Mass AI Job Losses Have Not Arrived: The Fear Has

A software engineer walks away from a decade of corporate code and puts on a police uniform. He has not been laid off, and nobody has told him his job is going. His reasoning, as he described it to The Guardian, is blunter than any redundancy letter: AI had drained much of the thinking and the fun out of coding, and corporate work no longer felt secure.

He is one of a growing cast. In a report published on 6 November 2026, The Guardian's Amelia Penniman described white-collar workers in the United States who have left established careers because they expect artificial intelligence to hollow them out. Vanessa Hsia, 30, spent nearly a decade in public relations before deciding to study for a master's degree in comparative literature. She described burnout and an industry chasing quantity over quality, and said that routine work such as emailing pitches to journalists was already being automated. A former advertising producer, identified only as Kara, left to concentrate on acting and now works as a barista. The software engineer, who asked to be called Chris, joined the police department of a major East Coast city.

Ten days earlier, The Globe and Mail had found the same pattern in Canada. Jim O'Quinn, 54, of Burlington, Ontario, left graphic design in 2023 and founded a personal-training business, Bro Pro Thrive. “What I have seen so far is AI absolutely devaluing what I brought to the table as a graphic designer,” he told the paper. “Graphic design is a chapter that is closed to me.”

None of this would be remarkable if the jobs were already vanishing in bulk. The strange thing is that, by most official measures, they are not. In early September 2026, The Economist estimated that AI had created roughly one million jobs in the United States since mid-2023, against about 200,000 layoffs attributed to the technology. The labour market apocalypse that executives, investors and podcasters have been forecasting has, so far, failed to turn up.

So a different question presents itself. What happens to people, and to the economy around them, when the fear of displacement does the displacing? If thousands of professionals rearrange their lives around a forecast, the forecast has already had consequences, whether or not it comes true. And when the people running AI companies cannot agree on what is coming, how is anyone supposed to tell a sensible hedge from a panic?

A Wave Measured in Anecdotes

The first thing to say about the pre-emptive flight is that nobody knows how big it is. The Guardian was careful on this point: its interviewees were individual stories, not evidence of an exodus. The Globe and Mail's Vanmala Subramaniam, writing on 27 October 2026 under the headline “'That field is dead'”, likewise built her report around people rather than a census.

Those people, though, are vivid. Eugene Kuznietsov, 34, worked in digital marketing as an SEO specialist at Dentalcorp in Toronto before co-founding Installix, a company that repairs and installs glass doors and windows across the Greater Toronto Area. “I don't think I'll ever go back into marketing,” he told the paper. “That field is dead.” There is an irony in his story that he readily acknowledges: he now uses Claude Code to run much of the business's administration. “The use of AI is really helping us cut administrative costs,” he said.

Rahman Moyed, 25, from Kitchener, Ontario, earned a graphic design diploma from Conestoga College in 2024. He has not yet made his move, but plans to become a handyman through home-maintenance apps such as Jiffy and TaskRabbit, and in the meantime does freelance web design and food delivery. His verdict on his own training is brutal: “I regret doing this graphic-design course.”

The Guardian had been tracking the same instinct in Britain earlier in the year. In February 2026, Lucy Knight, with additional reporting by Sumaiya Motara, profiled people retraining away from desk work: a Malmö-based academic language editor who went to culinary school and became a baker, a chartered occupational health and safety professional in Northampton who retrained as an electrical engineer, and a former University of Manchester geography student training in electrical installation.

Where survey data exists, it points to a large pool of people who are at least thinking about it. The Globe and Mail cited a FlexJobs survey, conducted in late 2025, of more than 3,000 white-collar professionals in North America, in which more than 60 per cent said they would consider switching to a skilled-trades job for greater stability. A separate February 2026 report from the payroll platform BorderlessAI found that 46 per cent of working Canadians believed AI would affect their long-term career plans, and that those with university degrees felt the least secure.

