What a Basic Income Cannot Buy: Work, Meaning and the AI Shock

Offer people their full salary to stop working and most will take it. Offer them a fifth more than their salary and a stubborn minority will still say no.

That second group is the most interesting finding in a paper posted to arXiv on 29 October 2026. Written by Julian Jacobs, a researcher at Google DeepMind and a doctoral candidate in politics at the University of Oxford, and supported by Google DeepMind, it rests on an original survey of 8,234 working-age adults in the United Kingdom, the United States, Denmark and India. The fieldwork ran in June and July 2026 in partnership with the polling organisation More in Common. The headline numbers are disarmingly warm. Eighty-one per cent of respondents said they were satisfied with their jobs. About one third were motivated intrinsically, valuing the work for its own sake. Nearly a quarter were driven chiefly by the social and community side of employment. A further 17 per cent were motivated by status, and just over a quarter by the plain transactional fact of the pay cheque.

Then came the hypothetical cheque. A majority said they would leave their role if offered their current salary as a guaranteed income. Yet one in five said they would stay even if the guaranteed income rose to 120 per cent of what they earned. People who valued their work for its own sake refused most often.

Jacobs is not arguing against a basic income. He writes that guaranteeing one may well be necessary to manage a future AI labour shock. His warning is about what the money leaves out: “Income transfers can fail entirely to compensate for this non-pecuniary value of work.”

That sentence carries more weight than it first appears to. Nearly every serious forecast of AI's effect on employment counts jobs, tasks, wages and hours. Nearly every serious policy response is a way of moving money: retraining grants, wage insurance, a universal basic income. If a large part of what a job gives a person was never in the payslip, then the forecasts measure the wrong loss and the remedies replace the wrong thing.

Four Countries, Four Kinds of Worker

The survey sorts respondents into what Jacobs calls Work Values Profiles: Intrinsic, Community, Status-Driven and Transactional. The labels matter less than the detail he puts underneath them.

The first detail is distribution. You might expect meaning to be a luxury good, the possession of architects and academics rather than warehouse pickers and care assistants. The data say otherwise. Jacobs reports that “non-monetary work attachment can be found in near-equal measure across income levels”, and that the share of people who would turn down a basic income is consistent across the income distribution and across occupational categories. Blue-collar workers are as likely as white-collar ones to value their jobs for reasons that have nothing to do with pay.

The second detail is what people say they would do with the time. Transactional workers mostly said they would pursue leisure. The other groups did not. According to the paper, 57 per cent of intrinsically motivated workers would fill their time with volunteering or caregiving, as would 51 per cent of community-oriented workers. People attached to their jobs for non-monetary reasons were also more likely to say they would look for other paid work or start a business. Give them freedom from wages and many of them would go looking for something that resembles a job.

The third detail is geography. The India sample, gathered online, showed the highest refusal rate of all: 28.7 per cent would decline a guaranteed income worth more than their salary. In both the Western and Indian samples, intrinsically motivated workers showed the lowest inclination to leave.

There are limits, and Jacobs states them. The answers are stated preferences about an imagined offer, not observed behaviour. The data are cross-sectional, so the paper's suggestive links between non-monetary attachment and political reactions to AI cannot establish cause. An online sample from India is not a portrait of Indian labour as a whole. The political thread is still worth noting. Among workers attached to their jobs for non-monetary reasons, those who saw AI as a threat to their industry showed steeper negative political reactions than other groups. It hints that the loss of meaning, unlike the loss of income, may not stay private. People who feel that something valuable is being taken from them, and that no cheque will return it, may look for someone to hold responsible.

None of this undermines the central point, which is a modest one: work gives people something that a cash-equivalent offer does not fully replace, and that something is not confined to the affluent.

It is also not a new discovery. It is one of the oldest findings in social science, and it was made in a village south of Vienna.

