NP Nithin Palal

Chapter 29: Fear, Ego, and Obsolescence

Part VI: The Inner Revolution

· 45 min read · Discuss · Plain text

AI-Proof: What Machines Can Never Be — Volume III: Transformation, Part VI: The Inner Revolution, Chapter 29: Fear, Ego, and Obsolescence

Volume III and Part VI begin here. After strategy and the sectors, the person: this first chapter of Part VI asks what fear and ego do when the work goes.

A test lead kept her job and lost the sentence that said who she was. An autoworker took a buyout, and his grief kept no order. The fear of becoming obsolete is right about some losses. It is wrong about the person.

In Brief For readers who fear that a machine will make them obsolete, or already feel it has. Chapter 29 opens Volume Three and asks what the fear of becoming obsolete is afraid of, and how much of it is right. A test lead in Bengaluru keeps her post and loses the sentence she used to say who she was. A Michigan autoworker takes a buyout in 2008 and finds that the stages of grief came in no order. Patañjali’s afflictions of ego and clinging, and Cicero’s four masks, help separate what can really be lost from what never rested on being useful. The evidence on job loss shows how much of the fear the institutions around a person decide.

A skill can become obsolete and a person cannot; the fear of obsolescence is often right about what you could lose and wrong about what you are, and seeing the difference neither ends the fear nor pays the rent.

Opening: The Question at Lunch

Two of her mother’s friends came to lunch in November 2025, and one of them asked what Priya did now.

Priya Chandran had known both women since she was a child, and had called them aunty for as long. They asked about work the way they asked about marriage, with the affectionate persistence of customs officials. Her mother answered first. “She is an architect now.”

“Of buildings?” the older one asked.

“Of testing,” Priya said. Everyone laughed, and she did not explain.

For years she had had a sentence for this: I’m in software testing. She was proud of it. She led the team that broke her company’s software before customers could, at its office in Koramangala, in Bengaluru. She was the one who found the bug that others missed. Then, in October 2024, a machine tested in eleven minutes what had taken her team a week, and found bugs they had missed. Her team later built the platform that now tests the company’s code. Since October 2025 she had led the effort to carry that platform across every product the company makes. The post came with a title, architect, and her pay had risen with it. The title gave her a seat in some rooms. It did not give her the right to decide what other teams’ products were for.

None of this fitted into a sentence an aunty could repeat. She lived in the same flat with her mother and her cat, Turing. By every measure her company kept, she was doing well. And she was afraid, in a way she could not have explained at that table, because she did not know what she was afraid of.

Fear like hers is common now. People call it the fear of becoming obsolete, a name it does not deserve. When a machine can do the work you took yourself to be, what is the fear of becoming obsolete afraid of, and how much of it is right?

The print edition answered with a consolation. Beneath every professional identity, it said, lies “the fact of being conscious”, and that fact is “the one thing about her that no machine can replicate”. This edition withdraws both halves. Whether any machine experiences anything is an open question, and nothing here depends on the answer. And no one is made safe by what machines lack, because what they lack changes every year. What this chapter can say is narrower. The fear is often right about what can be taken. It is wrong where it treats a person as a thing that wears out. Seeing the difference does not end the fear. It does not pay the rent either.

So the chapter’s title needs one correction. Obsolete comes from a Latin verb meaning “to wear out”, “fall into disuse”, “lose value”1. It is said of tools, words and machines, things that are for something. A skill can become obsolete. To call a person obsolete is to measure her by her output, which is the measure this book exists to refuse. Kant drew the line more than two centuries ago: skill has a market price, which an equivalent can replace, and a person has a dignity, which nothing can2. This book takes Kant’s line, not his ground for it. He placed dignity in the capacity for moral choice; here a person’s worth depends on no ability at all. The title’s third word names what the fear believes, not what is true.

This chapter also opens Volume Three and its first part, The Inner Revolution. The inner half of that title is real. How a person holds a loss changes what the loss does to her. The word revolution promises too much. An inner change is slow and partial, it can come undone, and it changes nothing about who decides the wage. Offered in place of an outer change, it insults someone who needs an income. This part of the book is about the inner half. It will keep saying where the outer half lies.

29.1 The Anxiety of Automation: What Is Being Lost

Clinicians draw a line that helps here. In the manual American psychiatrists use, fear is the emotional response to a threat close at hand, whether real or only perceived. Anxiety is the anticipation of a threat to come3. Fear has an object. Anxiety often has only a direction. Priya’s was mostly the second. Nothing had been taken from her. She was afraid of something that had not happened, and could not say what it was.

One way to find what such a fear is afraid of is to ask what a job gives. An earlier chapter of this book split it into four goods. They are income; practice, the daily work that keeps a skill alive; standing, being needed and counted; and voice, a say in how the work changes. A machine can take them apart, and they do not move together. Priya’s income was higher than ever. Her voice was larger, too: she now sat in rooms where the platform’s spread was planned. Her practice was gone. She no longer spent her days breaking software, because the platform did that, mostly better. Her standing had changed. She was no longer the one the team called when a release went wrong. So her fear was partly accurate. Two of the four goods had already moved, and the other two rested on decisions that were not hers. The four are not all a job gives. Her future had changed shape too: she could no longer picture what the next ten years of her skill would be.

Anxiety of this kind is not always a fault. A fear that looks ahead can be information. It can make a person save, learn or ask a question in time. The trouble, as I see it, is that anxiety does not sort what it fears. It can treat a loss that is likely, a loss that is possible and a loss that is imagined in the same way, with the same tightness in the chest. Sorting them is the first work, and it is done with facts.

Some of the facts are not about Priya at all. In the United States, at least, the fear is most accurate for people at the start of their working lives. In August 2026 economists at Stanford looked at workers aged 22 to 25 in the jobs most exposed to AI. By June 2026 their employment stood about 19 per cent below where it would have been had it kept pace with similar workers in less exposed jobs. That is a gap measured against similar workers, not an estimate of what AI caused, and not a count of people dismissed. The gap seemed to come mostly from young people not being hired4. In India, the five largest IT companies together employed about 7,400 fewer people in the year to March 2026 than a year before5. The report set the fall against changing demand and a turn to AI, without saying how much either caused. It is a net change in headcount, not a sum of announced lay-offs. Priya was not among the people these figures describe. Some of those who would once have joined her old team may be. Her own team had built part of what now does their first work. Chapter 4 calls this the missing rung: the loss of the junior work through which people used to become trustworthy. Where those juniors will learn now is a question her company had not asked.

In December the anxiety found an object, and became fear. The chief product officer’s office asked her to evaluate an agent a vendor was selling. Given a new product’s specifications, it drafted a whole test strategy: what to test, in what order, and where the risks lay. That was much of the work her post now consisted of. She ran it on a product her team had planned by hand in October. Its strategy was not hers, and it was not worse. It had taken an afternoon.

This call goes to the machine. Priya wrote in her evaluation that the agent was good enough for most new products, and she signed it. It was true, and against her interest: she wrote what she found, not what would protect her post. Whether to buy it was the CPO’s decision, and by the end of January no decision had been made. She had moved from testing into designing tests partly because the new work seemed safer. As of the end of 2025, a machine could do much of that work too. The new role was still built on professional capability, which meant it was still, in principle, vulnerable.

