu thought commerce made us better people, because it gave us a reason to improve ourselves and to make things of value for others. If AI makes human labor unnecessary, what happens to that reason? Mozart and Dickens wrote for money. The gentleman scientists of Victorian England did not need money, but they lived in a commercial society and they believed their work would not exist if they did not do it. So the question is not whether one person can live a good life of leisure. It is what a whole society looks like.

The answer side is optimistic. Humans have always chased excellence for its own sake. Grigori Perelman proved the Poincaré conjecture and turned down the Millennium Prize. Alex Honnold climbed El Capitan without a rope. People speed-solve Rubik's cubes. Free of economic pressure, we will see more of this, not less. The institutions already exist in mathematics: prizes, societies, recognition. Patrons will fund the rest, like the Medici did.

I think the answer side is right about the people it names. I also think it names the wrong people.

The outliers are not the question

Perelman, Honnold and da Vinci are proof that some humans have enormous intrinsic drive. Nobody doubts that. But they are the tail of a distribution. If we design a society around them, we design it for maybe one in a hundred people.

The question side asked about the general equilibrium, which means everyone. And the record for everyone is not good.

Marienthal is one data point. A larger one is what Anne Case and Angus Deaton found in the United States. In 2015 they showed that death rates for middle-aged white Americans without a college degree had stopped falling and started to rise, driven by drugs, alcohol and suicide. In their book they called these deaths of despair, and they tied them to the slow loss of good work and of everything that came with it: a role, a routine, a place in a community.

The optimist will say, correctly, that Marienthal was poverty and the American Rust Belt was decline. Those people lost their income, not only their work. A world of AI abundance takes the poverty away.

That is a fair point, and there is evidence for it. Erik Lindqvist, Robert Östling and David Cesarini followed Swedish lottery winners for years. A large win made people more satisfied with their lives, and the effect lasted more than a decade. Money did not ruin them. But the same researchers also looked at work, and the winners cut their earnings only a little. They did not walk away from their jobs. They kept the role. So the lottery studies show that money does not destroy people. They do not show that people do well without being needed.

That is the thing I think the debate misses. We can survive hardship. Humans have done it for most of history. What breaks people is meaninglessness, the feeling that nothing they do changes anything for anyone.

What the gentleman scientists had

The question side made one point that the answer side did not really answer. The great people of the past were driven by consequential agency. If they did not do the work, they believed it would not get done. Thucydides was writing about a war his city was losing.

Honnold has real stakes too, of course. If he makes a mistake, he dies. That is why the free solo is different from a climbing game on a screen. And the same is true for the sailor the question side mentions. Why sail at all in the age of motor boats? Because the sea is real. It does not care about you, and it does not get easier because you have a better phone.

This is the part I find hopeful, and it is not where most AI optimists look.

Abundance solves the problems that can be solved once. Make energy cheap and it stays cheap. Write the software and it stays written. But there is a large class of problems that never get solved once, because they are alive.

A reef does not get restored and stay restored. Seagrass has to be planted, watched and replanted. Soil has to be built up every season. Forests burn and grow back. Old people need someone to sit with them today, and again tomorrow. None of this is a puzzle you finish. It is care, and it never ends.

I wrote last week about Tom Chi and Reefgen, the robots that plant coral and seagrass back onto the sea floor (Don't Build the Company). Chi has said one robot planted 10,000 seagrass seeds in a single day. That sounds like the end of human work in the ocean. It is the opposite. A robot that plants an acre a day means a person can now take care of a whole bay. Somebody still has to decide where, watch what grows, and come back next year.

That is work with real stakes, and AI does not remove the stakes. If the bay dies, it is dead. The person who looks after it knows that the work would not exist without them. This is exactly what the gentleman scientists had, and what the median person in Marienthal lost.

Status moves, it does not disappear

There is a second problem with the optimistic picture, and it is about status.

The market was never only a way to get money. It was a ladder. You could see who was doing well and why, and most of the ways up the ladder meant making something that someone else wanted. That is Smith's point about commerce making us better, and it is also his older point, from The Theory of Moral Sentiments in 1759, about beneficence: we want to be admired, and we want to deserve it.

Take the market away and the need for status does not go away. It moves. People will compete for attention, for land in good places, for access, for recognition. We have a preview of what that looks like without any output attached to it. It is social media.

So the design question is not how to remove the ladder. It is which ladders we build. The answer side is right that prizes and societies work well in mathematics. The rule I would add is simple. A good status ladder is one you can only climb by making something for other people. A restored bay, a proof, a piece of music, a well-run village library. A bad one rewards attention for its own sake.

The Medici problem

The answer side points to patronage as the model for funding excellence after work. The Medici paid for da Vinci, so future patrons will pay for the next one.

I think this is half right. Patronage works. But Medici patronage put almost all of the taste of a city into a few rich hands. A small number of families decided which art and science existed. That is a fragile way to run a civilization, and it is the opposite of what made the commercial society the question side admires, where millions of buyers decided.

The better model already exists in small form. Thousands of small patrons instead of a few big ones. Members who pay a few euros a month for an open project, like the people who fund Precious Plastic. Volunteers who build Wikipedia and most of the open source software the world runs on, for no pay and with a clear system of who did what. None of these are perfect. But they spread the choice of what matters across many people, and that is closer to a market than a court of patrons ever was.

Where to look for precedent

The question side asked which historical groups we should study. Victorian gentlemen and Periclean Athens are the usual answers, but both were small elites on top of a society that worked. I would look at groups where ordinary people found purpose without a wage.

GroupWhat it shows
Monastic ordersA structured day and work as a discipline, with no pay and no market
Open source and WikipediaExcellence and status with no price attached
Learned societies, like the Royal SocietyRecognition from peers as the main reward
Sailing and expedition crewsReal stakes, a clear role for each person, and a choice to reject the easier technology

What these groups have in common is not money and not leisure. It is a role that other people depend on.

The transition is the dangerous part

I am not a pessimist about the end state. I think a world with cheap energy, cheap intelligence and people who take care of living systems because they choose to is a good world, maybe the best one we have had.

The danger is the space in between. That is the time after the old jobs go and before the new roles exist. Marienthal is what that gap looks like when nobody fills it. The library is right there, the books are free, and people stop reading.

So I would put the question from the debate a little differently. Not "will we still get the next Mozart?" We probably will. The better question is whether the ordinary person will still be needed by someone, and whether we build the places where that can happen before we need them.

Some of those places look like reefs and bays and soil. Some look like libraries. Most of them will need people to show up, again and again, because the thing they look after is alive.

AI can end labor. It cannot end the need to be needed. That part is still up to us.