Gate News reports that on March 19, Anthropic released its largest AI qualitative research report to date. In December last year (December 2025), the team invited Claude users to participate in one-on-one interviews with AI interviewers, collecting 80,508 interviews answered in 70 languages from 159 countries. Anthropic claims this is the largest and most linguistically diverse qualitative study in history.
Users’ visions for AI are categorized into nine groups, with the highest percentage (18.8%) hoping AI will handle chores so they can focus on high-value work. The next most common are “personal growth” (13.7%) and “life management” (13.5%). 81% of respondents say AI has, to some extent, progressed toward their described visions, with time-saving being the most frequently mentioned realized benefit (50%).
Regarding concerns, unreliability (hallucinations, citation errors, etc.) leads at 26.7%, followed by employment and economic impacts at 22.3%, and loss of human autonomy at 21.9%. Worries about employment and the economy are the strongest predictors affecting overall AI sentiment.
67% of global respondents hold a positive attitude toward AI, but regional differences are significant. Latin America (Peru 82%) and Sub-Saharan Africa (Nigeria 81%) are the most optimistic, viewing AI as a lever to overcome barriers in capital and education. Western Europe (UK 63%), North America (USA 66%), and Oceania tend to be more negative, closely correlated with higher concerns about AI’s economic impacts. East Asia’s concerns are unique; worries about governance and monitoring are below the global average, while cognitive decline (18%) and loss of meaning (13%) are more prominent.
The report introduces the core concept of “Light and Shadow”: the same AI capability can bring both benefits and risks, often coexisting within the same individual. Those who expect emotional companionship from AI are three times more likely to worry about becoming dependent on it.