It is widely agreed that AI’s potential to transform the global economy and the way we work is significant. But the results – what it means for work, for people’s lives and, to a large extent, the economy – are not automatic or guaranteed. A lot has to happen. To get there, we as a society must work together to positively shape how AI affects our lives, jobs and finances. For this joint work to be effective, it is crucial to have a rich understanding of how AI is being adopted and used in the economy. Society needs empirical insight and evidence-based research to inform decisions, initiatives and actions.
To help, Google is launching the first iteration of the AI & Economy ATLAS (Activity, Task, Landscape, and Adoption Study), an ongoing, large-scale, de-identified study of how people use Google’s AI products and tools. ATLAS’ first dataset (v1.0) is built from 15 million aggregated and de-identified human-AI interactions across the Gemini app, the AI mode, and the Gemini API, which are collectively used by more than 1 billion people monthly. ATLAS v1.0 insights span more than 150 countries, 140 languages, 800 occupations and 4,000 tasks; ATLAS is the most comprehensive look yet at how real people use AI at scale.
ATLAS sheds light on how people use Google’s AI tools for various tasks at work and in their daily lives. The ATLAS v1.0 report provides an early view of a fast-moving landscape: AI’s capabilities are evolving, its uses are evolving, and tools to observe its impact on the economy are still evolving.
What do we learn from ATLAS v1.0?
Here are a few of the most interesting observations so far:
- AI use in the workplace is broad but shallow: Workplace adoption spans all industry sectors and also 68% of all occupations, which together represent 90% of total US employment. But within jobs, people use AI selectively: in a typical job, AI is only used for ~21% of tasks.
- At work, most AI use is focused on collaboration and assistance with tasks, and so far task automation is uncommon: ATLAS data shows that the vast majority of AI interactions in the workplace 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”) appear in AI work interactions at a much higher rate than in the economy as a whole (65% vs. 35%). Less than 10% of these interactions fully automate tasks.
- AI use is not limited to white-collar workers, it also helps workers in predominantly physical and manual occupations with related tasks: AI use for work is not limited to jobs traditionally seen as knowledge work. Although not as widespread, workers in manual and technical trades (e.g., auto technicians, industrial mechanics) use conversational AI as a live collaborator for real-time diagnostics, troubleshooting, and learning on the go. When workers in these areas use our AI tools, they are 2x more likely to use multimodal AI (ie, using AI to create images or video). For example, automotive technicians and industrial mechanics use AI to interpret complex test results, troubleshoot electrical wiring, and inspect machinery for wear.
- AI delivers value at home that can be missed in standard financial metrics, especially around high-friction administrative tasks: Over 86% of interactions with AI tools in ATLAS happen outside of work. People are using AI in new and interesting ways not captured by standard economic metrics, including productive household activities (e.g., researching purchases, help using appliances and tools) and high-friction administrative tasks (e.g., navigating public services such as taxes, licenses, and fines).
- Global AI adoption tracks GDP per capita, with notable exceptions: AI use has spread globally. ATLAS data shows the use of artificial intelligence in over 150 countries and territories, representing 99% of the world’s population. We also see this in the diversity of languages used in ATLAS. English represents only about a third of global AI conversations, and users do not systematically abandon their native languages for complex tasks. Looking deeper, the use of artificial intelligence per population closely approximates a country’s relative wealth level, raising concerns about a persistent digital divide. However, this is not a universal rule: some middle-income countries in South America and the Middle East are adopting artificial intelligence at rates comparable to higher-income countries.
Here are some additional results: