Anthropic Economic Index
Confidence 0.75 · 2 sources · last confirmed 2026-04-28
A research initiative by Anthropic that measures real-world AI use through privacy-preserving analysis of Claude conversations on Claude.ai (consumer) and the 1P API (enterprise). Recurring report cadence (four reports through Jan 2026).
Stated mission (per Anthropic): provide ongoing, empirical measurement of how AI is changing tasks, occupations, and the labor market.
What it tracks
| Domain | Mapped to wiki concept |
|---|---|
| Task speedup, success, complexity | generative-ai |
| Automation vs. augmentation share | automation-vs-augmentation |
| Aggregate productivity impact | ai-employment-effects |
| Task-composition shift | ai-deskilling |
| Cross-country adoption | enterprise-ai-adoption |
| Task-horizon time scaling | ai-benchmarks |
Reports
| Report | Sample period | Status in this wiki |
|---|---|---|
| First | January 2025 | Not separately ingested; numbers cited |
| Second | (early 2025) | Not separately ingested |
| Third | August 2025 | Not separately ingested; numbers carried over |
| Fourth | November 2025 | Ingested — introduces “economic primitives” framework |
Economic primitives (introduced in fourth report)
Five measurements per conversation, derived by Claude classifying its own conversation samples:
- Task complexity — human time required without AI; whether multiple tasks were handled within one conversation.
- Human and AI skill level — years of education needed to understand prompts and Claude’s responses.
- Use case — work / education / personal.
- AI autonomy — degree of user delegation, from collaboration to fully directive.
- Task success — Claude’s own assessment of whether the task was completed.
See the fourth-report source page for definitions and applications.
Methodology notes
- Privacy-preserving — random samples (typically 1M conversations on Claude.ai + 1M API transcripts).
- Tasks mapped to O*NET taxonomy.
- Models change report-to-report — fourth report uses Claude Sonnet 4.5 predominantly; this affects comparability across editions.
Cited by external research
- Brynjolfsson et al. 2025 (Canaries) uses the Anthropic Economic Index automation/augmentation classification at the occupation level for Fact 3.
Open questions
- Earlier reports (1st through 3rd) are referenced indirectly through the fourth-report carry-over data; first-party ingestion of any prior report would clarify the longitudinal methodology.