Anthropic maps AI sentiment in 81K interviews

This is less a product launch than a reality check for the AI industry: people do want AI, but mostly as a helper for human problems, not a pure productivity toy. The biggest signal is that hope and fear coexist in the same users, which is exactly why trust matters more than raw capability.

• –Unreliability is the top concern at 26.7%, which says verification burden is still a core product problem, not a footnote.
• –The strongest desires cluster around personal transformation, life management, and time freedom, so the winning use cases look more like assistants than widgets.
• –Jobs and economy concerns at 22.3% and autonomy concerns at 21.9% show why deployment strategy shapes public sentiment as much as model quality does.
• –Regional sentiment tracks economic anxiety: wealthier regions are more negative, while lower- and middle-income regions are more optimistic.
• –The dataset is huge, but it is still a Claude.ai user sample, so it captures real usage at scale without being a random sample of the global public.