PaperlayerView exact revision on arXiv

Source-linked AI summary

Work, Wellbeing, and Choice: Empirical Lessons for AI Futures

Stephanie C. Y. Chan, Adam Bales, Katherine L. Hermann, Iason Gabriel

arXiv:2609.11019v1cs.CY

TL;DR

Evidence on people who do not work or do not need to work can inform questions about wellbeing as AI makes employment less necessary or less available. This paper synthesizes empirical research across populations and identifies agency, alternative sources of work’s benefits, and social and systemic context as factors shaping wellbeing in possible AI futures.

  • Problem

    Evidence on wellbeing across non-working populations has not been synthesized to inform AI futures, despite questions about how work status relates to wellbeing.

    Source: §1 · pp. 1, 3

  • Method

    The paper comparatively reviews psychological, sociological, and economic evidence from diverse populations, including unemployed people, retirees, lottery winners, and financially dependent spouses.

    Source: §1 · p. 3, §1.2 · p. 5

  • Results

    Wellbeing without traditional paid employment is possible but not guaranteed: involuntary job loss is linked to declines in mental and physical wellbeing, while voluntary retirees frequently report high, sustained life satisfaction.

    Source: §9 · p. 12

  • Takeaways & Limitations

    AI transitions should preserve agency, provide alternative sources of work’s latent benefits, and account for social norms and safety nets rather than relying on material redistribution alone.

    Source: §9 · pp. 12–13

    The reviewed studies largely concern modern postindustrial societies, whose norms, work alternatives, and safety nets may differ greatly from those of future societies.

    Source: §8 · p. 12

Abstract

from arXiv · show

Advances in AI-driven automation have raised questions about how humans might find wellbeing in a world where paid employment is less necessary or less available than before. Paid work has been variously characterized as both a contributor and an impediment to human wellbeing. What is already known about the relationship between paid work and wellbeing? What factors influence wellbeing among people who do not work---or who do not need to work? And how might these factors bear upon prospective AI-induced economic transformations? To help provide empirical grounding for these questions, we survey the psychological, sociological, and economic literature that investigates the relationship between wellbeing and work. We draw on evidence from multiple populations, including the unemployed, retirees, lottery winners, and financially dependent spouses. This comparative review draws from studies across OECD countries, China, India, and Gulf states. We identify three key factors that mediate the relationship between work status and wellbeing: (1) agency and choice---whether the exit from work is voluntary or involuntary, as well as long-term agency; (2) the availability of alternative sources of work's latent benefits---such as volunteering, hobbies, or state-provisioned employment; and (3) social and systemic context---including cultural norms around work and the robustness of social safety nets. We draw on these three factors to derive specific implications for different AI automation scenarios, connecting the empirical evidence to concrete policy considerations.

1 Introduction

Debates about AI range from fears that automation will remove fulfilling work to hopes it will free people from tedious jobs, while existing discussions often overlook evidence from people already living without traditional employment. This review synthesizes research across populations and contexts to identify factors shaping wellbeing and consider their implications for AI futures.

1 Introduction

  • Involuntary work loss tends to harm wellbeing, but retirees and some participants in state-provisioned jobs can have high wellbeing; agency, alternatives to work, and social context help explain this divergence.The review also highlights risks of dependence and argues that material transfers alone may not sustain wellbeing.Source: §1 · p. 3, §1.1 · p. 3

1.1 Related work and our contributions

  • Existing AI labor-economics, policy, and philosophical work offers no integrated empirical basis for anticipating how automation may affect wellbeing across non-working populations.Earlier empirical reviews are largely siloed by population, while this review compares populations and organizes findings around AI futures.Source: §1.1 · pp. 3–4

1.2 The populations studied

  • The review draws on deliberately heterogeneous populations, whose contrasting experiences can help isolate factors shaping wellbeing without work, although none directly represents a specific future.Its evidence includes studies across OECD countries, China, India, and Gulf states.Source: §1 · p. 3, §1.2 · p. 5

1.3 Measures of wellbeing

  • Wellbeing measures span subjective, eudaimonic, psychological, cognitive, and physical dimensions that overlap but may respond differently to external factors.The introduction also notes public doubt about wellbeing without work: 67% of respondents in a recent U.S. poll said jobs are needed for purpose and dignity.Source: §1 · p. 1, §1.3 · p. 5

1.4 Scope and approach

  • The paper uses an integrative empirical synthesis rather than a single systematic protocol, drawing on targeted database searches and citation tracing.It prioritizes meta-analyses, longitudinal panels, and stronger causal-identification studies, while acknowledging that selection and interpretation reflect author judgment and do not exhaustively cover any sub-question.Source: §1.4 · p. 5

2 The costs of unemployment

Evidence from longitudinal studies, meta-analyses, and natural experiments links unemployment to declines in mental health and evaluative wellbeing, with additional risks to physical health. The consequences for AI futures depend on whether displacement resembles present-day unemployment and whether it becomes prolonged.

