Who Gets the Time Back?

There is an uncomfortable question hiding underneath almost everything I have written so far, and I think I have avoided it for long enough.

I have spent this book looking for work that does not need to exist. Copying, re-keying, reconciling, waiting, apologising, checking the same thing twice, building reports nobody acts on and generally asking intelligent people to compensate for technology that was supposed to be helping them. I have argued that if I can understand that friction properly, I can often remove it, reduce it or at least stop pretending it is an unavoidable law of nature.

All of which sounds fairly harmless until I get good at it. What happens when an organisation becomes genuinely good at giving hours back?

History has asked versions of this question before. Machinery changed what one person could produce in factories, on farms and eventually in offices. Those changes altered the work, the skills required and the expectations placed on the people doing it. They did not simply remove a task and leave a peaceful empty space behind.

Word processors, spreadsheets and email each changed the amount of work a person could reasonably be expected to move through in a day. It became easier to edit a document, recalculate a forecast or send a message, and the expectation of what we should produce moved with the capability.

Technology has spent generations promising to save us time, and it has. It has saved astonishing amounts of it. Yet most of us have not been wandering around wondering what to do with all the spare hours. The saved time has a habit of disappearing into a larger expectation.

The typist did not generally get Thursday afternoon back because a document could be edited without retyping the entire page. The accountant did not finish at lunch because the spreadsheet recalculated the column instantly. Email did not create a beautifully relaxed working life because a message could cross the world in seconds. In many cases the opposite happened. Once communication became instant, we communicated more. Once documents became easier to produce, we produced more documents. Once analysis became faster, we expected more analysis.

The capacity became the new baseline.

There is an important counterpoint here, because history did not simply turn every productivity gain into more work. Across many industrialised countries, average working hours fell dramatically over the long run. Some of the gains did become shorter days, shorter weeks, holidays and retirement rather than extra output. But that did not happen automatically, and it did not arrive evenly. Productivity made shorter hours possible, while economics, law, bargaining, culture and politics helped decide whether people actually received them. That is the distinction I care about. Technology creates capacity. Humans decide where the capacity lands.

That does not make the technology bad. I am typing this on a machine while using technology that would have looked like science fiction for much of my career, so I would be on fairly thin ice trying to make that argument. It does mean that productivity has a destination, whether we choose it consciously or not.

Imagine I find a process in your organisation consuming one hundred human hours every week. I follow the work, strip away the nonsense, connect two things that should have been connected years ago and use AI where it genuinely helps. A few weeks later the same useful outcome takes twenty hours. Eighty hours have appeared. Where do they go?

That is not an accounting question, although somebody in Finance will quite reasonably make it one eventually. It is an organisational question and, increasingly, I think it is a moral one too.

Does the organisation remove a role? Does it leave a vacancy unfilled? Does it give the same people more work? Does it use the capacity to grow without adding headcount? Does somebody finally get time to train the team properly? Does a manager spend the recovered hours with guests instead of spreadsheets? Does service improve? Do wages improve? Does the working week get shorter? Does the organisation simply bank the eighty hours as margin?

None of those answers is automatically wrong. That is what makes the question interesting.

There will be situations where automation genuinely means fewer people are needed. Pretending otherwise would be sentimental rather than useful. Organisations have costs, markets change, jobs disappear and new jobs appear. That has been true through every technology shift I have just described. I am not arguing that every existing task should be preserved in amber because a person currently gets paid to do it. I have spent half this book trying to kill pointless tasks. It would be a strange moment to develop a sudden affection for them now.

But I do think something important happens when the task and the person get treated as though they are the same thing.

If somebody spends ten hours a week copying information between systems, I want the ten hours of copying to disappear. I do not automatically want the person to disappear with it. Those are two separate decisions and they deserve to be treated as such.

That distinction matters because organisations are very good at measuring the cost of labour and surprisingly poor at measuring the value of released human capacity. A spreadsheet can show me the saving from removing a role. It is much harder to show what happens when a manager has five extra hours to coach people, when a finance analyst gets to analyse rather than reconcile, or when somebody who has spent years surviving an inefficient process finally has enough space to improve it.

The second category is real value. It just doesn’t fit as neatly into a business case.