On the evidence about fears like hers, the print edition reported a finding without a source. It said that research “consistently finds” that people replaced by technology suffer more than people laid off for other reasons. I have found no such research. The nearest evidence points the other way, at least about the self. In experiments by Armin Granulo, Christoph Fuchs and Stefano Puntoni, almost two thousand people in North America and Europe imagined jobs being taken, sometimes other people’s and sometimes their own. Being replaced by “machines, robots or software”, rather than by other people, “is associated with reduced self-threat”. But being replaced by robots is “associated with a greater perceived threat to one’s economic future”6. The authors drew on theories of how people compare themselves with others: losing to another person can feel like a verdict on you, and losing to a machine less so. These were imagined losses, not real ones. Even so, they reverse the print edition’s order. In these experiments, losing a job to a machine threatened people’s sense of themselves less than losing it to a person did, and their economic future more. The experiments compared two kinds of replacement, and nothing in them says that losing work to a machine leaves the self untouched.

Real job loss has been studied more closely. In 2009 two psychologists, Karsten Paul and Klaus Moser, pooled the research on unemployment and mental health. Among unemployed people, 34 per cent had psychological problems, against 16 per cent of people in work. Studies that followed people over time showed that unemployment “is not only correlated to distress but also causes it”7. The harm was greater for men and for blue-collar workers, those in manual and factory jobs. It was greater, too, in countries with “weak unemployment protection systems”. That last finding matters for everything in this part of the book. How much a lost job hurts depends partly on what the society around it has decided a person who loses one is owed. The wage is not all that goes. The psychologist Marie Jahoda argued that a job also gives the day a shape, people to meet outside the family, a part in a shared purpose, status and regular activity. In the 1930s she and two colleagues studied Marienthal, an Austrian village whose factory had closed. The unemployed men walked more slowly along the village street than the women did, and stopped more often; the women still had households to run7. In 2023 Paul, Moser and three colleagues pooled the studies that tested Jahoda’s list. Unemployed people had less of all five than people in work, and the five, with income, explained about a fifth of the differences in mental health7. So her list explains part of the harm of losing work, not all of it. Nor does it mean that every job is good for a person. In an Australian survey, the mental health of people who moved from unemployment into a job of poor quality fell further than that of people who stayed unemployed7.

Priya had lost no job. Her fear was of the anticipating kind, and it was not about her income. It was about what she was, and that part of the fear needs taking apart.

29.2 When Your Identity Was Your Job

Losing a job that was your identity is not an ego death. What happens to a person whose work is taken is not what the contemplative traditions mean by the death of the ego.

The word ego does three jobs. It can mean the picture of yourself you have built from your roles: I am a tester. It can mean pride, the wish to be the one others need. And psychologists use it for something good: the strength to take a blow and keep functioning. The ego is not the enemy. Its third sense is what a person needs most when the first is under attack.

The traditions that loosen the ego loosen its first sense, and they do it on purpose. A practice is chosen, kept up slowly, and usually learned with a teacher and among others. Losing a role is not that. It comes without consent and without a teacher, and, as I read it, it often tightens the grip. Giving something up and having it taken are different events, even when the same thing goes. The one who gives up a role usually chose when, and kept a way to live.

Here is why I think so. A picture of oneself that is under threat gets defended, and the easiest defence is to work harder at being it. A role that a chosen practice might loosen over years is gripped harder when it is taken all at once. That is also why the traditions’ language of letting go can sound cruel to someone who is being let go. It describes a freedom she did not ask for, at the moment she is losing something she did not want to lose. Organisational psychologists have described something like this grip. In 1981 Barry Staw, Lance Sandelands and Jane Dutton argued that a threat narrows what a person or a group attends to7. It also pushes them back to the responses they know best. When the change is large, they warned, this rigidity can make the threat worse.

Priya’s grip tightened in the weeks after the lunch. Under the design her team had built, a small sample of the platform’s confident findings already went to a person each week. Priya began reading many more of them herself, in the evenings, though no team had asked her to. She joined every product review she could find. She rewrote documents that were finished. Some of this was diligence, since a platform spreading to every product needs someone reading what it finds. The test was whether the extra work answered anyone’s question, or only quieted her own. In December her technical lead asked her, kindly, who had asked her to read the findings. She had no answer. Much of what she was doing has a plain name. It was fear wearing the clothes of competence. This book’s first chapter, which tells the day the dashboard turned green, describes a fear that wears analysis as a disguise. This was the same fear in other clothes, and this time she did not notice it.

An Indian manual of meditation names something close to this pattern. The Yoga Sūtra of Patañjali is usually dated to the early centuries of the common era; some scholars place it as late as about 400. It lists five afflictions, kleśas, that keep a mind in pain. In James Haughton Woods’ translation: “Undifferentiated-consciousness (avidya) and the feeling-of-personality and passion and aversion and the will-to-live are the five hindrances.”8 Both of this chapter’s first words are in that list. What English calls the ego is close to the second affliction, asmitā, the sense of “I am”. The fifth is abhiniveśa, clinging to life, which shows itself as the fear of death.

Patañjali defines the second in one line: dṛgdarśanaśaktyor ekātmatevāsmitā. “When the power of seeing and the power by which one sees have the appearance of being a single self, [this is] the feeling-of-personality.”8 The power by which one sees is an instrument. Asmitā is taking the instrument for the one who uses it.

The Yoga Sūtra is a manual of release, not advice about careers, and I borrow only its diagnosis. Priya’s eye for the rare bug was a power by which she saw. For years she had taken it for herself. When a machine took over the seeing, the loss felt like a loss of self, because she had made it one. She had never taken her salary for herself, which is why the loss of practice and standing reached her sense of self and the rise in pay did not. The Yoga’s cure assumes a pure seer behind every instrument, a self that is never a role. Without the seer, the diagnosis still has two parts. The instrument is one power, hired out by the day. The one who uses it is the whole person across a life: a daughter, a friend, someone who chooses and answers. This book does not need the seer, and the next section says why.

Two things in her story need saying plainly. The first is that her identity broke before her employment did. She kept her post and lost her sentence. It can happen to anyone whose work gave them standing, and years before any job is lost.

The second is an objection, and it is about class. An earlier chapter put it bluntly. For most people in the world, work was never an identity. It was how they fed their children, and a lesson about the self offered to them in place of a wage is an insult. The objection is right about many people. It is not the whole truth. Paul and Moser found that blue-collar workers were more distressed by unemployment than white-collar workers, not less. Their study does not say why, and lost income and poorer chances of another job may explain much of it. Still, it does not fit the view that only professionals lose something of themselves with a job. On my reading, a factory job can carry a man’s standing as fully as a title carries a professional’s. Whether a person’s identity lives in his work then depends less on his collar than on whether the work gave him standing. The man in this chapter’s second story had work that did.

29.3 An Identity That Is Not a Skill

The tempting reply to Priya’s fear is a better identity. Lose the role of tester, and become an architect. That is a trap: the new identity is built the same way, on a capability, and can be taken the same way. The new work may be good work, and it may pay the rent. The trap is resting who you are on it. The print edition’s answer went deeper. Beneath every role, it said, there is “the one who plays them”, awareness itself, and no machine has it.

