Last December, Anthropic did something unusual. Instead of running another survey with multiple-choice questions, they built an AI interviewer into Claude.ai and invited users to have an actual conversation about AI—what they use it for, what they dream it could do, and what keeps them up at night.
Nearly 81,000 people took them up on it. Across 159 countries, in 70 languages. That’s not just a big survey; it’s the largest qualitative study of its kind ever conducted, and it’s the first time AI itself made that scale possible.
Hope and alarm, not hope vs. alarm
The headline finding, if you want one: people aren’t neatly split into “optimists” and “doomers.” Most people carry both hope and fear in their heads at the same time. A lawyer from Israel put it perfectly: “I use AI to review contracts, save time… and at the same time I fear: am I losing my ability to read by myself? Thinking was the last frontier.”
That tension runs through almost every response. People want AI to handle the boring stuff so they can focus on meaningful work, but they’re also worried that offloading too much will atrophy their own skills. It’s not a contradiction—it’s a tightrope.
What people actually want
Anthropic classified responses into primary categories. The top one, at 18.8%, is professional excellence. That’s the “AI as my tireless assistant” crowd—people who want to offload routine tasks and focus on higher-value strategic work. A healthcare worker in the US described getting 100-150 texts per day from doctors and nurses, and how AI lifted the documentation burden: “I have more patience with nurses, more time to explain things to family members.”
Personal transformation came in at 13.7%. These are people using AI as coach, therapist, or guide. One respondent from Hungary said “AI modeled emotional intelligence for me… I could use those behaviors with humans and become a better person.” That’s a use case that doesn’t get much airtime in the usual productivity-focused AI discourse.
Life management (13.5%) and time freedom (11.1%) round out the top four. People want AI to handle the cognitive overhead of modern life—schedules, mental load, executive function support—so they can be present with their families. A manager from Denmark said “If AI truly handled the mental load… it would give me back something priceless: undivided attention.”
The fear is real, and specific
The concerns are more scattered, which makes sense—people tend to articulate multiple distinct worries. Job displacement is the obvious one. A technical support specialist from the US said “I got laid off from my job in May because my company wanted to replace me with an AI system.” That’s not abstract; that’s a paycheck.
But there’s also a deeper, harder-to-articulate anxiety about what happens when we build something smarter than ourselves. A software engineer from South Korea said “Humanity has never dealt with something smarter than itself. We need to reflect on how to prepare for the AI age.” That’s not a complaint about losing a job; it’s a question about losing control.
The methodology matters
What makes this study interesting isn’t just the scale—it’s the approach. Traditional qualitative research forces a tradeoff between depth and volume. You can interview 50 people in depth, or you can survey 50,000 people with shallow questions. Anthropic used Claude itself to conduct conversational interviews, then used more Claude-powered classifiers to categorize the responses.
That’s a smart use of the technology, but it also means the results inherit whatever biases Claude has. The study was limited to Claude.ai users, which skews toward people who are already comfortable with AI. The interview was conducted by an AI, which might affect how people express sensitive concerns. Anthropic acknowledges these limitations in the appendix, and they’re real.
Still, 80,508 people in 70 languages is a lot of signal. The Quote Wall they built is worth browsing—filter by region, concern, or vision. It’s the closest thing we have to a global conversation about what AI should and shouldn’t be.
What’s missing
For all the richness of the data, I wish they’d pushed harder on one thing: what people want AI not to do. The study asked about fears, but the responses mostly stayed in the lane of personal consequences—job loss, skill atrophy, dependency. There’s almost nothing about systemic risks like disinformation, inequality, or political manipulation. Maybe that’s because the users surveyed are already in the AI bubble. Or maybe it’s because those risks are harder to feel in your daily life than the fear of losing your job.
Either way, this is a useful snapshot of where regular users—not pundits, not CEOs—actually stand. They want AI to be a tool, not a replacement. They want to use it without losing themselves. And they’re paying attention, even if they don’t have all the answers.
Comments (0)
Login Log in to comment.
Be the first to comment!