Where Generative AI Saves Time, by Industry
Which industries actually get time back. Workers in information, company management, professional services and finance report saving three to four percent of their hours; public administration, utilities, mining and hospitality report around one. The pattern is intuitive — the tools help most where the work is already words and numbers — but it is now measured rather than assumed. Bars are three-quarter averages, because splitting a survey by industry makes single quarters bounce for sampling reasons alone.
What does it show?
The ordering is the finding: reported savings track how much of a job is already information handling, not how modern the industry is.
Methodology
Reported share of work hours saved through generative AI, by industry, from the Generative Artificial Intelligence Adoption Tracker built by Alexander Bick, Adam Blandin and David Deming from the Real-Time Population Survey and distributed through FRED. Twenty industry series are shown, matching the survey's sector breakdown. Each bar is a three-quarter average rather than a single observation. Splitting a quarterly survey by industry produces small sub-samples that move sharply for sampling reasons alone — one sector reads 2.3% of hours saved and then 0.7% within a year, which is noise rather than a change in how that sector works — so averaging is what makes the comparison legible. Industries with fewer employed respondents have the thinnest cells and the least reliable estimates. The series begins in 2024 Q3, so both averages sit inside a short window and the changes between them are small relative to the sampling error. Time saved is self-reported throughout. Read the ordering of industries as the finding and the individual movements as provisional.