Projected US Job Displacement from AI by 2030

"U.S.-based employers announced 43,281 job cuts in September, down 18% from the 52,881 announced in August." "Artificial Intelligence was the fifth-most cited reason with 3,961 cuts in September, about 9% of the month's total." "So far this year, AI has been cited in 120,136 job cut announcements, approximately 21% of all cuts" - and it remains the leading reason year-to-date. "Through September, employers have announced 573,195 job cuts," down 39% from the 946,426 announced in the first nine months of 2025. "Technology companies announced 10,799 cuts in September, up 77% from 6,103 in August. The sector has announced 165,925 cuts this year."

Monthly edition of Revelio's five-lens AI labor tracker (profiles, postings, salaries, reviews, WARN notices), released October 1, 2026 with most workforce data through August 2026. "Employment for younger workers in the most AI-exposed occupations is down by 20% relative to the least exposed occupations, since pre-ChatGPT - compared with just 6% for older workers." "7.2% of U.S. job positions were held by workers with at least one reported AI skill in August 2026" (7.6% in July). Gap in job postings between the most and least AI-exposed occupations "-29%, narrowing from -40% in July." "Decline in the number of new AI-adopting firms per month, compared with July" of -17%, with cumulative adoption near 7% of eligible hiring firms. "90% of year-over-year activity change occurs within occupations - up from 89% in July." At adopting firms, senior employment is +32% versus +6% for junior roles since November 2022.

"Total employment goes up by a marginally significant 3.3 percent." Across the country-by-country meta-analysis, "senior employment rises by a marginally significant 3.4 percent on average, and no country shows a significant decline in total employment," with average junior-share decline of 1.1 percentage points. Cross-country dispersion is small enough that a common effect cannot be rejected. A global estimate at the firm level, not a US economy-wide one.

"The Budget Lab's labor market analysis, updated to incorporate August 2026 CPS microdata, does not provide clear evidence of labor market disruption associated with AI. Churn across occupations, AI exposure among the unemployed, and usage data all remain flat, lie within historical ranges, or continue along pre-AI trends." Key takeaways unchanged: "The occupational mix is not yet changing in ways that clearly align with the introduction of AI into the workforce"; "Measures of AI usage show no connection to changes in employment or unemployment"; "A synthetic differences-in-differences analysis of AI exposure does not yet clearly indicate an AI-related labor market footprint." This is the first update to extend the tracker past the Q1 2026 window covered by the May 2026 analysis.

The Economist estimates that AI has so far created around 1m new jobs in America. That easily exceeds the roughly 200,000 lay-offs attributed to AI since mid-2023

"Total nonfarm payroll employment rose by 162,000 in August, higher than the average monthly gain of 31,000 over the prior 12 months." The unemployment rate was unchanged at 4.1 percent. "Information employment declined by 23,000 in August, following losses that had averaged 8,000 per month over the prior 12 months." "Employment in health care continued to trend up in August (+13,000) but at a slower pace than the average monthly gain over the prior 12 months (+32,000)." "Employment showed little change over the month in other major industries, including ... professional and business services." "The change in total nonfarm payroll employment for June was revised up by 11,000, from +20,000 to +31,000, and the change for July was revised up by 44,000, from -23,000 to +21,000." Average hourly earnings rose 10 cents, or 0.3 percent, to $37.75, up 3.1 percent over the year.

The most AI-exposed firms are seeing fewer layoffs than the least-exposed firms, rather than more.

"U.S.-based employers announced 52,881 job cuts in August 2026, up 58% from July but the lowest August total since 2022." Artificial Intelligence "fell to the fourth-most cited reason with 3,462 cuts in August," ending a five-month run as the leading stated reason for cuts. Year to date, AI has been cited in 116,175 announced cuts, about 22% of the 529,914 cuts announced through August, against 892,362 over the same period of 2025 (down 41%). Technology announced 6,103 cuts in August, its lowest monthly total of 2026, for 155,126 year to date, or 29% of all 2026 cuts. Announced hiring plans were 12,325 in August against 1,494 in August 2025, and 119,825 year to date against 87,626 (up 37%).

Third annual AI module in the NY Fed's regional business surveys: "61 percent of service firms reported using AI this year, up from 40 percent last year and 25 percent in 2024," while among manufacturers "51 percent reported using AI as part of their business processes, roughly double the 26 percent from last year and triple the 16 percent in 2024." The bar is stricter than most adoption surveys — firms using AI "exclusively as an information search tool but nothing else" are coded as non-users. Adoption is wide but shallow: "Three-quarters of service firms and more than 90 percent of manufacturers characterize their AI investments as minimal to modest," and among adopters "the median share of workers using it was just 17 percent for service firms and 7 percent for manufacturers." On labor, "only 4 percent of service firms reported laying off workers" because of AI (up from 1 percent in 2025), "about 15 percent of service firms said they had hired fewer workers than they would have if not for AI use," and "about 13 percent of service firms said they had hired more workers due to AI" — while "just over a third of service firms and more than 20 percent of manufacturing firms report retraining workers." Recorded as overlays: the survey covers New York State and northern New Jersey only, so it is not comparable to the national Census BTOS series, and the workforce shares are of AI-adopting firms rather than of jobs or workers. The authors caution that "AI technology and its applications are still evolving rapidly, and these patterns could shift as adoption matures."

Modest change scenario (m2030 = 0.20, d2030 = 0.20, automation share 0.50, reinstatement 0.5). "GDP is 1.6 percent above its no-AI path in 2030... Cognitive employment is 0.5 percent below its mid-2026 level. The unemployment rate is 3.9 percent against a normal level of 3.8: the modest change scenario has only small labor market effects." The authors attach no probabilities to the scenarios: "The scenarios are not predictions, and we attach no probabilities to them; their purpose is to make the consequences of different assumptions comparable."

Substantial change scenario (m2030 = 0.30, d2030 = 0.40, automation share 0.75, reinstatement 0.25). "Cognitive employment is 3.9 percent below its mid-2026 level, and employment in all other occupations has risen by 4.6 percent. Because reallocation takes time, the unemployment rate of cognitive workers rises from 2.9 percent in mid-2026 to 4.5 percent in 2030." GDP is 8.3 percent above the no-AI path, the average wage 2.1 percent above it, and the labor share falls from 60 to 56 percent. The authors attach no probabilities to the scenarios.

Extreme change scenario (m2030 = 0.50, d2030 = 0.60, automation share 0.90, reinstatement 0). "By 2030, cognitive employment is 21.5 percent below its mid-2026 level, and the unemployment rate for workers who began in cognitive occupations is a stunning 17.9 percent, far above any postwar rate in the United States. The economy-wide unemployment rate rises to 11.9 percent." GDP is 32.4 percent above the no-AI path, the cognitive wage 11.5 percent below it, and the labor share falls from 60 to 45 percent. The authors attach no probabilities to the scenarios.

"Using CPS microdata, we provide the first estimates of the effects of AI on the unemployment of recent college graduates in June, July and August 2026. We provide evidence suggesting that unemployment rates are especially high for summer months and that 2026 might be the first year of widespread enough AI use in the workplace to detect impacts of AI on recent college graduates, the group argued to be most vulnerable to AI replacement." The unemployment rate for recent college graduates (bachelor's degree holders aged 22 to 25) was 7.3 percent in summer 2026, inside the 6.3 to 7.8 percent range observed since 2022. Rates did not rise significantly relative to older college graduates or to young workers without a degree; difference-in-differences and event-study specifications give similar results. The authors do find some positive correlation between 2026 unemployment and remote-work availability by occupation. Not plotted on the early-career chart: this is an unemployment rate level for a demographic group, not a relative percent employment decline measured against least-exposed occupations.