I have also started wondering about a phrase used so casually that I barely noticed it for years. We say technology will 'free people up for higher-value work'. It sounds excellent. I have said it myself hundreds of times. But higher-value work for whom?

For the organisation, presumably. More output, more sales, more analysis, more projects, more customers. Again, none of those things is bad. But if every hour technology gives back is immediately occupied by another hour of economically productive activity, I am not sure the human experienced a saving at all. The organisation did.

That is a much more uncomfortable way to look at productivity.

It also makes some of the language organisations use about people worth testing. If a company genuinely believes its people are its greatest asset, AI is going to give it an extraordinary opportunity to prove the point. When productivity rises, what share of the benefit reaches the people whose expertise, adaptation and knowledge made that productivity possible?

I am not suggesting a formula. I would distrust one immediately. Different businesses have different economics and different responsibilities. I am suggesting the question should be asked before the answer quietly defaults to 'more'. More output, tasks, targets and messages, followed by more meetings because everybody can now produce the analysis for the meeting faster.

There is a slightly comic possibility that AI could remove half the administrative work in an organisation and we respond by inventing enough new administration to fill the gap. Human beings have form here.

The smartphone changed how quickly we expected an answer because people could now be reached almost anywhere. AI may do something similar to the expected speed of thought, creation and response, and that worries me more than a faster calculation.

I do not want a world where the fact that I can produce a first draft in ten minutes means I am expected to produce six first drafts before breakfast. I do not want a manager using the hour recovered from paperwork to receive an extra hour of paperwork generated automatically by somebody else. I do not want to remove friction from work only to discover that the organisation has become more efficient at creating new friction.

When this gives time back, who gets the time? That question reaches beyond any one organisation. What happens if AI becomes as consequential as its most enthusiastic supporters believe it will?

I am deliberately saying if. I have lived through enough technology waves to know that the future has a habit of arriving less neatly than the keynote. Some jobs change, some disappear, some new ones emerge, some technologies plateau and a surprising amount of supposedly obsolete work carries on for decades because reality is untidy. I have no interest in pretending I can predict the labour market from a laptop at four in the morning.

I built a forecasting engine alone that would once have required more people and more specialist hand-offs. I can now move from an operational conversation to a working prototype at a speed that would have been absurd for most of my career. I can analyse, write, design, test, document and iterate with assistance that is available whenever I am. The key point is not that I suddenly became an engineer, designer, analyst and copywriter all at once. I did not. The point is that the expertise I already had can travel much further before I need to hand the work to somebody else.

For most previous waves of automation, it was relatively easy to imagine the human moving upwards into the cognitive work. Machines took more of the physical repetition, people supervised the machines. Software took more of the calculation, people interpreted the result. Self-service removed some transactions, people handled the complicated cases. There were painful transitions and plenty of exceptions, but the broad story was that human judgement sat somewhere above the automation. AI is awkward because it has started climbing the same ladder.

It can write, analyse, code, summarise, create and recommend, and increasingly it can act. It does not do all of those things perfectly, which matters enormously, but they are things many of us were comfortable describing as knowledge work not very long ago.

So the old reassurance that technology removes the boring bits and leaves humans with the clever bits starts to feel a little thin. What happens when the technology can help with some of the clever bits too?

What I have seen makes me more conscious of the judgement involved. Somebody still has to know whether a plausible answer makes sense in the actual operation, who carries the consequence and when the sensible response is not to build at all. More capability makes those decisions more visible, even if it does not guarantee we will make them well.

But I would be equally uncomfortable turning that into a comforting slogan and moving on. 'Humans will always be needed' may be true in the broadest possible sense and still avoid the difficult question of how many humans are needed to produce a particular amount of economic output.

If one experienced person with AI can do work that previously required five people, I can ask whether I now need only one person, or what the other four people could now do. Business has had plenty of practice with the first question. I am not sure we have put equal intellectual effort into the second.

What if we simply need less human labour to produce the life we want?

If technology eventually allows the same economic value to be created in four days rather than five, why is the automatic ambition to find enough additional work to refill the fifth day? If it becomes three days, does the question change? At what point does our definition of progress include some of the productivity gain becoming life rather than output?

