Will AI Replace Your Job or Just Your Raise? (2026)
10.0% of U.S. employer businesses now run AI in production as of 2025-09, according to the Census Bureau's Business Trends and Outlook Survey — a 2.7x increase since the survey began in September 2023. Challenger, Gray & Christmas tracked 139.9K AI-attributed job cuts across 2025, and tech sector job openings are down 72% from peak. Five indicators, measured separately. Whether they form a chain is the open question.
How Is AI Changing the U.S. Workforce?
10.0% of U.S. employer businesses ran AI in production as of 2025-09, according to the Census Bureau's Business Trends and Outlook Survey — up from 3.7% two years earlier. Frontier models can now autonomously complete tasks equivalent to 26 work-weeks of human time (1045 hours), per METR's autonomous task horizon benchmark. Challenger, Gray & Christmas recorded a single-month spike of 54.8K job cuts explicitly attributed to AI in 2025 12. The capability is growing faster than the labor market can adapt to it.
These workforce measures moved during the same period, but they do not by themselves show that AI caused the labor-market changes. Tech sector job openings have fallen 72% from their 2025 peak, while youth unemployment sits at 9.2%. The AI Displacement Tracker, a companion composite that groups these five signals, reads 79.4 (Serious). It is a contemporaneous monitoring tool, not a causal or predictive model of the American Distress Index; any claim that it moves before household distress remains a hypothesis for separate research.
Key Statistics at a Glance
The AI Displacement Tracker currently reads 79.4 (Serious), built from five components: AI-attributed layoffs (25%), youth unemployment (25%), tech job openings (20%), AI adoption rate (15%), and AI capability growth (15%). The tracker summarizes contemporaneous signals. A possible lagged relationship with household income is a hypothesis for separate research, not a result established by this page, while the American Distress Index currently reads 43.8 (Typical). On average, its inputs sit higher than in 44% of their own quarterly histories. The tracker does not establish that AI caused a change in household distress. For household-level measures, see the household financial health statistics and the unemployment statistics roundup.
How Widely Have Companies Adopted AI?
The Census Bureau's Business Trends and Outlook Survey (BTOS) asks U.S. employer businesses a direct question: are you using AI to produce goods or services? The share answering yes has climbed from 3.7% in September 2023 to 10.0% in 2025-09. A 2.7x increase in two years. This measures deployment in production workflows, not experimentation.
The thing about 10.0% is that the denominator includes every employer business in the country. Restaurants. Construction firms. Dry cleaners. Among large firms in technology, finance, and professional services, adoption rates run far higher. The national average is the floor, not the ceiling. Against the Census employer-business universe (roughly 6 million firms), each percentage point represents on the order of 60,000 businesses shifting tasks from human workers to automated systems. No BTOS wave has shown a decline. The curve is still accelerating.
U.S. Business AI Adoption Rate (% Using AI in Production)
Source: U.S. Census Bureau Business Trends and Outlook Survey (BTOS, biannual).
Full data and trend: The Adoption Curve (AI adoption rate)
How Much Work Can AI Do Without Human Help?
METR (Model Evaluation & Threat Research) benchmarks the duration of real-world tasks that frontier AI models can complete autonomously at a 50% success rate. Not what demos suggest. What models can actually do. As of 2026-04, the autonomous task horizon stands at 26 work-weeks of human task-time (1045 hours). Multi-week professional tasks, completed without human intervention.
I think the growth trajectory is the part that doesn't get enough attention. A year earlier, the horizon was approximately 60 hours. The jump from hours to days represents a qualitative shift in which jobs are vulnerable. A customer service shift, a content brief, a data cleaning job. Those are already within the autonomous frontier. Code sprints, research analysis, financial modeling. Those are entering the displacement zone now. The METR benchmark does not measure what will happen. It measures what current models can already do. That distinction matters.
AI Autonomous Task Horizon (Hours at 50% Success Rate)
Source: METR (Model Evaluation & Threat Research) — autonomous task duration benchmarks.
