Google announced a new interactive, open-access experience for its AI & Economy ATLAS on September 15, 2026, alongside research examining how scientists use AI. The updates explore AI adoption across workplaces, countries and scientific activities.
ATLAS shows distinct national patterns. In India, arts, design and media occupations account for 19% of work-related AI usage, or 1.6 times the global average. In the U.S., computer and mathematical occupations represent 30% of work-related AI usage—twice the share recorded in the rest of the world.
The new experience makes millions of ATLAS data points easier to explore. Users can examine AI usage across occupations ranging from electricians to purchasing managers, investigate household applications and compare adoption rates across countries.
Research from Google, Google DeepMind and MIT FutureTech also finds that scientists use AI more frequently than many other workers. Nearly half of surveyed scientists use some form of AI daily. Large language models and specialized models play complementary roles, and respondents report saving almost seven hours a week. However, constraints later in the research process are contributing to a growing queue of untested hypotheses.
How AI usage differs by occupation and region
In OECD countries, computer and mathematical occupations and business and financial operations lead AI usage. Outside the OECD, the leading categories are office and administrative support; arts, design, entertainment, sports and media; and educational instruction and library occupations.
National income generally tracks AI adoption, but there are exceptions. Brazil and the UAE have higher adoption rates than their GDP per capita would suggest.
Regional differences also appear in manual work, including real-time equipment diagnostics and troubleshooting. Manual tasks account for 7% of work-related AI usage in both Brazil and Germany, 1.4 times the global average, compared with 4% in Japan.
What scientists use AI for
The new scientific research draws on ATLAS data, an analysis of 2,600 specialized AI models and a survey of more than 600 scientists in the U.S. and U.K. It organizes scientific activities using a new taxonomy developed by MIT FutureTech.
Although both large language models and specialized models are widely used, their applications differ. LLMs such as Gemini are used across a broad range of scientific disciplines and task categories. Specialized AI models are relatively more prevalent in health and life sciences, and in domain-specific prediction, data generation and simulation.
Scientists report saving just under seven hours each week, making more time available for research. Those gains do not necessarily produce immediate increases in discoveries. The study finds substantial time devoted to checking AI outputs, more hypotheses awaiting tests, and bottlenecks in physical experimentation and clinical validation.
These findings suggest that, as in other occupations, realizing AI’s broader productivity potential may require changes to processes and workflows. Faster individual tasks alone may not translate into comparable gains in scientific output.
Further research planned
Google describes ATLAS as a long-term research effort. It plans to work with academic and other partners to identify further research areas and develop evidence on how AI is changing the economy.