Median Wage Impact from AI by 2030

The core premium is positive in all three, but smaller the more of the seniority shift the specification removes: 5.7% under occupation fixed effects (median pay), 4.7% comparing postings within the same job title, and 2.4% when the seniority mix is held constant within occupations.

Ms. Edlich and Mr. Slok found that occupations highly exposed to A.I. have experienced a 6.7 percent decline in real wage growth since 2023 and that the effect is harshest among lower earners. Service workers, often in administrative or support roles, have experienced a 24 percent decline in real wages. The bottom quarter of workers has seen an 11 percent decline. Top earners, however, have barely been affected. The headline 6.7 figure is already carried on this chart from the Apollo primary, which states it as percentage points of slower real wage growth rather than a percent decline.

The remaining 60%, or about 4.5 percentage points, is real, Goldman concludes, and it traces mostly to rising corporate markups, automation and the decades-long erosion of workers’ bargaining power.

Labor share—the fraction of economic output that accrues to workers as compensation in exchange for their labor—in the nonfarm business sector was 52.8 percent in the second quarter of 2026, the lowest level ever recorded.

Real hourly compensation, which takes into account consumer prices, decreased 3.3 percent in the second quarter of 2026 and decreased 0.1 percent over the last four quarters. The labor share, which is the percentage of output that accrues to workers in the form of compensation, was 52.8 percent in the second quarter of 2026, the lowest level in the series, which begins in the first quarter of 1947.

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.

Task-based model mapping AI capability, diffusion, productivity gain, automation share and reinstatement into paths for GDP, the labor share, wages, reallocation and unemployment to 2030. Cognitive occupations are SOC 11-29, 41 and 43 (management, professional, sales and office work), 62.4 percent of employment. "In the modest change scenario, AI acts like a 'normal technology' over the next five years and has small macroeconomic effects. In the extreme scenario, AI is transformative: by 2030 GDP is 32 percent above its no-AI path and nearly one in five cognitive workers is unemployed." Reviewed by Acemoglu, Autor, Moll, Nakamura, Restrepo, Romer and Steinsson among others, who were not asked to endorse its conclusions.

We find that moving from the 10th percentile to the 90th percentile of GenAI exposure is associated with a wage decline of 4.9% in OEWS data, with no statistically detectable change in employment. In Revelio data, moving from the 10th to the 90th percentile of exposure is associated with a decline of 8.71% in posted wages, and of 10.76% in starting wages. The estimates are -0.114 in 2024 and -0.083 in 2025. Moving from the 10th to the 90th percentile of exposure, these coefficients imply wage declines of approximately 6.7% and 4.9%, respectively. The pooled coefficient is -0.078, corresponding to an average decline of approximately 4.6% over the same exposure contrast.

Workers who are in high-exposure occupations saw their wages grow about 6.7 percentage points slower than workers not in high-exposure occupations after 2023. Difference-in-differences across 321 matched occupations, BLS OEWS 2015-2025, high exposure defined as an Anthropic Economic Index usage score of 0.5 or above. Service occupations had their real wage growth fall by about 24.3% relative to low-exposure occupations after 2023, though the author notes this rests on a small subsample (n=239) and should be interpreted with caution.

In those two scenarios, labor income shrinks by between 0.3 and 1.0 percentage points and tax revenue from labor income declines (in most cases).

More recently, the probabilities based on these two measures have picked up to varying degrees: around 57% based on labor productivity and 21% based on TFP as of the fourth quarter of 2025.

We also examine the log real hourly wages of AI-exposed and unexposed workers... we see no statistically or economically significant impact through 2026Q1.

While we expect US nonfarm business labor productivity growth to average around 2.3% over 2026-2030, most alternative forecasts currently cluster at or slightly below the 10-year trailing average of roughly 2%—consistent with the historical pattern we document. The gap partly reflects our more positive view on AI's likely contribution to future productivity growth.

positive outcomes are rare and driven chiefly by higher wages, as opposed to lower RTI scores due to shifted occupations. Moreover, wage gains may be the consequence of a structural 'catch-up' dynamic, whereby the most vulnerable program participants see a reversion to the mean.

