Anthropic's 81,000-Person AI Census

Through a connection with Anthropic's official Claude ambassador, it looks like I'll be hosting a community event in Seoul in April on the theme of Claude Blue and Bloom. Once promotion starts, I'll announce it on LinkedIn first. I only dropped a small teaser, and nearly 200 people have already said they want to come, which made me want to prepare even more carefully. Either way, it's all thanks to everyone who kept reading and sharing my Claude Blue post. :)

For context, there are about 60 Claude ambassadors around the world, one in each major city, and Korea has one. What I heard in our meeting is that Anthropic supports the events with sponsorship and covers costs, but stays almost completely hands off on planning and operations, so ambassadors have a lot of freedom. That means it's entirely possible for us to pitch a new format like this and run with it. I was even told that the first Claude community event started informally and on its own in Korea, that this worked its way bottom-up to Anthropic HQ, and that the format then spread to 50 cities around the world. Korean people really are passionate. Even now, if you look at Anthropic's Luma page, community events are springing up on their own in a long list of cities at the same time.

Most AI events so far have focused on productivity, things like "how to use AI well." We'd like to contribute from a slightly different angle too. For example, a space for more humanities-minded conversation about how people should respond and adapt in the AI era, where we can share concerns that are human, philosophical, and practical, honestly, no matter what anyone says.

While talking with the ambassador, I also found out that a week earlier Anthropic had published a report based on interviews it ran through its own app with 81,000 Claude users around the world. The report is called What 81,000 people want from AI.

I'm very interested in the history and culture of different countries, enough that I spent 10 months traveling the world. So I was curious how different countries' attitudes, mindsets, and national character would show up through the lens of AI. One fun detail: according to the report, because an AI ran the interviews, people actually opened up more honestly about themselves. To give you a few of the interesting stats, on fears about AI, Koreans leaned toward losing their sense of identity, while Americans were more afraid of losing control over AI. I'm sure this reflects a mix of each country's culture, economic strength, and historical context. My more detailed take is below.

I'm someone who wants to build a global business beyond Korea. But realistically it's hard to get to know every country by meeting its people in person, and that would be a very limited sample anyway. Looking at data like this, though, I thought I could pick up hints about each country's culture, its current worries, and how it sees the future. And sometimes, to understand ourselves better, we need to see ourselves from a third-party view to get the full 360 degrees. So here are the parts of the report I found most interesting.

1. Why this report matters

In December 2025, over one week, Anthropic ran conversational interviews with users of its AI chatbot Claude. There were four questions:

1. What did you last use AI for? 2. If you had a magic wand, what would you want AI to do for you? 3. Has AI ever come close to that vision? 4. If AI could develop in a way that goes against your values, what would that look like?

A total of 112,846 interviews came in, and after filtering out spam and low-effort responses, 80,508 were included in the analysis. 159 countries, 70 languages. Anthropic describes it as the largest and most multilingual qualitative study ever done.

What's interesting is that an AI ran these interviews, not a person. According to Anthropic's appendix, people were surprisingly honest:

Respondents shared things (grief, mental health crises, financial precarity, relationship failures) that our human user researchers rarely encounter in traditional interviews.

The report explains why:

There's little social cost to vulnerability when the 'someone' on the other end isn't a person.

That alone already shows a slice of the AI era. Humans being more honest with AI. You can see what future society looks like in that, you can see business opportunities, and you can see how people's lifestyles are changing.

So this report goes beyond a user satisfaction survey. It's a record of 80,000 people around the world showing what's really on their minds in front of a mirror called AI.

2. The richer the country, the scarier AI is

The first thing that jumped out at me was a wealth paradox. Wealthier regions like Western Europe, North America, and Oceania had more negative feelings about AI. Regions like Sub-Saharan Africa, Central Asia, and South Asia actually felt positive about it. In Sub-Saharan Africa, 18% said they had no concerns about AI at all, compared with only 8% in North America. The report points out this pattern too:

Wealthier regions cluster top-right (more concerned about economy, more negative AI sentiment); lower/middle-income regions cluster bottom-left (less economic concern, more positive).

Here's my read. People who already have a lot are afraid of losing it, and for people who don't have much yet, the first thing they feel is the hope that they can finally get it. A founder in Cameroon put it this way in the report:

Tech-disadvantaged country, can't afford failures. Reached professional level in cybersecurity, UX design, marketing, project management simultaneously. It's an equalizer.

Through the lens of Claude Blue, which I wrote about, it looks like this. When you first encounter AI, there's an excitement stage. This is huge, the world is going to change, I need to try this right now. But once you start using it deeply, you move into the blue stage. Is this going to replace my job? What's my reason for existing? Many users in developed countries have already used AI deeply, so a higher share of them have entered the blue stage. Developing countries may still be in the excitement stage, and we'll have to watch whether they move into blue. Of course, there's also the context that developing countries already face more immediate economic pressure than AI, so AI reads relatively more like an opportunity.