This edition cannot rest on that, for three reasons. First, the traditions disagree about any self beneath the roles. The Yoga and Advaita Vedānta say there is a pure witness. Early Buddhism, as an earlier chapter showed, declines to say either that there is a self or that there is none. The American philosopher Daniel Dennett called the self “an abstract object, a theorist’s fiction”: the main character in the story each of us tells about a life9. He did not mean that the idea of a self is useless. A centre of gravity is a fiction in the same sense, he wrote: it has no mass and cannot be touched, yet physics relies on it. Second, whether machines experience anything is open, and this book does not depend on the answer. Third, a ground defined by what machines lack would move every time they changed.

What cannot be automated is not something a person has. It is who she is. A machine will never be Priya, in the way that no one else will ever be her. That is true by what the words mean, not by what machines can or cannot do. It needs no self beneath the roles. Even if the self is the main character of a story, as Dennett held, the story is of this one life, and nobody else lives it. It is a thin truth. It protects no job, and it says nothing about which roles she will keep. It is still the place to begin, because it does not move. Thin is not empty. It means that the loss of a job, however large, is a loss to someone. It is never the loss of someone.

Four things are easy to run together here. The first is worth: what a person is owed as a person, which is not a price. The second is being this one: no one else is Priya or lives her life. The third is the story of her life, which changes, and which others can tell too. The fourth is the roles by which others know her: tester, test lead, architect. A machine can change the third and the fourth. It can even imitate her, writing in her style or answering in her manner. Imitating what a person is, to use the philosopher Hannah Arendt’s distinction from Chapter 4, does not make anything into who she is9.

A careful reader will object that a stone, too, is only itself. But nothing is owed to a stone, and a stone keeps no promises. A person has one life, lived forward, with debts, duties and people who wait for her. A copy of her work takes over none of them.

A Roman statesman described roles with care. In 44 BCE Marcus Tullius Cicero wrote a book on duties for his son, De Officiis. In Walter Miller’s translation: “We must realize also that we are invested by Nature with two characters, as it were”10. The word Miller translates as “characters” is persona, the mask a Roman actor wore on stage11. Cicero’s people are dressed in masks. The first, he says, “is universal, arising from the fact of our being all alike endowed with reason”. The second is each person’s own nature: one is fast, another strong. Then: “To the two above-mentioned characters is added a third, which some chance or some circumstance imposes, and a fourth also, which we assume by our own deliberate choice.” Under chance he lists “Regal powers and military commands, nobility of birth and political office, wealth and influence”. The fourth is a career: “what rôle we ourselves may choose to sustain is decided by our own free choice.”

On this map, Priya’s job was a fourth mask. The machine’s arrival belongs to the third, chance, which hands roles out and takes them back. The application is mine; Cicero had no machines in mind. Chance, here, means only what Priya did not choose. Others chose most of it: the company that brought the first testing machine in, and the vendor selling the agent. Some of it she helped to make. Her team built the platform, and her name is on the evaluation of the agent. A person can be the object of a change and one of its makers at once, and that has its own weight. She answers for her part, the platform and the evaluation, and not for what others decide to do with them. One part of Cicero’s picture must be refused. He grounds the first mask, the one everyone shares, in reason, in “that superiority which lifts us above the brute”. This book grounds a person’s worth in no ability at all. A child cannot reason as an adult does, and her worth is not smaller.

Cicero also says something the class objection needs. Most people, he admits, choose a career “in the years of early youth, when our judgement is most immature”. Choosing well takes time, and “There is one class of people that is very rarely met with”. It is made up of those with ability or education “who also have time to consider carefully what career in life they prefer to follow”. By his own account, everyone chooses a fourth mask, but a considered choice belonged to a few. It still does. Even a chosen mask is chosen from among those that nature and chance allow. The freedom to choose a new one is shared just as unequally: some people can afford a year to retrain, and others cannot afford a month.

The print edition said that its own “implicit promise” had been “that if you develop these qualities sufficiently, you will be safe”. That is a promise this chapter cannot keep. You will not be safe. No professional identity is permanent. The print edition offered “you will be awake” instead of safety, and that sentence does not stand. No state of mind replaces safety, and none is promised here. It also said that a person can lose a role and build another “because the builder was never the building”. That is true of the person. It is not true of the materials. The builder still needs an income, a place to practise and someone willing to hire her. Most of those are in other hands. In many places, health cover, a pension and even a loan come through a job. Losing it can take away the conditions in which a person’s worth is lived, without touching the worth.

The word obsolete has a second use, and this one is not the fear’s mistake. Employers call workers redundant, and a labour market can treat a man of fifty-four as finished. People make that judgement. A skill becomes obsolete when its work is no longer needed, or is done more cheaply. A worker is called obsolete when an employer decides that her abilities are no longer worth their cost. The second does not follow from the first. She may know how a production line fails, how a customer behaves in a crisis, or what a manual leaves out. Her employer may have no way to count it. A firm can also retrain her, redesign the job, move her, share out the hours, or let her go. These are different choices, and a machine makes none of them. A market does not measure what a person is worth. It records what someone will pay for one contribution, at one time and in one place.

29.4 Grief, Anger, and Stages That Keep No Order

Frank Mazur worked for Chrysler for twenty-five years. He was hired in 1983, the year Chrysler bought a plant in Sterling Heights to build cars12. Most of his years were spent there, at the Sterling Heights Assembly Plant on Van Dyke Avenue, in Macomb County, north of Detroit. Frank is a composite, like most of the people in this book. The company, the offer he took and the figures around him are real.

He and his wife, Donna, lived in Warren, the next city to the south, in a house they had bought when he was thirty-one. He had torn a shoulder on the line and worked through three holiday shutdowns. In January 2008 Chrysler offered buyouts to union workers at Sterling Heights and four other plants. Those with thirty years of service could take $70,000 and full retirement benefits. Anyone with a year could take $100,000 in one payment and six months of health care13. Frank had twenty-five years. He took the lump sum and left that spring, at fifty-three. He signed, and the choice was his. It was also made inside a company that was losing money, in an industry that was shrinking. A choice can be free on paper and narrow in fact. On 30 April 2009 Chrysler filed for bankruptcy14.

His story belongs in this book for a narrower reason than the print edition gave. His job went with a market, not with a robot, so it shows nothing special about machines. It belongs here because his grief is an old form of the fear this chapter is about, and because the institutions around his loss are in plain view. He blamed the robots. Over his years on the line he had watched them take the welding and the painting. When he drove past the plant, it was the machines he was angry at, not the company whose accounts he understood.

His anger had a real object, but it had the wrong address. A robot does not decide to replace a man. People decided: the managers who bought the robots, the customers who stopped buying the cars, the negotiators who wrote the terms of the offer. The machine stood in front of all of them, and it could not be argued with. Anger needs someone who can be asked why. When a machine is placed in front of a decision, the people who made it become harder to find, and anger goes where it can see. That is one more reason why the people who decide to automate work should say so in their own names, and explain.

