What Do You Say to Someone Who Has Lost Their Job to Artificial Intelligence?

By Matthew Parish

Friday 4 September 2026

There is an awkward conversation coming to millions of households. In some places it has already begun.

A person comes home and says that he has lost his job. Not because the company has gone bankrupt. Not because he was incompetent. Not because somebody younger, cleverer or cheaper has been appointed in his place. The work still exists. Indeed the company may be doing rather well. But a machine can now perform most of what he used to do, at negligible marginal cost and without holidays, illness, complaints, lunch breaks or a pension.

What do you say to him?

The first thing, perhaps, is what you should not say. Do not tell him to “reskill”. At least, do not say it immediately. It is one of those bloodless words beloved of economists, consultants and government departments because it turns a human catastrophe into a diagram. A fifty-three-year-old insurance administrator who has spent thirty years becoming exceptionally good at assessing claims does not necessarily want to become a Python programmer. Nor is there any particular reason to suppose that the economy requires several million new Python programmers. Artificial intelligence is becoming rather good at Python programming too.

Do not tell him that technological change has always created more jobs than it destroyed. This may be true as an observation about the past but history is not a contractual guarantee of the future. The mechanisation of agriculture released labour for factories. Industrial automation pushed labour towards services. Computers destroyed armies of clerks while creating software engineers, systems administrators and entire industries nobody had previously imagined.

But artificial intelligence is peculiar because it is moving directly into the territory to which previous technological revolutions displaced us — intellectual labour. For most of modern history, the comforting assumption has been that machines replace muscles while human beings retain minds. The tractor replaces the farm labourer but somebody designs the tractor. The industrial robot welds the car but somebody programmes the robot. The spreadsheet replaces the bookkeeper but accountants interpret the figures. Technology pushes humanity steadily upwards into more sophisticated cognitive occupations.

Generative artificial intelligence challenges this settlement. It writes. It translates. It programmes. It researches. It analyses documents. It produces illustrations. It answers customers’ questions. It prepares presentations. It summarises meetings. It can review contracts, draft advertisements, analyse medical images and imitate substantial portions of the work performed by junior professionals. The escalator may have reached the floor on which we were standing. So when somebody tells you that artificial intelligence has taken his job, the first useful thing to say is very simple:

I am sorry.

Then listen. Losing a job is not merely losing money. This is something economists sometimes forget because income is measurable whereas dignity is not. Work supplies structure to the day. It supplies acquaintances and friendships. It provides a reason to get out of bed. It tells people that somebody needs them somewhere at nine o’clock on Monday morning. Above all, work answers one of the most fundamental questions human beings ask themselves: What am I for?

Our societies have spent several centuries making occupation central to identity. At dinner parties we ask strangers what they do. Governments divide populations into employed, unemployed and economically inactive. Banks assess people according to occupations. Immigration systems classify them by skills. Children are repeatedly asked what they want to be when they grow up — as though adulthood were principally a profession. Then imagine telling somebody who has spent twenty-five years becoming excellent at something that a computer can now do it in eleven seconds. There is humiliation hidden inside technological redundancy even where there ought to be none.

That is why the second thing to say is: this is not a judgment upon your worth. A machine being able to perform an activity more cheaply does not make the human being who previously performed it stupid. The mechanical excavator did not prove that navvies were fools. The calculator did not reveal that mathematicians had wasted their lives learning arithmetic. Google Maps did not demonstrate that London taxi drivers who memorised thousands of streets were idiots.

Technology changes the economic price of skills. It does not retrospectively change their human value. Nevertheless sympathy cannot pay the electricity bill. Eventually the conversation must turn to what comes next. Here we should resist another seductive cliché — that everybody must somehow compete with artificial intelligence. That is probably the wrong objective. If artificial intelligence becomes dramatically better, faster and cheaper at certain categories of cognitive work, attempting to beat it at those activities may make as much sense as challenging a forklift truck to a weightlifting competition. The sensible question is instead: what can you do with it?

The distinction matters. The employee who loses his job because one person equipped with artificial intelligence can perform the work previously undertaken by ten people has discovered something unpleasant but economically important. There remains one person in that equation. The question is how to become that person. This may require learning new tools but it does not necessarily mean abandoning decades of accumulated knowledge. Quite the opposite. Artificial intelligence frequently becomes most useful when placed in the hands of somebody who understands the field in which it is operating. An experienced lawyer using AI may be vastly more valuable than either an inexperienced lawyer or an AI system alone. The same may prove true of engineers, accountants, architects, teachers, administrators and countless other professions.

