A survey found that the biggest source of anxiety for artificial intelligence (AI) users is not job replacement but AI errors (hallucinations).
On the 22nd, according to the Financial Times (FT), Anthropic surveyed more than 80,000 Claude users in 159 countries and found that 27% of respondents cited AI hallucinations as their top concern. Worries about job replacement and infringement on human autonomy followed at 22%, and concerns about deteriorating critical thinking skills at 16%.
The survey was conducted in 70 languages and designed so that Claude directly conducted interviews and analyzed responses. Anthropic said the goal is to use it to inform research direction and product development by reflecting real user experiences.
Productivity gains were most frequently cited as a positive effect of AI. Thirty-two percent of respondents said AI improved their work efficiency, and users in some countries said they used the time saved for family, hobbies, and creative activities. By contrast, about 19% said AI outcomes fell short of expectations.
Regional perception gaps also stood out. Users in South America, Africa, and Southeast Asia were more optimistic about AI than those in Europe, the United States, and East Asia. The research team analyzed that in low- and middle-income countries, where AI adoption at work is relatively low, concerns about job replacement may be perceived as less realistic.
Based on the findings, Anthropic plans to continue follow-up research to expand AI's positive impacts and reduce negative ones.
However, the methodology drew criticism. A large share of respondents were concentrated in North America and Western Europe, and some regions had small samples, raising questions about representativeness. There were also doubts about the study's reliability due to the short survey format and selection bias in participant recruitment.