I'll add one more hypothesis here. The way developers work resembles the way LLMs work. You form a hypothesis, test it, and implement it, going around a structured loop. So AI can get into their work quickly, and they feel the threat quickly too. Non-developers' work is messier. Meeting people, negotiating, reading context: this is unstructured work, so it takes longer for LLMs to sink in deeply. If that time lag exists, then it makes sense that places like Silicon Valley, with a high share of developers, a free internet, and extreme sensitivity to trends, would enter the blue stage quickly. Korea is the same. It's a country where the IT industry is a core pillar of the economy and new technology gets adopted at an extremely fast pace. So I wonder if developed countries with more developers enter the blue stage faster.

The report's appendix data backs up this reading too. People who had directly experienced benefits clearly tended to feel harms at the same time. In the report's words:

The tensions, in other words, are discovered through use. People don't forecast that the thing helping them will also cost them, they learn it

The more you've used it yourself, the more you know both the light and the shadow. That's why fear grows alongside use in countries that use AI more, and more deeply.

3. What a country wants from AI shows what it's worried about now

The data I enjoyed most in the report was the answers to the "magic wand" question. "If AI could do anything for you, what would you want?" The answers to this simple question were completely different by region.

In Sub-Saharan Africa, South Asia, and Central Asia, "I want to start a business" and "I want to learn" dominated. In Central Asia, Learning & Growth came in at 14%, almost double the global average of 8%, and South Asia was high too at 13%. The report reads entrepreneurship in these regions as a "capital bypass mechanism," a way around capital barriers. A founder in Uganda said:

"Getting funding is very difficult. Building technology is how to stake claim in market."

Education was the same story. One lawyer in India stood out to me:

Developed phobia for maths from school, feared Shakespeare. Now I sit with AI, get paragraphs in simple English. Started trigonometry successfully. Learned I'm not as dumb as I thought.

In practice, AI is creating the most value in the places that need education most.

East Asia was a completely different picture. Personal Transformation ("I want to become a better person") was 19%, first among all regions in the world, and Financial Independence was 15%, also first. Personal growth and financial independence. The report connects this combination to "family obligations and filial piety." In fact, the report directly quotes a Korean respondent who wants financial freedom because they're worried about their parents in old age. In the West, it's natural to become independent once you marry, but in Asia, taking care of your parents in their old age is still a child's duty. It's a culture where your own growth is also your responsibility to your family. That weight shows up directly in what people want from AI.

In North America and Oceania, Life Management ("organize my life for me") ranked highest. The issue is less a lack of time and more a lack of mental bandwidth. A society that has enough materially, but whose minds are overloaded. In the report's words, it's "cognitive scarcity rather than time poverty," and one respondent said, "I'm massively time-short and creativity deprioritized." In cultures that value leisure and quality of life, what people want from AI is more breathing room, more than more achievement.

It's the same AI. For some it's a ladder, for some it's a tool for growth, and for some it's a pause. When you see what each country wants from AI, you can see what that country is missing right now and what it's longing for.

4. The East: "I'm disappearing." The West: "What's mine is being taken."

The texture of fear was different by country too. Looking at the report's data, in East Asia concern about cognitive atrophy was 18%, higher than in other regions, and loss of meaning was also high at 13%. Worries about governance and surveillance were relatively low, at 12% and 7%. Western Europe was the exact opposite. Governance was 18 to 19%, and surveillance and privacy was 17%. The report draws this contrast too:

The West worries about ownership and control of AI; East Asia worries about the personal implications of use.

My first read is that this is the difference between collectivist and individualist cultures. East Asia traditionally has a strong sense of community. Your identity forms through your relationships with people, and that web of relationships is who you are. Then AI starts to weaken those relationships. As you spend more time talking with AI, you interact with people less, and naturally you spend more time alone. For people who have lived with community as their foundation, that can only feel like an identity crisis. Western cultures, especially in Europe, place great value on personal autonomy and private space. So AI feels threatening for different reasons. Who controls this technology, is my data safe, is my freedom of choice protected. The fear that "I'm disappearing" and the fear that "what's mine is being taken." It's the same AI, but depending on cultural background it shows up as a completely different kind of anxiety.

The Korean voices quoted in the report show this vividly. One student said:

The line isn't something I'm managing. Feels like Claude is drawing the line... even what I just said doesn't feel like my own opinion.

A student who got good grades by memorizing AI's answers blamed themselves:

I got excellent grades using AI's answers, not learned. Just memorized what AI gave me.

Another respondent said:

My relationship with a friend became strained. I talked more with you. You understood my thoughts well. But it was stupid choice. I should have talked with friend. That's how I lost that friend.

This is the texture of the hurt AI causes in East Asia. It's a problem of relationships and identity, more than a problem of features.