This call goes partly to the machines. Robots made work safer. Rania Gihleb and three colleagues found that in the American places more exposed to industrial robots, injuries at work fell15. Injuries at work, like Frank’s, became less common where robots spread. The same study found that those places also saw “a significant increase in drug- or alcohol-related deaths and mental health problems”. In Germany, where the authors followed workers over years, robots brought less physical strain and less disability, and no clear effect on mental health. Rourke O’Brien, Elizabeth Bair and Atheendar Venkataramani found that automation between 1993 and 2007 raised deaths among Americans aged 45 to 5416. The effect was stronger in states with “right to work” laws, which weaken unions, and with lower minimum wages. Stronger safety nets, the authors argued, could soften it. Both studies describe places, not any one person. The machines took some of the damage out of bodies. What happened to the people they replaced depended on the society around the plant.

His daughter, Lisa, is a social worker in Ann Arbor. In 2010 she showed him the five stages of grief: denial, anger, bargaining, depression and acceptance. Elisabeth Kübler-Ross described them in On Death and Dying (1969), from interviews with dying patients17. They were not drawn from the grieving, still less from the laid-off. Lisa told him that the stages do not always come in order, and that not everyone goes through all of them. Frank listened. Then he said that his own experience had been messier than any model, but that the words were the right words.

“Models are for people who didn’t go through it,” he said. “The words are for people who did.”

His line is better than the model, and the research is on his side. In 2017 three researchers of bereavement, Margaret Stroebe, Henk Schut and Kathrin Boerner, reviewed the theory of stages. They found “the absence of sound empirical evidence, conceptual clarity, or explanatory potential”. Worse, “the expectation that bereaved persons will, even should, go through stages of grieving can be harmful to those who do not”18. The psychologist George Bonanno had already shown that resilience after a loss “is more common than is often believed”19. Many people bend without breaking. Resilience is common, but it is not a duty. One large study found it more common among people whose income held, who had fewer chronic illnesses and who had support from others19. Those are not shared equally. A person who struggles more than another may simply have had less to lean on. Nobody should be told that he is grieving wrongly because he is not grieving in order. So the stages stay here as Frank kept them: words for states a person may recognise in himself, not a road he must walk.

In Frank’s words, then, in his order. Denial came first, as busyness. He took up woodworking, fixed everything in the house, and told Donna he was fine. Anger came at the machines. Bargaining came as retraining. In 2009 he enrolled at Macomb Community College, whose South Campus is in Warren. At fifty-four he earned a certificate in computer-aided design and applied to the new manufacturers, the ones that used the robots he was angry at. None of them hired him.

That was not a stage of grief. It was the labour market. In January 2010, the Bureau of Labor Statistics found, 49 per cent of the long-serving workers displaced in 2007 to 2009 were back in work. Among those displaced from manufacturing the share was 39 per cent, and among those aged 55 to 64 it was also 39 per cent20. Machines had a part in that, though not the part Frank gave them. Economists have found that in the United States, in recent decades, almost all the jobs lost in routine work, the kind that follows set steps, were lost in recessions. Those jobs did not come back in the recoveries20. Frank’s job ended as a recession began. Years of machines, and the choices of the people who bought them, helped decide that jobs like it would not return. Retraining does help on average. A study of displaced workers in Washington State found that a year of community college raised men’s later earnings by about 9 per cent21. A year of technical courses raised them by about 14 per cent. Frank was not the average. At fifty-four he could learn a new skill. Turning it into a wage depended on employers, on his age and on where the jobs were, and none of that was his to decide. A certificate can give a person a new ability. It cannot make a job for it. When a displaced worker is told to retrain, someone should also say what the training leads to, who pays for it and what happens if no job follows. His six months of health care ran out in the autumn of 2008. After that, Donna’s job in a school office carried their insurance.

So the loss was not only his. A job is lost by one person, and its costs are carried by a household. Donna’s job was now kept for its insurance as much as its wage. Their spending, their savings and the work each of them did at home all changed. The statistics counted one displaced worker. The house in Warren carried more than that.

In the middle of all this came a winter when he hardly left the house, and the television ran from morning to night. Donna was frightened. He saw his doctor, and was treated. This book does not diagnose him, and it does not turn his worst months into a step towards anything. The stage Kübler-Ross called depression is a name for sorrow. Depression the illness is something doctors treat, and his doctor did.

The print edition said that Frank and Priya shared one psychological architecture, and that her kind of loss was harder because it “attacks your self-concept”. That is too neat. Frank lost his income first, and his practice, standing and voice with it. Priya lost practice and standing and kept the rest. Their losses differ in kind, and in who could help with them. Her fear looks forward, and his grief looked back. She has a salary, a title and rooms to sit in. He had a lump sum, six months of health care and then the open market. What they share is narrower: goods a job had bundled, and a picture of oneself that rested on some of them. This chapter’s two people also had some means to bear a loss. Many people who lose work have less, and nothing in this part of the book replaces what they are owed.

29.5 The Ground Beneath the Floor

The image this section is known by still holds. “The tide is rising. You can build higher. You cannot build on sand forever.” In this image the floors are identities built from skills, each raised higher as the water rises. The ground is what all of them stand on. Every identity built on a capability can be disrupted again. A tester becomes an architect, and the machine follows her into the new work, as it followed Priya in December. Building a better identity of the same kind only postpones the question.

So the question is what remains when the job goes. The print edition’s answer was the bare fact of experiencing, which it said no machine could replicate. This edition puts something different beneath the floor: whatever never rested on being useful.

The first is worth. A person’s worth is not a price, and it does not depend on being needed. Skill has a price; a person does not, as Kant said. Frank was worth no less on the morning he left the plant than on the morning it hired him.

The second is that each life is someone’s. The agent that drafts test strategies may do Priya’s work. It will never be her, because no one else will ever be her. That is true by what the words mean, and it names no ability, so no new machine can take it away.

The third is the people to whom you are someone. Frank was Donna’s husband and Lisa’s father before he was an autoworker, and he was both after. The third is not a condition of the first. A person with no one, isolated or bereaved, is worth no less. The people who know us show us our worth; they do not create it.

None of the three is defined against machines, and that matters. The old ground was whatever machines lacked, so every new machine could shrink it. These three do not depend on what any machine can do. They would be as true in a world without machines, and they are no smaller in a world full of them.

Knowing that your worth is not a price does not make you feel it. A person can believe it and still feel useless in the week after she is dismissed. That is not a contradiction. Wages, titles and the question asked at every lunch, “What do you do now?”, teach people every day what they are worth. The feeling follows that teaching more than it follows any argument. So the outer half matters here too. A society that treats people without work as failures will be believed, whatever a book says.

An old Indian picture makes the same point from another side. In the Kaṭha Upaniṣad, the self rides in a chariot, and the intellect is its driver22. The journey has an end, the text says, and it is the rider’s. On my reading, a person who takes his skills for himself has mistaken himself for the driver. A better driver does not unseat the one the journey is for. But the picture removes a consolation too. If you were never the driver, the identity you built on your driving was never your self, and nothing built on it was ever safe. The picture also leaves open who says where the chariot goes. A company can hand the driving to a machine and leave nobody a say over the road. That is the loss of voice, seen from the chariot. The Upaniṣad means a larger journey than a career, and this use of it is mine.

