The Anthropic Economic Index report: New building blocks for understanding AI use

Source: Anthropic Research Date: 2026-01-15 URL: https://www.anthropic.com/research/economic-index-primitives

Summary

This is the same Economic Primitives report as the January 2026 entry filed under “economic-primitives” — two signal files appear to cover the same report from different URL slugs. Five primitives (task complexity, skill level, purpose, AI autonomy, success) applied to 1M Claude.ai + 1M API conversations from November 2025. Augmentation (52%) exceeds automation (45%) on Claude.ai. Productivity estimates 1.8pp → 1.0-1.2pp after task success rate adjustment.

Implications

Duplicate coverage of the Economic Index measurement infrastructure thread. The augmentation > automation finding on Claude.ai (vs. 77% automation on API) is the key split: consumer Claude is used collaboratively, enterprise API is used to replace tasks. The deskilling concern (Claude covers higher-skilled task components within occupations) is the emerging concern — even when humans are still in the loop, the parts AI handles may be the parts that build expertise over time. Watch the augmentation/automation ratio as a leading indicator of how rapidly the labor displacement story moves from theoretical to empirical.