Introducing the Anthropic Economic Index

Source: Anthropic Date: 2025-02-10 URL: https://www.anthropic.com/news/the-anthropic-economic-index

Summary

Anthropic launched the Economic Index, analyzing AI’s labor market effects through millions of Claude.ai conversations. Key findings: AI usage concentrates in software development and technical writing (37.2% of queries); augmentation (57%) outpaces automation (43%); mid-to-high wage occupations (programmers) show heaviest adoption; both lowest and highest-paid roles show minimal AI use. Dataset open-sourced on Hugging Face. Positioned as “first-of-its-kind data” on real-world AI-labor integration.

Implications

• Safety/policy / labor economics thread. The Economic Index is Anthropic’s primary defense against “AI is destroying jobs” narratives — publishing that augmentation (57%) beats automation (43%) in actual usage data is a directly policy-relevant finding.
• Mid-wage concentration. AI use concentrated in mid-to-high wage tech roles (not lowest-wage workers) inverts the fear narrative that AI hits the most vulnerable first. The data suggests the opposite — high-skill, high-wage software development is the primary beneficiary so far.
• Open-source dataset. Publishing the dataset on Hugging Face enables external researchers to independently verify Anthropic’s findings — and to find things Anthropic didn’t highlight. This is genuine transparency with accountability risk.
• Policy leverage. The Index has been cited in Anthropic’s Paris AI Summit statement, OSTP submission, and UK MOU scope. It functions as proprietary research that Anthropic deploys in regulatory contexts where it has an interest in a specific narrative about AI’s economic impact.
• Watch: subsequent Index releases (does the augmentation/automation ratio shift as models get more capable); what external researchers find in the open dataset; whether the Index’s findings are cited in Congressional testimony or academic literature.