Given AI usage rates and automation scores across occupations and Texas’ industry composition, the estimates imply that automation exposure to generative AI reduced total Lightcast job postings in Texas by approximately 1.8 percent in 2024 and by 2.6 percent in 2025.

The 88 million US workers aged 22 and over without a bachelor's degree — more than half the labor force — sit near the strongest unemployment readings of the past twenty years, while degree-holders have weakened. Levanon ranks each group's 12-month moving-average unemployment rate as a percentile against that group's own history back to 2003, so the numbers are relative rankings rather than rates: young graduates aged 22-34 sit at the 70th percentile of their own history, young non-graduates at the 25th, with the two lines decoupling in 2024 after moving together for two decades. Blue-collar and manual-service workers (65 million) sit at the 18th percentile. STEM is the only occupational group above its own median. Non-metro areas sit at the 12th percentile while the top 50 metros climbed roughly 30 points, and 47 of 51 jurisdictions sit below their own median. Levanon names two non-AI confounds for the non-graduate strength: immigration enforcement shrinking labor supply in the occupations these workers hold, and a construction and data-center building boom raising demand for them. Recorded as overlays: every headline figure is a percentile rank against a group's own history — an index score, not a rate — and the raw rates given are not AI-attributed.

Census has twice asked firms "In the last six months, how did the use of Artificial Intelligence affect this business's total employment?" Tabarrok sets the two waves side by side. In December 2023 to February 2024, when about 5% of firms used AI, "2.8% increased, 2.6% decreased and 94.6% reported no change." In November 2025 to February 2026, with adoption roughly tripled, "2.3% increased, 2.0% decreased, and 95.7% reported no change." Adoption rose sharply and the share reporting any AI-attributed employment effect fell slightly; those reporting one stay evenly split between hiring more and hiring fewer. He notes the answers were similar by firm size and that "employment-weighting gives essentially the same result," and that Information is the only sector where fewer than 92% report no change. He reports the countervailing trend himself: among adopters where AI has displaced employee tasks, the share saying it took over "a large number" rose from 2.4% to 7.1% and "a moderate number" from 13% to 22% — though that group is about a tenth of adopters, who are about a fifth of firms. Recorded as overlays: the estimates count firms, not jobs, so 2.0% is not a share of employment lost, and the graph's unit is % of US jobs. The underlying 2026 figures are already plotted from Census directly; what is new here is the 2024 baseline and the comparison.

Guest essay for EIG's American Worker Project by Chad Syverson, co-author of the original productivity J-curve paper. On the macro series: labor productivity "grew throughout the 2010s by about 1.5 percent per year, roughly half of its growth rate during the productivity boom of 1995-2004," and "productivity from mid-2022 on has maintained a faster-than-2010s trajectory involving annual growth of about 2.2 percent." He declines to attribute this to AI: "The timing leans against AI being the sole initial cause," since the acceleration began when AI investment was small and coincided with pandemic-era churn and business formation; his stated test is duration. A sustained 1.5-to-2.2 move would leave GDP per capita "7 percent higher than otherwise" after a decade, against nearly 3.0 percent annual growth in the 1995-2004 boom and "in excess of 2.5 percent for a decade or longer" in past GPT booms. Original cross-sector analysis (Figure 2) plots each sector's change in contribution to economy-wide productivity growth (2019-25 minus 2007-19) against its employment-weighted Census BTOS 2025 AI adoption rate: the correlation is positive but "weak enough, relative to the number of sectors, that one cannot rule out chance as its source"; excluding retail it is "more than twice as large," a specification he calls "treading on thin statistical ice." On employment: "At least over the first couple years of its use, AI has had remarkably little employment effect, positive or negative." J-curve restatement: understatement periods "could extend well beyond a decade, and cumulative productivity mismeasurement of double-digits percent," and "Where are we on the AI J-curve now? It is too early to know that with much precision."

"AI will take on work in law, customer service, medicine, software, and manufacturing. It will hit these industries rapidly, over the course of a decade rather than a few generations. There will be some new jobs, but without the right policies there will be far fewer than exist today." Gates argues the agriculture-to-office analogy misleads because that shift "proceeded over several generations and created new jobs where human cognition was required. In this case, the technology can substitute for human cognition." His policy proposals are a Human Reserved domain of jobs set aside for people, a tax on AI tokens and robots to rebalance the payroll-vs-capital tax asymmetry, and new national and international institutions for the transition. Commentary essay; no original data and no quantified displacement estimate.

"Existing federal data sources cannot be used yet to conclusively show whether and how AI might be driving changes in the U.S. labor market." The binding constraint is linkage, not collection: "The Business Trends and Outlook Survey and Annual Business Survey adoption data are not linked to the Longitudinal Employer-Household Dynamics data, wage records, or other worker-level outcomes, even though both datasets sit within the Census Bureau." Specific gaps: CPS response rates have declined and it cannot resolve individual occupations; LEHD carries a 4-6 month lag and has industry but no occupation codes; JOLTS samples roughly 16,000 establishments with no occupational detail; O*NET averages only 71 observations per occupation with some collection dating to 2006; UI claims exclude independent contractors, and BLS estimates that in some years most unemployed people did not apply for benefits. A measurement critique rather than a displacement estimate, so it is recorded as an overlay.

"Our main result is that automation of junior tasks can be immiserizing: when its cost savings are small, automation lowers long-run output. Immiseration occurs because junior automation shifts earnings to later in life. This shift lowers the present value of experience-intensive careers, so that fewer workers pursue them, and the supply of experienced workers falls." Overlapping-generations model with career ladders, calibrated to the 1980 US economy and tested against US Census/ACS industry data 1970-2020. "As the transition to the new steady state is Pareto efficient, any policy that averts immiseration hurts at least some workers." A mechanism and long-run output result, not a share-of-jobs estimate.

Goldman's first cross-country read on AI and employment, covering major developed and emerging markets. "Industries with greater exposure to AI automation have generally seen slower job openings growth since the second half of 2022, with the relationship particularly pronounced in Germany, Australia and the U.S." Analyzing employment growth across more than 800 occupations: "Across the broader labor market, a 10% occupational exposure to AI was associated with only a 0.1 percentage point drag on annual headcount growth in France, Canada and the U.S." Goldman concludes that "AI-related hiring pressures are clearly visible in employment data globally, but remain limited to a relatively narrow set of industries and workers." A per-unit-exposure elasticity rather than a share of US jobs, so it is recorded as an overlay rather than plotted.

But their share of total unemployment is small, and their contribution to the aggregate decline is correspondingly limited, at most about one-third of the total.

We find no evidence of widespread, economy-wide job displacement. Average employment in the ADP sample rose by about 6% between November 2022 and June 2026, while employment for the most exposed quintile grew by about 4%. This aggregate stability is itself informative: claims of economy-wide AI-driven job losses are not visible in payroll data through June 2026.

