Artificial intelligence is reshaping daily work in offices, labs, and studios worldwide. New visualizations in the AI & Economy ATLAS make millions of global data points easier to explore. You can see how professions and countries adopt AI, and how people use it at home. Examples from India and the U.S. show different paths across sectors and roles. New research also reveals how scientists apply models. They save time, yet face new bottlenecks. Ready to dive into the numbers and context?
What’s new in ATLAS, and why does it matter?
The answer is clear: a new interactive, open-access experience now simplifies ATLAS exploration. It turns millions of global data points into a few easy clicks. You can study AI adoption by profession, country, and at-home use cases. It works for quick overviews and deeper dives.
ATLAS lets you compare how frequently different occupations use AI. The range is wide, from electricians to purchasing managers. You can also explore how people apply tools at home. You see adoption across countries and across task types.
A couple of precise examples show the scale of change. India’s creative industry shows higher AI usage than the rest of the world. And the U.S. leads technical adoption, with strong contributions from computer and mathematical roles.
Creative occupations in India account for 19% of work-related AI usage — 1.6 times the global average. In the U.S., 30% of work-related AI usage is in computer and mathematical occupations — double the share in the rest of the world.
Why should you care? This format helps you quickly spot trends and differences. You can form hypotheses about growing segments and shifting roles. It supports planning for skills, investments, and organizational priorities. The data is open, and comparisons are transparent.
Which professions use AI most, and where?
ATLAS shows that AI adoption varies by occupation and region. In OECD countries, computer and mathematical roles, and business and financial operations lead. In non-OECD countries, office and administrative support, creative fields, and educational instruction and library roles are on top.
This view helps avoid overgeneralization. Some markets move through technical roles, others through service and creative work. Educational and library occupations also rank highly in non-OECD nations. That points to practical learning, search, and knowledge-structuring scenarios.
Another lens is income level versus adoption. In general, adoption correlates with a country’s income level. At the same time, some countries stand out. Brazil and the UAE show higher adoption than their GDP per capita would predict.
Regional differences also appear in manual tasks. Part of work-related AI usage goes to real-time equipment diagnostics and troubleshooting. In Brazil and Germany, that share reaches 7%, or 1.4 times the global average. In Japan, it is 4%.
What does this granularity deliver? It shows where AI captures routine and speeds manual workflows. This is not only about offices. It is about shop floors, production, and service operations. You see how hands-on work is changing on the ground.
How are scientists using AI in daily work?
New ATLAS-based research from Google, Google DeepMind, and MIT FutureTech brings fresh insight. It draws on an analysis of 2,600 specialized AI models and a survey of over 600 scientists in the U.S. and U.K. MIT FutureTech provided a new taxonomy that maps scientific tasks. That supports consistent analysis of roles and actions.
The results are clear: scientists use AI at a higher rate than many other jobs. Nearly half of surveyed scientists use some form of AI every day. They use both LLMs and specialized models. Their combination creates reinforcing effects.
Scientists are using AI at a higher rate than many occupations; nearly half use it every day. Both LLMs and specialized models are used widely for science, in mutually reinforcing ways.
LLMs like Gemini are spread widely across fields and task categories. They cover broad workflows and simplify many steps. Specialized models are relatively more common in health and life sciences. They stand out in domain-specific prediction, generation, and simulation tasks.
This pairing yields flexible workflows. LLMs support broad queries and drafting. Specialized models provide precision in narrow scenarios with specific data. Together they speed preparation, analysis, and early interpretation of results.
Productivity versus bottlenecks: what does the research show?
The direct answer: scientists report significant time savings from AI — almost seven hours a week. That frees more time for research. However, benefits do not always translate immediately into discoveries. New bottlenecks emerge further down the pipeline.
Scientists spend notable time validating AI outputs. That ensures quality and reliability of results. A backlog of hypotheses is building up, awaiting testing. Physical experimentation and clinical validation are also becoming choke points.
The pattern looks familiar across science and other occupations. AI offers strong potential for productivity growth. Yet large-scale impact on outputs requires redesigning processes and workflows. Steps must adapt to realize fast-evolving capabilities.
Does that signal a slowdown in progress? Not necessarily. It signals a need to rethink sequences and priorities. Where validation is critical, capacity should be planned in advance. That helps turn saved hours into proven outcomes.
What’s ahead for ATLAS and AI-economy research?
The answer remains open: many questions about AI and the economy persist. ATLAS is a long-term research project. The team will work with academic and other partners. The goal is to identify new areas and deliver insights on how AI transforms the economy.
That implies continued data collection and structuring. It also means deeper comparisons across professions, countries, and tasks. The tool’s openness improves access and verification. Users can see not only summaries, but also the shape of trends.
Should you expect quick answers? The data already suggests useful directions. It shows where demand for skills grows and which roles are being rethought. At the same time, methods and categories will keep evolving. That is the natural path of long-term research.
For practitioners, this is a call for careful, decisive action. Run experiments, but allocate time for validation. Use occupational and regional maps to plan training. Watch ATLAS updates to see changes in near real time.
Based on Google AI & Economy ATLAS.
