Anthropic Economic Index economic primitives
3 months ago
- #AI-economy
- #productivity
- #task-automation
- AI usage metrics reveal geographic variation and task complexity in interactions with Claude.
- Claude usage is concentrated in coding-related tasks, with top 10 tasks accounting for 24% of conversations.
- US, India, Japan, UK, and South Korea lead in Claude.ai usage, with adoption influenced by GDP per capita and workforce composition.
- Claude's success rate decreases with task complexity, impacting job exposure and productivity estimates.
- Higher-income countries use Claude more for personal and work purposes, while lower-income countries focus on educational use.
- Claude tends to handle higher-skill tasks, potentially deskilling some jobs while upskilling others.
- Productivity gains from AI are higher for complex tasks but are offset by lower success rates.
- Global AI adoption patterns suggest that higher education levels correlate with more effective AI use.
- API usage is more automation-dominant, while Claude.ai sees more collaborative, multi-turn interactions.
- The report provides comprehensive data on AI's economic impact, enabling further research and policy development.