“Would consider” is doing a lot of work in that FlexJobs sentence. Considering a change is not making one. But the gap between the sentiment and the hard data on job losses is exactly where the interesting questions live.

The Numbers That Refuse to Panic

The Economist's September estimate, published under the headline “The jobs apocalypse is postponed”, is the starkest piece of counter-evidence. Roughly one million jobs created against roughly 200,000 attributed layoffs is a five-to-one ratio in AI's favour. The two figures are not measured in the same way. The layoff number tracks job cuts in which employers cite AI, while the creation figure is an estimate built from employment trends in AI-related occupations and infrastructure hiring, and the paper itself cautioned that not all of those jobs can be credited to AI. Still, the direction is hard to miss. Growth has clustered in data-centre construction, manufacturing and technical work, and in new AI-specific professional roles, even as routine administrative and service work declines.

Academic work points the same way. In October 2025, the Yale Budget Lab, working with researchers at Brookings, examined the 33 months after ChatGPT's launch in November 2022 and found no discernible economy-wide disruption. “While the occupational mix is changing more quickly than it has in the past, it is not a large difference and predates the widespread introduction of AI in the workforce,” the authors wrote. They compared the shift with the arrival of personal computers and the internet and found nothing out of the ordinary.

Executives, asked directly, report the same thing. A 2026 Federal Reserve Bank of Atlanta working paper, “Firm Data on AI”, surveyed nearly 6,000 executives across the United States, the United Kingdom, Germany and Australia. More than 90 per cent reported no impact of AI on their own firm's employment over the previous three years, and 89 per cent reported no impact on labour productivity. Looking ahead, they expected AI to trim their own headcount by an average of 0.7 per cent. That is a striking gap between boardroom reality and public mood. Executives who actually run payrolls report almost nothing happening so far and expect only small effects. Yet much of the public conversation is set by a handful of leaders at frontier AI companies, who have every commercial reason to describe the technology as transformative, whichever direction that transformation is said to take.

Then there are the official projections, which sit oddly beside the anecdotes. The US Bureau of Labor Statistics projects employment of software developers to grow by 10 per cent between 2025 and 2035, which it classes as much faster than average. Median pay for the occupation was $135,980 in May 2025. The career Chris left is, on paper, among the more secure in the American economy.

None of this proves the fear is wrong. Labour statistics lag, and a decade-long projection is only as good as its assumptions. But it does establish that, at the moment people like Chris and Hsia made their decisions, the aggregate data was not telling them to run.

Where the Fear Is Not Imaginary

It would be glib, though, to treat the leavers as simply misinformed. Averages can hide a great deal, and some of what they hide supports the people heading for the exit.

The most cited evidence of real damage comes from Stanford. In August 2025, Erik Brynjolfsson, Bharat Chandar and Ruyu Chen used ADP payroll records to show a 13 per cent relative decline in employment for workers aged 22 to 25 in the occupations most exposed to AI, after controlling for firm-level shocks. Subsequent revisions have made the finding starker. A November 2025 revision put the decline at about 16 per cent, and an August 2026 revision, extending the data through June 2026, found that the employment gap for 22 to 25 year olds in the most exposed occupations had widened to about 19 per cent below where it would have been had it kept pace with less-exposed peers, operating mainly through reduced hiring rather than increased separations. Older workers in the same occupations did not suffer the same fall. Crucially, the declines were concentrated where AI automates work rather than augments it. Where AI acts as an assistant, employment held up or grew.

Specific fields show specific strain. The Globe and Mail reported Indeed data showing that, in the year to August 2026, Canadian job postings in marketing fell 8.2 per cent and postings in media and communications fell 15.2 per cent, while total postings fell just 0.4 per cent. The Bureau of Labor Statistics projects employment of graphic designers to decline by 2 per cent between 2025 and 2035. Kuznietsov's marketing and O'Quinn's design work sit squarely inside those numbers.