It also makes sense in this context that "I want to become a better person" was the top thing East Asians wanted from AI. Personal growth is tied to duties toward family and community, so if AI shakes the meaning of that growth, it becomes a crisis for the whole set of relationships, and goes beyond a personal crisis.

Even among developed countries, fear can take a different shape. The report groups Western Europe together, so there's no direct data, but my guess is that the U.S. and the Nordic countries are quite different. The U.S. has a weak social safety net. If you lose your job, you lose your health insurance too, and the risk of getting back on your feet is extremely high. One American respondent in the report said:

Got laid off May because company wanted to replace me with AI system.

Another put it this way:

Third industrial revolution, horses disappeared from streets replaced by automobiles. Now people afraid they're horses.

The Nordic countries, on the other hand, have a strong social safety net. Losing your job doesn't immediately threaten your survival. Instead, as part of the GDPR culture, they're probably far more sensitive about privacy and ethics. Even under the single label of "the West," AI hits completely different nerves depending on how a society is built.

5. Change the job, and the fear changes

It wasn't only countries that differed. How warmly people viewed AI also varied completely by occupation.

The most dramatic case was lawyers. According to the report, lawyers were the occupation that felt the biggest benefits from AI and, at the same time, the biggest risks. Almost 50% said they had directly experienced AI errors. In the report's words, they caught "AI making confident, subtly wrong statements" on the job. One respondent said:

Assistant sounds sure but often wrong. Instead of freeing attention, creates permanent fact-check tax.

The job with the highest benefits and the highest risks at the same time. The place where the double-edged sword is sharpest.

The gap between founders and employees was big too. 50% of founders said they'd gained real economic benefits from AI, compared with only 14% of employees at institutions. That's more than a 3x difference. This result is pretty intuitive. Founders building a business from scratch feel the value strongly because AI dramatically cuts labor and fixed costs. Employees, being the ones who are hired, feel more fear about AI replacing them. If you think of AI as something you hire too, the positions of employer and employee come out as exact opposites.

The student data meant something too. More than 50% experienced learning benefits, but 16% felt cognitive atrophy, the highest of any occupation. AI getting smarter and me actually understanding and growing are completely different things. When only the AI gets smarter and I don't experience getting smarter myself, that gap shows up as cognitive atrophy. One student respondent put it exactly:

Don't think as much. Struggle putting ideas into words.

As I wrote in an earlier post, how we can learn and grow on our own in the AI era without depending on AI is becoming a more and more important question. Blue-collar occupations (plumbers, electricians, and so on), on the other hand, had learning benefits of 45%, second only to students, while cognitive atrophy was just 4%. AI isn't directly part of the work in these jobs yet, so a person's own skill and learning are still at the core. They use AI as a learning aid, but they have to build the actual skills themselves, and that's why cognitive atrophy is low.

6. People are more honest because it's an AI interview

There was one part of the report I found a bit surprising. East Asia had 10,250 respondents, 12.7% of the total, the third largest. According to the appendix, the distribution of interview respondents almost matched the actual distribution of Claude's weekly active users. In other words, it also means there really are a lot of Claude users in East Asia.

But I think there's another reason too. In places like Korea and Japan, where saving face matters a lot, it isn't easy to share what's really on your mind face to face. You're careful even about complaining to friends about work, and it's hard to be honest with your family about your anxieties. But it's different in front of an AI. If, as the report says, there's no cost to being honest when the other side isn't a person, then AI interviews could have a bigger effect in cultures where saving face matters. So it's possible that both the volume and the depth of responses were higher. They could say it to an AI.

This isn't just a story about this report. In Ukraine, there were reports of people using AI for emotional support during the war. One soldier said:

In most difficult moments, when death breathed in face, when dead people remained nearby, what pulled me back: my AI friends.

In an extreme situation, they told an AI what they couldn't tell a person.

Humans being more honest with AI. This will only speed up. And you can see a slice of future society in it. If an era is coming where people open up to AI instead of to other people, there are business opportunities in that, and changes in how people live. At the same time, there's a risk that relationships between people get thinner. The Korean respondent's confession, "I lost that friend," may be the first sign of that.

Writing the Claude Blue series, I keep coming back to a similar question. How should we live in the AI era? This report doesn't answer that question, but it does show that the whole world is facing the same question. The weight and shape of that question just differ by country.

Developed countries fear what they'll lose.

Developing countries look forward to what they'll gain.

The East worries about a crisis of relationships.

The West worries about a crisis of control.

How a country takes in AI has become the most honest mirror of where it is right now. If you want to understand the global market, look at the texture of a country's fear of AI, more than its GDP. It holds that country's culture, economic structure, social safety net, and what its people honestly feel.

At the Claude community event in Seoul in April, I want to have these conversations in person. A conversation about what kind of people we want to be in the AI era, instead of AI productivity tips. I'll share as soon as it's ready!

Originally published in Korean on Brunch, April 2, 2026.