The Village Where Time Went Slack

In the winter of 1931 to 1932, a team of young researchers from Vienna went to Marienthal, a workers' settlement built around a single textile factory. The factory had closed. According to the study's historians, roughly half of the workers were unemployed and about three-quarters of families were affected. The researchers were Marie Jahoda, Paul Lazarsfeld and Hans Zeisel, with much of the fieldwork carried out by Lotte Schenk-Danzinger. Their book, Die Arbeitslosen von Marienthal, was published in Leipzig in 1933. Jahoda wrote the main text, Zeisel the sociographic appendix and Lazarsfeld the introduction.

Its method was unusual for its time. The team combined interviews and household statistics with close observation of everyday behaviour. They studied the accounts of the co-operative store, the library and the local shops. They even recorded how fast people walked through the streets and how long they stood about. They were not measuring poverty, which was obvious. They were measuring what happens to a community when its routines collapse.

What they found was not revolt. It was a slackening. The book's reviewers have long summarised it as a diminution of expectation and activity, a disrupted sense of time and a steady slide into apathy. The image the study left behind was of a tired community. Jahoda wrote that the workers, cut off from their work and from contact with the outside world, had lost the material and moral means to use their time. The emphasis falls on the word moral. Free time did not become leisure. Without the scaffolding of the working day, it became something closer to drift.

Half a century later Jahoda turned that observation into a theory. In a 1981 article in American Psychologist and a 1982 book, Employment and Unemployment: A Social-Psychological Analysis, published by Cambridge University Press, she borrowed the sociologist Robert Merton's distinction between the manifest and latent functions of institutions. The manifest function of employment is income. The latent functions are the things a job provides as a by-product, which nobody signs a contract to receive. Jahoda named five: time structure, social contact, collective purpose (the sense of being useful to other people), status and regular activity.

Her claim was that these are psychological needs, and that modern societies supply them overwhelmingly through jobs. Remove the job and you remove all five at once. The distress of unemployment, on this account, is only partly about money. It is the experience of losing the structure of the day, the people you saw in it, the sense that someone needed you, your answer to the question “what do you do?” and the reason to get up.

For decades the latent deprivation model was more quoted than tested. That has changed. In 2023, Karsten Paul, Hannah Scholl, Klaus Moser, Andrea Zechmann and Bernad Batinic published a meta-analysis in Frontiers in Psychology bringing the evidence together. Employed people reported higher levels than unemployed people on all five latent functions as well as on income. Together the five functions accounted for 19 per cent of the variation in mental health. The authors also flagged the question this article turns on: it is still not clear whether employment is the best provider of those functions, or merely the most common one.

Unemployment Hurts More Than the Money

If Jahoda is right, joblessness should hurt beyond what lost income would predict. A large body of research says it does.

In 2009 Karsten Paul and Klaus Moser published a meta-analysis in the Journal of Vocational Behavior covering 237 cross-sectional and 87 longitudinal studies. On average, 34 per cent of unemployed people had psychological problems, compared with 16 per cent of employed people. Longitudinal studies and natural experiments supported the view that unemployment causes distress rather than simply being more common among people who were already struggling. Selection effects existed but were weak. Two moderators matter for what follows. The damage was greater in countries with weaker economic development, more unequal incomes or weaker unemployment protection. And men and blue-collar workers were more distressed by unemployment than women and white-collar workers.

The harm also lasts. Using a 15-year German panel of more than 24,000 people, Richard Lucas, Andrew Clark, Yannis Georgellis and Ed Diener reported in Psychological Science in 2004 that people reacted strongly to losing a job and then moved back towards their earlier life satisfaction, but that prior unemployment did not make a new spell any easier to bear. An earlier working-paper version of the study went further, reporting that on average people did not completely return to their former life satisfaction even after finding work again. Clark, Georgellis and Peter Sanfey, writing in Economica in 2001, found that life satisfaction was lower not only among the currently unemployed but also among people with more unemployment in their past. Economists call it scarring, and the word is apt.

At the far end, the harm shows up in death records. Daniel Sullivan and Till von Wachter matched Pennsylvania earnings records to Social Security death records and published the results in the Quarterly Journal of Economics in 2009. For high-seniority men who lost their jobs in mass layoffs, mortality in the following year was 50 to 100 per cent higher than it would otherwise have been. Twenty years later, annual death rates were still 10 to 15 per cent higher. The authors calculated that a worker displaced at 40 could expect to lose between one and one and a half years of life. They also showed that the effect was not explained by less healthy workers being the ones selected for layoff.