Here the ground must be described honestly. It holds up a person. It does not hold up a household, and it does not pay the rent. Frank’s worth did not pay for his health care when the six months ran out; Donna’s job did. A book that offers this ground in place of a wage has made the mistake the class objection names. The ground is where a person stands while the wage is argued for, by her and by the institutions that decide what she is owed. What she is owed can be named. It includes warning before the change, income while she looks, and health cover that does not end with the job. It includes training that leads somewhere, and a say before the decision rather than after it. Some of it falls to the employer, and some to the state. Frank got a lump sum and six months of health care. Whether that was enough, others decided.

The aim is not to keep every job. Robots made work on lines like Frank’s safer, and some work is worth losing. The aim is that a person can lose a job without losing her place among others. If less human work is needed as machines spread, whether that brings more free time or more insecurity is not for the machines to decide. It depends on who owns them, and on how the gains are shared.

Cicero has a rule for the next role, and it fits this ground. A person who finds that he chose his career wrongly “must change his vocation and mode of life”. “If circumstances favour such change, it will be effected with greater ease and convenience.” If they do not, “it must be made gradually, step by step”. And then: “we must take all possible care to make it clear that we have done so with good reason”10. The change is to be explained to those it touches. That is close to what this book means by answering for a choice.

Cicero knew, too, that a person can be pushed into a part he never chose. “But if at some time stress of circumstances shall thrust us aside into some uncongenial part,” he wrote, we must give it all our effort, so as to play it “if not with propriety, at least with as little impropriety as possible”. That is not a promise that the part will fit. It is a duty to the part one is given. Frank’s certificate was an attempt to meet it, and it failed for reasons that were not his. Cicero’s advice is about how to change a role. It says nothing about whether the world offers one.

Cicero and Patañjali disagree, and the disagreement is useful. For Patañjali, taking any instrument for the self is an affliction. The cure is a long discipline that separates the seer from what it sees, and a better role would only be a better disguise. For Cicero, a life is properly lived in masks. The fault is a mask that does not fit, and the cure is a better one, chosen slowly and explained. They partly answer different questions: why losing a role terrifies, and how to change a role decently. Still, each has a reply to the other. To Patañjali, Cicero’s well-fitted role is asmitā well dressed. To Cicero, a seer behind every mask is a philosopher’s luxury, and he says himself that time to choose is a luxury. Both wrote for the few, the renunciant and the Roman gentleman. Neither offers a wage. This chapter takes Patañjali’s diagnosis without his seer, and Cicero’s way of changing roles without his grounding of worth in reason.

One evening in January 2026, at the dining table, Priya opened the list she kept on her laptop. She added three items at the bottom: learn the vendor’s agent; ask the CPO about the post’s next year; reread November’s findings. Above them were items it had held for months. The first, unchanged since August, asked whether the developers who used the platform were becoming better engineers or only faster ones. She read down the list and saw that nothing on it was the thing she was afraid of. Each item was a task. The fear was not a task. She was trying to solve a feeling with a system.

She closed the laptop. Nothing dissolved. The fear was still there in the morning, but it had stopped looking like work.

29.6 The Other Side of Fear

Frank remembers a morning in November 2011, more than three years after he left the plant. He was on the back porch in Warren with a cup of coffee, watching a squirrel cross the telephone wire above the yard. He had no plans. For two years, having no plans had felt like having no meaning. That morning it felt like room. He remembers thinking: this is enough.

That is his report, and it is told here as his, not as the end of a road. His circumstances did not change. He never worked for wages again. He volunteers two mornings a week at a veterans’ centre in Roseville. He built his granddaughter a bookshelf that is, by any measure, overengineered. On some days he still misses the plant: the noise, and the men he ate lunch with. Donna still worries about money. Frank did not go on to found a company, write a memoir or discover a hidden talent.

The Yoga Sūtra says something about that morning which a promise of liberation misses. Of its five afflictions, it says of the fifth that it survives learning. “The will-to-live sweeping on [by the force of] its own nature exists in this form even in the wise.”8 Svarasavāhī viduṣo ’pi: literally, flowing by its own nature, even in the learned. The verse is about the fear of death, and I apply it, by analogy, to a smaller fear. The fear of becoming obsolete is a small form of the same wish: not to stop being. If the clinging remains even in the wise, the far side of fear is not freedom from fear. It is a life lived beside it. Frank’s fear came back after 2011, on bad days. No practice in this book promises calm.

The print edition said that the fear of obsolescence is “a fear of losing something you never were”. That is half true. Frank was never his job, and the porch showed him so. But he was also afraid of losing things he really had: an income, a craft, a place among men who knew his work, a say in it. Those were not illusions, and their loss was not a lesson.

What the morning showed was smaller than a liberation. The centre of a life, if a life has one, is not a task. Nobody could have had that morning for Frank, or been Donna’s husband in his place. A machine can do much of a person’s work. It does not live a person’s life for him, because a life is not a service one person performs for another. That is not a comfort about jobs. It is a fact about what a job never was.

So the chapter’s question has an answer in two parts. The fear of becoming obsolete is right about the goods a job carries. An income, a practice, standing and a voice can all be taken, and how much their loss hurts depends on what a society has decided people are owed. The fear is wrong about the person. No one becomes obsolete, because no one was ever for something in the way a tool is. Knowing the second part does not cancel the first, and it does not end the fear, which stays even in the wise. What it changes is what the fear is allowed to say. It may say: I could lose my work. It has no right to say: then I will be nothing.

What would prove this wrong?

The two parts of that answer are claims of different kinds. The second, that no person becomes obsolete, is true by what the words mean. No study can confirm it, and none can overturn it. The first is a claim about the world, and it can fail.

I hold that how much a lost job hurts depends partly on what a society has decided people are owed. Three findings would count against me. The distress might come mostly from people who were already unwell losing their jobs, rather than from the loss. It might be no smaller where the floor is higher. Or generous support might change the money and leave health untouched.

The evidence so far points the other way, with limits. Paul and Moser found that people in poor mental health were somewhat more likely to lose work or to stay without it, but they called that effect weak7. Jonathan Cylus and two colleagues studied American states from 1968 to 2008. Where unemployment benefits were more generous, a rise in unemployment brought a smaller rise in suicide among adults of working age, though the effect was small16. That is a study of states, not of any one person. A reader who wants to test me can look at places that raise or cut what they give people who lose work. Then she can look at what happens to those people’s health.

In the summer of 2026, at the kitchen table in Warren, Lisa asked him what had changed. “Took me a while to figure out there was a difference,” he told her. “Between what I did and what I was. Everybody told me I’d find something new to do. Nobody told me I’d have to figure out I was already something.”

Donna, at the counter behind him, said: “I told you.”

“She told me,” Frank said. “I wasn’t listening.”

Ending: The Sentence a Machine Wrote

At the end of January her mother’s friends came to lunch again. The older one asked the same question in the same tone.

This time Priya had a sentence ready. In November, the evening after the first lunch, she had asked a machine how to explain her work to two women who had known her since she was six. It gave her one sentence: she decides what her company’s testing machines should look for, and checks that they are looking for the right things. The sentence was accurate, and clearer than anything she had found herself. She had not used it then. She said it now, word for word.

The aunty nodded, satisfied, and asked about a cousin’s wedding. Her mother passed the rice.

Priya listened to the talk of the wedding and turned the sentence over. It was true. A machine had written it, and it described what she did better than she could. She could not tell whether it said who she was, or only what she was paid for.