2.1 The negative effect of unemployment on wellbeing

  • Unemployment causally decreases mental health and evaluative wellbeing, with effects that worsen over time and extend beyond income loss.Mass layoffs and plant closures provide strong causal evidence in those contexts.Source: §2.1 · p. 5
  • Average daily affective wellbeing may remain similar for employed and unemployed people even as the unemployed report lower life satisfaction, potentially because they substitute leisure for work.The passage distinguishes evaluative wellbeing from day-to-day emotional experience.Source: §2.1 · p. 5
  • A later German study using the same methods found long-lasting negative effects that persisted for at least five years after re-employment, consistent with psychological scarring after individuals regain work.Separately, job loss has been found to trigger a measurable increase in physiological stress markers such as cortisol or interleukin-6.Source: §2.1 · pp. 5–6
  • Job loss was associated with 79% higher mortality in the year of displacement and 11% higher mortality 20 years later in a Danish plant-closure study.Negative effects included mortality, suicide and attempts, and deaths or hospitalizations linked to accidents, alcohol-related disease, and mental illness; the effects persisted despite Denmark’s robust welfare system.Source: §2.1 · p. 6

2.2 Implications for AI futures

  • If automation produces job loss under conditions resembling current unemployment, evaluative wellbeing and mental and physical health may be at risk even if leisure preserves daily affective wellbeing.Because these tolls may begin immediately and prolonged unemployment is associated with worsening wellbeing, the review argues that responses may need to be swift and structurally enduring.Source: §2.2 · p. 6

3 The benefits of work beyond income

Work provides benefits beyond income, including status, time structure, purpose, and cognitive stimulation, while unemployment can constrain both material resources and access to these benefits. The review considers how non-work activities and policy might sustain them.

3.1 Latent and manifest benefits

  • Financial strain is the largest predictor of unemployed people’s psychological wellbeing, while status, time structure, and collective purpose also predict it.The reviewed evidence suggests both manifest and latent benefits are needed to explain work’s contributions to wellbeing; social contact and regular activity were not significant predictors in this study.Source: §3.1 · p. 6

3.2 Purpose and meaning

  • Retirement is associated with a small negative effect on eudaimonic wellbeing, but purpose can increase for some lower-SES people and those leaving physically demanding, low-autonomy jobs.Bridge employment or volunteering can help mitigate losses of purpose and meaning.Source: §3.2 · p. 7

3.3 Cognitive health

  • Early retirement is causally linked to significant declines in cognitive ability, with evidence from cross-country policy variation and subsequent U.S. and U.K. studies.Source: §3.3 · p. 7

3.4 Implications for AI futures

  • Employment can provide status, time structure, purpose, and cognitive-health benefits, while lost social participation appears driven more by financial deprivation and loss of standing than by losing the workplace as a social venue.Source: §3.4 · p. 7
  • Material transfers such as UBI or social dividends may replace work’s income but not its broader benefits, so post-labor policy may need to sustain or replace those benefits.Options include guaranteed or reduced-hour jobs and activities such as volunteering that can mimic some benefits of paid work.Source: §3.4 · p. 7

4 The preference for work

Lottery-winner evidence suggests that wealth often changes how people work rather than ending employment altogether. Work centrality, rewarding roles, age, and purchases of leisure help qualify that pattern.

4.1 Lottery winners and employment

  • 85%–89% of lottery winners in U.S. and Swedish studies continued working after their windfalls.This pattern held across different work cultures and prize sizes, including an average of $5.2 million among U.S. winners who continued working in one study.Source: §4.1 · p. 7
  • Winners were more likely to stay employed when work was central to their identity or their roles were psychologically rewarding; older winners were more likely to retire.Source: §4.1 · p. 7
  • Rather than leave work entirely, winners often bought flexibility through unpaid leave or reduced hours, and spent about 11% of unearned income on leisure time over their lifetimes.Source: §4.1 · p. 7

5 Decoupling wellbeing from traditional market-driven employment

Wellbeing can remain high outside traditional full-time employment, but outcomes vary with agency, alternative activities, and institutional conditions.