Eleven years ago today, on 2 November 2015, Anne Case and Angus Deaton published a paper in the Proceedings of the National Academy of Sciences showing rising death rates in midlife among white non-Hispanic Americans, driven by drugs, alcohol and suicide. Their 2020 book with Princeton University Press, Deaths of Despair and the Future of Capitalism, traced the surge among Americans without a university degree to the decline of working-class economic and social life, including the weakening position of labour. Theirs is not a story about a single redundancy. It is about whole communities in which steady, respectable work stopped being available, which is Marienthal's story told again over decades instead of a single winter.

None of this proves that AI will produce the same outcomes. Some displaced workers will find better jobs, some will retire comfortably and some will be glad to go. But it sets a baseline. When work disappears involuntarily, the research consistently finds losses in mental health, life satisfaction and even lifespan that are larger and longer-lasting than the lost wages alone would suggest.

What the Basic Income Trials Actually Measured

The obvious objection is that much of that harm is financial, and that a decent unconditional income would remove it. The evidence from cash trials is genuinely encouraging on this point, and it is worth reading carefully, because it also shows exactly where money stops.

Finland ran the best-known European test. From January 2017 to the end of 2018 its social insurance agency, Kela, paid 2,000 randomly selected unemployed people 560 euros a month, tax-free and without conditions. The employment effect was small: between November 2017 and October 2018, recipients worked six more days on average than the control group, with interpretation complicated by a new activation model introduced in 2018. The wellbeing effects were clearer. According to the final results reported in May 2020, recipients “were more satisfied with their lives and experienced less mental strain, depression, sadness and loneliness.” They reported more trust in other people and in institutions, and more confidence in their own future. Anthony Painter of the RSA told the Big Issue at the time: “Even getting a small amount of money has a big effect on people's agency and sense of control.” Agency and control are not on Jahoda's list, but they are close cousins of it.

Kenya hosts the most ambitious test. Since 2017 the charity GiveDirectly has paid adults in 295 villages across two rural counties about 75 US cents a day via mobile money, with one group of villages promised payments for 12 years. Other groups received two years of payments or an equivalent lump sum, and a control group received nothing. In a September 2023 working paper, Abhijit Banerjee, Michael Faye, Alan Krueger, Paul Niehaus and Tavneet Suri reported that the long-term basic income did not reduce total work effort. Instead, people shifted from wage labour towards self-employment, and villages saw more enterprises and higher revenues. Recipients in the long-term and short-term arms were also less depressed.

The United States supplied the sobering counterpart. The OpenResearch Unconditional Cash Study gave 1,000 low-income people in Texas and Illinois $1,000 a month for three years, from November 2020 to October 2023, while 2,000 controls received $50 a month. In the employment paper, now published in the Quarterly Journal of Economics, Eva Vivalt, Elizabeth Rhodes, Alexander Bartik, David Broockman, Patrick Krause and Sarah Miller report that recipients reduced their work by one to two hours a week, their partners by a similar amount, and that participation in the labour market fell. The largest increase in time use was leisure. The authors conclude that the reduction in work does not appear to have been offset by other productive activities.

Put the three together and a pattern appears. Unconditional money reliably reduces stress, improves reported wellbeing and gives people room to choose. It does not make people stop working en masse. In Kenya it nudged them into running their own businesses, which supply plenty of time structure, contact and status. But none of these trials replaced work. Every participant in Finland was already unemployed and remained free to take a job. Kenyan villagers kept farming and trading. American recipients mostly stayed employed. The trials tested cash as a supplement to a world of work. They tell us little about cash as a substitute for one.

This is where Jacobs's survey adds something. The OpenResearch finding that freed time went mostly to leisure echoes his Transactional workers. His Intrinsic and Community workers say they would do something different: volunteer, care for others, start something. If they are right about themselves, the question is not whether a basic income would make them idle. It is whether the activities they would turn to could give back what the job gave.