In Britain, a King's College London study led by Bouke Klein Teeselink, published in October 2025 and cited by The Guardian, predicted the sharpest AI-driven declines in employment and wages in software engineering and management consultancy.

So the pre-emptive leaver is not responding to nothing. They are responding to a signal that is real in some corners, faint in others, and often louder in their own workplace than in the national data. Chris told The Guardian that AI had taken much of the thinking and enjoyment out of coding and that his employer pushed it hard on his team. Kara's employer, according to the paper, encouraged her team to use generative tools such as Runway and ElevenLabs while warning about the legal risks in the same breath, a contradiction that landed on her desk because she handled legal clearance.

That detail matters. Many of these departures are not purely about future unemployment. They are about what the job has already become: a role that still pays, still exists, and still appears in the statistics, but no longer feels like the work the person trained for. A job can survive automation and still lose the parts that made someone want it.

When the Prophets Disagree

If workers are confused about the scale and timing of disruption, they are in good company. The people building and buying AI are openly contradicting one another.

At one pole sits Dario Amodei, chief executive of Anthropic. In May 2025 he told Axios that AI could eliminate half of all entry-level white-collar jobs and push US unemployment to between 10 and 20 per cent within one to five years, and urged companies and governments to stop sugar-coating the prospect. The Guardian's November report placed that warning near the centre of the anxiety its interviewees described.

At the other pole sits Jensen Huang of Nvidia. Asked about Amodei's forecast at a press briefing during VivaTech in Paris in June 2025, Huang said he “pretty much” disagreed with “almost everything” his fellow chief executive says. Anthropic replied that Amodei stood by his concerns about entry-level jobs. Even the terms of the disagreement are disputed. Huang's position has only hardened. Speaking at Y Combinator's Startup School, in remarks reported by Fortune on 28 July 2026, he said: “The narrative about AI destroying jobs is exactly backwards.” His argument is that tasks and jobs are different things. “AI automates tasks away, but it doesn't necessarily eliminate jobs,” he said, adding that “every single job will change, and there'll be a whole bunch of new jobs.”

In between sit executives who say both things at once. In a June 2025 memo to staff, Amazon's chief executive Andy Jassy wrote that “we expect that this will reduce our total corporate workforce as we get efficiency gains from using AI extensively across the company”, and that Amazon would need “fewer people doing some of the jobs that are being done today, and more people doing other types of jobs.” The memo gave no timeline and no numbers. It is the kind of statement that can reasonably frighten an employee without telling them anything they could act on.

Corporate signals have kept hardening since. In late October 2026, Microsoft made a voluntary exit offer to roughly a quarter of its roughly 223,000-strong workforce, its largest ever buyout programme, while framing the restructuring as preparation for an AI era rather than AI replacement; press reports anticipate further layoff rounds in November and January. It is a voluntary offer, not a confirmed cut — but, as with Jassy's memo, a message that can frighten without telling anyone anything they can act on.

Then there are the reversals. Klarna, the Swedish payments company, froze hiring in late 2023 and promoted its AI assistant as doing the work of hundreds of customer service agents. In May 2025 its chief executive, Sebastian Siemiatkowski, told Bloomberg that the company was hiring human agents again, conceding that a focus on cost had led to a decline in quality. The company did not abandon AI, but it did demonstrate that an announced automation can be partially undone in well under two years.

The disagreement even reaches inside the labs. In February 2026, Mrinank Sharma, who led Anthropic's Safeguards Research team, resigned with a public letter warning that “the world is in peril”, and said he would be “moving back to the UK and letting myself become invisible for a period of time”. Several outlets reported that he intended to study poetry. The Guardian opened its November piece with his departure. There is something telling about a public relations specialist and an AI safety researcher both heading towards literature: a shared sense that the humanities offer a kind of ground that machines have not yet levelled.

For an individual worker, this cacophony creates a specific problem. When forecasts range from “half of entry-level jobs” to “exactly backwards”, choosing which expert to believe becomes a matter of temperament. Anxious people find the anxious forecasts more plausible. Optimists find the optimists more persuasive. The forecast a worker acts on may say more about the worker than about the labour market.