Full data and trend: The Horizon (AI task horizon)
How Many Layoffs Are Being Caused by AI?
Challenger, Gray & Christmas, the outplacement firm that has tracked U.S. layoff announcements since 1993, began tagging AI-attributed cuts as a separate category in late 2023. The data is lumpy — some months record zero, while December 2025 spiked to 54.8K in a single month as year-end restructuring announcements cited AI transformation.
The monthly latest of 14.0K (2026-06) understates the trend for two reasons. Many AI-driven eliminations happen through hiring freezes and attrition rather than announced layoffs. And employers increasingly frame AI restructuring as "efficiency" or "strategic realignment" without citing AI explicitly, meaning the Challenger tag captures only the subset where companies name the cause. The total for 2025 was 139.9K announced cuts where AI was the stated reason — a new category of economic displacement that traditional unemployment statistics were not designed to isolate.
Monthly AI-Attributed Layoff Announcements (Thousands)
Source: Challenger, Gray & Christmas (monthly, AI-attributed subset of total announced cuts).
Full data and trend: The AI Cut (AI-attributed layoffs)
How Much Has Tech Hiring Slowed Down?
BLS JOLTS data for the information sector (NAICS 51) shows job openings at 76K as of 2026-05. Down 72% from the 274K peak in 2022 04. The decline is sharper than any single-quarter drop during the post-pandemic normalization, and it extends beyond the tech sector's own hiring cycle.
Here is where the adoption data meets the labor market data. Two explanations fit the same decline in openings. One is substitution: AI tools absorb capacity that would otherwise have required new hires. The other is ordinary demand contraction moving through the tech hiring cycle. These six indicators measure both worlds the same way, so they cannot tell the two apart.
Separating them would take firm-level evidence on why specific roles went unfilled. We do not have that evidence. Until we do, substitution is a hypothesis this page tracks rather than a finding it reports.
The decline itself is not in question, and it reaches past the tech sector. Information sector hiring is contractually tied to marketing, consulting, legal, and creative services. Movements here have shown up in unemployment figures for the affected age cohorts in the same periods. Read the openings chart as a white-collar demand story. The tech framing only captures the first tranche.
Information Sector (NAICS 51) Job Openings (Thousands)
Source: BLS Job Openings and Labor Turnover Survey (JOLTS), NAICS 51 — Information.
Full data and trend: The Tech Drought (tech job openings)
Are Young Workers Being Displaced by AI?
Youth unemployment (ages 16-24) stands at 9.2% as of 2026-06. Down from the COVID peak of 27.5% in April 2020, but still above the pre-pandemic low of 7.9%.
The part that I think matters most is the composition of what young workers are doing when they find work. The headline rate moves within a familiar range. The composition underneath has shifted. Entry-level positions, the first rung of the career ladder, are the most substitutable by current AI capabilities. Content writing, basic coding, customer support, data entry, junior analysis. Those roles once absorbed new graduates into the knowledge economy. Now they are handled by AI tools at a fraction of the cost. Even when young people find work, they are increasingly funneled into service and manual labor roles that AI cannot yet reach, while the knowledge-economy positions that defined upward mobility contract. The unemployment rate captures the quantity of jobs. It says nothing about whether those jobs lead anywhere. For the broader unemployment picture, see the unemployment statistics roundup.
Youth Unemployment Rate, Ages 16-24 (Monthly)
Source: BLS Current Population Survey via FRED (LNS14024887, monthly, seasonally adjusted).
Full data: Youth Unemployment Rate (16-24)
A Displacement Hypothesis
The five indicators on this page measure different points in a possible displacement pathway: AI capability, business adoption, disclosed AI-attributed job cuts, tech openings, and youth unemployment. They come from different sources and periods. Movement across the set does not establish that one stage caused the next or that the pattern will continue. The tracker treats them as concurrent signals of workforce pressure.