Can Anything Else Do What a Job Does

Jahoda never claimed that only employment could supply the latent functions. She claimed that in industrial societies it was the institution that did so for most people, most reliably. That leaves room for alternatives, and there are three obvious candidates.

The first is volunteering. It looks like the perfect replacement: structured, social, useful to others. The research is more cautious. In a 2018 study in Work, Employment and Society, Daiga Kamerāde and Matthew Bennett analysed responses from nearly 2,500 unemployed people across 29 European countries. Volunteering during unemployment partly improved wellbeing, but generous unemployment benefits mattered more for mental health. In countries with less generous benefits, regular volunteering was even associated with worse mental health. The authors warned that compulsory community work for the unemployed was unlikely to deliver the benefits at all. Volunteering appears to supply some latent functions when it is chosen freely by people who are financially secure. It is not a cure that can be prescribed.

That matters for reading Jacobs's results. His intrinsically motivated and community-oriented respondents said they would turn to volunteering and caregiving. The European evidence suggests that plan could work, but only under two conditions: that their income is secure and generous, and that nobody forces them into it. A displaced workforce pushed into mandatory community service in exchange for benefits would be running the experiment Kamerāde and Bennett warned against.

The second is community institutions built deliberately to restore what work used to provide. The Men's Sheds movement, in which mostly older men gather in shared workshops to make and mend things together, is the clearest example. The evidence is promising but thin. A scoping review led by Christine Milligan in Ageing & Society found limited evidence of benefits for mental health and wellbeing and little evidence for physical health, and described studies in the field as few and of variable quality. A later mixed-methods systematic review found benefits for self-rated health, social isolation and wellbeing, while calling for longitudinal studies to establish cause. Sheds probably help. Nobody yet knows whether they could carry the weight of a whole displaced workforce.

The third candidate is less work rather than no work. Here the most striking finding comes from Kamerāde again, this time with Senhu Wang, Brendan Burchell, Sarah Ursula Balderson and Adam Coutts. Their 2019 paper in Social Science & Medicine used UK Household Longitudinal Study data from 2009 to 2018 to ask how much paid work people need for the mental health benefits of employment. The answer was surprisingly little. Moving from unemployment or inactivity into between one and eight hours of paid work a week brought significant gains in mental health and wellbeing. Standard full-time hours of 37 to 40 a week brought no additional benefit. The authors noted that there is no single optimum, and the gains were concentrated among people moving out of unemployment.

If that result holds, it has an uncomfortable implication. The latent functions of work may behave less like a dial and more like a switch. A small amount of employment seems to provide most of the benefit. Having none removes it.

Put the evidence together and the honest answer to the brief's central question is this: the non-monetary benefits of work are not uniquely bound to employment in principle, but they are bound to it in practice. Other institutions can supply pieces of them, especially when people are financially secure and taking part by choice. None has yet been shown to supply all five at once, at scale, for people who did not choose to leave work. Jahoda's latent functions arrive bundled in a job, and nobody has yet worked out how to bundle them any other way.

Meaning Is Spread Evenly, and So Is the Exposure

Jacobs's finding that non-monetary attachment is spread evenly across incomes and collar types matters because of how AI exposure is distributed.

Debates about automation have often treated displacement as a problem for the lower-paid, whose routine manual and clerical tasks were the easiest to mechanise. Generative AI complicates that picture. When the International Monetary Fund published its analysis in January 2024, Managing Director Kristalina Georgieva said that almost 40 per cent of global employment was exposed to AI. In advanced economies, where cognitive work dominates, the figure was about 60 per cent, compared with about 40 per cent in emerging markets and 26 per cent in low-income countries. The IMF was careful to note that exposure includes jobs AI would complement rather than replace. Even so, the centre of gravity has moved up the income scale.

Combine the two findings and the picture changes. If meaning were a luxury good concentrated among the well-paid, AI's reach into professional work would mean that those with the most to lose in meaning also had the most savings, networks and options to cushion them. If meaning is spread evenly, as Jacobs reports, then the loss of meaning is no more concentrated among the affluent than among anyone else. It follows exposure. Blue-collar workers, whom Paul and Moser found more distressed by unemployment, carry as much non-monetary attachment as their white-collar counterparts, and in sectors where AI and robotics converge they may face exposure too.

