anthropic.com
|
ksl
|
|
Anthropic ran what it calls the largest multilingual qualitative AI study ever – 80,508 Claude users interviewed across 159 countries in 70 languages, with Claude itself conducting the conversations and classifying responses. The top desired outcome was professional excellence at 18.8%, and 81% said AI had already advanced their personal goals. But the tensions are more interesting than the toplines: 26.7% cited hallucinations as their primary concern, 22.3% worried about job displacement, and 16.3% flagged cognitive atrophy – often the same people who reported productivity gains. Lower-income countries were significantly more optimistic than North America and Western Europe, with double the rate of respondents expressing zero concerns. Using your own model to interview users about your own model is a methodological choice worth noting. OpenAI and Google have relied on traditional surveys for user research – Anthropic’s approach generates richer data but introduces obvious selection and framing effects.