Sorting the unemployed by Webb (2019) occupational AI-exposure quartiles: "Historically, outflow rates by AI-exposure quartile co-moved tightly with the business cycle, with only small level differences across quartiles... Since 2023, outflow rates have diverged: Workers in highly AI-exposed occupations have seen the largest declines in the job-finding rate." The authors read this as "consistent with a literature on automation that emphasizes displacement effects in exposed occupations alongside reallocation toward less-exposed work." Their summary: "Most of the recent decline in unemployment outflow rates is being borne by strongly attached workers, particularly those whose occupations are exposed to AI. This is not the typical recessionary pattern." The brief measures job-finding rates rather than a share of jobs lost, and explicitly defers the decomposition of aggregate versus technological components to a forthcoming third article, so it is recorded as an overlay.

"Overall employment is projected to fall 1.3% by 2030 and fall 3.7% by 2035 relative to 2025 due to AI-driven displacement. This sharply contrasts with government baselines projecting +3.1% growth over the decade when accounting for aging-adjusted population trends." Community forecast timeline readings: 2027 +0.365%, 2030 -1.28%, 2035 -3.73%. This revises the April 2026 vintage of the same forecast (-1.9% by 2030, -3.4% by 2035): the 2030 decline is now shallower and the 2035 decline deeper.

BLS Employment Situation, July 2026, seasonally adjusted, retrieved from the BLS Public API v2 on August 20, 2026. Total nonfarm payroll employment was 158,858,000 in July, down 23,000 from June (158,881,000). June's gain was revised to +20,000 from the +57,000 initially reported. The unemployment rate was 4.1% in July, down from 4.2% in June and 4.3% in May. Month-over-month by sector: professional and business services +18,000 (22,518,000); information +11,000 (2,780,000); health care and social assistance +22,600 (23,912,200); leisure and hospitality -40,000 (16,931,000).

"U.S.-based employers announced 33,429 job cuts in July", the lowest monthly total in two years. "Artificial Intelligence (AI) led all reasons for job cuts, with 10,970 announced during the month, or 33%" — the fifth straight month AI has led all stated reasons. "So far this year, AI has been cited in 112,713 job cut announcements, approximately 24% of all cuts." "Technology again led all sectors, announcing 9,867 job cuts in July for a total of 149,023 in 2026." "Through July, employers have announced 477,033 job cuts, down 41% from the 806,383 cuts announced in the first seven months of 2025." Announced hiring plans reached 107,500 YTD, up 25%.

while AI adoption is rising quickly across industries, firms report very few AI-driven layoffs

As shown in the regression results in Table 3, no significant employment effect was found from an occupation having a high AI exposure score. Coefficient on AI Exposure x Post for log(Employment) was -0.0639 (SE 0.0870), not statistically significant, versus -0.0667 (SE 0.0181, p<0.01) for log(Real Wage). Suggests firms are capturing AI productivity gains through wage compression rather than workforce reduction.

Demand for the most AI-exposed roles has declined 42% relative to the least exposed roles since October 2022 (down 8.6 percentage points year over year). AI-adopting firms grow headcount 27% more than non-adopters since November 2022; they were also growing faster before adoption. Employment grows at adopting firms across seniority levels, but the gains are concentrated in senior roles: 31% compared with 6% for junior roles.

Stanford SIEPR policy brief reviewing observed AI labor market effects. Using IPUMS-CPS data, the unemployment rate for the top quintile of AI-exposed workers has risen by 0.77 percentage points since 2022, while the unemployment rate for the least-exposed workers rose slightly more, by 0.85 percentage points over the same period — a differential implying no measurable net AI displacement in aggregate US employment. Employment growth in coding-heavy occupations has slowed somewhat but remains positive, and online job postings for software developers have grown faster than for other occupations over the last year. Among firms that adopted enterprise AI, employment grew by 10 percent in the two years following adoption. In Census BTOS data only 5 percent of firms report any employment impact, with equal numbers reporting gains and losses. Key takeaway: “AI’s effects on overall employment is likely small, though a tough job market for new graduates may be partly due to AI.”

Google's AI & Economy ATLAS v1.0 (Iscenko, Strand, Chen, Imas, Manyika et al., Google/Google DeepMind, Jul 23 2026; reviewed by Diane Coyle and David Autor). 15M de-identified interactions across Gemini App, AI Mode, and Gemini API (Apr 6-19, 2026), mapped to 800+ occupations, 4,000 O*NET tasks, 300 ATUS activities, 150 countries, 140 languages. Work: 'AI adoption spans occupations covering just above 88% of US employment' (68% of detailed occupations) but 'penetration remains shallow' — 'AI is used for only 21% of total tasks in the median occupation with any AI use'; only 3% of occupations show usage for >75% of tasks. 'Attempts to automate tasks end-to-end represent less than 10% of AI conversations in non-routine cognitive work'; >25% for routine cognitive work. Non-routine cognitive tasks = 35% of O*NET universe but 65% of work interactions. Wages: 'a 1% increase in an occupation's median earnings is associated with a more than 2.5% increase in AI usage intensity'; Gemini-weighted median salary $82,919 vs $62,252 employment-weighted national median. Home: 86% of conversational use is non-work; government/civic queries over-represented ~20x vs time spent; household time-savings valued at $14.9B-$149B/yr (0.5-5% scenarios). Global: 1% GDP/capita ↑ → 0.9% usage ↑; English only ~1/3 of conversations across 143 languages.

Since the introduction of ChatGPT in November 2022, all exposure groups see employment growth, but the rate of expansion is slowest for the two most-exposed occupation groups. However, these differences remain modest. Published dashboard data (vintage 2026-07-16): most-exposed quintile employment -0.2% YoY as of June 2026, vs +0.6% YoY for least-exposed; annualized since Nov 2022, +1.1%/yr most-exposed vs +2.0%/yr least-exposed.

Federal Reserve Board FEDS Note cataloguing publicly available indicators for tracking the generative AI buildout across three stages: capabilities and costs; firm investment and adoption; and productivity and labor. Notes that headline adoption figures “do not reflect usage intensity, which surveys suggest remains shallow even where reported adoption is broad,” and that “the absence of a large aggregate signal as of 2026 would not necessarily invalidate future productivity impacts, given the shallow adoption outlined in the previous section.” Tracks sectoral productivity by AI exposure level, unemployment and labor force participation for workers aged 20-24 against prime-age workers, and layoffs, discharges and job openings in information-processing sectors as a white-collar displacement signal.

But over the past year, the most AI-exposed occupations have generally seen the largest rebound in postings.

The most direct cross-country test yet of whether the entry-level AI story survives outside the US. Across Australia, Canada, the EU and the US, "the role of LLMs in explaining the unemployment gap of young labour market entrants remains limited," with no trend break around the mid-2023-to-early-2024 onset of intensive firm LLM use in either unemployment gaps or LLM-exposed job postings. OECD's alternative explanation is compositional: exposed occupations sit disproportionately in information, finance and professional services, sectors "particularly sensitive to capital cost and general economic uncertainty," so "what may seem an effect of the diffusion of LLMs is in fact simply the result of worsening macroeconomic environment" (citing Curto Millet and Iscenko 2026; Frank et al. 2026). OECD hedges appropriately: "The conclusion that LLMs diffusion has not significantly impacted the employment prospects of young entrants remains preliminary," and the finding "may also mask substantial impacts in specific occupations or sectors." A null attribution result with no share-of-jobs estimate, so it is recorded as an overlay.