Practice

What the fear is of. Write the sentence you say when someone asks what you do. Then separate what has already changed in your work from what you only expect: what has your employer actually announced? Next, imagine a machine doing that work well next year, and write what you are afraid of losing. Use four headings. Income. Practice: the daily work that keeps a skill alive. Standing: being needed and counted. Voice: a say in how your work changes. Beside each, write who decides it: you, your employer, your clients or the state. Mark each fear accurate, uncertain or mistaken, and write one fact for each mark. For an accurate fear that someone else decides, write whom you could tell, and whether it is safe to tell them. Last, write one sentence about yourself that uses no job title and no skill. If you cannot write it, that is worth knowing; it is not a failure. This takes half an hour; if the half hour is your employer’s, note who gave it. It will not make the fear go, and it promises no job, no wage and no calm. It shows how much of the fear is about the world and how much is about a sentence. It is not treatment. If you are struggling to cope with fear or worry, or what you try yourself is not helping, speak to a doctor. In a crisis, get urgent help23.

Words to keep

  • abhiniveśa: Sanskrit, clinging to life; in the Yoga Sūtra, the fifth affliction, which shows itself as the fear of death, present “even in the wise”.
  • anxiety: in clinicians’ use, the anticipation of a threat to come; fear is the response to a threat close at hand, real or perceived.
  • asmitā: Sanskrit, the sense of “I am”; in the Yoga Sūtra, the second affliction, taking the power by which one sees for the one who sees.
  • ego: in this chapter, three things: the picture of yourself built from your roles; pride; and the strength to take a blow and keep functioning.
  • four goods: what a job bundles and a machine can take apart: income, practice, standing and voice.
  • kleśa: Sanskrit, affliction; the Yoga Sūtra names five.
  • latent functions of employment: Marie Jahoda’s name for what a job gives besides a wage: a shape to the day, contact with people outside the family, a part in a shared purpose, status and regular activity.
  • obsolete: worn out, fallen out of use; said of tools and skills, never of persons.
  • persona: Latin, the mask a Roman actor wore; in Cicero, each of the four “characters” a person wears.
  • resilience: in George Bonanno’s research, bending under a loss without breaking, which is “more common than is often believed”.
  • stages of grief: Kübler-Ross’s five (denial, anger, bargaining, depression, acceptance), described from dying patients; words for states, not a road.
  • threat rigidity: in Staw, Sandelands and Dutton’s account (1981), the narrowing of attention and the return to well-learned responses that a threat produces in a person, a group or an organisation.

