Anthropic surveys 80,508 Claude users about hopes and fears on AI

In short: Anthropic ran a large survey of Claude users to understand what people expect from AI and what they worry about.

What's going on

Anthropic said it interviewed 80,508 people who use Claude in December 2025. Respondents were in 159 countries and spoke 70 languages. The company used an AI tool it calls the “Anthropic Interviewer” to ask open questions and then ask follow-ups based on each person’s answers.

This approach is closer to a long conversation than a multiple-choice poll. Think of it like a human interviewer who listens and asks the next question based on what you just said, but here the interviewer was software. Anthropic describes it as the largest qualitative study of AI users, meaning it focused on detailed written responses rather than just numbers.

When users described what they want from AI, the most common themes were practical. About 19% wanted to do better at work, for example by getting specialized help in areas like law, accounting, business, and research. Another 14% focused on personal growth, such as talking through problems, getting advice, or using AI like a tutor or business partner.

Anthropic said 81% felt AI had delivered on their goals at least somewhat, often by helping them finish tasks faster or learn more quickly. The remaining 19% said current tools can be too obedient, and may not take initiative or challenge the user’s thinking.

What to watch

The biggest worries were about reliability and knock-on effects. In the survey, 27% mentioned unreliable results, 22% worried about jobs and loss of human control, and 16% feared becoming less capable by leaning on AI too much. How AI companies respond to these concerns may shape whether more people keep using tools like Claude or stop after early experiments.

Source: Financial Times