Experience may therefore change its function. Instead of producing every component of the work personally, the experienced worker becomes the director, critic and editor of machine-generated labour. Yet we must be careful not to turn this observation into another false consolation. There may simply be fewer such jobs. If ten accountants become one accountant supervising machines, nine people have still lost their employment. Telling all ten to become AI-enhanced accountants does not solve the arithmetic.

This is where the conversation ceases to be merely personal and becomes political. The great artificial-intelligence debate is too often framed as a contest between optimism and pessimism. Optimists promise astonishing productivity growth. Pessimists warn of mass unemployment. Both things might happen simultaneously. Indeed that combination is precisely what should concern us.

Imagine an economy capable of producing vastly more goods and services using vastly fewer human working hours. In purely technological terms this would be an extraordinary triumph. Humanity has spent its entire history trying to accomplish more with less labour.

Yet under our existing social arrangements it might simultaneously be a disaster, because employment is the principal mechanism through which most people obtain claims upon the abundance the economy produces.

We could therefore arrive at the strangest economic crisis in history — extraordinary productive capacity accompanied by widespread personal insecurity. The machines are producing more than ever. Companies are exceptionally profitable. Society is richer. Yet millions of people cannot pay their rent because nobody needs their labour. This would not principally be a technological failure. It would be a failure of distribution.

That is why telling the newly unemployed AI casualty merely to try harder is inadequate. Some responsibility belongs to him — he must adapt where he can, learn unfamiliar tools, exploit his experience and look for fields in which human judgement, trust, responsibility and physical presence retain value. But some responsibility belongs to the rest of us.

Governments may have to reconsider taxation where extraordinary quantities of economic value are generated with comparatively little labour. Education systems must stop pretending that teaching a person a marketable skill at twenty-one guarantees economic security until sixty-five. Social insurance systems designed around occasional unemployment may need adaptation for repeated technological displacement. We may also have to reconsider the moral status of leisure. For centuries increased productivity has gradually reduced the amount of human life surrendered to work. The six-day week became five. Working days shortened. Holidays expanded. Retirement became normal. None of these developments represented civilisation becoming lazier. They represented civilisation becoming richer.

If artificial intelligence genuinely permits society to produce the same wealth with half the human labour, perhaps the answer is not desperately to invent enough unnecessary work to keep everybody occupied for forty hours each week. Perhaps we should work less. A four-day week, shorter working days, earlier retirement or other forms of distributing necessary human labour across more people may eventually appear less radical than trying to preserve an employment structure designed for an economy in which human cognitive labour was scarce.

There remains, however, the problem of meaning. Human beings need to feel useful even if they do not need to spend forty hours each week processing invoices. Families, communities, charities, arts, sport, scholarship, local government and voluntary organisations contain enormous quantities of socially valuable activity that markets price poorly or not at all.

Caring for an elderly parent is work. Raising a child is work. Coaching a football team is work. Writing a local history nobody will buy is work. Visiting lonely people is work. Restoring an old building is work. Learning a language, painting pictures, organising a community garden and helping refugees are all purposeful human activities. Perhaps artificial intelligence will eventually force us to confront an uncomfortable proposition: the market price of a person’s labour was never a reliable measurement of the value of his life.

For the person who has just been dismissed, however, these grand reflections may be of little immediate comfort. He has a mortgage. His confidence has been damaged. He may be frightened to tell his family. He may privately wonder whether the future has decided that it has no further use for him. So sit with him. Buy him a drink if he wants one. Let him be angry. Do not immediately produce a list of online courses. Do not explain comparative advantage. Do not tell him that artificial intelligence will create exciting new opportunities. Tomorrow there will be time for plans.

And then help him make one. Ask what he knows that a machine does not. Ask whom he knows. Ask which parts of his old profession require judgment, relationships, responsibility or physical action. Ask whether he can use the technology that displaced him rather than compete against it. Ask whether there is something he has always wanted to do but never dared because the monthly salary kept him safely imprisoned. Most importantly, remind him that redundancy is an economic event, not a verdict upon a human life.

Artificial intelligence may become better than us at an astonishing number of things. It may calculate faster, remember more, write more quickly, detect patterns invisible to us and operate continuously at costs against which human beings cannot realistically compete.

If that happens, the moral challenge will not be proving that human beings remain economically superior to machines. We may not.

The challenge will be remembering that economic superiority was never the reason human beings mattered in the first place. And perhaps that is ultimately what you say to somebody whose job has been taken by artificial intelligence. Your job has disappeared. You haven’t.