NYT chief economics correspondent Casselman synthesizes the measurement problem: different data sources give contradictory answers on basic questions (how many companies use AI, which workers are most vulnerable, whether AI is helping or hurting employment, whether the productivity boom is real). Reports on Nathan Goldschlag's (Economic Innovation Group) new report documenting the measurement challenge; a bipartisan Senate bill (led by Sen. Mark Kelly, D-AZ, introduced June 2026) that would expand federal AI labor data collection; Yale Budget Lab's new monthly 'occupational churn' analysis as an early-warning system; Stanford DEL's ADP-based Canaries dashboard; and new Ramp/Revelio research finding companies using AI most intensely are adding jobs FASTER than laggards — the opposite direction from displacement narratives. Frames the current confusion as J-curve territory: companies still on the downward experimentation phase before productivity gains materialize. Notes federal statistical system is under stress from falling survey response rates and funding cuts; former BLS Commissioner Erika McEntarfer (fired by Trump last year) says $10M/yr would meaningfully expand the monthly labor market survey.

EIG applies five different measures of AI exposure (drawn from four different research papers) to test whether current labor market data shows an AI signature. The startling finding: which exposure measure you pick determines not just the SCALE but the DIRECTION of AI's estimated effect on employment. Under some measures AI appears to be hurting employment; under others it appears to be helping. Companion piece 'A New Threat to Economic Data' documents the deteriorating federal statistical system (falling response rates, funding cuts) that Casselman NYT (July 2) also foregrounds. Goldschlag is EIG's Director of Research and formerly Principal Economist at Census's Center for Economic Studies.

BLS Employment Situation, June 2026 (released July 2, 2026). Total nonfarm payroll employment changed little in June (+57,000), roughly in line with the average monthly change over the prior 12 months (+36,000). Unemployment rate 4.2% (dipped slightly). Health care +22,000 (below prior 12-month average of +38K); leisure and hospitality declined by 61,000 (weaker seasonal hiring). Prior months revised down: April -31K (to +148K), May -43K (to +129K); combined -74K. Average hourly earnings for private nonfarm payrolls +13 cents (0.3%), to $37.64. Wall Street had expected 115,000 jobs — actual well below.

Daron Acemoglu, asked how large AI's labor market impact will be: "I would expect some net job losses, limited, but net job losses within the next five years... none of this, in my estimation, will be of a scale anywhere close to the kinds of things that some expect. I would say less than two to four percent." The 2-4% is plotted as a midpoint with the range as its confidence band; note it is stated as an upper bound over a five-year horizon rather than a point estimate to 2030. Acemoglu attributes the ceiling to missing deployment infrastructure: "we don't have easy to use applications based on the foundation models that can be adopted by many large-scale employers," with coding an exception. He puts the most vulnerable pool - cognitive routine work such as customer service reps and back-office staff - at "eight million in the United States, nine million," and expects "bigger net job losses within the next 10 to 15 years if things continue like this." Goldman's Joseph Briggs, on the same episode, sizes the current imprint at "probably a 10 to 15,000 drag on month-over-month job growth from AI impacts," concentrated in tech, management consulting and graphic design, and reiterates the 9%/15M ten-year reallocation forecast under a 15% productivity uplift, adding that "the unemployment rate increase in any given year would be less than one percentage point." Recorded June 2026, published July 2, 2026.

Bloomberg (Boesler & Prakash, Jul 2 2026): 'A decline in payrolls in the financial-activities and information sectors — where AI adoption rates have been fastest — has accelerated in 2026, to 28,000 per month on average based on government data.' 'The weakness stands out against an otherwise robust labor market that created more than 113,000 jobs monthly this year through May.' Challenger data: 'almost 102,000 announced job cuts attributed to AI so far this year.' 'Overall, the tech sector accounted for a third of all layoffs announced in 2026.' Finance workforce: 'Office and administrative support occupations — including customer service representatives, bank tellers and insurance claims processors — account for about a quarter of employment in financial activities.' California Policy Lab: 'Finance and insurance had the highest concentration of unemployment claims in the state coming from workers in highly AI-exposed occupations.'

Overall, employment in these occupations increased by 335,800 jobs, or 4.05 percent, between 2019 and 2024.

First study to link observed firm-level AI spending to workforce outcomes at scale. Tracked AI spending across 21,599 US firms via Ramp expense-management data, matched to Revelio Labs employment records. Companies that invest heavily in AI grew headcount 10% over the two years following adoption; entry-level headcount grew 12%. Gains are ENTIRELY driven by high-intensity adopters — low-intensity adopters see no statistically significant change. Direction is opposite to the displacement narrative: heavy AI adopters ADD jobs faster than laggards. Caveats: AI adopters are already larger, more engineering-intensive, more likely to be venture-backed, and faster-growing than non-adopters; effect may partly reflect selection into adoption rather than causal effect of adoption.

AI is not eliminating jobs across the economy. Instead, it is raising the bar for young workers trying to secure their first foothold in the labor market,

Chandar — coauthor of the Brynjolfsson-Chandar-Chen 'Canaries' paper — explains why he was one of 5 of 16 economists on a WSJ panel to predict AI would cause net job loss (others: Acemoglu, Henderson, Restrepo, Wolfers; 8 said no change, 2 said net growth). All 16 agreed AI would boost productivity. The 5 net-loss economists also unanimously said AI would replace rather than complement workers and would reduce demand for white-collar jobs. Chandar's argument: his net-job-loss prediction is not a 'jobs bloodbath' story — he expects AI to make people rich enough (via capital income or transfers) that the income effect dominates the substitution effect, lowering labor force participation in the long run (~50yr horizon). He explicitly cites Kinder's 'messy middle' framing as the short-to-medium-run risk. Useful disambiguation of what economists mean by 'net job loss' — long-run LFP decline driven by post-scarcity income, not displacement-driven destitution.

A bipartisan consortium called RAISE US launches with a 'people strategy' for the AI era — led by former Commerce Secretary Gina Raimondo (CEO) and former Indiana Gov. Eric Holcomb. Founding employers include Amazon, Microsoft, Bank of America, and Eli Lilly; OpenAI and Anthropic are involved; MIT economist David Autor sits on the advisory board. The group has raised $500M+ (about half its multiyear goal) and will initially work with Arkansas, Maryland, Utah, and Connecticut. Mandate goes beyond retraining: revisiting unemployment insurance so displaced workers can keep benefits while starting AI-enabled businesses, and developing corporate incentives for employers to retain and reskill rather than lay off.

Yale Budget Lab's preferred econometric strategy finds no strong evidence of AI's impact on aggregate US employment or unemployment 33 months post-ChatGPT. Measures of AI usage show no detectable connection to changes in employment or unemployment. The labor market in early 2026 features low layoffs but also low hiring, with payroll growth ~20,000 net new jobs per month and unemployment 4.3% in March 2026. Authors caution that historically technological disruption unfolds over decades, not months — AI is likely to leave its mark eventually, but the macro signature is not yet visible. A companion to the lab's March 2026 CPS update.

many economists are more concerned about a different, larger group of white-collar workers: customer service representatives, bookkeepers, payroll clerks and human resources specialists who fly under the radar but collectively account for tens of millions of jobs.

