AI adoption is rising, but broad US labor-market displacement remains limited so far

Overview

Deutsche Bank examines whether rising firm AI adoption is becoming visible in US labor-market data. Its evidence points to rapidly expanding but uneven adoption, with limited aggregate evidence thus far of AI-driven layoffs, alongside more notable pressure in selected job-opening and young-graduate indicators.

Core views

The report begins with firm adoption. Most sectors remain below 50% adoption, but usage is rising, especially at larger firms. Adoption also varies substantially by state: high-adoption states have rates more than twice those of low-adoption states, while the national average is approaching one-quarter. Survey evidence indicates that firms using AI mostly employ it to supplement or enhance employees rather than solely replace them, and a significant share report no employment-level change in the preceding six months. In higher-adoption sectors, AI use is more often directed toward performing an existing employee task than introducing a new task; information is identified as an outlier. On labor demand, the report notes that JOLTS job openings have risen in recent months and Indeed postings have edged higher over the past few months. At the same time, openings in some AI-sensitive occupations have declined faster than the overall market over the past few years, even though they have subsequently risen. Trends differ across occupations, and software development is cited as an AI-exposed area showing strong gains in job postings. The share of postings containing AI terms has risen sharply over the past year and is increasing across countries. Cross-sectional evidence suggests an emerging negative relationship between changes in job openings and AI adoption rates, but the report presents this as an emerging relationship rather than a settled conclusion. The report finds little indication that AI has yet changed the broad low-firing environment. Key layoff indicators are limited or declining, WARN layoff announcements remain well below their 2019 average, initial jobless claims remain low, and continuing claims are near multi-year lows. There is an emerging positive relationship between changes in layoff rates since 2019 and AI adoption, but labor-market churn has little relationship with adoption. State-level adoption does not show a clear relationship with changes in WARN notices, initial claims or continuing claims. A more specific distinction matters: WARN indicators are somewhat higher in states where firms are more likely to use AI to replace tasks, and initial claims show a modest positive relationship with task-replacement use, while continuing claims show only a limited relationship. On pay, wage growth is moderating but remains highest for workers with at least a bachelor's degree, and the college-degree wage premium persists. Average hourly earnings show a limited relationship between wage growth and AI adoption, although sectors with higher adoption have, on average, experienced more significant wage-growth deceleration; utilities are an outlier. Younger workers aged 16–24 have shown some pickup in wage growth, while deceleration continues elsewhere. The final section highlights young-worker outcomes. Employment-to-population ratios for younger individuals are reasonably close to 2019 levels, and unemployment-rate underperformance for 16–24-year-olds is no longer as clear as it had been because their unemployment rate has fallen. However, unemployment among younger workers with bachelor's degrees remains elevated, including relative to the 2000–2019 history for non-enrolled 16–24-year-olds. New York Fed evidence cited by the report shows unemployment rising most for recent college graduates, although underemployment among college graduates has improved somewhat in recent months.

Analysis framework

The report combines business AI-adoption survey data with labor-market indicators including JOLTS openings and layoffs, Indeed postings, WARN notices, jobless claims, wage data and young-worker employment measures. It compares trends over time, including against 2019 averages, and examines cross-state, cross-sector and occupation-level relationships between AI adoption, task replacement and labor outcomes.

Methodology notes

• Other

  Cross-sectional comparison of AI adoption or AI task-replacement use with labor-market indicators

  The report compares adoption rates across states and sectors with changes in openings, layoffs, claims and wage growth to identify emerging relationships rather than establish causation.

Key data

• National firm AI adoptionApproaching 1/4National average; high-adoption states have rates more than twice those of low-adoption states.
• State AI-adoption dispersionHigh adoption states >2x low adoption statesData as of 2026 week 32.
• WARN layoff announcementsWell below 2019 averageThe report sees no sign of increased firing in this indicator.
• Initial jobless claimsLowClaims remain low despite rising AI adoption.
• Continuing claimsNear several-year lowsThe report finds no clear relationship with state-level AI adoption.

Impact & implications

The report's evidence suggests that AI diffusion is advancing faster than any broad labor-market displacement signal. Its more cautious findings concern selected AI-sensitive occupations, task-replacement use, wage-growth deceleration in higher-adoption sectors, and relatively weak outcomes for recent college graduates.