His choice of countries makes the point sharper. The sample spans a Nordic welfare state, two large liberal market economies and a fast-growing developing economy, and non-monetary attachment shows up across all of them. Paul and Moser found that stronger unemployment protection softens the mental health damage of unemployment. It does not remove it. If people in a high-security welfare state such as Denmark still report strong non-monetary attachment to work, that attachment is not simply fear of poverty dressed up as virtue.

Rerunning the Forecasts in a Different Currency

Consider how the main AI employment forecasts read once meaning, not money, is the thing being counted.

In March 2023 Goldman Sachs estimated that generative AI could expose the equivalent of 300 million full-time jobs worldwide to automation. Roughly two-thirds of US and European jobs were exposed to some degree, and up to a quarter of all work could be done by AI entirely, though most exposed workers would see less than half their workload automated. A 2023 preprint by Tyna Eloundou, Sam Manning, Pamela Mishkin and Daniel Rock estimated that around 80 per cent of the US workforce could have at least 10 per cent of their tasks affected by large language models, and about 19 per cent could see at least half their tasks affected. In 2025, Goldman economists Joseph Briggs and Sarah Dong put their baseline estimate of US workers displaced at 6 to 7 per cent if AI were widely adopted, within a range of 3 to 14 per cent, and argued that displacement tends to be temporary, with the effect fading after about two years. By mid-2026, according to widely reported coverage of a new Goldman note, that estimate had risen to around 9 per cent of US workers, or roughly 15 million people, over about a decade.

Measured in money, these numbers are tolerable. Task exposure suggests a reshaping rather than a removal of most jobs, and a temporary displacement of 6 to 9 per cent of workers is the kind of shock that wage insurance, retraining and, if needed, a basic income are designed to absorb. Measured in meaning, three things change.

First, the unit of loss changes from hours to jobs. If Kamerāde and colleagues are right that a few hours of work deliver most of its psychological benefit, then a large reduction in hours spread across many workers costs relatively little meaning. A smaller number of whole jobs lost costs a great deal. Task exposure figures, which describe the first kind of change, tell us almost nothing about the second. The 9 per cent displacement estimate is the more important number, and each percentage point represents people losing all five latent functions at once.

Second, temporary is not the same as harmless. The research on scarring and displacement suggests that a spell of involuntary joblessness leaves marks after it ends. Sullivan and von Wachter found raised mortality two decades after a layoff. Clark and colleagues found that past unemployment depresses current life satisfaction. A forecast that counts a worker as reabsorbed after two years records the end of the income loss. It does not record the end of the harm.

Third, the most important loss may be the one nobody experiences as a loss. Erik Brynjolfsson, Bharat Chandar and Ruyu Chen at the Stanford Digital Economy Lab have tracked payroll data from ADP covering millions of US workers. In their August 2026 revision, using data through June 2026, they report that employment of 22-to-25-year-olds in the most AI-exposed occupations stands 19 per cent below where it would be had it kept pace with less-exposed peers, a gap that has widened since they first documented it in August 2025. The adjustment has come mainly through reduced hiring rather than layoffs. Those young people were never fired. They simply never received the time structure, colleagues, status and sense of contribution that a first job brings. No unemployment statistic records a latent function that was never acquired, yet that may be where the cost of meaning first appears.

Seen this way, a hiring freeze is not a gentler version of a layoff. It may be the version whose costs are hardest to see, because they fall on people who have not yet built the routines, friendships and identities they would otherwise lose.

The money-based reading and the meaning-based reading agree on the size of the shock. They disagree on which parts of it matter most, and therefore on which remedies count as sufficient.

Building What the Cheque Cannot Buy

None of this is an argument against a basic income or similar cash guarantees. The trials show they reduce distress, raise trust and widen choice, and Jacobs himself sees a basic income as possibly necessary for managing an AI labour shock. The argument is that cash should be treated as the floor of a response, not the whole of one. The evidence suggests several principles for what should sit on top.

