Executives report little net AI job change. Many still expect staffing shifts.

What the study measured

The summer 2026 survey covers firms in the U.S., U.K., Germany, Australia, Canada, Mexico and South Africa. Seventy-eight percent report using AI. The respondents are CFOs, CEOs and other senior executives asked about adoption, past effects and expected effects. The authors also provide an aggregate-data dashboard.

This October version expands the earlier four-country study to seven countries and nearly 9,000 respondents. It is a revision of the same research program, not a second independent estimate. Earlier figures should not be carried into the new version as though they describe the same sample.

What the near-zero net does not mean

The survey records executives’ attribution of past employment changes to AI, not independently verified hiring, separations or worker histories. A near-zero total can hide jobs gained at some firms and lost at others. More than 40% of firms anticipate a positive or negative AI employment effect; the paper’s abstract says these expectations roughly offset. It does not identify who would face a cut or whether those forecasts will be realized.

The finding can coexist with the Stanford–ADP young-worker payroll decline, which compares employment in occupations with different AI exposure, and with Yale’s broad U.S. result. They study different places, people, outcomes and measures of AI. The survey is closer to firms’ own reports of use, but its causal attribution is still unverified.

What would change the answer

Linked employer-adoption records and worker-level hiring, exits, hours and pay could test whether the reported aggregate net conceals losses for young or otherwise exposed workers. Only the October abstract was available for this review; detailed sampling and country weights remain to verify. This survey alone is not a reason to change jobs, buy training or make an investment decision.