How Is AI Adopted Across Countries and Occupations? Early Evidence from One Million Usage Conversations in Over 100 Countries

Abstract

How is AI adopted across countries and occupations? This paper provides the first large-scale early evidence, using a sample of one million Claude AI conversations from more than 100 countries during one week in February 2026. Three findings emerge. First, per-capita AI usage varies more than 200-fold across countries; income level, economic structure, scientific output, and AI-policy readiness jointly determine the level of AI penetration. Second, occupational breadth follows volume: countries with more total AI conversations use AI across more occupations. Third, countries naturally follow a global diffusion path, entering occupations that are structurally proximate to their existing AI-occupation portfolio and for which AI tools are widely available globally. Compute capacity, policy vision, and targeted sector-specific interventions determine which occupations adopt AI next.