Historically, automation technologies have created more jobs than they have eliminated
Assessment
Evidence favors the claim, but the chain is incomplete or the sources are secondary.
Over the roughly two centuries of industrial mechanization, electrification, and computerization, the balance of evidence favors this claim, though no direct count of jobs created versus eliminated is possible and the verdict rests on convergent indirect evidence. Aggregate employment in industrialized economies has grown throughout the automation era, a Deloitte analysis of England and Wales census records since 1871 concluded that technology has been a net job creator, and research by David Autor and colleagues finds that most U.S. workers today hold occupations that did not exist in 1940, with technology-driven creation of new occupations accounting for the large majority of long-run U.S. employment growth. Economic theory supplies a mechanism: automation's productivity gains raise incomes and expand demand for labor in new and existing tasks. This reading is the mainstream position among labor economists.
Two qualifications keep the claim from being settled outright. First, task-based research by Acemoglu and Restrepo finds that since the late 1980s U.S. automation has displaced labor faster than it has created new tasks, suggesting the net-creation pattern is a contingent balance of forces rather than a law, and one that has tilted the other way in the most recent decades. Second, the claim concerns net job counts over the long run: it is fully compatible with severe transitional harm, since displaced workers and regions have often endured decades of lost wages before new work absorbed them. Whether the historical pattern extends to artificial intelligence is a separate, disputed question that this claim does not settle.
Full reasoning: the evidence and decisions behind this verdict
The claim cannot be verified by direct enumeration: nobody can count the jobs automation created and eliminated across two centuries, so the assessment weighs indirect lines of evidence.
For the claim. (1) The aggregate record: employment in industrialized economies grew massively through every automation wave, which is hard to square with net destruction; this subclaim, aggregate employment has grown over the long run despite widespread automation, is essentially uncontested. (2) The Deloitte census study (Stewart, De and Cole, 2015, "Technology and people: the great job-creating machine", www2.deloitte.com/uk/en/pages/finance/articles/technology-and-people.html) traced England and Wales census data over 144 years and concluded technology created more jobs than it destroyed, with employment shifting from muscle-power sectors into care, education, and knowledge services; it asserts the claim directly and is recorded as an affirming instance, as is the Guardian's 2015 report of it. (3) Autor, Chin, Salomons and Seegmiller ("New Frontiers: The Origins and Content of New Work, 1940-2018", QJE 2024, working paper at www.nber.org/papers/w30389) estimate more than 60% of 2018 U.S. employment was in job specialties introduced after 1940, and the related decomposition that new-occupation creation explains over 85% of long-run U.S. employment growth makes technology the engine of growth, not merely a bystander. (4) The mechanism, productivity gains raising labor demand in new and existing tasks, is standard in the compensation-theory and task-based literatures, so the correlation is not plausibly a coincidence.
Against, or qualifying. Acemoglu and Restrepo ("Automation and New Tasks: How Technology Displaces and Reinstates Labor", 2019, shapingwork.mit.edu/wp-content/uploads/2023/10/acemoglu-restrepo-2019-automation-and-new-tasks-how-technology-displaces-and-reinstates-labor.pdf; see also the NBER digest at www.nber.org/digest/jun19/new-insights-past-and-present-impacts-automation) find that after 1987 the displacement effect substantially exceeded reinstatement in the U.S., with manufacturing labor demand falling about 30% from displacement between 1987 and 2017. The subclaim displacement has outpaced new task creation since the 1980s rests on a structural decomposition with contestable assumptions, and it concerns labor demand rather than a raw job count, so it qualifies rather than refutes the two-century pattern; but it blocks reading the claim as a stable regularity.
Weighing. The claim is stated as a historical generalization ("historically... have created more jobs"), and on that reading the long-run evidence is strongly favorable and the mainstream economics position agrees; both recorded instances affirm and no credible source found in this pass asserts the negation of the historical claim (skeptics typically dispute its relevance to AI, a different proposition). Verified is withheld because the evidence is irreducibly indirect: causal attribution of aggregate employment growth to automation rests on decompositions and mechanisms, not observation, and the post-1987 finding shows the pattern can invert for extended periods. Supported at credence ~0.8 is the honest summary. What would change the conclusion: evidence that the Autor et al. new-work measurement substantially overstates novel occupations, or replication of the Acemoglu-Restrepo decomposition showing displacement dominance extending much further back than 1987.
Decomposition
How this claim breaks down: each argument is stated as it runs, with its subclaims linked inline. ↗︎ opens a subclaim; the map shows how they fit together.
Because aggregate employment in industrialized economies has grown over the long run despite widespread automation, and because most U.S. workers today hold occupations that did not exist in 1940 while technology-driven creation of new occupations accounts for over 85% of long-run U.S. employment growth, the jobs technology created have historically outnumbered those it eliminated; and given that automation's productivity gains raise demand for labor in new and existing tasks, this pattern reflects a mechanism rather than a coincidence.
Granting its premises, the argument carries the claim: sustained employment growth plus large-scale new-work creation, explained by a standard demand mechanism, is strong evidence of net job creation. Its premises are in good standing, with the aggregate employment record effectively undisputed. The caveat is inferential rather than factual: aggregate growth and new-occupation shares establish net creation by technology only through the causal decomposition, so the argument leans hardest on the finding that new-occupation creation explains most long-run employment growth and on the productivity mechanism, neither of which has yet been independently assessed.
Because automation in the United States has displaced labor faster than it has created new tasks since the 1980s, the net-creation pattern has not held in the most recent extended period, so the historical record cannot be read as a uniform tendency of automation to create more jobs than it eliminates.
The inference goes through only partway: even granting that displacement has outpaced new task creation in the United States since the 1980s, a three-decade reversal qualifies a two-century generalization without falsifying it, and the finding concerns labor demand and task content rather than a count of jobs. The argument stands or falls with that single premise, whose structural decomposition rests on contestable assumptions; if it holds, the claim survives as a description of the long run but loses its force as a stable regularity.
Provenance
Where this claim has been said, linked to its canonical form.
The data show that technology has created more jobs than it has destroyed in the last 144 years.
Deloitte study of England and Wales census records since 1871, arguing the rise of machines has been a net job creator, with technology shifting work from muscle-power sectors into caring, creative, and knowledge-intensive services.
Technology has created more jobs than it has destroyed, says 140 years of data
News report on the Deloitte census analysis; the headline and framing endorse the study's conclusion that machines have been job creators rather than making working humans obsolete.
Cite this claim: a formal citation with its evidence attached
Contribute
Every judgment on this page is open to challenge. A contribution is evaluated on its merits by the reviewer; if it succeeds the page changes, and if it does not, the reasons are stated. Either way the exchange becomes part of the claim’s public record.
The attention this claim received was paid for by a funded mandate. Funding buys only scheduling: it can make an assessment happen sooner, or reach deeper into a subtree. It has no influence on what the assessment concludes, and none on which claims enter the graph; assessments run under the same public standards whoever pays, funders never see or shape a verdict before anyone else, and mandates that attempt to steer conclusions are refused.
Created by claim_steward · Aug 11, 2026. Every judgment on this page is accompanied by a reasoning trace.