What 81,000 people want from AI
Anthropic AI Attitude Surveys
What 81,000 people want from AI
- Published: 2026-03-18
- Sample: 80,508 Claude users (159 countries, 70 languages)
Method and sample
- Over one week in December 2025, everyone with a Claude.ai account was invited to an interview.
- The interviewer was Anthropic Interviewer, a version of Claude prompted to conduct a conversational interview.
- 80,508 people took part, across 159 countries and 70 languages.
- Open-ended answers were categorized with Claude-powered classifiers.
What people hope for
- Professional excellence: 18.8%
- Personal transformation: 13.7%
- Life management: 13.5%
- Time freedom: 11.1%
- Financial independence: 9.7%
- Societal transformation: 9.4%
- Entrepreneurship: 8.7%
Inside personal transformation
- Cognitive partnership and collaboration: 24%
- Support with mental health: 21%
- Support with physical health: 8%
- Romantic connection with AI: 5%
Are people getting what they want?
- 81% said AI had taken a step toward their vision.
- Where it delivered: productivity 32.0%, cognitive partnership 17.2%, learning 9.9%.
- AI hasn't delivered: 18.9%.
What people are concerned about
- Unreliability: 26.7%
- Jobs and economy: 22.3%
- Autonomy and agency: 21.9%
- Cognitive atrophy: 16.3%
- Governance: 14.7%
- Respondents voiced 2.3 distinct concerns on average; about 11% expressed no concern.
Benefits and harms come in pairs
- Learning as a benefit 33%; cognitive atrophy as a harm 17%.
- Time-saving as a benefit 50%; illusory productivity as a harm 18%.
- Economic empowerment 28%; economic displacement 18%.
- People who value emotional support are three times more likely to fear dependence.
How views vary around the world
- Globally, 67% expressed net positive sentiment toward AI.
- No concern: Sub-Saharan Africa 18%, Central Asia 17%, South Asia 17%.
- Roughly double the rate in North America (8%), Oceania (8%) and Western Europe (9%).
- Learning matters more in Central Asia (14%) and South Asia (13%) than globally (8%).
- Concern about jobs and the economy was the strongest predictor of overall sentiment.
What it means
- Anthropic calls this a new form of social science: qualitative research at massive scale.
- Surveys and usage analysis show what people do with AI; open-ended interviews help explain why.
- The usefulness is real; the question is how to claim the benefits without incurring undue costs.
Source
- https://www.anthropic.com/features/81k-interviews
- Published: 2026-03-18
- All figures are as stated in the source. Summary by this site.