Monthly-updated dashboard built on anonymized ADP payroll data covering 4.6 million workers across 730+ occupations. April 2026 reading: most AI-exposed occupations contracted 0.2% YoY, least-exposed grew 0.1% YoY — a small but consistent gap. For workers ages 22-25 in highly AI-exposed occupations, employment is now shrinking at 3.8% per year, with the early-career decline sharpening from -2.8% to April 2024 to over -4% per year since. Builds on Brynjolfsson, Chandar, Chen (2025) 'Canaries in the Coal Mine.' Live update of the underlying employment story behind the 'overall not yet, early-career already' pattern.

an expanding series of economic dashboards that connects policymakers, business executives, and individual workers to timely and reliable information on the economic impact of AI

Goldman Sachs revised its US labor displacement forecast upward from 6-7% to ~9% (about 15 million workers) over the next decade. The change reflects a new methodology measuring total flow of workers leaving jobs due to AI-driven productivity gains, rather than steady-state unemployed count. Briggs estimates each 1% increase in technology-driven productivity yields a 0.5-0.6% increase in the job-destruction rate over the following two years. Goldman maintains the long-run benefits outweigh the disruption (US churns 25-35M jobs annually, and AI will generate new employment), and that unemployment rises less than 1 percentage point at peak under standard adoption timing. As of mid-2026, Goldman estimates AI is erasing ~16,000 net jobs per month, concentrated in entry-level and administrative roles.

Goldman Sachs's economists estimate that, over the next decade, A.I. may automate 25 percent of current work hours. If our estimate proves correct, A.I. won't eliminate 25 percent of jobs. What's more likely is that people will find more productive ways to spend their time.

While job postings show a relative decline in vacancies in occupations with greater exposure to AI, that divergence began before the release of ChatGPT in late 2022. These patterns make it difficult to attribute the recent slowdown in entry-level hiring to AI alone.

despite all the exciting progress in AI, the U.S. unemployment rate remains a healthy 4.3%

Using synthetic differences-in-differences (SDID) to compare AI-exposed (top tercile) to a synthetic-control group built from unexposed (bottom tercile) occupations through 2026Q1: 'this approach provides no clear evidence of AI effects on the labor market, but this could change quickly.' Wages: 'we see no statistically or economically significant impact.' Unemployment rose ~0.5pp in latest quarter for the AI-exposed group (more for 16-34 subsample), but statistically insignificant. Treatment date: 2022Q4 (ChatGPT release). Demographic note: AI-exposed occupations are 55.1% women and 57.3% BA+ vs. unexposed 32.7% and 10.3% — making naive comparisons unreliable.

WIOA rarely supports worker resilience to automation, with 45% of all WIOA participants returning to their prior industry of work, and 27% staying in the same occupation. Successful outcomes driven mostly by wage gains, possibly due to catch-up mean reversion, rather than changes in occupation.

A one-SD increase in RL exposure is associated with a 2.9% decline in job openings after the introduction of ChatGPT. This effect is marginally significant (p=0.085).

in a difference-in-differences framework, individual-level coefficient magnitudes vary 2.4-fold across annotators, and county level estimates flip from a significant negative to an insignificant positive depending on annotators

Dario Amodei, CEO of Anthropic, went on Fox News and said that AI will eliminate up to half of all entry-level white-collar jobs within one to five years. He named finance, consulting, law, and tech... the argument that 'half of entry-level white-collar jobs be gone in five years' confuses task automation with the extinction of jobs.

Between 2011-2023, 18% of US workers were employed in jobs introduced since 1970. New work differs from more jobs in existing occupations because it requires novel, scarce expertise, and therefore pays higher wages.

One fifth of the respondents in our survey voiced concern about economic displacement. Respondents' perceived threat from AI was correlated with our own measure of observed exposure, which reflects the percentage of a job's tasks for which Claude is used.

Most users (66%) rely on AI solely to augment tasks, while AI-related employment decreases are rare, occurring in only 2% of firms.

Output growth slowed from 3.1% to 2.6% per year. But hours growth fell from 1.9% to just 0.7%. The result: productivity growth rose from 1.3% to 2.2%. There are no signs of a slowdown in 2025, and early 2026 job growth points to continued flat employment outside healthcare and social services.

Our proprietary survey of AI integration efforts finds that only 11% of management teams expect AI to replace existing software subscriptions. Far more expect the move to AI-centric architectures to reduce maintenance spending on legacy infrastructure, save on consulting fees, and increase the productivity of existing workers, reducing future headcount needs. Figure 4 (Carlyle Proprietary Survey, April 2026): Internal headcount cited as AI spending offset by 26% of respondents — the largest single category.

Overall employment is projected to fall 1.9% by 2030 and fall 3.4% by 2035 due to AI-driven displacement. Unemployment for new 4-year college graduates is projected to more than double to 12% by 2035.

All 921 occupations (147.9M jobs): 18% of jobs are at a higher short-term automation risk; 46% of jobs are less likely to experience near-term change; 12% of jobs could grow because of AI; 24% of jobs may see declining employment as their task composition shifts, but the remaining jobs will still need workers. People that are or were employed in jobs we predict to have less immediate change have actually seen the largest increase in unemployment at +0.6 p.p. For comparison, people employed in jobs at higher automation risk or jobs that will reorganize have seen +0.3 p.p. and +0.3 p.p. increases in the unemployment rate over the same period.

What A.I. company executives are saying about their products — that they might lead to human extinction and almost certainly will lead to large-scale permanent disemployment — is so obviously 'bad messaging' that I would really urge people to consider that it's not a 'message' at all.

The overall reduction in hires fully explains a 15% decline in employment in the most AI-exposed industries since large language models became available, representing a loss of over 150,000 early career jobs.

The addition of the March 2026 CPS and the introduction of Anthropic's February usage metrics do not suggest any substantial changes. Occupational dissimilarity, industry dissimilarity, and our exposure and usage metrics all remain flat, lie within historical ranges, or continue along the trends they were already exhibiting. Currently, measures of exposure, automation, and augmentation show no sign of being related to changes in employment or unemployment.

Around 62 percent of all respondents believe the unemployment rate will increase over the next twelve months due to AI, while around 11.6 percent expect it will decrease due to AI.

Recessions worsen outcomes: The effects of technology-related displacements are amplified (by three weeks of additional unemployment and a 5-percentage-point likelihood of subsequent joblessness). Goldman Sachs previously estimated that 6% to 7% of US workers (about 11 million people) could have their jobs displaced by AI.

Our analysis implies that AI substitution has reduced monthly payroll growth by roughly 25k and raised the unemployment rate by 0.16 percentage points over the past year, while augmentation has added about 9k to monthly payroll growth and lowered the unemployment rate by 0.06pp. This implies a net drag of 16k per month on payroll growth and a 0.1pp boost to the unemployment rate.

AI will definitely eliminate some jobs, while it enhances others. There is a possibility that AI deployment will move faster than workforce adaptation to new job creation. In prior technological transformations, labor had time to adjust and retrain.

New research by Goldman Sachs economists finds that AI is already a measurable drag on the U.S. job market—erasing roughly 16,000 net jobs per month over the past year, with the pain falling hardest on Gen Z and entry-level workers. Goldman's breakdown shows AI substitution wiped out roughly 25,000 jobs per month in the past year, while augmentation added back about 9,000.

Total nonfarm payrolls rose by 571K in March 2026 to 157,775K.

Unemployment rate 4.3% in March 2026, down 0.1 percentage points.

researchers at Boston Consulting Group estimated that more than half of the jobs in the United States would be 'reshaped' by artificial intelligence over the next two to three years but that far fewer would be replaced entirely.