Anticipation as Its Own Injury

One of the clearest bodies of evidence on what fear of job loss does to people predates generative AI by three decades, and it comes from Whitehall.

In 1995, Jane Ferrie, Martin Shipley, Michael Marmot, Stephen Stansfeld and George Davey Smith published a study in the BMJ using data from the Whitehall II cohort of London-based civil servants. One department, with 666 staff in the study, was threatened with early privatisation. The researchers compared these employees' self-reported health before and during the period of anticipation with that of staff in 19 other departments. The self-reported health of those facing privatisation tended to decline. The authors' conclusion was direct: “anticipation of job loss affects health even before employment status has changed.”

That finding reframes the pre-emptive flight. Even workers who stay put are not unharmed by a forecast; the waiting itself has a cost. Some workers will absorb that cost by remaining in place and worrying. Others will try to escape it by moving first. Neither group has actually been displaced, and both are paying something.

The scale of the worry is now measurable. In a Pew Research Center survey conducted in late June 2026 among 3,488 US adults, 71 per cent said they expected AI to lead to fewer jobs over the next 20 years, up from 64 per cent in 2024. Just 5 per cent expected more jobs. Among adults under 30, 73 per cent expected losses, up from 61 per cent two years earlier.

A more granular picture comes from Anat Bracha and Jenny Tang at the Federal Reserve Bank of Boston. Using a special module of the New York Fed's Survey of Consumer Expectations, fielded in December 2024, they found that only about 5 per cent of workers feared losing their job to AI. Yet about 21 per cent expected AI to worsen their financial situation within one to five years. Concern varied sharply by industry, exceeding 30 per cent in leisure services, while doctorate holders were the least worried. Only around 29 per cent felt they could adjust to AI on their own; 49 per cent said they could with proper training.

Put those figures side by side and a pattern emerges. Most workers do not think the axe is about to fall on them personally. They think the ground is shifting, that their earnings and standing will erode, and that they lack the tools to respond alone. That is precisely the psychological condition in which a dramatic, self-directed leap starts to look attractive. It converts a slow, shapeless dread into a decision.

The Price of Starting Again

Leaping has a price, and economists have been measuring it for years.

In a 2009 paper in the International Economic Review, Gueorgui Kambourov and Iourii Manovskii found that, other things being equal, five years of occupational tenure is associated with wage gains of roughly 12 to 20 per cent. Once occupational experience is accounted for, tenure with a particular employer or industry matters comparatively little. Their interpretation is that much of what makes a worker valuable is tied to the occupation itself. Leave the occupation and a large part of that accumulated value stays behind.

The official pay data makes the trade-offs concrete. According to the Bureau of Labor Statistics, the median annual wage for software developers in May 2025 was $135,980. For police and sheriff's patrol officers it was $76,210. For fitness trainers and instructors it was $47,160, against $62,960 for graphic designers. Medians are not destinies, and many individuals will earn more or less than these figures. But for the occupational moves in the reporting, the official numbers point to a step down in typical pay, in Chris's case a substantial one.

Some changers will judge that worth it, and they may be right. Chris told The Guardian he found satisfaction in working directly with his community after more than a decade of corporate work. O'Quinn's personal-training business, the BLS suggests, sits in an occupation projected to grow 7 per cent between 2025 and 2035, much faster than average. The trades the FlexJobs respondents fancy are, as George Brown College's Adel Esayed told The Globe and Mail, “an area where AI cannot touch.”

But the costs are not only financial. Retraining eats years. Apprenticeships and degrees carry fees and forgone earnings. And the decision is hard to reverse: a former software engineer who wants to return in five years will find that the field has moved without them, and that their occupational capital has depreciated. If the fear proves exaggerated, the pre-emptive leaver absorbs the full cost of displacement without ever having been displaced. That is the purest form of anticipatory harm: a loss produced entirely by the forecast.