Since the early 2000s, the 'labor share' – the share of a country's national income that goes to workers, rather than to owners of machines and capital – has declined 10% in the US.

Workers initially employed in occupations that later decline by at least 25% demonstrate 0.4 lower future years of work, although this employment difference is mostly explained by other individual traits. These workers, conditional on controls, experience a 4.7% reduction in future cumulative earnings relative to starting earnings, akin to losing one year's worth of earnings over 2007–2024.

Metaculus community forecasts for 2030 and 2035: overall US employment -1.9%/-3.4%; most vulnerable AI-exposed occupations -11.4%/-17.2%; software developers -15.1%/-22.3%; financial specialists -8.1%/-15.3%; services sales -11%/-14%; lawyers -5.4%/-9.6%; designers -4%/-8.4%; K-12 teachers -1.3%/+1.3%; overall median wage -0.6%/+1.4%; percent workers using AI daily 52.5%/70.9%; new-grad unemployment ~10%/12%.

Long-lasting impacts: 10 years after a job loss, technology-displaced workers' real earnings were 10 percentage points below that of non-displaced workers. Short-run impacts: It can take one month longer for technology-displaced workers to find a new job; and their inflation-adjusted earnings take bigger hits (more than 3%) versus other workers (negligible effect).

Workers displaced from technology-disrupted occupations face more difficult short-run transitions back into employment. They take approximately one month longer to find a new job and suffer real earnings losses of more than 3% upon reemployment, compared with negligible losses for workers displaced from more stable occupations.

The mean is 0.17 (SD 0.11), implying that for the average occupation, GenAI could reduce total work time by roughly 17 percent under this conservative calibration.

Workers using AI might well agree that it's increasing their productivity without believing they're seeing the benefits.

Using realized worker-firm matches data for 138 million U.S. workers, Azar et al find the arrival of ChatGPT reduced average wages of employees at firms that can substitute AI for labor by 4.5% relative to firms that cannot.

In a rapid AI scenario, economists forecast labor income share dropping from ~58% to 45% by 2050, implying significant downward pressure on median wages even if aggregate GDP grows.

whether it is, on net, an equalizing force remains a genuinely open question—one that the current evidence, important as it is, does not yet answer.

total income loss of between $200 billion to $1.5 trillion, representing 2% to 15% of overall income from wages and salaries across the U.S.

This estimate of the value of tasks in Claude.ai has dropped slightly from $49.3 to $47.9. US avg. hourly wage: $37.3.

average individual wages are roughly unchanged for incumbents of occupations with no AI exposure, rise by around 4% at moderate exposure, and drop by as much as 35% for workers in the most exposed occupations

In research forthcoming in the Quarterly Journal of Economics, they examine how automation often targets tasks that pay workers wages above their outside options. When those tasks are automated, those wage 'rents' disappear, amplifying wage losses and contributing to rising inequality.

Are we finally seeing AI in the productivity data? A big upward revision to earlier data and strong Q4 bring us 2.2% above CBO's pre-pandemic forecast. Annual rates: 1 year: 2.8%, 2 years: 2.5%, 6 years: 2.2%

Benefits of steering technology depend on complementarity to labor, relative income, and labor factor share.

For occupations with the median experience premium (40 percent), a one-standard-deviation increase in AI exposure is associated with a 0.05 percentage point decline in wage growth.

AI's impact depends on whether it automates or augments. For zero-experience-premium occupations, increased AI exposure is associated with -0.28pp reduction in wage growth. Wages in AI-exposed jobs not uniformly declining.

AI adoption causally increases labour productivity levels by 4% on average. Employees in AI-using companies benefit through higher wages, both in total and per employee.

Small but statistically significant positive wage effects found across AI-exposed occupations; 35.9% of US workers used generative AI by December 2025.