U.S.-based employers announced 60,620 job cuts in March, according to Challenger, up 25% from 48,307 cuts announced in February. AI was the leading reason for cutting jobs, cited in 25% of announcements, followed by closings, restructuring and economic conditions.

Headcount reduction was the largest outcome in 45% of the deployments, but alternatives (hiring avoided, redeployment, no reduction) accounted for 55%.

If progress continues, we can expect systems to be capable of carrying out projects that currently take people months. This shift will reshape how organizations run, how knowledge is created, and how people find meaning and opportunity.

Conditional on a 'rapid' AI progress scenario, in which AI systems surpass human performance on many cognitive and physical tasks, experts forecast substantial, though not historically unprecedented, economic shifts: annualized GDP growth rising to around 4% and the labor force participation rate falling from its current level of 62% to 55% by 2050, with roughly half of that decline—equivalent to around 10 million lost jobs—attributable to AI. The average economist assigned a 61.4% probability to moderate or rapid progress, but only 14.0% to the rapid scenario. Unconditional economic forecasts are close to historical trends: median GDP growth 2.5% for 2030, LFPR 61.0% for 2030 vs 62.6% in 2025.

Johnston & Makridis (2026, SSRN 6460358) use US data through 2024 and find that exposure to AI is associated with SECTOR-LEVEL INCREASES in employment, not declines. Cited in Stanford DEL Research Note #1 as a key counter to the Brynjolfsson-Chandar-Chen Canaries finding of early-career declines — same period, different unit of analysis, opposite sign at the sector level. Illustrates the current dataset-and-methodology dependence of the AI-labor debate.

In a rapid AI progress scenario, economists forecast a drop in labor force participation from its 2025 baseline of 62.6% to 59.1% in 2030 and 55% in 2050, with roughly half of this decline—equivalent to 10 million jobs—attributable to AI.

The paper maps both the opportunities GenAI creates and the structural challenges it poses to workers across skill levels. It underscores the uneven distribution of risk, with routine cognitive roles facing the greatest exposure.

Despite faster growth, treated firms do not scale inputs proportionally. Their demand for external capital investment falls by 39.5% relative to the control group, while their demand for labor remains unchanged.

75% of AI application market value concentrated in software and information tasks. Physical work activities remain largely unaffected. Think/Do/Interact ontology of 39,603 activities based on O*NET 29.1.

Industry-wide vulnerability to job displacement is approximately 6%... 9.3 million jobs are vulnerable to job loss due to AI under our median adoption path, with a plausible range of 2.7 to 19.5 million

The workers at greatest risk are not necessarily those with the highest average exposure, but those whose jobs are built around a small number of core tasks that AI can automate.

In weak-bundle occupations, AI automates some tasks and narrows the boundary of the job, leading to the standard task-substitution channel. In strong-bundle occupations where tasks are not independently reallocable, AI improves performance inside the job, but does not remove the human from the bundle.

They find that among 22-25 year olds employment in the top two quartiles of AI exposure fell about 12 percent relative to employment in the bottom quartile. Starting from total private employment of 130 million, and assuming about 7.6 percent of the workforce is 22-25 years old (based on the CPS), a 12 percent job loss for two quartiles works out to about 475,000 jobs lost.

The addition of the January and February 2026 CPS and the introduction of Anthropic's 'Observed Exposure' metrics do not suggest any substantial changes. Occupational dissimilarity, industry dissimilarity, and exposure and usage metrics all remain flat, lie within historical ranges, or continue along pre-existing trends.

If it takes place over a decade, Goldman Sachs Research expects to see a 0.6 percentage point increase in the unemployment rate.

A lot of these claims are premature. Many of the people making them have an interest in selling a product, so you have to take them with a grain of salt. Right now I would describe the labor market as more of a wait-and-see environment. You cannot automate everything tomorrow because we simply don't have enough computing power to do it.

Two points of general agreement stand out: There's no measurable evidence so far that AI is putting Americans as a whole out of work, economists say. And while the victims of past workplace automation were mostly factory and trade workers, it's white collar jobs that are first in line for AI shake-ups today.

An event study documents an accelerating decline in employment of 22–25-year-olds in high-AI-exposure occupations, reaching 5.5 per cent by early 2025 relative to less exposed occupations within the same employers, while employment of workers over 50 rose by 1.3 per cent.

Survey of ~750 CFOs finds firm-size-and-sector-weighted aggregate employment is expected to decline by less than 0.4% due to AI in 2026, implying ~502K workers. 50% of firms report no AI job replacement. Large firms expect workforce reductions; small firms expect modest gains.

The evidence on how AI is affecting the labor market today is inconclusive, and claims about harmful impacts on particular groups of workers are premature. Initial evidence suggests that transitional disruption from AI to date is not outpacing recent technological changes.

I'm skeptical that simply slotting AI into human-shaped jobs will have the results people seem to expect... as long as we are in that regime, I expect disappointing productivity gains and relatively little real displacement.

89% of the respondents said they have no plans to cut staff or lay off workers because of AI. Just 3% said they have already cut staff, while 7% said they have plans to reduce their workforce within the next year.

So far in 2026, AI has been cited for 12,304 job cut announcements, or 8% of job cut plans. In 2025, companies referenced AI for 54,836 announced layoff plans, 5% of total cuts during the year.

We find no systematic increase in unemployment for highly exposed workers since late 2022, though we find suggestive evidence that hiring of younger workers has slowed in exposed occupations

Companies with frequent AI use are about 4% more likely to take on additional staff. Only 15% of firms that use AI cite reducing labour costs as a factor.

there is little evidence of AI disruption in labor market data as of today. In fact, the forward-looking components of our labor market tracking have improved and AI data center construction appears to be driving a pick-up in construction hiring.

After the public launch of ChatGPT in November 2022, job postings for occupations that involve lots of structured and repetitive tasks, likely replaceable by generative AI, decreased by 13%.

While we still do not find a meaningful relationship between productivity and AI adoption at the economywide level, companies that quantified productivity impacts of AI on specific tasks reported a median productivity gain of around 30%.

We first analyze 43 benchmarks and 72,342 tasks, measuring their alignment with human employment and capital allocation across all 1,016 real-world occupations in the U.S. labor market.

We do not find clear evidence that industry-level AI adoption is associated with employment changes.

Unemployment level 7,239K in March 2026.

As technology devalues labor, optimal policy shifts from steering innovation toward redistribution.

In particular, firm-size-and-sector-weighted aggregate employment is expected to decline by less than 0.4% due to AI in 2026. [...] little evidence of near-term aggregate employment declines due to AI. [...] Over the next three years, the share of routine clerical employment is expected to decline by more than 2 percentage points (mostly among large firms), with partially offsetting increases in skilled technical roles. [...] The mean reported increase in labor productivity attributable to AI investment was 1.8% in 2025 and is expected to reach 3.0% in 2026. Survey of nearly 750 corporate executives.

AI's labor market impact still small — only 5-10K/month drag on net job growth. Only 2½% of jobs exposed to automation today. Baseline forecast: 6-7% of workers displaced (range 3-14%), lowering annual hiring by 1M jobs and raising unemployment ~½pp. US creates 30M gross new jobs/year; does not anticipate job apocalypse. Impact so far largely confined to tech sector but expects it to grow materially.

employment has declined 1 percent since late 2022 in the 10 percent of sectors most exposed to AI

No evidence that AI reduces employment in the short run. AI augments worker output — enabling employees to complete tasks faster and make better decisions — without displacing labour.

