What 81,000 people told us about AI economics

Anthropic AI Attitude Surveys

What 81,000 people told us about AI economics

  • Published: 2026-04-22
  • Sample: About 81,000 Claude users (re-analysis of the December 2025 interviews)

Method and sample

  • The December 2025 interviews with 81,000 people were re-analyzed for economic themes.
  • Respondents were users of Claude.ai personal accounts.
  • Occupation, career stage and sentiment were inferred from answers with Claude-powered classifiers.
  • The study related each occupation's AI exposure to worry about job displacement.

Key findings

  • People in roles more exposed to AI have more concern about job displacement.
  • These concerns are higher among early-career respondents.
  • The highest- and lowest-paid occupations report the largest productivity gains.
  • Those with the largest speedups express higher concern about displacement.

Who worries about displacement

  • One fifth of respondents voiced concern about economic displacement.
  • For every 10-point increase in exposure, perceived job threat rose 1.3 points.
  • The top 25% by exposure mentioned the worry three times as often as the bottom 25%.

Reported productivity

  • The mean productivity rating was 5.1, corresponding to substantially more productive.
  • 3% reported negative or neutral impacts.
  • 42% gave no clear indication on productivity.

Who benefits

  • Most cited benefits to themselves: faster tasks, expanded scope, freed-up time.
  • 10% of those who named a recipient said employers or clients were asking for and getting more work.
  • Only 60% of early-career workers indicated they personally benefited, versus 80% of senior professionals.

Scope or speed

  • The most common gain was in scope, cited by 48% of users who mentioned productivity.
  • 40% emphasized speed.
  • Doing new tasks was mentioned more often than doing tasks faster.

What it means

  • Benefits and worries sit side by side: those who gained the most speed worry more.
  • Early-career workers report fewer benefits and more concern.
  • Caveats from Anthropic: self-selected Claude.ai users only, occupation inferences may be inaccurate, and findings need confirmation through structured surveys.

Source

Back to list