I do not know the answer, and I am immediately suspicious of anyone who claims to have the entire economic model neatly solved. There are questions about wages, competitiveness, taxation, inequality, ownership, public services and who captures the value created by these systems. A four-day week is not magically affordable because somebody put AI into the CRM. I am not trying to smuggle a manifesto into the final pages of a book about operational friction.

For most of modern history, earning a decent life has been tightly connected to selling a substantial portion of our waking week as labour. That arrangement made sense in economies where human labour was a major constraint on production. If AI and automation weaken that constraint dramatically, some of the assumptions built around it may eventually need the same treatment I have been giving old software throughout this book. The questions become familiar, even if the scale does not. Why do we do it this way, what constraint made it necessary, and is that constraint still there?

Those are dangerous questions when applied to a clunky restaurant process. They become enormous when applied to work itself.

AI will not make those choices for us either, although it may make them look deceptively technical.

A model can tell an organisation that a process now requires fewer hours. It cannot decide what a good organisation owes the people whose hours are no longer required. It can estimate productivity. It cannot tell us how much of a productivity gain should become profit, lower prices, investment, higher wages, shorter hours or something else entirely. Those are human decisions wearing economic clothes.

Earlier in the book I described the graduation point, the moment a little prototype becomes important enough that the responsibilities around it have to grow up. I think there is a version of that happening with AI itself. When it helps me draft an email faster, the social consequence is fairly small. When it changes the number of people required to run a department, the responsibility has changed category.

That should change the questions leaders ask. Which work should disappear, and which human capability do I want to amplify? What should better productivity feel like for the people doing the work? Who remains accountable, and what am I trying to create on the other side of the reduction?

Those questions are not anti-business. I think they are unusually pro-business, because organisations ultimately need customers, employees, trust and societies stable enough to buy what they produce. A future in which technology creates enormous productivity while most people experience only insecurity would not strike me as a particularly impressive piece of systems design.

And that is perhaps the part that connects most strongly to everything before it. I have spent years looking at systems where one part has been optimised while the whole has become worse. The report is faster, but the manager has more admin. The software is efficient, but the guest journey is clumsy. The forecast is accurate, but nobody understands it. Improving one part of a system can be seductive because the metric goes green while the human quietly invents a workaround.

It would be a spectacular mistake to do the same thing with society.

We could optimise output, headcount and cost per transaction, then look up twenty years later and discover that human beings had been treated as an inconvenient variable in the model.

I do not think the answer is to slow the technology down until we have solved philosophy. That horse has left the stable, found a way to connect itself to other software and probably built itself a dashboard by now. Nor do I think fear is a useful strategy. I am excited about what AI makes possible. It has allowed me to create things I could not realistically have created alone before, and it has made the experience accumulated over my career more useful, not less.

That is exactly why I think the human question matters now rather than later.

The best version of this future, to me, is not one where humans heroically protect every task from automation. It is one where we become much more deliberate about what deserves human time.

Maybe that means less time moving information and more time making decisions or talking about what the information means. Perhaps it means a manager can spend more time with guests, or someone providing care can get back to caring. And if the productivity gains become large enough, it may also mean less time at work.

I can already hear the finance director clearing their throat, so let me repeat that I am not claiming to know the mechanism. I am simply unwilling to accept that the only imaginable destination for every technological saving is more economic activity squeezed into the same human week. Progress should at least give us the option to ask for something better.

There is a personal reason I care about this too. I have spent most of my career around operational people. Restaurant managers, team members, support teams, finance people, technicians, the people who keep things moving while strategy is being discussed somewhere else. The best technology I have seen does not make those people feel smaller. It removes the nonsense around them and gives their judgement more room to breathe.

For me, the time I want back is time to choose which work deserves the energy, then to think, question and actually complete it. I enjoy working on something that matters. What wears me down is the painful repetition around it, the parts of a role that keep getting in the way of the reason the role exists. If technology removes some of that, I want enough room to follow a worthwhile question properly rather than fitting the thinking around another round of administration.

That is the future I want to build towards, with humans deciding what matters, machines taking on what they are unusually good at, and organisations being deliberate about where the benefit lands.

I think every organisation adopting AI should be able to answer one deceptively simple question.

When the machine gives us an hour back, who gets the hour?

If we can answer that deliberately rather than accidentally, then perhaps the human really does stay in the middle of the story.