AI, robotics and autonomy won’t literally add another hour to the day, but they may create the psychological equivalent by removing routine work. The enormous question becomes what humans do when technology begins returning meaningful quantities of time to them.

Almost a century ago, John Maynard Keynes made one of the most famous wrong predictions in the history of economics. Writing in 1930, he argued that rising productivity would shrink the workweek to just fifteen hours by the early 21st century, and that humanity’s biggest looming problem would be figuring out what to do with all that leisure. He called it, without a trace of irony, “the greatest change which has ever occurred in the material environment of life.”

Keynes got the productivity part exactly right. He got the leisure part almost entirely wrong. Americans worked 38 hours a week on average in 1950. Today, nearly a century into the productivity revolution Keynes foresaw, we average 34. Four hours. That’s the entire dividend of a hundred years of electrification, computers, the internet, and cloud software.

I think about that gap constantly as I watch AI, robotics, and autonomous systems start doing to routine cognitive work what machines already did to routine physical work a century ago. The technology really is capable of handing back meaningful chunks of time. The open question — the one Keynes got catastrophically wrong — is whether we’ll actually keep any of it.

The Hour That Keeps Vanishing

Here’s the uncomfortable pattern showing up in the earliest hard data on AI and work: the hours aren’t disappearing into leisure. They’re disappearing into more work. Researchers studying nearly two decades of time-use data found that employees in AI-exposed occupations are working longer hours, not shorter ones, and that the effect has gotten stronger since generative AI tools went mainstream. Workers in high-AI-exposure jobs picked up roughly three extra hours of work per week in the period right after ChatGPT’s public release, compared to workers in occupations AI barely touches. The freed time didn’t evaporate into a long weekend. It got quietly reabsorbed into the job.

This isn’t just an academic finding — it’s something researchers have documented across a strikingly consistent range of studies, with one European analysis putting the number at only about a fifth to a quarter of employees actually redirecting AI-saved time toward their personal lives. The rest simply gets reinvested into more output. Economist Cyril Northcote Parkinson diagnosed the underlying mechanism back in 1955, decades before anyone had heard of a large language model: work expands to fill the time available for its completion. Hand a motivated professional four newly efficient hours on a Monday, and watch the calendar quietly swallow them whole by Thursday.

Technology may create the 25th hour, but it won’t give it to you. The future of free time belongs to those who claim it.

Why the 25th Hour Isn’t Automatic

This is the central insight I want you to sit with: the 25th hour doesn’t arrive as a gift. It has to be actively claimed, the same way a raise has to be negotiated rather than simply expected. Technology creates the capacity for more free time. It does not, on its own, create the free time itself. Those are two entirely different things, and confusing them is exactly the mistake Keynes made.

Consider a concrete example. A financial analyst who used to spend six hours building a quarterly report by hand can now generate the same report with AI assistance in ninety minutes. In a world where that 25th hour gets automatically banked as personal time, the analyst goes home four and a half hours early. In the world we actually live in, the analyst’s manager notices the capacity, and by the following quarter that analyst is producing four reports instead of one, working the exact same total hours as before — just at a much higher rate of output. The productivity gain is entirely real. The personal benefit is entirely absent. One researcher who has coached hundreds of professionals through exactly this pattern calls it “the refill reflex” — and notes, encouragingly, that a trained reflex can also be retrained.

The Two Kinds of Freed Time

Here’s where I think the conversation needs more precision than it usually gets. There are actually two very different things happening when AI removes routine work, and they deserve two different names.

The first is what I’d call structural time — hours that get freed at the level of an entire industry or economy, whether or not any individual worker ever sees the benefit. A trucking company that automates its dispatching, a hospital that automates its paperwork, a law firm that automates its document review — these organizations are all sitting on enormous pools of newly available capacity. What happens to that capacity is almost entirely a management and policy decision, not a technology decision. When 61 UK organizations tested a four-day week for six months, 71% of employees reported measurably lower burnout while company revenue actually rose slightly. The freed structural time was there either way. The only variable was whether leadership chose to bank it as rest or reinvest it as growth.

The second is what researchers studying happiness call time affluence — the deeply personal, felt experience of having genuine ownership over unscheduled hours in your own life. Harvard Business School’s Ashley Whillans has shown that time poverty erodes wellbeing even among people earning high salaries, and that people who deliberately spend money to buy back time from chores and obligations report measurably higher life satisfaction. Structural time and time affluence sound similar, but only one of them actually makes a person happier, and the entire tragedy of the last century of productivity gains is that we kept producing the first while almost never converting it into the second.

What an Hour Actually Looks Like When It’s Really Free

It helps to picture this at the scale of an individual life rather than an abstract statistic. One business coach who restructured her own working life around a permanent 3.5-day week has spent the years since tracking what her clients actually do with hours AI genuinely hands back, once they stop refilling them automatically. A chief technology officer she worked with started spending recovered afternoons at a driving range instead of in his inbox — and within five weeks had lost weight, strengthened his marriage, and still closed three new deals. Several of her other clients, all founders who hadn’t taken more than a long weekend away from work since having children, took their first real three-week family vacations in years.

None of these outcomes required a single additional dollar of AI capability beyond what the tools already do today. What they required was a deliberate decision, made by an actual human being, to treat a freed hour as something other than an empty slot waiting for the next task. That distinction — between capacity and choice — is the entire ballgame.

AI can create the 25th hour—but only those who protect it will ever truly own it.

My Prediction: The 25th Hour Becomes a Skill, Not a Gift

I don’t believe AI, robotics, and autonomy are going to hand humanity a shorter workweek the way a vending machine dispenses a candy bar — automatically, the moment enough productivity coins get inserted. A century of technological history, and the earliest hard data on AI specifically, both point firmly in the other direction. The gains flow overwhelmingly toward output, not toward calendars, unless something else actively intervenes.

That “something else” is going to become one of the defining competitive skills of the next two decades — for individuals, for companies, and for entire economies. The organizations and cultures that figure out how to deliberately convert freed capacity into genuine human time, rather than simply reinvesting every recovered hour into higher output targets, will end up with healthier, more loyal, more creative workforces than their competitors. The individuals who learn to treat an unexpectedly open hour as something worth protecting, rather than something waiting to be filled, will be the ones who actually experience the abundance this technology keeps promising.

Keynes wasn’t wrong that productivity would soar. He was wrong to assume leisure would simply follow along for the ride. The 25th hour is real, it’s arriving faster than most people realize, and it is entirely ours to lose — exactly as it has been for the last hundred years of technological progress that quietly, repeatedly, promised us rest and delivered us more work instead.

Related Articles

• “AI Will Not Give You Fridays Off” — Fortune — fortune.com/2026/08/21/ai-4-day-workweek-fridays-dimon-gates-keynes
• “As AI’s Power Grows, So Does Our Workday” — CEPR / VoxEU — cepr.org/voxeu/columns/ais-power-grows-so-does-our-workday
• “What to Do With the Time AI Frees Up” — Ridiculously Efficient — ridiculouslyefficient.com/what-to-do-with-the-time-ai-frees-up

The post The Invention of the 25th Hour appeared first on Futurist Speaker.