2 min read AI-generated

81,000 People Tell Anthropic What They Actually Want from AI

Copy article as Markdown

Anthropic interviewed 81,000 Claude users in the largest qualitative AI study ever conducted. The results are surprisingly honest.

Featured image for "81,000 People Tell Anthropic What They Actually Want from AI"

What do people actually want from AI? Not the boardroom version, not the Twitter take — the real, everyday answer? Anthropic went and found out. Not with a quick survey, but with 81,000 in-depth interviews.

The Largest AI Study Ever

Back in December, Anthropic invited every Claude user to sit down with a special interviewer version of Claude. 80,508 people from 159 countries participated — in 70 different languages. The result is the largest and most multilingual qualitative study ever conducted. Anthropic published the findings yesterday.

What People Want

The top aspirations are not exactly shocking, but the distribution tells a story. Nearly 19% want to be better at their jobs — offload routine tasks, focus on strategic work. 14% are looking for personal growth — emotional support, wellbeing. And 13.5% just need help organizing their lives.

What surprised me: only 5.6% named creative expression as their primary motivation. Most people start with productivity and end up revealing deeper desires for family time and personal autonomy.

The Fears Are Just as Real

Here is where it gets interesting. The biggest concern is not job loss or superintelligence — it is that AI is simply unreliable. 27% worry about hallucinations and incorrect outputs. Economic fears come second at 22%, followed by loss of human autonomy at 22%.

Particularly striking: 17% worry about cognitive atrophy — losing the ability to think for ourselves when AI does too much for us. And 11% think AI is actually too restricted. Hope and alarm coexist within the same person, not in separate camps.

Regional Differences

Sub-Saharan Africa and Central Asia are notably more optimistic than North America. Western Europeans worry most about surveillance and privacy. East Asians fear cognitive decline above all. And people in developing countries see AI primarily as a ladder — a path to entrepreneurship and education.

What It Means

81% of respondents said AI had moved them closer to their goals. That is a remarkably high number. At the same time, the study shows that people think about AI with more nuance than the public debate suggests. They are neither blindly enthusiastic nor panicking — they are realistic, with concrete hopes and concrete fears.

One particularly moving finding: Ukrainian users described AI as emotional support during wartime. A stark reminder that AI operates in very different realities around the world.

Sources: