Time ideas May 30, 2026 11:00 AM CUTPhoto-Illustration by TIME; Source Image: David Trood—Getty Imagesby Christopher Marquis is a professor at the University of Cambridge and the author of "The Profiteers." May 30, 2026 11:00 AM CUTIn 1930, in the depths of the Great Depression, John Maynard Keynes wrote a short essay called Economic Possibilities for our Grandchildren. It is often remembered for one striking prediction: by 2030, people in wealthy countries might only need to work about 15 hours a week.What Keynes imagined was a society advanced enough to solve what he called the “economic problem” of basic material provision. If technology kept improving, and societies kept growing richer, then fewer hours of human labor would be needed to produce the necessities and comforts of life.Nearly a century later, that old prediction has returned in distorted form amid the AI boom. Tech executives, investors, and many politicians speak as if a new wave of intelligent machines is about to transform labor—automatically and at a planetary scale. While many warn of mass unemployment, some promise liberation from drudgery—that AI will free up time for workers to do as they please. And then, others make the opposite claim: that AI’s efficiency will create more work, not less. Drawing on the Jevons paradox—the idea that when a technology becomes more efficient and cheaper to use, total demand for what it enables can rise rather than fall—this view posits that the routinization of cognitive work will expand the need for “relational work,” and the amount of human communication and coordination organizations want to do.Underpinning all of these competing forecasts is the same false assumption: that technological capability itself determines social outcomes. AI may be able to automate tasks, generate new demand, or reduce the need for human labor. But by focusing on what AI can make possible, the deeper questions these predictions miss is who will control those possibilities, and beyond job loss or gain, what kind of work is left behind. History has shown that gains in productivity within firms rarely expand workers’ freedom or improve working conditions, but instead are used to make them more measurable, replaceable, and controlled.Machines and the means of productionWhat Keynes underestimated, and what many of today’s AI evangelists tend not to confront, is that progress unfolds inside social systems structured by power. In today’s capitalist system, machines do not simply lighten labor; they are deployed according to the interests of firms, investors, and managers. The important question we face now is how the transformation from AI will be organized, and crucially, whether it will widen human freedom or tighten control, concentrate wealth, and extract still more from labor. Like Keynes, we fail to acknowledge that technology is more than a simple means of producing more with less effort. It is also a means of reorganizing labor so that business owners can extract more value, rely less on worker skill, and extend managerial control. New AI tools can displace labor (which recent high-profile layoffs demonstrate), but we also need to consider more fully how AI tools can also cheapen labor, fragment it, and make workers more interchangeable. Early machines, for instance, could spin, weave, stamp, and cut far faster than individual artisans. Yet these technologies did not lead to long-term elevated unemployment, nor lighten the burdens put on workers. Factory production moved work out of skilled, relatively autonomous craft systems and into centralized regimes of supervision, discipline, and timing. Workers did not just do less of the old labor; they entered a world of clock-based work, managerial oversight, and repetitive tasks. The machine changed who controlled labor and the terms on which it was done.Who reaps the benefits of increased productivity?The same pattern repeated when assembly-line production expanded in the early 20th century. It’s easy to imagine that, because productivity rose dramatically, such efficiency should have meant much less human labor. In practice, the system of mass production pioneered by Henry Ford made work more fragmented, monotonous, and physically and psychologically demanding. Craftsmanship was stripped away, and the pace of labor was subordinated to the line itself. Technical progress produced a new regime of intensified extraction.The computer revolution followed this pattern, too. When I started work in the 1990s, many people talked about the paperless office and a dramatic reduction in routine work. Some clerical tasks certainly disappeared or changed. Yet office work expanded, as recent invocations of the Jevons paradox rightly note. But the lesson is not simply that cheaper tools create more and different types of demand and with that new occupations. It is that, inside firms, efficiency gains were rarely passed on to workers as free time. Instead, they were typically used to help businesses become more profitable. For instance, computers and email shifted administrative and communicative labor onto professionals themselves: managers, academics, and lawyers were expected to format documents, manage calendars, monitor inboxes, and respond continuously to a growing stream of messages. These tools made individual tasks more efficient, but the savings were quickly absorbed into constant availability and an increase in the volume of work that could be demanded. Jevons paradoxThis leads us to recent discussions of the Jevons paradox, which can obscure as much as it reveals. True, in Jevons’ original example, Watt’s steam engine made coal more productive, and Britain consumed more coal, not less. A more recent