Anthropic asked 81,000 people what they want from AI

Anthropic interviewed 80,508 Claude.ai users across 159 countries in 70 languages during December 2025, which it calls the largest multilingual qualitative study of its kind. The method is part of the story: an AI conducted the open-ended interviews and AI classifiers grouped the answers, which is how a study this size is even possible.

People's stated goals sorted into nine groups, led by professional excellence at 18.8%, personal transformation at 13.7%, and life management at 13.5%, with financial independence, societal change, entrepreneurship, learning, and creative work filling out the rest. The sentiment numbers are the interesting tension. 81% said AI had moved them toward their stated goal and 67% reported net positive sentiment, with people in developing regions notably more optimistic than those in wealthy countries. At the same time 26.7% raised unreliability and hallucination as a concern. Read the full study on Anthropic's site.

Why it matters

If you build AI products, the split is the actionable part: strong optimism about productivity and learning sitting next to a quarter of users worried about trust. That is not a contradiction to resolve in a blog post, it is the exact gap your product roadmap has to address.