5.1 Sustained wellbeing in retirement

  • Over 90% of retirees report life satisfaction, and transition patterns are mostly stable: ∼70% maintain wellbeing, ∼25% recover after a decline, and ∼5% improve.The reviewed U.S. panels find high satisfaction despite health decline, loss of work, and reduced income.Source: §5.1 · pp. 7–8
  • Retirement can support high subjective wellbeing, but 20–30% forced into retirement do not share the positive trend, and retirement is causally linked to cognitive decline.Retirement’s benefits are mediated by agency and should be weighed against cognitive costs.Source: §5.1 · p. 8

5.2 Alternate activities as sources of wellbeing benefits

  • Volunteering can supply work-like structure and identity while improving wellbeing; benefits appear at 1 hour weekly, with marginal gains beyond 2 hours.Benefits are stronger with varied activities and when volunteers feel appreciated.Source: §5.2 · p. 8
  • Caregiving outcomes depend on intensity: occasional grandchild care is associated with better mental health, while intensive care can raise stress and reduce wellbeing.The strain is more pronounced among grandfathers than grandmothers across American and Chinese populations.Source: §5.2 · p. 8

5.3 State-guaranteed employment

  • Gulf public-sector workers report greater happiness and life satisfaction than private-sector workers, amid job security, wage premiums, and lighter workloads.Public-sector openings are desirable enough that some young citizens wait for them rather than take private-sector jobs.Source: §5.3 · p. 8
  • The Civilian Conservation Corps increased lifetime longevity by about one year and earnings by 4.6%; a 1990s Job Corps experiment found similar results.The study found no short-run changes in wages or labor-force participation.Source: §5.3 · pp. 8–9

5.4 Implications for AI futures

  • AI-transition policy should cultivate and subsidize activities that replace work’s non-financial benefits, not focus solely on income replacement.Some studies find that 1–2 weekly volunteering hours can provide benefits, at least for older adults.Source: §5.4 · p. 9
  • State-guaranteed employment may suit people who want or need job structure, but Gulf evidence may reflect unusually generous wages and lighter workloads, and dependence can erode agency.A portfolio of options can accommodate differing preferences and preserve agency.Source: §5.4 · p. 9

6 The importance of agency and choice

Wellbeing depends not only on work status but also on whether people choose that status and retain autonomy amid material or financial dependence.

6.1 Voluntary vs involuntary retirement

  • Across countries, 20-30% of workers perceive retirement as forced, while voluntary retirees consistently report significantly higher physical and mental wellbeing than involuntary retirees, even controlling for health, income, job characteristics, and life events.Source: §6.1 · p. 9
  • Retirement driven by personal “pull” motives avoids the steep mental-health declines associated with external “push” factors such as layoffs or poor health.Pull motives include pursuing personal interests or spending time with family.Source: §6.1 · p. 9

6.2 Stay-at-home mothers

  • Mothers’ wellbeing is most strongly tied to whether employment status matches their wish to work; stay-at-home mothers who wanted jobs reported the poorest adjustment.Their adjustment was worse even than for mothers who worked from financial necessity but preferred to stay home.Source: §6.2 · p. 9

6.3 Financial dependence

  • Financial independence enables women to leave unhappy marriages, while fully dependent married men report greater psychological discomfort than men contributing roughly equally to household income.For women over 50, divorce is associated with an average 45% decline in living standard, compared with 21% for men.Source: §6.3 · p. 10
  • Gulf-state dependence on public-sector employment has made economic diversification difficult, and subsidy reductions have been met with public pushback and increased state repression.Changing the state’s economic role can challenge both individuals and societies.Source: §6.3 · p. 10

6.4 Implications for AI futures

  • AI transitions should preserve choice and autonomy because involuntary exits can diminish wellbeing even when material circumstances are held constant.Long-term systemic dependence may also leave people less able to navigate change or exit unfavorable arrangements.Source: §6.4 · p. 10

7 Social and system-level factors

Work-related wellbeing is shaped by social expectations, peers’ employment situations, and the safety nets that cushion job loss.