Sources

  1. Charlton T. Lewis and Charles Short, A Latin Dictionary (Oxford: Clarendon Press, 1879), s.v. obsolesco: “to wear out, to grow old, decay, fall into disuse, lose value, become obsolete” (alatius.com edition).
  2. Immanuel Kant, Groundwork of the Metaphysics of Morals (1785), Second Section, Akademie edition 4:434–435, trans. Thomas Kingsmill Abbott (1873): “Skill and diligence in labour have a market value”; what has a price can be replaced by an equivalent, what has dignity cannot (paraphrased); “morality, and humanity as capable of it, is that which alone has dignity” (4:435); see Chapter 5.
  3. American Psychiatric Association, Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) (Arlington, VA: American Psychiatric Publishing, 2013), p. 189, introduction to “Anxiety Disorders”. Paraphrased, because quotations of the sentence differ slightly between secondary sources.
  4. Erik Brynjolfsson, Bharat Chandar and Ruyu Chen, Stanford Digital Economy Lab, “Canaries in the Coal Mine” update, 12 August 2026: employment of workers aged 22–25 in the most AI-exposed occupations “about 19% below where it would be if it had kept pace with employment among similarly aged workers in less-exposed occupations” (as of June 2026; the authors call these “descriptive patterns, not causal estimates of the effect of AI”), the adjustment coming “primarily through reduced hiring of young workers rather than increased separations”; data from ADP payroll records; see Chapters 4 and 6.
  5. IANS, “Top IT firms’ hiring turns negative as headcount falls over 7,000 in FY26,” 25 April 2026: combined headcount of the five largest firms down 7,389 in FY26 (April 2025 to March 2026) after a rise of 12,718 in FY25, “amid an increasing focus on Artificial Intelligence (AI) and changing demand in the industry”, with layoffs of “around 12,000 employees” announced by TCS; see Chapter 4.
  6. Armin Granulo, Christoph Fuchs and Stefano Puntoni, “Psychological reactions to human versus robotic job replacement,” Nature Human Behaviour 3, no. 10 (2019): 1062–1069 (online 5 August 2019), abstract (“In contrast, being replaced by robots is associated with a greater perceived threat to one’s economic future”); the participants and the observer and own-job scenarios as described in “People prefer their jobs to be taken by robots, not other workers,” BPS Research Digest (British Psychological Society), 8 October 2019. The theories drawn on: the paper cites Leon Festinger’s theory of social comparison (1954) and Abraham Tesser’s self-evaluation maintenance model (1988); the Technical University of Munich’s release “Employees less upset at being replaced by robots than by other people” (9 August 2019) reports: “People tend to compare themselves less with machines than with other people.”
  7. Karsten I. Paul and Klaus Moser, “Unemployment impairs mental health: Meta-analyses,” Journal of Vocational Behavior 74 (2009): 264–282, abstract. Jahoda’s list and its test: Karsten I. Paul, Hannah Scholl, Klaus Moser, Andrea Zechmann and Bernad Batinic, “Employment status, psychological needs, and mental health: Meta-analytic findings concerning the latent deprivation model,” Frontiers in Psychology 14 (2023): 1017358 (the five latent functions, “time structure, social contact, status, collective purpose, and activity”; employed persons “always reporting a higher level of fulfillment of the latent and manifest functions than unemployed persons”; together the five and the manifest function, income, “explained 19% of variation in mental health”); Marie Jahoda, Employment and Unemployment: A Social-Psychological Analysis (Cambridge: Cambridge University Press, 1982). Marienthal: Marie Jahoda, Paul F. Lazarsfeld and Hans Zeisel, Die Arbeitslosen von Marienthal (Leipzig: S. Hirzel, 1933); in English, Marienthal: The Sociography of an Unemployed Community, translated by the authors with John Reginall and Thomas Elsaesser (Chicago: Aldine-Atherton, 1971; reprinted New Brunswick, NJ: Transaction, 2002, with an introduction by Christian Fleck), chapter 7, “The Meaning of Time”, pp. 66–77 (the women: “They have the household to run, which fully occupies the day”, p. 74); the walking observation as summarised in Fleck’s introduction (the men “walked more slowly across the main street and stopped more often on their way than women”). Threat and rigidity: Barry M. Staw, Lance E. Sandelands and Jane E. Dutton, “Threat-Rigidity Effects in Organizational Behavior: A Multilevel Analysis,” Administrative Science Quarterly 26, no. 4 (1981): 501–524. A poor job against none: Peter Butterworth, Liana S. Leach, Lyndall Strazdins, Sarah C. Olesen, Bryan Rodgers and Dorothy H. Broom, “The psychosocial quality of work determines whether employment has benefits for mental health: results from a longitudinal national household panel survey,” Occupational and Environmental Medicine 68, no. 11 (2011): 806–812, abstract (“the transition from unemployment to a poor quality job was more detrimental to mental health than remaining unemployed”). Selection (added with “What would prove this wrong?” in 29.6): Paul and Moser’s abstract (2009; 237 cross-sectional and 87 longitudinal studies) also reports mental-health-related selection during job loss and job search, and calls these effects weak (FAU research portal, cris.fau.de, publication 117729304).
  8. Patañjali, Yoga Sūtra 2.3, 2.6 and 2.9, trans. James Haughton Woods, The Yoga-System of Patañjali, Harvard Oriental Series 17 (Cambridge, MA: Harvard University Press, 1914), translation of the sūtras without the comment, p. xxxiii (the sūtras with Vyāsa’s comment begin sections at pp. 106, 115 and 117) (checked in the archive.org copies of the 1914 edition and the 1927 reprint). On the date: Philipp A. Maas (about 400 CE); Edwin F. Bryant, The Yoga Sūtras of Patañjali (New York: North Point Press, 2009), introduction (first to second century CE, by most scholars’ reckoning).
  9. Daniel C. Dennett, “The Self as a Center of Narrative Gravity,” in F. Kessel, P. Cole and D. Johnson (eds.), Self and Consciousness: Multiple Perspectives (Hillsdale, NJ: Erlbaum, 1992): “A self is also an abstract object, a theorist’s fiction.” Of a centre of gravity: “It has no mass; it has no color; it has no physical properties at all, except for spatio-temporal location”, yet it is “a fiction that has nicely defined, well delineated and well behaved role within physics”. What and who: Hannah Arendt, The Human Condition (Chicago: University of Chicago Press, 1958), §24, “The Disclosure of the Agent in Speech and Action” (what a person is, “qualities, gifts, talents, and shortcomings”, against who a person is); see Chapter 4.
  10. Cicero, De Officiis I.107, 114, 115, 117, 119, 120 and 121, trans. Walter Miller, Loeb Classical Library (London: William Heinemann; New York: Macmillan, 1913), pp. 109–125, LacusCurtius; Latin (I.107: duabus quasi nos a natura indutos esse personis) from The Latin Library.
  11. Lewis and Short, A Latin Dictionary, s.v. persona: “A mask, esp. that used by players”; “the part or character which any one sustains in the world” (alatius.com edition).
  12. Stellantis North America media site, “Sterling Heights Assembly Plant” (plant fact sheet, updated January 2024): built in 1953 as a jet engine plant and later run for the Army as a missile plant; “converted to an automobile plant in 1980 by Volkswagen and purchased by Chrysler Corporation in 1983”; production of the Chrysler LeBaron GTS and Dodge Lancer began in September 1984; 38111 Van Dyke, Sterling Heights. Chrysler as the Army’s contractor there, “building Redstone and Jupiter missiles”: Chrysler Group LLC, “Chrysler Group LLC Celebrates Groundbreaking of State-of-the-Art Paint Shop at Sterling Heights (Mich.) Assembly Plant,” news release, 20 June 2011.
  13. “More buyouts for Chrysler plant workers,” Autoblog, 9 January 2008, citing the Detroit News: buyouts at Jefferson North, Belvidere, Toledo North, Sterling Heights and Brampton; $70,000 and full retirement benefits for UAW workers with thirty or more years of service, or a $100,000 lump sum with six months of health care for workers with at least one year.
  14. “Chrysler files for Chapter 11 bankruptcy,” MPR News, 30 April 2009.
  15. Rania Gihleb, Osea Giuntella, Luca Stella and Tianyi Wang, “Industrial Robots, Workers’ Safety, and Health,” Labour Economics 78 (2022): 102205, abstract: a one-standard-deviation rise in commuting-zone robot exposure “reduces work-related annual injury rates by approximately 1.2 cases per 100 workers”; in Germany, a 4 per cent decline in physical job intensity, a 5 per cent decline in disability, and no significant effect on mental health.
  16. Rourke O’Brien, Elizabeth F. Bair and Atheendar S. Venkataramani, “Death by Robots? Automation and Working-Age Mortality in the United States,” Demography 59, no. 2 (2022): 607–628, abstract; the policy findings as reported in “Robots are increasing mortality among U.S. adults,” Yale News, 24 February 2022 (effects “more pronounced in states with ‘right to work’ laws” and “lower minimum wages”). Benefits and suicide (added with “What would prove this wrong?” in 29.6): Jonathan Cylus, M. Maria Glymour and Mauricio Avendano, “Do Generous Unemployment Benefit Programs Reduce Suicide Rates? A State Fixed-Effect Analysis Covering 1968–2008,” American Journal of Epidemiology 180, no. 1 (2014): 45–52, abstract (United States, adults aged 20 to 64; generous state benefits offset part of the effect of unemployment rates on suicide, “though estimated effects are small in magnitude”) (eScholarship copy; LSE Research Online record).
  17. Encyclopaedia Britannica, “Elisabeth Kübler-Ross”: the stages “experienced by the dying”, from seminars in which “the terminally ill were interviewed”; Elisabeth Kübler-Ross, On Death and Dying (New York: Macmillan, 1969).
  18. Margaret Stroebe, Henk Schut and Kathrin Boerner, “Cautioning Health-Care Professionals: Bereaved Persons Are Misguided Through the Stages of Grief,” OMEGA: Journal of Death and Dying 74, no. 4 (2017): 455–473, abstract.
  19. George A. Bonanno, “Loss, trauma, and human resilience: Have we underestimated the human capacity to thrive after extremely aversive events?,” American Psychologist 59, no. 1 (2004): 20–28, abstract. On what predicts it: George A. Bonanno, Sandro Galea, Angela Bucciarelli and David Vlahov, “What predicts psychological resilience after disaster? The role of demographics, resources, and life stress,” Journal of Consulting and Clinical Psychology 75, no. 5 (2007): 671–682, abstract (New York residents after the attacks of 11 September 2001; resilience “uniquely predicted by” factors including “income change, social support, frequency of chronic disease”).
  20. U.S. Bureau of Labor Statistics, “Worker Displacement: 2007–09,” news release, 26 August 2010 (long-tenured workers: three or more years with their employer; 6.9 million displaced; the 39 per cent for manufacturing, Table 4, and for ages 55 to 64, Table 1, are also for long-tenured workers). Routine jobs and recessions: Nir Jaimovich and Henry E. Siu, “Job Polarization and Jobless Recoveries,” Review of Economics and Statistics 102, no. 1 (2020): 129–147; NBER Working Paper 18334 (August 2012), abstract: “essentially all employment loss in routine occupations occurs in economic downturns”, and the slow recoveries in jobs are accounted for by jobless recoveries in those occupations. The recession “began in December 2007”: National Bureau of Economic Research, Business Cycle Dating Committee, announcement of 20 September 2010.
  21. Louis Jacobson, Robert J. LaLonde and Daniel G. Sullivan, “Estimating the returns to community college schooling for displaced workers,” Journal of Econometrics 125, no. 1 (2005): 271–304, abstract (workers displaced in the first half of the 1990s in Washington State; about 13 per cent for women; for more technically oriented courses, about 14 per cent for men and 29 per cent for women).
  22. Kaṭha Upaniṣad 3.3–4 and 3.9 (1.3.3–4 and 1.3.9), trans. Patrick Olivelle, Upaniṣads (Oxford: Oxford University Press, 1996), pp. 238–239 (paraphrased); see Chapter 2.
  23. NHS, “Get help with anxiety, fear or panic” (nhs.uk, page last reviewed 13 July 2026; accessed 29 October 2026): see a GP if “you’re struggling to cope with anxiety, fear or panic” or “things you’re trying yourself are not helping”; the page also directs anyone in “a mental health crisis or emergency” to urgent help.

