Anthropic maps AI hopes and fears through 81,000 user interviews

Anthropic published What 81,000 people want from AI on March 18, 2026, summarizing open-ended interviews collected through an AI interviewer inside Claude.ai. The company says 80,508 people took part across 159 countries and 70 languages, making it the largest and most multilingual qualitative study of its kind. The release is framed as an attempt to understand not just what AI can do, but what people want “AI going well” to look like in their actual lives.

Anthropic’s classification puts professional excellence first at 18.8%, followed by personal transformation at 13.7% and life management at 13.5%. Time freedom reached 11.1%, financial independence 9.7%, and societal transformation 9.4%. Anthropic also reports that 81% of respondents said AI had already taken at least one step toward the vision they described. A notable finding is that optimism and anxiety often appeared together: users described gains in productivity, learning, and health support while also voicing concerns about job loss, dependence, and broader social disruption.

Why it matters

• Major AI labs are starting to complement benchmark narratives with large-scale qualitative research about real user goals and tradeoffs.
• The findings suggest future AI products will be judged not only by raw capability, but by whether they improve wellbeing without creating new forms of risk or dependency.
• Anthropic is positioning the study as input for future Claude development and follow-up wellbeing research, which could influence product, safety, and policy decisions.

The study is not a population-representative survey. Participants were Claude users who chose to respond, and Anthropic used Claude-powered classifiers to organize the answers. Even with those limits, the work is still a useful signal because it combines unusual geographic breadth with open-ended testimony rather than narrow multiple-choice polling. For operators, regulators, and product teams, the report offers a more grounded picture of what active AI users currently value, fear, and expect from the next generation of systems.