Google Releases 15 Million AI Interaction Data Points in First Global AI Usage Study

Google has unveiled its inaugural "AI & Economy ATLAS" research, offering an unprecedented look into how artificial intelligence is being used globally. The study analyzes 15 million de-identified AI interaction data points, aiming to build a long-term framework for understanding AI's influence on economies and labor markets worldwide.

The ATLAS initiative, an ongoing large-scale anonymous research project, examines de-identified interaction data from products like the Gemini App, AI Mode, and Gemini API. This allows Google to track the practical application of AI across different industries, job roles, and countries. The v1.0 dataset encompasses interactions from over 150 countries, 140 languages, 800 professions, and 4,000 job tasks, drawing from products that collectively serve over a billion users monthly.

Google highlights that ATLAS's unique strength lies in observing how people genuinely integrate AI into their work, daily lives, studies, and administrative tasks. This provides an early, comprehensive view of AI adoption and usage patterns. The company emphasizes that while AI holds immense potential to reshape the global economy and work, its ultimate impact requires societal participation and shaping, underscoring the importance of empirical data for informed policy, advocacy, and decision-making.

To manage the vast conversational data, Google DeepMind developed OCTO, an analytical tool that automatically organizes and categorizes unstructured text data, such as large language model dialogues, into distinct activities, tasks, and usage contexts. Google assures robust privacy protection measures, including the removal of all personally identifiable information, automatic deletion of sensitive content, dissociation of ATLAS data from user records, and summarization of text data into aggregated groups, ensuring no individual user can be traced.

ATLAS v1.0 is just the beginning, with plans to incorporate more AI products, application scenarios, and research methodologies. Google aims to expand the research scope and analytical capabilities to create a more complete database of the AI-driven economy. Future expansions will include AI products like Google Workspace, Google Translate, AI Overviews, enterprise platforms such as Gemini for Google Cloud and Gemini Enterprise, and cutting-edge technologies like Agentic AI and World Models, providing a clearer picture of AI's transformative economic impact.