Google’s AI & Economy ATLAS project has launched a new, interactive, open-access experience to make it easier to examine millions of data points showing how AI is transforming working life around the world. Users will be able to explore usage rates across occupations ranging from electricians to purchasing managers, how AI is used in homes, and adoption levels across countries.
A new study by Google, Google DeepMind and MIT FutureTech, based on ATLAS data, also examined scientists’ use of AI. The research is based on an analysis of 2600 specialized AI models and a survey of more than 600 scientists in the US and the UK. According to the findings, scientists use AI at a higher rate than people in many other occupations, with around half using an AI tool every day.
While LLMs such as Gemini are widespread across scientific fields and tasks, specialized models stand out in health and life sciences, as well as in forecasting, generation and simulation. AI saves just under seven hours per week. However, the need to verify outputs, hypotheses awaiting testing, and bottlenecks in physical experimentation and clinical validation may limit its impact on discoveries.
Why it matters
Open-access visualizations make it possible to compare not only the overall prevalence of AI use, but also differences across occupations, households and countries. This distinction may help reveal which jobs and regions are seeing the greatest concentration of economic transformation. The findings on scientists, meanwhile, show that widespread use does not rely on the same tools or tasks in every field. Weekly time savings do not directly translate into faster scientific discovery because of bottlenecks in verification, physical experimentation and clinical processes. In addition, the survey’s focus on scientists in the US and the UK leaves open the question of how well these results represent researchers worldwide.
Background
Google is not a new name in the FikirPilot archive: over the past 90 days, we have published 36 articles mentioning the name; the latest was dated September 13, 2026.