The Child Born Today May Never Need a Job

A child born today, in September 2026, will turn eighteen in 2044. What should we tell that child to prepare for? For generations the answer was obvious: study hard, learn a profession, acquire skills that society values, exchange them for an income, build a career.

There is now a serious possibility that by the time today’s newborn enters adulthood, that framework will be obsolete.

The reason is not simply that artificial intelligence will take some jobs. We have seen technological unemployment before. Agriculture displaced hunters and gatherers, machines displaced farm labor, computers automated clerical work. What is different now is which resource is being automated. AI is beginning to replace the one input every previous revolution continued to depend on: human intelligence itself.

From muscle to intelligence

The economic history of civilization can be read as a succession of constraints. For hunter-gatherers, survival depended almost directly on human effort. Agriculture let one person’s labor feed others, and that surplus made cities, governments, armies and scholars possible. The Industrial Revolution replaced human and animal muscle with coal, oil and electricity, so that one machine could do the physical work of thousands.

But industrial civilization still needed people to decide what the machines should do. The premium shifted to the brain. Engineers designed, scientists discovered, physicians diagnosed, managers coordinated, programmers instructed, investors allocated. The modern white-collar economy is, to a remarkable extent, built around the scarcity of competent human cognition.

Artificial intelligence potentially removes that scarcity too.

OpenAI’s GPT-6 Astra, released in September 2026, is designed to perform complicated computer-based work rather than merely answer questions. On OpenAI’s own figures it scores 97.6% on FrontierMath Tier 4, a benchmark of research-level mathematics, against 87.8% for its predecessor GPT-5.6 Sol, and 99.9% on ARC-AGI-3, a reasoning test built to stay ahead of AI, where Sol scored 7.8%. On OSWorld 2.0, which measures operating a computer unaided, it finishes tasks in 47% less time than Sol.¹ It is also the first OpenAI model for which previous AI models played a large role in supervising training. OpenAI described that as progress toward recursive self-improvement, even though Astra is not itself an autonomous recursively self-improving system.² The company’s stated goal is an AI system that can “accelerate and increasingly automate the research process itself,” with a significant fraction of its own research done by AI by March 2028.³

If it succeeds, the loop closes. AI develops better AI. The improved AI develops better algorithms, chips, materials and robots. Those improve manufacturing and energy production, which permits still larger AI systems, which perform more science and engineering. Human beings progressively drop out of the productive feedback loop. That is not automation in the familiar sense. It is the emergence of an increasingly autonomous technological civilization.

What happens when nobody needs our labor?

The endpoint of that loop is machines that outperform humans at virtually every economically productive activity, from diagnosing disease and designing structures to writing software, discovering drugs, managing companies and allocating capital, and robots that do the physical work of farming, mining, manufacturing, construction and transport.

At that point the conventional reassurance, that humans will supervise the AI, stops making sense. If an AI is vastly better than a physician at diagnosis, it will also be better at checking another AI’s diagnosis. If it is better than the best scientists at science, it can evaluate scientific research better than they can. Human oversight might remain because society legally or morally requires it, but not because humans possess superior expertise. Eventually there may be nothing economically necessary for a human being to do.

For hundreds of thousands of years, humanity’s central problem has been producing enough: enough food, shelter, energy, medicine and manufactured goods. Economics exists because useful resources were scarce and human wants were not. An automated system that produces food, housing, transport, medicine, education, software and manufactured goods at very low marginal cost, with abundant energy and increasingly automated resource extraction behind it, changes the problem. It is no longer how to produce enough. It is how to distribute abundance.

Marx’s technological assumption

This leads unexpectedly back to Karl Marx. The communist societies of the twentieth century are poor models here. The Soviet Union and Mao’s China operated under severe material scarcity. They desperately needed farmers, miners, factory workers, engineers and physicians, and scarcity forced rationing, production targets and administrative decisions about who received what. Whatever their ideology, they were economies in which human labor was indispensable.