The first is to protect the switch rather than the dial. If small amounts of work provide most of the psychological return, policy should favour spreading reduced work across people over concentrating total loss on a few. Shorter working weeks, job-sharing and partial retirement keep more people on the right side of the threshold. Kamerāde and her colleagues framed their study around exactly this idea, under the title “A shorter working week for everyone”.

The second is to keep support unconditional. Finland's recipients reported greater trust, which researchers tentatively linked to the payments coming without sanctions, and Kamerāde and Bennett found that compulsory volunteering was unlikely to help. Latent functions appear to work best when chosen. A system that forces displaced workers into make-work to earn their income risks recreating the form of a job without its meaning.

The third is to fund the places where the latent functions might live outside work. Jacobs's respondents said they would turn to caregiving, volunteering and enterprise. The Kenyan evidence suggests secure income can push people towards self-employment. The Men's Sheds evidence suggests shared, purposeful activity helps. These are institutions that need investment, staff and premises. They will not appear automatically just because people have spare time.

The fourth is to count what is being lost. Labour forecasts could report, alongside exposure and displacement, estimates of how many people would lose employment entirely rather than lose hours, and for how long. Wellbeing data, which governments including the UK's already collect, could be tracked against AI adoption as closely as payroll. Forecasts that measure only money will keep recommending only money.

Keynes's Permanent Problem Comes Due

In 1930, in an essay titled “Economic Possibilities for our Grandchildren”, John Maynard Keynes predicted that within a century the economic problem of scarcity might be solved. Humanity would then face what he called “his real, his permanent problem”: how to live well with freedom from pressing economic cares. He suspected the transition would be hard, and he offered a practical stopgap. “Three-hour shifts or a fifteen-hour week may put off the problem for a great while.”

Keynes's deadline is four years away, and his stopgap looks more perceptive than his optimism. He seems to have sensed what Jahoda would document in Marienthal and what Kamerāde would measure nearly ninety years later: that people need some work long after they stop needing all of it. Work is where most people find their routine, their colleagues, their usefulness and their answer to a stranger's first question.

Jacobs's survey suggests that this need is held by the warehouse worker as firmly as by the professional, in India as in Denmark. A basic income may be the right answer to the question of how people will pay their bills if AI displaces their jobs. It is not an answer to the question of what they will do on Tuesday morning, whom they will see, or what they will say they do. Those questions are older than AI. AI may simply be the force that stops us postponing them.