There is a quieter inequality hidden in all this. A pre-emptive move is a luxury good. It requires savings to cover a retraining period, a household that can absorb a pay cut, and enough confidence to bet on a new field. The workers best placed to leap early are often not the ones most exposed. The Stanford data shows that the sharpest declines have fallen on 22 to 25 year olds, precisely the group least likely to have a financial cushion, while Moyed, at 25, is filling the gap between careers with food delivery. Those who can afford to hedge do so; those who cannot simply wait and absorb whatever comes.

There is also a risk of herding. If a meaningful share of the 60 per cent who would consider the trades actually move, they will arrive in labour markets with their own constraints on training capacity, licensing and local demand. Safety in numbers is not guaranteed when everyone is running towards the same shelter.

Identity After the Exit

Money is the cost that shows up on a payslip. The harder cost to measure is who you become. Careers are not only income streams. They are answers to the question “what do you do?”, and for many professionals they are the scaffolding of a self.

Herminia Ibarra of London Business School, whose 2003 book “Working Identity” remains one of the most cited studies of mid-career change, argued that people rarely change careers by first discovering a true self and then acting on it. They change by doing: experimenting with what she called provisional selves, testing possibilities in small ways and letting what they learn shape the next step. In her account, one big, decisive leap made in a single go is usually the wrong approach.

The pre-emptive flight often runs the other way. Fear compresses the timeline. Instead of trying on a new identity, the worker abandons the old one, sometimes in language that sounds almost like grief. “Graphic design is a chapter that is closed to me,” O'Quinn said. “That field is dead,” said Kuznietsov. Moyed regrets the course that was supposed to launch him. These are not the words of people sampling options. They are the words of people closing doors behind them.

There is a case for that kind of finality. Some workers clearly find relief in it. “Business is going well so far,” Kuznietsov told The Globe and Mail of his glazing company. Hsia left an industry she felt had come to value AI output over critical thinking. For people whose work had already been reorganised around tools they distrusted, exiting may be less a forecast than a verdict on the present.

But identity built on an exit is fragile in a particular way. It defines the new self by what it escaped. If the predicted collapse of the old field never arrives, or arrives in a milder form, the story the person has told themselves can wobble. A professional identity has to be rebuilt in public, among colleagues who did not leave and who may be doing perfectly well.

When the Forecast Writes Itself

In 1948, the sociologist Robert K. Merton named the self-fulfilling prophecy, illustrating it with a parable of a bank run. A false belief about a situation triggers behaviour that makes the belief come true. A solvent bank collapses because enough depositors believe it will and rush to withdraw their money.

The AI labour market has the ingredients for a version of this. Brendan Moore, an economist at the University of Notre Dame who studies worker sentiment about AI, told The Guardian that executive warnings and companies blaming layoffs on AI were likely fuelling workers' fears, even where firms might have downsized anyway. He went further: “just the fear of job loss alone, even if mass job loss never comes, can affect the economy.” If enough people change course, he argued, the effects on wages, prices and growth could resemble the turbulence of the pandemic-era “great resignation”.

The mechanism is not hard to sketch. If experienced professionals leave a field pre-emptively, employers lose skills and institutional knowledge. Some will respond by leaning harder on automation to fill the gaps, which accelerates exactly the substitution the leavers feared. Meanwhile, students read the same headlines and avoid the field altogether, thinning the pipeline. A profession can be hollowed out by expectations before it is hollowed out by software.

The prophecy can also run in reverse. Companies that announce AI-driven efficiencies raise worker anxiety, which raises turnover, which makes AI investment look more necessary. Klarna's experience suggests that the loop can be broken by evidence: when quality fell, humans came back. But not every firm will be as quick, or as public, in reversing course.

None of this means the leavers caused the disruption they fled. It means that anticipation is not a neutral act of observation. A labour market is made of people's expectations as much as of their tasks, and a widely shared forecast becomes an economic force in its own right.