Named people in this chapter are composites, and their conversations and circumstances are constructed; the studies, laws and figures cited are real and sourced.

Revised from the print edition (2026)

  • What the print edition claimed: that losing one’s work to a machine is an unchosen “ego death”, the contemplatives’ dissolution arriving uninvited (“In the AI era, the dissolution is not chosen. It arrives.”); that every tradition finds “the one who plays” the roles, and that tafakkur names the self’s dissolution; that “the ground beneath all professional identities is the fact of being conscious”, “the one thing about her that no machine can replicate”, “the experience that no machine has and no machine can simulate”; that a learning system “does not experience receiving” feedback, that “Feeling requires a subject”, and that “a meteorological model of a storm is not wet” (John Searle’s argument of 1980, used without his name); that what makes Priya’s judgement hers is “consciousness rather than computation”; that the reader, though not safe, “will be awake”; that research “consistently finds” that people replaced by technology suffer more than people laid off for economic reasons; that Kübler-Ross’s model is “a useful cultural framework”; that medication “helped with the chemistry” and “did not help with the meaning”; that Frank’s and Priya’s losses share one “psychological architecture”; that the fear of obsolescence is “a fear of losing something you never were”; and that a teacher’s classroom was a place “that no algorithm had ever entered”. It called Dennett an “eliminativist”. It also opened “At three in the morning” on six weeks of broken sleep and closed on “She sleeps.”, named Priya’s manager Sarita, put her on the 500D bus and promoted her into a role “that reported directly to the Chief Product Officer” after “two years” as an architect, gave a second teacher’s story laptops, several weeks of denial and a mystical afternoon, misplaced her town, gave Frank twenty-seven years on the line at a plant that had built cars only since Chrysler bought it in 1983, and a house “three miles east” of it, told his life as if from an interview (“he says, in the kitchen”), named him DeLuca and his daughter Karen, and ended on a bridge (“The next chapter is about what comes next.”).
  • What I now hold: the fear of becoming obsolete is often right about what a job’s loss can take (income, practice, standing and voice), and how much that loss hurts depends partly on institutions: unemployment causes distress, more for men, blue-collar workers and people with weak unemployment protection, and places exposed to robots fared worse where unions, wages and safety nets were weaker. It is wrong where it treats a person as a thing that wears out: skills have a price, persons a dignity. “Ego” means three things, and losing a role tightens the first rather than dissolving it. The Yoga’s diagnosis (asmitā, the instrument taken for the self) is borrowed without its seer, and its abhiniveśa, present “even in the wise”, withdraws any promise of calm. Cicero’s four masks show a career as a chosen role, a machine’s arrival as chance (chance for her; others chose it), and the free choice of a role as a privilege of the few; his rule for changing roles is to change slowly and explain why. Nothing here depends on whether machines experience anything. What cannot be automated is who a person is, by meaning, which protects no job. The ground beneath the floor is what never rested on being useful (worth, being the one whose life it is, the people to whom one is someone), and it does not pay the rent. The stages of grief are words for states, not a sequence: stage theory lacks sound evidence and can harm those who do not follow it, and resilience is common. Resilience is not a duty, and it depends on what a person has to lean on. A worker is called obsolete by an employer’s judgement, which does not follow from a skill’s decline; a market records a price, not a worth. Worth, being this one person, a life’s story and its roles are four things, and a machine reaches only the last two. A person can be the object of a change and one of its makers. Retraining is a preparation, not a promise of work, and a lost job is carried by a household. The aim is not to keep every job, but that losing one need not mean losing a place among others.
  • What changed my mind: the Prologue’s rule against defining people by what machines cannot yet do; Granulo, Fuchs and Puntoni’s experiments, which reverse the print’s order of fears; Paul and Moser’s meta-analyses; Gihleb et al. and O’Brien et al. on robots, injuries and deaths; the Bureau of Labor Statistics’ 2010 figures and the Washington State study of retraining; Stroebe, Schut and Boerner’s review and Bonanno’s work on resilience; Chrysler’s January 2008 offer; the Yoga Sūtra’s own list of afflictions, which puts ego and fear side by side and keeps the second in the wise; Cicero’s admission that few have time to choose a career; the audit’s corrections; and the class objection raised in an earlier chapter; three outside readings of an earlier version of this chapter, in October 2026; Jahoda’s account of what a job gives besides a wage, and Marienthal; Staw, Sandelands and Dutton on threat and rigidity; Jaimovich and Siu on routine jobs lost in recessions.
  • What survives: identity breaking before employment does, and the four goods as the test of a fear; “the affectionate persistence of customs officials”; “fear wearing the clothes of competence”; “The ego is not the enemy”; the better identity “still built on professional capability ... still, in principle, vulnerable”; “a promise this chapter cannot keep”; “You will not be safe. No professional identity is permanent.” verbatim; the stages of technological displacement, kept as words; Frank’s “Models are for people who didn’t go through it. The words are for people who did.”; the CAD certificate and the empty applications; “The tide is rising. You can build higher. You cannot build on sand forever.”; the ground beneath the floor, redefined, with “what remains when the job goes”; “She was trying to solve a feeling with a system.”; the porch and “this is enough”, told as Frank’s report; the overengineered bookshelf; “Nobody told me I’d have to figure out I was already something.”; Donna’s “I told you.”
  • Section titles in the print edition: three kept withdrawn claims: 29.2 was “Ego Death in the AI Era: When Your Identity Was Your Job”, 29.3 was “The Identity That Cannot Be Automated”, and 29.6 was “The Liberation on the Other Side of Fear”. Two more changed in draft 2: 29.1 was “The Anxiety of Automation: What It Feels Like to Be Replaced” (Priya had not been replaced), and 29.4 was “Grief, Anger, and the Stages of Technological Displacement” (the stages are not a model of displacement).

↑ Back to top

Discuss

Loading the discussion…

More in Artificial Intelligence
  •  ·  26 min read

    Epilogue

    Late in the Gītā, the one who knows most says: ponder it thoroughly, then act as you like. A machine can now set out the reasons on every side. Saying “I answer for this” then is not pride. It is a reply, for your part, to those owed one.

  •  ·  37 min read

    Chapter 40: What This Book Cannot Tell You

    The print edition’s last chapter reduced the book to one sentence, and rested human worth on what machines cannot do. I withdraw both. A book can date its claims and correct them in public. It cannot sit in the room where you decide.

  •  ·  6 min read

    Part IX: The Unfinished Chapter

    The print edition’s last Part said that consciousness never concludes. This one is a single chapter that turns the book on itself: what a book can stand behind, what it owes those who read the print edition, and what it leaves open.