References and Sources

  1. Jacobs, Julian. “What is Work For? Labor, Politics, and Meaning in the Age of AI.” arXiv:2610.39375 [econ.GN], submitted 29 October 2026. University of Oxford and Google DeepMind; supported by Google DeepMind. https://arxiv.org/abs/2610.39375 (HTML version: https://arxiv.org/html/2610.39375v1). Accessed 2 November 2026.
  2. Oxford Martin AI Governance Initiative. “Julian Jacobs” (profile). University of Oxford. https://aigi.ox.ac.uk/people/julian-jacobs/. Accessed 2 November 2026.
  3. Jahoda, Marie; Lazarsfeld, Paul F.; Zeisel, Hans. Die Arbeitslosen von Marienthal: Ein soziographischer Versuch über die Wirkungen langandauernder Arbeitslosigkeit. Leipzig: S. Hirzel, 1933. English edition: Marienthal: The Sociography of an Unemployed Community. Chicago: Aldine-Atherton, 1971.
  4. University of Graz, Archive for the History of Sociology in Austria (AGSÖ). “The Marienthal Factory and Workers' Settlement” and “The Marienthal Study.” https://agso.uni-graz.at/archive/marienthal/e/pictures/15_marienthal_study.htm. Accessed 2 November 2026.
  5. Jahoda, Marie. “Work, Employment, and Unemployment: Values, Theories, and Approaches in Social Research.” American Psychologist, 36(2), 184-191, 1981. DOI: 10.1037/0003-066X.36.2.184. And: Jahoda, Marie. Employment and Unemployment: A Social-Psychological Analysis. Cambridge: Cambridge University Press, 1982. ISBN 9780521285865.
  6. Paul, Karsten Ingmar; Scholl, Hannah; Moser, Klaus; Zechmann, Andrea; Batinic, Bernad. “Employment status, psychological needs, and mental health: Meta-analytic findings concerning the latent deprivation model.” Frontiers in Psychology, 14, 1017358, published 2 March 2023. DOI: 10.3389/fpsyg.2023.1017358. https://www.frontiersin.org/articles/10.3389/fpsyg.2023.1017358/full
  7. Paul, Karsten I.; Moser, Klaus. “Unemployment impairs mental health: Meta-analyses.” Journal of Vocational Behavior, 74(3), 264-282, 2009. DOI: 10.1016/j.jvb.2009.01.001.
  8. Lucas, Richard E.; Clark, Andrew E.; Georgellis, Yannis; Diener, Ed. “Unemployment Alters the Set Point for Life Satisfaction.” Psychological Science, 15(1), 8-13, 2004. DOI: 10.1111/j.0963-7214.2004.01501002.x. Working paper version: Brunel University, https://bura.brunel.ac.uk/handle/2438/926
  9. Clark, Andrew E.; Georgellis, Yannis; Sanfey, Peter. “Scarring: The Psychological Impact of Past Unemployment.” Economica, 68(270), 221-241, 2001. DOI: 10.1111/1468-0335.00243.
  10. Sullivan, Daniel; von Wachter, Till. “Job Displacement and Mortality: An Analysis Using Administrative Data.” Quarterly Journal of Economics, 124(3), 1265-1306, 2009. DOI: 10.1162/qjec.2009.124.3.1265. Working paper: NBER Working Paper 13626, https://www.nber.org/papers/w13626
  11. Case, Anne; Deaton, Angus. “Rising morbidity and mortality in midlife among white non-Hispanic Americans in the 21st century.” Proceedings of the National Academy of Sciences, 112(49), 15078-15083, published online 2 November 2015. DOI: 10.1073/pnas.1518393112. https://deaton.scholar.princeton.edu/publications/rising-morbidity-and-mortality-midlife-among-white-non-hispanic-americans-21st ; and Deaths of Despair and the Future of Capitalism. Princeton, NJ: Princeton University Press, March 2020. https://discovery.princeton.edu/2020/12/18/deaths-of-despair-and-the-future-of-capitalism
  12. Finland Toolbox (Ministry for Foreign Affairs of Finland). “Finland's Basic Income Experiment 2017-2018.” https://toolbox.finland.fi/life-society/finlands-basic-income-experiment-2017-2018/. Accessed 2 November 2026. See also Kela, “The Basic Income Experiment 2017-2018 in Finland: Final Results,” May 2020, https://toolbox.finland.fi/wp-content/uploads/sites/2/2018/04/basic_income_experiment_2017-2018_results_2020.pdf ; Big Issue, “Universal basic income boosted wellbeing and trust in Finland trial,” 6 May 2020, https://bigissue.com/news/universal-basic-income-boosted-wellbeing-and-trust-in-finland-trial/
  13. Banerjee, Abhijit; Faye, Michael; Krueger, Alan; Niehaus, Paul; Suri, Tavneet. “Universal Basic Income: Short-Term Results from a Long-Term Experiment in Kenya.” Working paper, September 2023. https://conference.nber.org/conf_papers/f192616.pdf ; summary: https://voxdev.org/topic/social-protection/universal-basic-income-kenya