Telling Adaptation From Alarm

Given all this, how might a worker distinguish a rational hedge from an anxiety-driven bolt? The evidence suggests a handful of practical tests, none of them perfect.

The first is to look at tasks, not titles. Huang's distinction between jobs and tasks is self-serving coming from the man who sells the chips, but it is backed by Brynjolfsson and his colleagues, who have found employment falling where AI automates work and holding up where it augments it. A worker whose core tasks are being handed wholesale to software faces a different risk from one whose tools are simply getting faster.

The second is to read local evidence rather than national rhetoric. Job postings for one's own role and region, like the Indeed data showing Canadian marketing and communications advertisements falling far faster than the market as a whole, are more informative than any chief executive's keynote. So are official projections, which in the case of software developers point upward and in the case of graphic designers point slightly down.

The third is to notice what is actually driving the decision. The Boston Fed data suggests most workers are not afraid of imminent dismissal but of slow erosion. Several of the leavers in the reporting were reacting to how their work already felt, under mandates and tools they disliked. That can be a perfectly good reason to change careers. It is simply a different reason from a forecast of mass unemployment, and it deserves to be judged on its own terms.

It is also worth noticing what workers say would help. In the Boston Fed survey, nearly half of respondents said they could adapt to AI with proper training, compared with fewer than a third who felt they could manage alone. That suggests the more rational response for many people is not to abandon their occupation but to demand, or seek out, the training to work alongside the tools within it.

The fourth is to prefer reversible moves. Ibarra's research suggests that experiments, such as evening courses, side projects and part-time work in the new field, reveal more than leaps. They also keep occupational capital intact for longer, softening the losses Kambourov and Manovskii measured.

The fifth is humility about forecasts, including the optimistic ones. In 2013, Carl Benedikt Frey and Michael Osborne of Oxford estimated that 47 per cent of US jobs were at risk of computerisation; Frey has since acknowledged, as The Guardian noted, that technologies such as autonomous vehicles took far longer to arrive than expected. In 2016, Geoffrey Hinton suggested radiologists would soon be made obsolete; according to figures Huang cited, their numbers grew by about 12 per cent between 2010 and 2022. Forecasts of rapid displacement have a poor record. But the Stanford payroll data — whose young-worker gap has widened with each revision — is a reminder that some forecasts land, first at the bottom of the ladder.

What the Early Leavers Are Really Telling Us

It is tempting to read the pre-emptive flight as either prescience or panic. The evidence suggests it is neither, or rather both at once. The aggregate numbers are calm, and on present data a former software engineer in uniform has probably given up more than the labour market was about to take from him. Yet some fields really are contracting, young workers in exposed jobs really are losing ground, and the experience of work in many offices really has changed in ways that people find alienating.

So, to return to the questions the leavers pose. Does anticipation produce its own disruption? Yes, and measurably: in health, in lost occupational capital, in identities closed off before the old field has actually closed. Can workers tell rational adaptation from anxiety? Only partly, and only if they look past the headline forecasts to the tasks they actually perform and the hiring data for their own role.

What the leavers reveal most clearly is a failure of signal. Workers are being asked to make decades-long decisions on the basis of forecasts from executives who openly contradict each other, statistics that lag reality, and workplace mandates that arrive without explanation. In that fog, the forecast itself does the damage: to health, as Whitehall showed; to accumulated skill, as the wage data shows; and, if Moore is right, to the wider economy, even if the predicted job losses never fully arrive.

Better information would not end the anxiety, but it would make the choices less blind. Employers could be specific about which tasks they intend to automate and on what timetable. Policymakers could publish faster, finer-grained data on occupational demand. And the people building these systems could weigh the fact that their forecasts are not commentary. For a PR specialist in a seminar room, a designer in a gym and an engineer on patrol, those forecasts have already become life decisions.