Marx’s theoretical endpoint was different. In his 1875 Critique of the Gotha Programme he argued that the principle “From each according to his ability, to each according to his needs” could apply only after productive capacity had increased enormously and society’s cooperative wealth flowed abundantly.⁴ His mature communist society implicitly required something close to post-scarcity economics. That condition never existed.

Perhaps AI changes that.

There is also a modification Marx could not have anticipated. If machines perform essentially all economically useful work, the first half of his formula becomes unnecessary. Human productive ability is no longer required; the machines supply it. What remains is: to each according to need, or desire, subject to whatever remains genuinely scarce.

But Lake Tahoe still exists

Even an extraordinarily abundant civilization cannot eliminate every kind of scarcity. There can be billions of beautiful homes, but only so much shoreline on Lake Tahoe. There can be unlimited recordings of a concert, but a finite number of seats in the room. There is one original Mona Lisa.

How should those scarce things be allocated? Today the mechanism is money, and money is obtained through some combination of labor, entrepreneurship, investment and ownership. If AI is better than every human at all four, why should Alice get the Tahoe mansion rather than Bob? Because she is a better engineer? The AI is vastly better than either. Because she worked harder? Her extra work adds essentially nothing to civilization’s output. Because her grandfather happened to own the company that owns the robots? That is legally coherent under today’s capitalism. But when automated capital generates nearly everything society consumes, inherited ownership of that machinery begins to resemble inherited ownership of society itself.

New mechanisms for allocating the remaining scarce goods would be needed: auctions with equal resource credits, lotteries, rotating access, collective ownership, or markets for positional luxuries alongside free provision of ordinary needs. Which of these, nobody knows. But one principle changes: economic contribution can no longer justify unequal access, because human economic contribution itself has ceased to matter.

Capitalism may have a problem too

None of this means Marx was right or that a Soviet-style economy becomes desirable. Something stranger could happen.

Capitalism is an extraordinarily effective system for coordinating scarcity. Prices reveal what is scarce, profits reward good allocation, markets coordinate billions of decentralized decisions. But what if superhuman AI can optimize production, logistics, discovery and allocation at a scale beyond human institutions, and the marginal cost of many important goods approaches zero? Markets would remain useful for genuinely scarce resources. Large areas of everyday life could become decommodified simply because charging for them serves little purpose. We already see a primitive version with information: Wikipedia does not meter articles because one person’s reading imposes no scarcity on another’s. Future generations might regard basic computation, education, transport or manufactured goods the same way.

The eventual system might resemble neither twentieth-century capitalism nor twentieth-century communism: a post-labor economy with universal access to abundant goods and markets for scarce ones. We do not yet have a name for it.

So what should the child born today learn?

If recursive AI progress continues for another eighteen years, asking which occupation that child should prepare for may make as little sense as asking a modern American child which subsistence crop they intend to grow.

The instinctive answer is that children should learn what AI cannot do: creativity, leadership, empathy, judgment. That answer may only reflect our difficulty imagining intelligence substantially beyond our own. Why should those capabilities uniquely remain human? If AI surpasses human cognition broadly, there may be no economically valuable refuge left to retreat into.

Education would then need a different purpose. A child would study mathematics not because it provides a career but because it is worth understanding. Study history to understand humanity. Learn science because the universe is fascinating. Learn music because making it is pleasurable. Exercise, travel, build things, compete, make art, fall in love, raise children. None of these requires being better than an AI. People still run though cars are faster, still play chess though computers surpassed the world champion decades ago, still cook though machines can prepare food. Human activity does not need economic superiority to have meaning.

For almost all of human existence the question has been what a person must do in order to live. A child born today may belong to the first generation for whom the question becomes what to do with a life when nothing requires them to do anything.

If that happens, AI will have completed a journey that began when the first human fashioned a tool to reduce the effort required to survive. It will have done what technology has always been trying to do: eliminate the necessity of human labor. And humanity will face a problem for which neither capitalism nor communism has prepared us:

What is civilization for once scarcity and work are no longer what organize human life?

Sources