  14. Vivalt, Eva; Rhodes, Elizabeth; Bartik, Alexander W.; Broockman, David E.; Krause, Patrick; Miller, Sarah. “The Employment Effects of a Guaranteed Income: Experimental Evidence from Two U.S. States.” Quarterly Journal of Economics, 141(4), 2675-2734, 2026. Working paper: NBER Working Paper 32719, July 2024 (revised August 2026), https://www.nber.org/papers/w32719
  15. Kamerāde, Daiga; Bennett, Matthew R. “Rewarding Work: Cross-National Differences in Benefits, Volunteering During Unemployment, Well-Being and Mental Health.” Work, Employment and Society, 32(1), 38-56, 2018. https://research.birmingham.ac.uk/en/publications/rewarding-work-cross-national-differences-in-benefits-volunteerin
  16. Kamerāde, Daiga; Wang, Senhu; Burchell, Brendan; Balderson, Sarah Ursula; Coutts, Adam. “A shorter working week for everyone: How much paid work is needed for mental health and well-being?” Social Science & Medicine, 241, 112353, 2019. DOI: 10.1016/j.socscimed.2019.06.006. https://www.iser.essex.ac.uk/research/publications/publication-525783
  17. Milligan, Christine; et al. “Older men and social activity: a scoping review of Men's Sheds and other gendered interventions.” Ageing & Society, Cambridge University Press. DOI: 10.1017/S0144686X14001524. https://www.cambridge.org/core/product/76C6C5DBD76D55AD2717BC4BB421AA4A
  18. Albrecht, Lisa; et al. “Men's sheds as community-based health promotion for men aged 50 plus: protocol for a mixed-methods systematic review.” 2021. https://pmc.ncbi.nlm.nih.gov/articles/PMC8336413/ ; and the subsequent mixed-methods review in American Journal of Men's Health, 2022, https://pmc.ncbi.nlm.nih.gov/articles/PMC8928410
  19. Cazzaniga, Mauro; et al. “Gen-AI: Artificial Intelligence and the Future of Work.” IMF Staff Discussion Note SDN/2024/001, International Monetary Fund, January 2024. https://www.imf.org/en/Publications/Staff-Discussion-Notes/Issues/2024/01/14/Gen-AI-Artificial-Intelligence-and-the-Future-of-Work-542379
  20. Hatzius, Jan; Briggs, Joseph; Kodnani, Devesh; Pierdomenico, Giovanni. “The Potentially Large Effects of Artificial Intelligence on Economic Growth.” Goldman Sachs Global Economics Analyst, 26 March 2023. Reported in Workplace Insight, https://workplaceinsight.net/generative-artificial-intelligence-set-to-boost-gdp-but-affect-300-million-jobs-across-major-economies
  21. Briggs, Joseph; Dong, Sarah. “How Will AI Affect the Global Workforce?” Goldman Sachs Research, 2025. https://www.marcus.com/us/en/resources/heard-at-gs/how-will-ai-affect-the-global-workforce- ; HR Leader, “Displacement, not replacement: Goldman Sachs on how AI will affect workers,” https://www.hrleader.com.au/tech/27194-displacement-not-replacement-goldman-sachs-on-how-ai-will-affect-workers
  22. Coverage of the revised 2026 Goldman Sachs displacement estimate: Allwork.Space, “Goldman Sachs says AI could displace 9% of the U.S. workforce,” July 2026, https://allwork.space/2026/07/goldman-sachs-says-ai-could-displace-9-of-the-u-s-workforce/ ; PYMNTS, “Goldman Sachs Says AI Will Eliminate 15 Million US Jobs,” 2026, https://pymnts.com/economy/2026/goldman-sachs-says-ai-will-eliminate-15-million-us-jobs ; OECD.AI Incidents Monitor entry, 3 July 2026, https://oecd.ai/en/incidents/2026-07-03-1717
  23. Eloundou, Tyna; Manning, Sam; Mishkin, Pamela; Rock, Daniel. “GPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Models.” arXiv:2303.10130, March 2023. https://arxiv.org/abs/2303.10130 (published in revised form in Science, 384(6702), 1306-1308, June 2024).
  24. Brynjolfsson, Erik; Chandar, Bharat; Chen, Ruyu. “Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence.” Stanford Digital Economy Lab working paper, first released August 2025, revised 12 August 2026. https://digitaleconomy.stanford.edu/publications/canaries-in-the-coal-mine
  25. Keynes, John Maynard. “Economic Possibilities for our Grandchildren.” The Nation and Athenaeum, 11 and 18 October 1930. Reprinted in Essays in Persuasion. London: Macmillan, 1931.

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...