Google opens AI & Economy ATLAS, scientists save 7 hours weekly

Google's AI & Economy ATLAS has been tracking how AI shows up in work around the world, and the new interactive experience is meant to let anyone explore its millions of global data points directly — from electricians to purchasing managers. The underlying patterns vary by region and income: in OECD countries, computer and mathematical and business and financial operations lead AI usage, while non-OECD countries show office and administrative support, arts and media, and education at the top. Adoption broadly tracks income level, though Brazil and the UAE exceed what their GDP per capita would predict, and manual-task usage such as real-time equipment diagnostics runs at 7% of work AI usage in Brazil and Germany versus 4% in Japan.

The accompanying study with Google DeepMind and MIT FutureTech, drawing on a new taxonomy mapping what scientists do, explains why AI's effect on science may not be immediate. Specialized models and LLMs reinforce each other across different tasks, and the reported time savings are substantial, but the same research finds significant time going into validating AI outputs and a growing pile of hypotheses waiting to be tested. That pattern — gains at one stage of a pipeline exposed as constraints at another — is what Google says makes science like most other occupations.

The stakes for research institutions and the teams running labs would seem to turn on the downstream stages: physical experimentation and clinical validation are the bottlenecks the research flags, and whether the saved hours become actual discoveries is likely to depend on whether those processes and workflows are redesigned around AI. Google frames ATLAS as a long-term project and says it will keep working with academia to identify where further research is needed.