Survey of ~6,000 executives across US, UK, Germany, Australia. 90%+ report no employment impact from AI over the past 3 years. Firms expect AI to reduce employment by 0.7% over the next 3 years (US: -1.2%). Employees, by contrast, expect +0.5% job creation.

Job openings fell to their lowest level since December 2017. Quits rate held steady at 2%, remaining below its 2019 average.

Survey respondents said that AI adoption had led to elimination of 11% of jobs and an additional 12% were left unfilled. This is partially offset by 18% new hires, resulting in a 4% net job loss globally.

No measurable change in job openings or aggregate employment in AI-exposed occupations; adjustment happening through task reallocation, not broad displacement.

a 10 percent increase in the minimum wage increases robot adoption by roughly 8 percent relative to the mean

Synthesizes firm-level survey data from the US, UK, Germany, and Australia measuring AI's realized and expected impacts on sales per employee and employment. Data collected Feb-Apr 2025 and Nov 2025-Jan 2026.

Job openings at 6,882K in February 2026, gradual decline from 2024 peaks.

65% expect 11%-30% of their workforce to be redeployed or reskilled due to AI [over the next two years]. Employee concern about job loss due to AI has surged from 28% in 2024 to 40% in 2026. 98% of executives are planning organizational design changes over the next two years. The report draws on insights from nearly 12,000 business executives, HR leaders, investors, and employees worldwide gathered via a survey conducted from September to October 2025.

the Indeed Job Postings Index, which was down 5.2% year-over-year as of December 31

In het derde kwartaal van 2025 werkten ruim dertien procent minder jongeren in de meest vatbare beroepsgroepen dan in het vierde kwartaal van 2022, terwijl de werkgelegenheid in andere beroepsgroepen juist drie procent steeg.

A one-standard-deviation increase in Generative AI exposure is associated with an 8% decline in job postings and a 0.6% decline in the hourly wage rate.

Of the 37.1 million workers in the top quartile of AI exposure, 26.5 million are in occupations that also have above-median adaptive capacity.

Of the 37.1 million workers in the top quartile of AI exposure, 26.5 million are in occupations that also have above-median adaptive capacity. At the same time, 6.1 million workers (4.2% of the workforce) are both highly exposed and in the bottom quartile of adaptive capacity, concentrated in clerical and administrative roles.

49% of jobs now use AI for at least 25% of their tasks (up from 36% in early 2025); augmentation (52%) has overtaken automation (45%) as primary use pattern.

Unemployment risk in AI-exposed occupations rose beginning early 2022; graduates with AI-exposed curricula have higher first-job pay and shorter job searches post-ChatGPT.

Education task exposure jumped from 11% to 49% — a 4.5x increase driven by multimodal AI, advanced reasoning, and agentic systems. Based on reassessment of 18,000 tasks across 1,000 O*NET occupations.

By 2040, output is only 4% higher than it would have been without the growth acceleration, and by 2060 the gain is still only 19%. A key reason for the slow acceleration is the prominence of 'weak links' (an elasticity of substitution among tasks less than one).

Employment 3.6% lower in AI-vulnerable occupations in regions with high AI-skill demand after 5 years; geographic divergence driven by skill concentration.

Within a year, more than a third of surveyed companies (36%) expect at least 10% of their jobs to be fully automated.

Nearly 40% of AI time savings are lost to rework, including correcting errors, rewriting content, and verifying outputs from one-size-fits-all AI tools. Only 14% of employees consistently get clear, positive net outcomes from AI.

AI and automation could account for 6% of total US job losses by 2030, equating to 10.4 million roles. Widespread AI-driven job replacement remains unlikely. AI will augment 20% of jobs rather than eliminate them.

19% of respondents said they are actively pursuing different careers (7%) or considering changing plans in the near future (12%) due to AI-driven transformation.

automation can depress demand and push AI-driven growth toward the lower end of current forecasts. [...] the conditions needed for growth to actually turn negative are likely too unrealistic to hold in practice

If fully translated to unemployment, AI-driven employment declines among young workers would explain only 0.1 percentage point rise in aggregate unemployment.

Employment share for AI-exposed occupations fell from 16.4% (Nov 2022) to 15.5% (Sep 2025). Workers age 22-25 in most AI-exposed occupations experienced 13% employment decline since 2022, driven by fewer workforce entrants rather than layoffs.

Perceived 'identity fit' — whether a new skill feels compatible with a person's sense of self — dominated re-skilling decisions, often outweighing beliefs about wages or employer demand.

EU firms see +4% productivity from AI adoption; no employment decline detected.

Task-by-task substitution logic is incomplete because automating one task changes the return to automating others. Automation decisions are discrete and can require bundled adoption even when automation quality improves smoothly.

35.9% of US workers used generative AI by December 2025; small positive wage effects, no significant decline in job openings.

AI can already replace 11.7% of the U.S. workforce, or as much as $1.2 trillion in wages across finance, health care, and professional services. The visible tip — tech sector layoffs — represents just 2.2% of total wage exposure ($211B). Simulated 151 million workers across 32,000 skills and 3,000 counties.

Currently demonstrated technologies could automate activities accounting for about 57 percent of US work hours today. AI agents could perform tasks occupying 44 percent of US work hours, while robots could handle another 13 percent. Roles with the highest potential for automation make up about 40 percent of total jobs.

Overall employment continues to grow robustly, but employment growth for young workers has been stagnant since late 2022. Declining employment in AI-exposed jobs drives stagnant overall employment growth for 22- to 25-year-olds.

32% of companies expect AI to reduce workforce by 3%+ within next year. Median 17% of respondents report workforce declines within functions from AI use. 88% of organizations now use AI in at least one business function.

Job postings for high-AI-substitution occupations fell 12% relative to low-substitution roles post-ChatGPT. Effect grew from 6% in year one to 18% by year three. This measures posting-level demand shifts, not direct employment changes.

Study of 11 occupations in Denmark: essentially zero effects on earnings and hours worked; confidence intervals rule out effects larger than 2%. AI adoption linked to ~4% occupational switching.

AI agents perform near the floor on RLI, with the highest-performing agent achieving an automation rate of 2.5%. 240 real freelance projects across game dev, design, architecture, data analysis, and animation. Best agent earned $1,720 of $143,991 total project value.

Analysis finds no significant macro-level job displacement from AI as of late 2025, though structural shifts are emerging in specific sectors.

Despite fears of imminent AI jobs apocalypse, the overall labor market shows more continuity than disruption since ChatGPT's launch.

More than half of businesses using freelancers in 2022 have stopped entirely; freelance marketplace spend fell from 0.66% to 0.14% of total business spend.

Productivity gains 20-60% in controlled RCTs, 15-30% in field experiments; AI exposure measures converge toward high-wage jobs being most exposed.

Roughly 10% of US work affected in the short run; AI will increase productivity and GDP by 1.5% by 2035. Jobs that AI can completely replace saw 0.75% employment fall 2021-2024.

In jobs with high AI exposure, employment for 22- to 25-year-olds fell 6% between late 2022 and July 2025. Software developers saw a 20% early-career decline. Employment among workers 30 and older grew 6-13%.

In the absence of active countermeasures, transformative AI may lead to significantly more centralization of decision-making. Some types of information might defy codification, and some types of information processing might be better done locally.

Within highly exposed occupations, like office and administrative roles, workers specialized in information-processing tasks leave and suffer wage losses, while those specialized in customer-facing and coordination tasks stay and experience wage gains as work rebalances toward their strengths.