One plausible mechanism connecting workforce pressure to household distress is income loss followed by weaker savings and debt-payment capacity. That is a hypothesis to test, not a conclusion established by this page. ADI separately measures household financial conditions. The AI Displacement Tracker at 79.4 (Serious) summarizes five signals that traditional unemployment statistics do not capture in one measure; it is not a forecast of a future displacement wave.
Explore the AI Displacement Tracker score and methodology →Data Sources and Methodology
Census Bureau BTOS
Business Trends and Outlook Survey, measuring AI adoption among U.S. employer businesses. Biannual collection since September 2023. Asks whether businesses use AI to produce goods or services, not whether they are experimenting with it. Census BTOS data.
METR
Model Evaluation & Threat Research benchmarks the autonomous task horizon — the duration of real-world tasks frontier AI models can complete at 50% success rate. Updated with each major model release. Measures practical capability, not synthetic benchmarks. METR autonomy evaluations.
Challenger, Gray & Christmas
Monthly announced layoff tracker operating since 1993. Began tagging AI-attributed cuts as a separate category in late 2023. Captures publicly announced cuts where the employer cites AI or automation as the stated reason.
BLS JOLTS and CPS via FRED
Job Openings and Labor Turnover Survey for tech sector (NAICS 51) openings. Current Population Survey for age-cohort unemployment. Both monthly, seasonally adjusted. FRED series LNS14024887 for youth unemployment.
Frequently Asked Questions
How many jobs has AI eliminated?
Challenger, Gray & Christmas tracked 139.9K announced job cuts explicitly attributed to AI across all of 2025, with a single-month spike of 54.8K in 2025 12. These are announced cuts where the employer cited AI or automation as the reason. The actual displacement is likely higher — many AI-driven role eliminations happen through attrition, restructuring, or hiring freezes that don't generate public announcements. The AI Cut tracks the announced subset.
What percentage of U.S. businesses use AI?
10.0% of U.S. employer businesses reported using AI in production as of 2025-09, according to the Census Bureau's Business Trends and Outlook Survey (BTOS). That is up from 3.7% when the survey launched in September 2023 — a 2.7x increase in two years. The BTOS measures actual deployment in production workflows, not experimentation. The adoption rate varies dramatically by firm size and sector, with large tech firms far ahead of the national average.
How fast is AI capability advancing?
METR (Model Evaluation & Threat Research) benchmarks the autonomous task horizon — the duration of tasks that frontier AI models can complete at 50% success rate. As of 2026-04, the horizon stands at 26 work-weeks of human task-time (1045 hours), up from 2 work-weeks (60 hours) one year prior. This means current models can autonomously execute multi-week work tasks that previously required human labor. The Horizon tracks this benchmark over time.
Are tech companies still hiring?
Tech sector (NAICS 51) job openings fell to 76K in 2026-05, down 72% from the 274K peak in 2022 04. The collapse in openings reflects both the end of the pandemic hiring surge and the substitution effect of AI tools replacing roles that would previously have been filled by human workers. BLS JOLTS data shows information sector openings contracting faster than the broader labor market. The Tech Drought tracks the full series.
Which workers are most at risk from AI displacement?
Entry-level and early-career workers face the steepest displacement risk. Youth unemployment (ages 16-24) sits at 9.2% as of 2026-06 — above the pre-pandemic low of 7.9% and more than 2x the overall unemployment rate of 4.2%. Junior roles in content, customer service, coding, and data analysis involve routine cognitive tasks that current models handle competently, which is why they are the roles most often named in substitution arguments. We track AI adoption, rising capability, and labor market effects as three separate measures. That adoption leads capability leads employment is a testable hypothesis, not a relationship these six indicators establish.
What is the AI Displacement Tracker and how does it connect to the ADI?
The AI Displacement Tracker is a companion composite of five workforce signals. It currently reads 79.4 (Serious), built from AI-attributed layoffs (25%), youth unemployment (25%), tech job openings (20%), AI adoption rate (15%), and AI capability growth (15%). The tracker is not part of the American Distress Index (currently 43.8), and its co-movement with household distress does not establish that AI caused a change in the ADI.