New insights from Google’s AI & Economy ATLAS

Google has released a new interactive, open-access way to explore data from its AI & Economy ATLAS, a long-term research project. The launch was announced by Zanna Iscenko, AI & Economy Lead in the Chief Economist's Office, and Scott Strand, Head of StratOps and Special Projects for Technology & Society. The experience lets users examine AI usage rates by occupation, at home, and by country.

The ATLAS data show regional patterns. In India, arts, design and media occupations account for 19% of work-related AI usage, 1.6 times the global average. In the United States, computer and mathematical occupations make up 30% of work-related AI usage, double the share elsewhere. In OECD countries, computer and mathematical plus business and financial operations lead, while non-OECD countries are led by office and administrative support, arts/design/media, and educational instruction and library roles.

Adoption generally tracks national income, though Brazil and the UAE adopt AI at higher rates than their GDP per capita would predict. For real-time equipment diagnostics and troubleshooting, Brazil and Germany devote 7% of work AI usage to manual tasks, 1.4 times the global average, versus 4% in Japan.

New research from Google, Google DeepMind and MIT FutureTech analyzed 2,600 specialized AI models and surveyed more than 600 scientists in the U.S. and the U.K., organized with a new taxonomy mapping scientific work. Nearly half of surveyed scientists use some form of AI every day, a higher rate than many occupations. LLMs such as Gemini are spread across fields and task categories, while specialized models are relatively more common in health and life sciences and in domain-specific data prediction, generation and simulation. Scientists report saving just under seven hours a week, but the research also finds time spent validating AI outputs, a growing backlog of untested hypotheses, and bottlenecks in physical experimentation and clinical validation.

Why it matters: Developers and product teams can now access ATLAS data interactively, while the research indicates that realizing AI's productivity potential in science may require redesigning workflows around validation and experimentation bottlenecks.