Everything in this book comes down to a single decision, made over and over, in small moments throughout your week. AI hands you back an hour. What do you do with it? By default, without you choosing, it becomes more work. The entire argument is about how to make a different choice, and why it is worth making.
What is the windfall, and what is the trap?
The windfall is real. AI is the largest efficiency gain in a century, and it is genuinely saving people time, forty to sixty minutes a day for many, more for some tasks.1 Keynes predicted a windfall of exactly this kind would arrive, and he was right that it would; he was only wrong that we would take it as leisure.2 We took it, every previous time, as more output, more income, more stuff, and almost never as time.
The trap is that saved time does not stay saved on its own. It is capacity, and capacity gets filled. Parkinson’s Law expands the remaining work to occupy the gap. The Jevons rebound means that when a task gets cheaper, we do more of it rather than less. And a culture that reads busyness as virtue and status rushes to convert any freed hour back into visible work.3 These forces do not need your permission and do not wait for your good intentions. Left alone, they reabsorb the windfall completely, which is why faster tools have so rarely produced shorter days.
Why take it as time at all?
Because the evidence on what makes a life good points away from more output and toward the things work crowds out. The single strongest predictor of long-term health and happiness is the quality of your relationships, ahead of wealth and achievement.4 People who value time over money are reliably happier. The deepest satisfaction comes from active engagement and connection, not from another increment of production. Taking the dividend as time is not self-indulgence at the expense of a good life. On the evidence, it is the good life, and more output is the thing being indulged at its expense.
How do you actually do it?
The method is five moves, in order. See your real week, because memory lies and you cannot reclaim time you cannot measure. Shed the work that should not exist, eliminating before you automate, because doing pointless work faster is the signature mistake of the age. Shift the right tasks to the machine, delegating without abdicating, keeping your core craft and your judgment. Shield the reclaimed time with structure rather than willpower, because a block others can schedule over is not a boundary. And spend the time deliberately, on people and on what absorbs you, because time you do not consciously claim will be claimed for you.
Each move is structural, not heroic. None requires you to become a person of superhuman discipline. They require you to arrange your work so that ordinary discipline is enough, and to decide, in advance, where the saved time goes.
What about the limits?
Honesty to the end: an individual can reclaim meaningful time this way, and many do, but the playbook asks the person with the least power to manage a problem the system created. The deepest forces, the hours arms race, the coordination trap, the way gains flow to capital unless bargained otherwise, are collective, and they yield only to collective answers: four-day-week policies, results-based cultures, the laws that gave us the weekend in the first place.5 The AI dividend will be taken as time, or as more work, depending on choices made at both levels, individual and collective. Neither alone is sufficient.
But the choice is genuinely open, which is the hopeful core of all of it. The fifteen-hour week did not fail because we could not afford it. It failed because we never decided to take it. AI reopens the decision. You cannot legislate your own shorter week into existence, but you can stop handing every gain straight back as output, you can reclaim real hours through the moves in this book, and you can add your weight to the collective push that makes shorter hours stick for everyone. The instruction the whole book reduces to is just this: when the machine hands you the time, do not work harder. Take it as time.
Frequently asked questions
What is the core message of Don’t Work Harder?
That AI and productivity gains should be taken as reclaimed time rather than converted into more output. The windfall is real, but every previous efficiency wave was reabsorbed into more work, and changing that requires deliberate individual and collective choices.
Why won’t AI shorten our hours automatically?
Because saved time is capacity, and capacity gets filled by Parkinson’s Law, the Jevons rebound, and a culture that rewards busyness. Without a deliberate decision to bank the gain as time, it defaults to more work, as it always has.
How do you take the AI dividend as time?
See your real week, shed the work that should not exist, shift the right tasks to AI, shield the reclaimed time with structure, and spend it deliberately on what matters, while supporting the collective change that makes shorter hours stick.
About the author
Tom Goodwin is the author of Don’t Work Harder, a book about taking the time AI gives back as time rather than more work. He is a co-founder of GAMEPLAN and writes on productivity, technology, and the economics of the working week.
Footnotes
Footnotes
-
Goldman Sachs AI Adoption Tracker (December 2025): average savings of 40 to 60 minutes per day among AI users; Indeed Hiring Lab (2025): the most common use of saved time was taking on more tasks. ↩
-
John Maynard Keynes, “Economic Possibilities for our Grandchildren” (1930): the prediction of abundance, and the argument that humanity’s enduring challenge would become how to use freedom and leisure well. ↩
-
C. Northcote Parkinson, Parkinson’s Law (1958); William Stanley Jevons, The Coal Question (1865); Silvia Bellezza, Neeru Paharia, and Anat Keinan, “Conspicuous Consumption of Time,” Journal of Consumer Research (2017). ↩
-
Harvard Study of Adult Development; Robert Waldinger and Marc Schulz, The Good Life (2023); Ashley Whillans et al. on valuing time over money (2016). ↩
-
On the collective and structural dimension: the Fair Labor Standards Act of 1938, the contemporary four-day-week movement and 32-hour-week proposals, and analyses noting that AI productivity gains may accrue first to capital unless redistributed through bargaining and policy. ↩