Post-labor in five minutes
1. An old dream, a new deadline
Section titled “1. An old dream, a new deadline”In 1930 Keynes predicted that within a century his grandchildren’s generation would work “a fifteen-hour week”, and he named the side effect: “technological unemployment” (Keynes 1930). Fifty years later Wassily Leontief warned that human labour could go the way of the draft horse after the tractor (Leontief 1983).

2. What the data already shows
Section titled “2. What the data already shows”- EvidenceThe global labour share of income has declined significantly since the early 1980s (Karabarbounis & Neiman 2014).
- EvidenceIn the US, productivity has grown far faster than the pay of typical workers since the 1970s (Bivens & Mishel 2015).
- EvidenceUS local labour markets more exposed to industrial robots in 1990–2007 lost employment and wages (Acemoglu & Restrepo 2020).
- EvidenceAbout 80% of US workers have at least 10% of their tasks exposed to large language models, and about 19% have at least half (Eloundou et al. 2024, abstract). Exposure is not job loss: the authors make no prediction about timing or adoption.
- EvidenceIn the field, an AI assistant raised customer-support productivity by 15% on average, mostly for less experienced agents (Brynjolfsson et al. 2025, abstract).
3. The Post-Labor Economics answer
Section titled “3. The Post-Labor Economics answer”David Shapiro’s Post-Labor Economics starts from what he calls the Great Decoupling: economic output no longer runs through wage employment (Shapiro 2026c). If that holds, the question is not how to save jobs but how households get an income without them.
ProposalHis answer is to make ownership, not wages, the main way income is distributed: sovereign wealth funds paying automation dividends, wider employee ownership, universal basic capital grants and public stakes in AI built on public data (Shapiro 2025b). He calls basic income “necessary but not sufficient”: prosperity measures need power measures beside them (Shapiro 2025a).
4. What real experiments say
Section titled “4. What real experiments say”EvidenceCash without conditions has been tested for decades. The headline results:
| Programme | What people got | Headline result |
|---|---|---|
| Mincome, Manitoba, 1974–79 | A guaranteed annual income | Hospitalisations 8.5% lower than controls (Forget 2011, abstract) |
| Alaska Permanent Fund, since 1982 | A yearly dividend for every resident | No effect on employment; part-time work up 1.8 points (Jones & Marinescu 2022, abstract) |
| Finland, 2017–18 | €560 a month to 2,000 unemployed people | Small employment effects, better wellbeing (Kela 2020) |
| Kenya, GiveDirectly | About $1,000 per poor household, once | Each dollar raised local output by about $2.50, with minimal inflation (Egger et al. 2022, abstract) |
| OpenResearch, Texas and Illinois, 2020–23 | $1,000 a month for three years | Earned income down about $1,900 a year; work down 1–2 hours a week; leisure up most (Vivalt et al. 2024, abstract) |
None of these tested a world without jobs. They tell us how people respond to cash while jobs still exist.
5. The strongest doubts
Section titled “5. The strongest doubts”- EvidenceEvery wave of automation so far has also created work; automation complements labour as well as replacing it (Autor 2015).
- EvidenceFear that “this time is different” is as old as industry, and has been wrong before (Mokyr et al. 2015).
- ForecastA task-based estimate puts AI’s ten-year boost to total factor productivity at no more than 0.66% (Acemoglu 2024, abstract). Post-labor advocates reply that such estimates leave out AI agents and humanoid robots (Shapiro 2025a).

Where to go next
Section titled “Where to go next”The full arguments, data and history get their own sections as the wiki grows. Until then, every source on this page is in the bibliography, and How we source explains the labels.