7.1 Social norms around work

  • A meta-analysis of 29 studies found a country’s “norm to work” significantly correlated with unemployment’s negative wellbeing effects; a cross-national European study found significantly higher unemployment wellbeing costs in more individualistic countries.Source: §7.1 · p. 10
  • Moving from unemployment into retirement can raise life satisfaction without income or routine changes, and crossing Early Retirement Age is associated with a 23% decrease in depression.The authors suggest that the stigma of being able-bodied and jobless may lift when unemployment becomes socially acceptable.Source: §7.1 · p. 10

7.2 Others’ unemployment

  • Unemployment is less harmful to psychological wellbeing when relevant others are also unemployed, while joint retirement is generally associated with positive outcomes.Relevant others include partners, household adults, and same-sex peers in the region.Source: §7.2 · p. 11
  • Higher unemployment among employed people’s peers can lower wellbeing through insecurity or guilt, although comparison with worse-off peers may increase job satisfaction.The passage also notes that high local unemployment can make re-employment harder and mortality effects more pronounced.Source: §7.2 · p. 11

7.3 Social safety nets

  • Plant closures caused an almost four-fold greater rise in depression among American than European workers; pre-existing wealth buffered U.S. workers, while European safety nets offered protection.The findings suggest individual financial means matter more where comprehensive safety nets are absent.Source: §7.3 · p. 11
  • More generous U.S. unemployment benefits can reduce unemployment’s negative effects on suicide risk and self-reported health.Source: §7.3 · p. 11

7.4 Implications for AI futures

  • AI policies should reflect local work norms and individualism; UBI may not restore lost social standing where the “norm to work” is strong.Broadly shared displacement could shift work expectations, whereas uneven displacement may preserve norms that burden displaced people and affect those who keep jobs.Source: §7.4 · pp. 11–12
  • Safety nets can support wellbeing during economic change, but policy must also address social standing and identity alongside material needs.Source: §7.4 · p. 12

8 Limitations

The review’s evidence is bounded by the societies studied and by populations it could not comprehensively cover.

8 Limitations

  • Most reviewed studies concern modern postindustrial societies, whose norms, work alternatives, and safety nets may differ greatly from future societies.These arrangements reflect industrial-era work traditions and specific institutional contexts.Source: §8 · p. 12
  • The review could not comprehensively survey all relevant populations, including the Financial Independence, Retire Early movement.Source: §8 · p. 12
  • The review excludes universal basic income studies because participants generally do not fit its “not working or not needing to work” inclusion criterion, and because such studies already have high visibility in AI discussions.The authors say non-Western societies would be valuable to study in future work.Source: §8 · p. 12

9 Conclusions

Wellbeing without paid employment is possible but not guaranteed: outcomes diverge with agency, access to work’s latent benefits, and social and systemic context. The review draws policy implications for automation transitions, including agency, alternative activities, and institutional support.

9 Conclusions

  • Involuntary job loss is linked to declines in mental and physical wellbeing, while voluntary retirees often report high, sustained life satisfaction.Source: §9 · p. 12
  • The review explains divergent outcomes through agency and choice, alternatives to work’s latent benefits, and social and systemic context.Source: §9 · p. 12
  • Involuntary retirees and stay-at-home mothers report lower wellbeing than voluntary counterparts, while financial dependence can make it difficult to exit unhappy marriages.Historical reliance on state-provisioned jobs also created vulnerabilities for Gulf State citizens when economic conditions shifted.Source: §9 · p. 12
  • Work’s latent benefits include status, time structure, collective purpose, and cognitive health, while volunteering, hobbies, care roles, and state-provisioned jobs can support wellbeing.Source: §9 · p. 12
  • Work norms and safety nets shape outcomes: wellbeing rises at retirement age among unemployed people even without income or routine changes, while high regional unemployment and robust benefits buffer distress.Source: §9 · pp. 12–13
  • Automation policy should preserve agency and offer varied alternatives, including structured civic roles, volunteering, or state-provisioned jobs, to accommodate differing preferences and needs.Source: §9 · p. 13
  • Near-term work norms may persist despite longer-term cultural shifts, while material-only UBI proposals miss latent benefits, risk eroding agency, and neglect social pressure.Source: §9 · p. 13

Appendix

Appendix Table 3 is titled “Types of studies and analytical techniques.”

Appendix

  • Appendix Table 3 is titled “Types of studies and analytical techniques.”Source: Table 3 · p. 21
Loading 2609.11019v1…