Anthropic Economic Index: Tracking AI's Role in the US and Global Economy (www.anthropic.com)

Anthropic’s third Economic Index maps how its Claude models are being used across US states and countries, and for the first time includes anonymized first‑party API traffic and an interactive dataset. Key geographic findings: after adjusting for working‑age population (the Anthropic AI Usage Index, AUI), wealthier, knowledge‑work economies (e.g., Israel, Singapore, DC) show disproportionately high Claude adoption; across countries a 1% higher GDP per capita is associated with a 0.7% higher AUI, while within US states a 1% higher per‑capita GDP correlates with a 1.8% higher use. Lower‑use countries tend to favor automation more than higher‑use ones, hinting at different adoption phases and potential global divergence effects from AI. Technical and economic implications: software engineering remains dominant (37–40% of conversations), but “knowledge‑intensive” uses are rising—education tasks up ~40% (9%→13%), physical/social sciences up ~33% (6%→8%) while management and finance shares fell. Interaction patterns shifted sharply toward automation: directive conversations rose from 27% to 39%, making automation the plurality at 49.1% vs augmentation 47%. API customers drive this trend most strongly—77% of sampled API traffic shows automation (mostly directive) versus ~12% augmentation—raising meaningful labor‑market and productivity implications as businesses embed Claude into workflows. Anthropic also published an interactive explorer and open dataset for further analysis.