Google has published the first iteration of the AI & Economy ATLAS — Activity, Task, Landscape, and Adoption Study — built from 15 million aggregated, de-identified human-AI interactions across the Gemini App, AI Mode, and the Gemini API, products used by more than a billion people monthly. Version 1.0 spans more than 150 countries, 140 languages, 800 occupations, and 4,000 tasks.
The headline finding is a split between width and depth. Workplace adoption reaches every industry sector and 68% of all occupations, which together represent 90% of U.S. employment. Inside a typical job, though, AI is used for roughly 21% of tasks.
What it does there is mostly collaborate. The bulk of work interactions involve ideation, strategy, information retrieval, and learning. Non-routine cognitive tasks — creative design, hypothesis testing — show up in AI work interactions at nearly twice their share of the economy overall, 65% versus 35%. Fewer than 10% of those interactions fully automate a task.
Two findings cut against the knowledge-work assumption. Workers in manual and technical trades — auto technicians, industrial mechanics — use conversational AI as a live collaborator for diagnostics, troubleshooting, and on-the-fly learning, and are twice as likely to use multimodal AI when they do. And more than 86% of all interactions in the dataset happen outside work entirely: researching purchases, figuring out appliances, and grinding through high-friction administrative tasks like taxes, licensing, and fines — activity standard economic metrics do not capture.
The dataset covers Google’s own Gemini surfaces and excludes Workspace, Translate, AI Overviews, and enterprise platforms — a large share of economically relevant AI usage sits outside the study.
Read more at Google.