technological example people will cite to support this line of thinking is the LED lightbulb. As these energy-saving bulbs became cheaper and more efficient, societies found more occasions for lighting rather than simply consuming less electricity.But when considering work, the crucial question is not only whether efficiency lowers cost and so creates more demand. It is who controls (and benefits from) the efficiency gain and how work is reorganized as a result. AI may well create more tasks and new forms of demand. But within today’s business environment, that expansion is most likely to arrive through the familiar mechanisms we saw with mechanization in the industrial revolution, the assembly line, and the computer: deskilling, monitoring, and accelerating work. Perhaps the most concerning side effect: the transfer of judgment—and power—from workers to systems.Early use of algorithmically controlled work shows similar patterns. Gig platforms were sold as a new “free agent” economy, promising autonomy, flexibility, and freedom from the traditional boss. In practice, many gig platforms have become systems of control without the protections of employment: software sets prices, routes workers, and disciplines behavior through ratings, penalties, and opaque incentives. All the while, many of the companies have shed their need to provide benefits and other human resources, and their investors have become rich. The politics that Keynes missedThe lesson we should take from all of these examples is that Keynes’ error was not that he underestimated the transformative power of technology, but that he underestimated the power of capital’s drive for expansion and its tendency to put financial return above all other considerations. He assumed that we may reach a point when “needs are satisfied in the sense that we prefer to devote our further energies to non-economic purposes.” But capitalism has rarely treated “enough” as a stopping point. Each productivity gain becomes a fresh opportunity for business owners to maximize their own gains while using the same technology to further control their employees. Indeed, one could argue that the legal structures of modern capitalism requires leaders to maximize gains to boost shareholder value. Whether the advent of AI leads to mass unemployment, degraded working conditions, or a shorter and better working week will depend on much more than the capabilities of the technology. Rather, institutions, bargaining power, ownership, and managerial strategy will truly shape our society in the AI era.This is where authority matters. Sadly, too many of the people building and promoting AI portray their private-sector ambitions to dominate markets, build government-subsidized infrastructure, capture consumer data, and weaken labor power as historical inevitability, a supposed attempt to normalize disruption as a natural force instead of a political and institutional choice. This is why recent appeals to Jevons paradox, and how the need for human interaction may mean less job loss than prior predictions by AI leaders such as OpenAI’s Sam Altman are so revealing. By centering the debate on whether AI will create or destroy jobs, these perspectives obscure the deeper question at the heart of the issue: whether AI’s productivity gains will expand workers’ freedom or deepen firms’ power over them. Political leaders and regulators have failed in parallel, whether through misunderstanding, passivity, or fascination with the technology itself. Rather than treating AI as a matter of labor standards, surveillance, competition, and accountability, they have often framed it as a race to be won or a sector to be lightly managed.Recent episodes of students openly booing AI-themed graduation speeches shows the unease today’s youth have about whether an economy organized around machine intelligence will still protect human creativity and judgment. Pope Leo XIV has also raised a similar warning, framing AI as a moral and political test of whether technology will serve humans or concentrate power in ways that diminish our dignity and flourishing.The central question, then, is whether democratic societies will shape AI before firms and technologists define its terms for everyone else. There is still an opportunity to recover the better part of Keynes' vision, but only if politicians are willing to govern AI before its power becomes entrenched. This would include stronger protections for workers whose tasks are being automated or degraded, firmer limits on surveillance and data extraction, protections against a few dominant firms setting the terms of AI adoption, and rules that direct productivity gains toward shared security rather than deeper precarity. Without taking these kinds of actions, warnings about job loss will remain in vain and incomplete. AI may eliminate some jobs, but it is also likely to deepen a quieter transformation: the movement of judgment, discretion, and power out of workers’ hands.Forecasts about technology throughout history have repeatedly failed because they treated machines as the engines of social change while neglecting the institutions, power, and interests shaping their use. Keynes’s lost 15-hour workweek was not simply a failed forecast about productivity. It was a question about whether societies would use technological abundance to enlarge freedom or to deepen accumulation. Keynes saw the possibility. We invented the technology. The issue is that we still have not invented the necessary politics.TIME Ideas hosts the world's leading voices, providing commentary on events in news, society, and culture. We welcome outside contributions. Opinions expressed do not necessarily reflect the views of TIME editors.