References and Sources

  1. Penniman, Amelia. “Fearing AI disruption, white-collar workers are quitting their jobs.” The Guardian, 6 November 2026. https://www.theguardian.com/technology/2026/nov/06/ai-disruption-job-quitting Accessed 9 November 2026.
  2. Subramaniam, Vanmala. “'That field is dead': Bracing for AI disruption, white-collar workers are pre-emptively changing careers.” The Globe and Mail, 27 October 2026. https://www.theglobeandmail.com/business/technology/article-bracing-ai-disruption-white-collar-pre-emptively-changing-careers/ Accessed 9 November 2026.
  3. Knight, Lucy, with additional reporting by Sumaiya Motara. “The big AI job swap: why white-collar workers are ditching their careers.” The Guardian, 11 February 2026. Republished by Portside, 13 February 2026. https://portside.org/node/38967/printable/print Accessed 9 November 2026.
  4. The Economist. “The jobs apocalypse is postponed. An AI jobs boom is here.” September 2026. Summarised via Dealroom, https://dealroom.co/news/econ-6xq5bi-the-jobs-apocalypse-is-postponed-an-ai-jobs-boom-is-here/ Accessed 9 November 2026.
  5. Gimbel, Martha; Kinder, Molly; Kendall, Joshua; Lee, Maddie (The Budget Lab at Yale and Brookings). “Evaluating the Impact of AI on the Labor Market: Current State of Affairs.” The Budget Lab at Yale, 1 October 2025. https://budgetlab.yale.edu/research/evaluating-impact-ai-labor-market-current-state-affairs Accessed 9 November 2026.
  6. Federal Reserve Bank of Atlanta. “Firm Data on AI.” Working Paper 2026-03, March 2026. https://atlantafed.org/research-and-data/publications/working-papers/2026/03/24/03-firm-data-on-ai Accessed 9 November 2026.
  7. US Bureau of Labor Statistics. Occupational Outlook Handbook: Software Developers. https://www.bls.gov/ooh/computer-and-information-technology/software-developers.htm Accessed 9 November 2026.
  8. US Bureau of Labor Statistics. Occupational Outlook Handbook: Graphic Designers. https://www.bls.gov/ooh/arts-and-design/graphic-designers.htm Accessed 9 November 2026.
  9. US Bureau of Labor Statistics. Occupational Outlook Handbook: Police and Detectives. https://www.bls.gov/ooh/protective-service/police-and-detectives.htm Accessed 9 November 2026.
  10. US Bureau of Labor Statistics. Occupational Outlook Handbook: Fitness Trainers and Instructors. https://www.bls.gov/ooh/personal-care-and-service/fitness-trainers-and-instructors.htm Accessed 9 November 2026.
  11. Brynjolfsson, Erik; Chandar, Bharat; Chen, Ruyu. “Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence.” Stanford Institute for Economic Policy Research working paper, August 2025; revised November 2025 and August 2026 (data through June 2026). https://siepr.stanford.edu/publications/working-paper/canaries-coal-mine-six-facts-about-recent-employment-effects-artificial Accessed 9 November 2026. Revised version: Stanford Digital Economy Lab, https://digitaleconomy.stanford.edu/publication/canaries-in-the-coal-mine-six-facts-about-the-recent-employment-effects-of-artificial-intelligence/ Accessed 14 November 2026.
  12. Fortune. Report on Jensen Huang's VivaTech remarks on Dario Amodei and AI jobs. 11 June 2025. https://fortune.com/2025/06/11/nvidia-jensen-huang-disagress-anthropic-ceo-dario-amodei-ai-jobs Accessed 9 November 2026.
  13. Quiroz-Gutierrez, Marco. Report on Jensen Huang's remarks at Y Combinator Startup School (AI killing tasks, not jobs). Fortune, 28 July 2026. https://fortune.com/2026/07/28/nvidia-jensen-huang-ai-killing-tasks-not-jobs/ Accessed 9 November 2026.
