One in Five Workers Would Not Quit Even at 120% of Their Pay

Offer workers in the UK, US, Denmark and India a guaranteed income equal to their full salary, with pension and health benefits kept, and about 75% say they would leave their job. The other 20% turn down every offer up to 120% of their pay. [Fact] A survey of 8,234 working-age adults asks what those holdouts are holding on to, and what a basic income would and would not replace if AI removed the job.

The paper is "What is Work For? Labor, Politics, and Meaning in the Age of AI," posted to arXiv on October 29, 2026, by Julian Jacobs of the University of Oxford and Google DeepMind. [Fact] The survey was fielded in June and July 2026 by the polling organisation More in Common, with roughly 2,000 respondents per country, and was supported by Google DeepMind. [Fact]

Four reasons people work

Respondents rated nine work priorities, and the paper sorts the 6,224 classifiable workers into four profiles. 33% are intrinsic, valuing the job for its purpose, learning and contribution to society. 26% are transactional, ranking pay and benefits first. 23% are community-oriented, and 17% are status-driven. [Fact] The split barely moves with income: intrinsic shares are 32%, 33% and 34% across the low, middle and high income terciles. [Fact] Across ISCO-08 occupation groups, intrinsic attachment appears in 38% of professionals, 34% of managers, 31% of services and sales workers and 33% of craft and trades workers. [Fact] Country moves the numbers more than class does: the United States has the largest transactional share at 38%, Denmark the smallest at 19%. [Fact]

What a guaranteed income buys

The survey ran an auction: would you quit for 60% of your salary as guaranteed income, then 70%, and so on up to 120%. A slight majority leave at 80%, about 65% at 90%, and about 75% at full parity. [Fact] Among those who accept any offer, the median reservation wage is 80% of salary for intrinsic workers against 70% for transactional and community workers and 60% for status-driven workers. [Fact] After controls for earnings, occupation, demographics and country, intrinsic workers require a 1.76 to 1.92 percentage-point higher replacement rate than transactional workers; models that keep the 20% holdouts in the sample raise the premium to 4.58 to 5.77 points. [Fact] Intrinsic workers are also the most likely to refuse every offer, 23% against 14% of transactional workers. [Fact]

Income does not buy the same time for everyone. Told they could no longer work but would receive a guaranteed income, 79% of transactional workers would spend the time on rest, hobbies and travel. 57% of intrinsic workers would volunteer or care for family, 51% of community workers say the same, and 46% of the community group would look for new work or start a business. [Fact]

Who fears AI, and who reacts

White-collar workers in ISCO groups 1 to 4 are more likely to expect AI to replace their job within ten years, 40% against 29% of manual and service workers. [Fact] The paper reads this ranking as consistent with the task-exposure estimates of Eloundou et al. (2024). [Claim] Community- and status-driven workers report the most concern about their own job; transactional workers worry most about society. [Fact] Among workers who expect AI to hurt both their job and their industry, the intrinsic, community and status profiles show a steeper negative turn in political attitudes than transactional workers, which the author labels potential evidence pending further research. [Claim]

What it means for your occupation

The 40% figure covers the roles our site lists under managers, professionals and clerical support: general and operations managers, software developers, registered nurses and office clerks. The 29% group includes retail salespersons and electricians. The paper's conclusion is that a basic income may be necessary in an AI labour shock but would not compensate the roughly three quarters of workers whose attachment to work is non-monetary. [Claim]

What to prepare

Decide which of the four profiles describes your own attachment; in this data, income and job title do not predict it. If the answer is intrinsic or community, what automation puts at risk is the activity and the people, and respondents in those groups plan to replace it with volunteering, caregiving or new work rather than leisure. Workers in the 40% white-collar band can open their occupation page and read the task list to see which tasks carry the exposure, then build the next role around the tasks that remain.

Limits

This is a cross-sectional survey released as a working paper without peer review, and stated willingness to quit is not observed behaviour. The India sample is online-only and skews urban, educated and white-collar (77% graduates, 84% white-collar), and 24% of respondents could not be assigned a profile. The author describes the political-attitude results as needing further research.

Sources

• Jacobs, J., "What is Work For? Labor, Politics, and Meaning in the Age of AI," arXiv:2610.39375, working paper, October 29, 2026: https://arxiv.org/abs/2610.39375

About this analysis

This article was written with AI assistance from the full text of the arXiv paper (HTML version). Occupation links point to our job pages; the mapping from ISCO-08 major groups to those pages is ours.