What 81,000 people told us about the economics of AI

Source: Anthropic Research Date: 2026-04-22 URL: https://www.anthropic.com/research/81k-economics

Summary

Open-ended survey of 81,000 Claude users; Claude-powered classifiers infer occupation, career stage, productivity gains, and job concerns from free-form text. Key findings: workers in high-exposure occupations express 3x more displacement concern than low-exposure roles; early-career workers significantly more anxious than senior professionals; 48% cite scope expansion vs. 40% speed as primary benefit; U-shaped anxiety curve — both severe slowdowns and massive speedups correlate with high job threat concern.

Implications

Connects the Economic Index measurement thread to ground-level worker sentiment. The 3x concern differential in high-exposure occupations validates that workers’ intuitions track the actual usage data — people are worried about the right jobs. The early-career anxiety finding is consistent with the Economic Index learning curves data (younger worker hiring slowdowns) and is the most politically salient result. The U-shaped anxiety curve is analytically interesting: extreme speedups cause concern, not just slowdowns, likely because rapid capability expansion makes workers uncertain about their future role. The Claude-classifying-Claude methodology is efficient but worth noting as a potential self-report bias vector — watch for independent validation.