  14. SiliconANGLE. “Andy Jassy says generative AI will reduce Amazon's workforce in the coming years.” 17 June 2025. https://siliconangle.com/2025/06/17/andy-jassy-says-generative-ai-will-reduce-amazons-workforce-coming-years/ Accessed 9 November 2026.
  15. The Verge. “Microsoft's AI future has fewer jobs.” November 2026. https://www.theverge.com/tech/1022023/microsoft-exit-offer-layoffs-notepad Accessed 14 November 2026.
  16. Entrepreneur. “Klarna Is Hiring Customer Service Agents After AI Couldn't Cut It on Calls, According to the Company's CEO.” May 2025. https://www.entrepreneur.com/business-news/klarna-ceo-reverses-course-by-hiring-more-humans-not-ai/491396 Accessed 9 November 2026.
  17. Global News. “Anthropic AI safety researcher Mrinank Sharma quits, citing concerns.” February 2026. https://globalnews.ca/news/11664538/anthropic-ai-safety-researcher-mrinank-sharma-quits-concerns Accessed 9 November 2026.
  18. Ferrie, Jane E.; Shipley, Martin J.; Marmot, Michael G.; Stansfeld, Stephen; Davey Smith, George. “Health effects of anticipation of job change and non-employment: longitudinal data from the Whitehall II study.” BMJ, 311(7015), 1264-1269, 11 November 1995. DOI: 10.1136/bmj.311.7015.1264. https://pmc.ncbi.nlm.nih.gov/articles/PMC2551182 Accessed 9 November 2026.
  19. Pew Research Center. Survey of US adults on AI and jobs, conducted June 2026. https://www.pewresearch.org/?p=356101 Accessed 9 November 2026.
  20. Bracha, Anat; Tang, Jenny. “Shaping the Future of Work: Workers' Optimism and Pessimism about AI.” Federal Reserve Bank of Boston, Current Policy Perspectives No. 25-16, 17 December 2025. https://www.bostonfed.org/publications/current-policy-perspectives/2025/workers-fears-and-hopes-about-ai.aspx Accessed 9 November 2026.
  21. Kambourov, Gueorgui; Manovskii, Iourii. “Occupational Specificity of Human Capital.” International Economic Review, 50(1), 63-115, 2009. https://ideas.repec.org/a/ier/iecrev/v50y2009i1p63-115.html Accessed 9 November 2026.
  22. Ibarra, Herminia. “Working Identity: Unconventional Strategies for Reinventing Your Career.” Harvard Business School Press, 2003.
  23. Ibarra, Herminia. “Provisional Selves: Experimenting with Image and Identity in Professional Adaptation.” Administrative Science Quarterly, 44(4), 764-791, 1999. https://web.mit.edu/curhan/www/docs/Articles/15341_Readings/Self-presentation_Impression_Formation/Ibarra_1999_Provisional_selves.pdf Accessed 9 November 2026.
  24. Merton, Robert K. “The Self-Fulfilling Prophecy.” The Antioch Review, 8(2), 193-210, 1948. DOI: 10.2307/4609267.
  25. Frey, Carl Benedikt; Osborne, Michael A. “The Future of Employment: How Susceptible Are Jobs to Computerisation?” Oxford Martin School working paper, 2013; published in Technological Forecasting and Social Change, 114, 254-280, 2017. DOI: 10.1016/j.techfore.2016.08.019.

Tim Green

Tim Green UK-based Systems Theorist & Independent Technology Writer

Tim explores the intersections of artificial intelligence, decentralised cognition, and posthuman ethics. His work, published at smarterarticles.co.uk, challenges dominant narratives of technological progress while proposing interdisciplinary frameworks for collective intelligence and digital stewardship.

His writing has been featured on Ground News and shared by independent researchers across both academic and technological communities.

ORCID: 0009-0002-0156-9795 Email: tim@smarterarticles.co.uk

Listen to the free weekly SmarterArticles Podcast

Support What We Do

Discuss...