For much of the past few decades, workers in technology and finance assumed their jobs were secure. They no longer do. In 2025 and 2026, large layoffs have been announced at Amazon, Meta, Oracle, Snap, Block, Aumovio, Intel, Salesforce, Nokia, Ericsson, ASML, Atlassian, Pinterest, Autodesk, and eBay, among others, enough to spawn a cottage industry of competing tech layoff trackers. Finance has followed a similar pattern, with significant cuts even at banks reporting record earnings. Yet the picture is not uniformly grim: many tech and finance firms continue to hire, and the sectors in aggregate remain large employers.

Each employer has its own circumstances, but a common thread runs through the recent announcements: artificial intelligence (AI). Some observers see AI as a catalyst that will create millions of new jobs. Others see a dystopian future of widespread unemployment, particularly among the college-educated. The truth is almost certainly somewhere in between, and this article argues that AI might be the rare technological innovation whose displacement effects fall disproportionately on highly educated workers.

AI has existed for more than a decade, but its economywide effects are only now becoming visible. This article examines what federal employment data reveal about those effects in the U.S. economy, and, equally important, what the data cannot yet tell us.

The central findings are four:

First, the federal data show tentative evidence of AI-related slowing in computer and finance employment, but nothing close to the dramatic displacement some forecasts envision. Between 2018 and 2022, employment in computer fields and finance grew more rapidly than the workforce as a whole. From 2022 to 2024, that pattern reversed. Of 16 three-digit occupation categories showing this reversal, only five (financial specialists, computer occupations, art and design workers, media and communications workers, and media and communications equipment workers) have a plausible AI connection. The combined estimated shortfall in these five categories was 321,000 to 441,000 jobs over two years, depending on the estimation method. This estimation is meaningful, but an order of magnitude smaller than the World Economic Forum’s projection of roughly 11 million U.S. jobs lost to AI between 2025 and 2030.

Second, Bureau of Labor Statistics (BLS) projections issued as recently as August 2025 still anticipate more rapid employment growth in computer and finance fields than in the economy as a whole over the next decade. Those projections might prove correct, but the layoff wave of early 2026, not yet reflected in government occupational data, suggests the forecasts might be revised downward. The more timely industry data, which extend through early 2026, show an elevation in information-sector layoffs that might be the earliest visible industry-level signal of the AI layoff wave.

Third, the industries showing employment patterns most consistent with AI displacement are narrower than the popular narrative suggests. The information sector overall shows the expected pattern; so do parts of finance, legal services outside law firms, and computer systems design. But law firms, graphic design services, and the performing arts do not. BLS’s Job Openings and Labor Turnover Survey (JOLTS) data show elevated layoffs in the information sector in early 2026, but little unusual movement in the broader economy.

Fourth, AI’s displacement effects, though real, differ from prior technological transitions in an important way. Earlier waves of technological change (mechanized agriculture, industrial automation, word processing) generally harmed workers with less education and rewarded those with more. AI appears to reduce demand for some highly educated workers, such as programmers, software engineers, finance professionals, lawyers, and accountants. Employment in these fields will not disappear, and education will retain substantial returns, but the nature of the displacement is unusual.

The purpose of this article is to examine what the federal employment data can and cannot tell us about AI’s effects on the U.S. labor market as of early 2026. The data can tell us that employment in certain occupations and industries has grown more slowly than would be expected based on pre-2022 trends, and that this pattern is concentrated in occupations and industries where AI exposure is plausible. The data cannot yet tell us how much of the slowdown is attributable to AI as opposed to other factors, and they cannot yet tell us whether the slowdown will continue, accelerate, or reverse. The analyses in this article are therefore preliminary. What seems clear, however, is that AI is the rare technological innovation whose displacement effects are likely to be concentrated among workers who have historically been least affected by technological change. That is a distinguishing feature of the current transition, and the United States is only beginning to understand its implications.