AI reaches 68% of occupations but automates under 10% of tasks: Google

New Delhi: Google has released the first edition of its AI & Economy ATLAS, a large-scale study examining how people use artificial intelligence at work and in everyday life. The study is based on 15 million aggregated and de-identified interactions across the Gemini app, AI Mode, and Gemini API. According to Google, these products are used by more than one billion people every month. The ATLAS dataset covers users across more than 150 countries, 140 languages, 800 occupations and 4,000 tasks. The findings suggest that workplace adoption of AI is broad but remains limited to select activities within individual jobs. Google said AI tools were being used across all industry sectors and in 68% of occupations, which together account for around 90% of employment in the US. However, within a typical occupation, AI was used for only about 21% of tasks. Most workplace interactions involved assistance and collaboration rather than complete automation. Users primarily relied on AI for ideation, strategy, information retrieval, and learning. Less than 10% of workplace interactions resulted in the full automation of a task, the report said. Non-routine cognitive activities, including creative design and hypothesis testing, accounted for 65% of work-related AI interactions, compared with their estimated 35% share in the wider economy. The study also found that AI adoption was not limited to white-collar professions. Workers in manual and technical roles, including automotive technicians and industrial mechanics, were using conversational AI for diagnostics, troubleshooting and on-the-job learning. According to Google, workers in these occupations were twice as likely to use multimodal tools that can process or generate images and videos. Examples included interpreting test results, examining electrical wiring and checking machinery for wear. The report said more than 86% of AI interactions in the dataset occurred outside the workplace. People were using AI for household activities such as researching purchases, operating appliances and completing administrative tasks related to taxes, licences, fines and government services. Google said the economic value generated through these activities may not be captured by traditional measures of workplace productivity. ATLAS also found that AI usage had spread across countries representing 99% of the global population. English accounted for about one-third of conversations, while users continued to rely on their native languages for complex tasks. Per-capita AI usage broadly followed a country’s level of income, indicating that wealthier markets were generally adopting the technology faster. However, the report identified exceptions, with some middle-income countries in South America and the Middle East recording adoption levels comparable with higher-income economies. Google said the findings were generated using DeepMind’s Observation Clustering and Taxonomy Organisation tool, which categorises large volumes of unstructured conversational data. The company said personally identifiable and sensitive information was removed, while the remaining data was summarised and aggregated across groups of users. Google plans to expand ATLAS as part of a long-term research programme examining AI’s impact on workers, businesses and the global economy. The current dataset does not include usage from several other Google products, including Workspace, Google Translate, AI Overviews, Gemini for Google Cloud and Gemini Enterprise.