A one-standard-deviation increase in Generative AI exposure is associated with a 0.6% decline in the hourly wage rate. For occupations with exposure derived solely from core tasks, a one-standard-deviation higher Generative AI exposure corresponds to a significantly greater decline in relative wage rates by 1.4%.

raising average wages by 21 percent

Despite the accelerating growth, the effects of A.I. on GDP per person are remarkably small for the next 20 to 40 years.

wages fall by 25%, and long-run welfare drops by 16.5%

Labour income can rise under partial automation because automation scales the value of remaining bottleneck tasks.

falling labor share accounted for 16% of U.S. real wage growth between 1954 and 2019

Each year of AI model progress reduced task time by 8%. Access to any AI model increased Earnings Per Minute by 81.3%. Continued scaling could boost US productivity by ~20% over the next decade.

Small positive wage effects identified for workers in AI-exposed occupations; adoption concentrated among younger, college-educated, higher-earning employees.

Since January 2023, salaries at the top quartile of the wage distribution have grown by more than 30%. Over the same period, salaries at the bottom quartile have risen by only around 10%.

The United Kingdom and the United States may experience higher gains, ranging from 0.4 to 1.3 percentage points annually for the next 10 years.

Median annual wage for all occupations: $49,500. AI-exposed administrative and clerical wages grew 1.1% vs. 3.8% for non-AI-exposed occupations.

LLMs erode labor market signaling on Freelancer.com: worker surplus declines 4%, employer surplus rises <1%, total surplus falls 1%. Signal degradation is a net-negative welfare channel.

Claude speeds up individual tasks by about 80%. Extrapolating suggests 1.8% annual US productivity growth over the next decade.

AI-skilled workers command a 56% wage premium, up from 25% in prior year. AI roles median compensation at $157K vs. $49,500 all-occupation median.

Large-scale field experiments find treatment effects 0-16.3% on sales, mapping directly into TFP improvements.

The overall effect of industrial robots on wages is close to zero and insignificant. 53 papers, 2,143 estimations.

Audit of 500 AI vendors finds surveillance pay spreading across healthcare, customer service, and logistics.

automation can simultaneously replace experts in some occupations while augmenting expertise in others.

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%.

For the empirically plausible case that capital and labor are gross complements, raising wages requires capital-augmenting innovation rather than labor-augmenting.

Access to AI assistance increases worker productivity, as measured by issues resolved per hour, by 15% on average, with substantial heterogeneity across workers.

Automation AI negatively impacts low-skilled wages; augmentation AI raises high-skilled wages.

Posted wages for AI-exposed occupations (admin, data entry, basic accounting) grew 1.2% YoY vs. 4.1% for non-AI-exposed roles, suggesting wage suppression.

Tenured workers cumulatively lose about 3,800 Euros in wage and salary earnings over five years on average (about 9% of one year's income). We find little change in wage rates.

When someone transitions to a job with high exposure to AI, they work 3.5 hours longer per week. Productivity gains captured by firms, not wages.

Walmart's minimum starting wage is $14/hr while average U.S. associate wage reached $18.25 by 2025. The company invested in checkout and inventory automation but has not disclosed specific per-location staffing reduction figures.

UPS announced 12,000 job cuts, primarily management and salaried positions, as part of a $1 billion cost-reduction effort. Full-time driver pay raised to $49/hr top rate under 2023 Teamsters contract.

In most scenarios, AI will likely worsen overall inequality. Higher-income workers may see disproportionate wage gains.

The paper reports no statistically significant average effect on revenues or profits. But effects are highly heterogeneous: high‑performing businesses at baseline appear to improve (roughly 15 percent), while low performers do worse (roughly 8-10 percent worse)

Technology does not automatically benefit workers; shared prosperity requires deliberate policy choices. AI could follow historical pattern of concentrating gains among capital owners.

No single technology in modern memory has caused mass job loss among highly educated workers. Will generative AI really be an exception? No one can answer this question, given how new the technology is and given how slowly employment can adjust in response to technological change.

I find that AI is projected to reduce inequality measured as the ratio of the 90th to the 10th percentile of wages. On the other hand, it is projected to increase inequality at the top of the distribution, measured as the ratio of the 99th to the 90th percentile. Applying the estimated software coefficient to AI, I project a 4% decrease in 90:10 inequality; using the robot coefficient, the decrease is 9%.