Google published the first edition of its AI & Economy ATLAS report on July 23, revealing that people are using the company’s AI tools primarily as a collaborator rather than a replacement for human work.
The dataset behind the study covers 15 million aggregated and de-identified human-AI interactions across the Gemini App, AI Mode, and the Gemini API. ATLAS v1.0 insights span more than 150 countries, 140 languages, 800 occupations, and 4,000 tasks.
What ATLAS is and how it was built
ATLAS stands for Activity, Task, Landscape, and Adoption Study, and is an ongoing, large-scale, de-identified study of how people are using Google’s AI products and tools.
The dataset analyzed 14.65 million de-identified interactions across the Gemini app, Google’s AI Mode search experience, and the Gemini API over a two-week period in April.
Automated systems classified interactions as work-related or non-work-related, then mapped work-related activity to 4,000 tasks across 800 occupations in 150 countries and 140 languages, with non-work tasks mapped to the categories used in the Bureau of Labor Statistics’ American Time Use Survey.
The analytical taxonomy was generated using Google DeepMind’s proprietary Observation Clustering and Taxonomy Organisation framework.
Broad reach, shallow depth
Workplace adoption spans all industry sectors and 68% of all occupations, collectively representing roughly 90% of U.S. employment, yet within a typical job, AI is used for only around 21% of tasks.
At work, most AI use is focused on collaboration and assistance with tasks, and task automation remains uncommon: the vast majority of AI interactions at work focus on collaborative uses such as ideation, strategy, information retrieval, and learning.
Tasks like creative design and hypothesis testing, categorized in ATLAS as “non-routine cognitive,” show up in AI work interactions at a much higher rate than in the economy as a whole (65% vs. 35%). Less than 10% of those interactions fully automate tasks.
Only 3% of occupations showed AI usage for over 75% of their tasks. Those findings align with how open-source coding tools have already embedded AI as an assistant in developer workflows rather than as a wholesale replacement.
Most AI use happens at home, not at work
More than 86% of interactions occur in personal contexts, with users relying on AI for activities ranging from researching purchases and operating household appliances to navigating government services, taxes, and licensing processes.
Although work-related activity comprises a minority of conversational AI use at 14% of total usage, the Gemini App, AI Mode, and the Gemini API together are used by more than 1 billion people monthly.
As of Google I/O in May 2026, the Gemini app serves more than 900 million monthly active users, while AI Mode surpassed 1 billion monthly active users.
The personal-use dominance also fits patterns visible in how companies are integrating AI across internal app development toolkits, where AI functions as a layer that helps workers rather than executing tasks end-to-end.
Limitations and what comes next
The study did not include usage data from Google Workspace, AI Overviews, Gemini for Google Cloud, or Gemini Enterprise. Google does not maintain logs for those products, and it is possible that additional data could show greater levels of automation.
Google confirmed that ATLAS will operate as a multi-year research program tracking AI’s macroeconomic trajectory.
The full report and dataset are available through Google’s official blog, alongside the underlying ATLAS v1.0 paper published by Google’s AI & Economy Research Program.