AI could displace 6-7% of the US workforce if widely adopted (range 3-14%). Currently 2.5% of US employment at risk of displacement from current AI use cases; unemployment projected to increase by 0.5pp during transition.

Occupations with higher AI exposure experienced larger unemployment rate increases between 2022 and 2025 (r=0.47). Computer/math occupations (exposure ~80%) saw steepest rises. GenAI-intensive occupations: r=0.57.

Amazon deployed its 1 millionth robot across over 300 fulfillment centers worldwide. CEO Andy Jassy acknowledged that AI will result in fewer people doing some automated jobs, while new roles in robotics maintenance grew 30% at next-generation sites.

Resolves the long-standing puzzle of why routine task automation lowered employment but often raised wages in routine task-intensive occupations. Builds 'expertise framework' on 3 pillars: hierarchical expertise, occupational task bundling, automation. Key counterintuitive prediction: changes in occupational expertise requirements have OPPOSITE-signed effects on wages and employment — rising expertise raises wages but reduces qualified workers (employment falls), falling expertise lowers wages but expands the eligible labor pool (employment rises). Empirically (303 harmonized Census occs, 1980-2018): 0.1σ rise in occupational task expertise predicts a 5-7 log point fall in employment per decade, while a 10% expansion in task quantity predicts a 14% employment rise — opposite directions. Concrete case: Accounting clerks 1980-2018 saw employment fall 1.62M→1.11M (-31%) while wages rose $13.07→$18.17 (+39%); Stock/Inventory clerks saw employment rise 0.54M→1.48M (+175%) while wages fell $14.30→$12.43 (-13%). Routine task share fell 50.4%→32.2%; abstract share rose 33.2%→53.6%. Cross-occupation expertise wage premium nearly doubled: 1σ expertise = +16 log pts in 1980 vs. +31 log pts in 2018. Methodology: content-agnostic expertise measure via Efficient Coding Hypothesis on Standard Frequency Index; OpenAI embedding-based task matching (caliper 0.95) to identify removed/retained/added tasks between 1977 DOT and 2018 O*NET.

Job numbers are growing in every industry analysed. Growth remained robust even in more exposed occupations (38%). Productivity growth nearly quadrupled in AI-exposed industries.

27.1% of companies expect artificial intelligence to lead to job cuts in the next five years. Only 5.2% of companies anticipate additional jobs, while two-thirds expect no change.

Calibrated to U.S. data, the model predicts more than threefold improvements in productivity in the some-AI steady state, alongside a long-run employment loss of 23%, with half this loss occurring over the initial five-year transition.

Using O*NET and LightCast job postings data (2019-June 2024), the study found a 24% decrease in generative AI-exposed skills per firm per quarter for automation-prone occupations, while augmentation-prone occupations saw a 15% increase. Demonstrates the dual displacement/complementarity impact of generative AI.

Workplaces that adopted AI chatbots showed no significant difference in employment, early-career jobs, job churn, or worker mix. Difference-in-differences estimates for earnings and wages are all precisely estimated zeros, ruling out effects larger than 2%.

Calibrated to U.S. data, the model predicts more than threefold productivity improvements alongside a long-run employment loss of 23%, with half (~11.5%) occurring over the initial five-year transition period.

Framework for guiding innovation to increase labor demand; steering technology becomes more desirable the less efficient social safety nets are.

24% decrease in generative AI-exposed skills per firm per quarter in top-quartile automation-exposed jobs. 15% increase in AI-exposed skills for augmentation-prone jobs. Augmentation-prone job postings increased 22% per quarter per firm.

IBM CEO stated plans to replace ~7,800 back-office roles with AI. Company investing in reskilling remaining workforce for AI-augmented roles.

BLS projects employment of software developers to increase 17.9 percent between 2023 and 2033, much faster than the average for all occupations (4.0 percent).

Workers in AI-exposed occupations face significantly higher unemployment risk.

Only 0.7% of all workers on average leave their employers each year due to automation. The risk of losing a job due to automation is much smaller than the risk of a mass layoff (3.5-7.2%/yr in the Netherlands).

Most adversely impacted occupations (business, financial, engineering) experienced a decline of 2% to 2.5% over a five-year period. Reduced demand in exposed occupations is offset by productivity-driven increases in labor demand at AI-adopting firms.

Total US job postings are 15% below their February 2022 peak. Categories most exposed to AI (data entry, basic admin, customer service) show steepest declines.

17% decrease in job postings for highly automatable occupations, but 22% increase for augmentation-prone ones — net effect depends on occupation mix.

Among respondents actively using or experimenting with AI, 93% agree that it allows them to focus more on higher-level responsibilities. The leading response (53%) was 'AI will augment human capabilities, leading to increased productivity and new forms of innovation.'

92 million roles will be displaced by 2030, equivalent to 8% of current jobs, offset by the creation of 170 million new roles for a net gain of 78 million (7%). Global, all drivers of change. 40% of employers expect to reduce headcount where AI can automate tasks.

To have a sense of the magnitude of the effect, if the CZ with average AI adoption over the sample period (0.004) had hypothetically had no AI adoption at all, its employment-to-population ratio would have grown by 0.6 p.p. more, i.e., 20% faster than observed growth. While these numbers cannot be interpreted as a counterfactual exercise, they neverthess suggest that AI adoption has contributed to slowing down job creation in the US over the last two decades. US commuting zones, 2000-2020, before generative AI.

High-education, high-cost metros are most vulnerable to AI displacement.

GenAI increased STEM incorporated entrepreneurship; displacement may be offset by new firm creation.

Using existing estimates on exposure to AI and productivity improvements at the task level, these macroeconomic effects appear nontrivial but modest—no more than a 0.66% increase in total factor productivity (TFP) over 10 years. [...] Consequently, predicted TFP gains over the next 10 years are even more modest and are predicted to be less than 0.53%.

White-collar layoffs growing as companies link cuts to AI. Google, Duolingo, UPS all reduced workforce. Nearly two-thirds of white-collar workers report improved productivity from AI use.

Almost 40% of global employment is exposed to AI, with advanced economies more affected.

27% of jobs are in occupations at high risk of automation across OECD countries.

Roughly two-thirds of current jobs are exposed to some degree of AI automation; generative AI could substitute up to one-fourth of current work.

The main impact of automation in the near future may be to cause a major reallocation of jobs, even if it does not permanently eliminate large numbers of jobs. During the 19th century, technologies had automated 98% of the labour required to weave a yard of cloth. Yet, the number of weaving jobs actually increased for decades over this period.

If demand is sufficiently elastic and AI does not completely replace humans, then technical change will create jobs rather than destroy them. In manufacturing, technology sharply reduced jobs in recent decades. But before that, for over a century, employment grew, even in industries experiencing rapid technological change. What changed? Demand was highly elastic at first and then became inelastic.

The evidence is that technological unemployment did not occur on a large scale during the Industrial Revolution. The fears of the Luddites that machinery would impoverish workers were not realized, and the main reason is well understood: mechanization could only replace a limited number of human activities, while technological change increased the demand for other types of labor that were complementary to the capital goods embodied in the new technologies.

We examine how susceptible jobs are to computerisation… implementing a novel methodology to estimate the probability of computerisation for 702 detailed occupations, using a Gaussian process classifier… According to our estimates, about 47 percent of total US employment is at risk.