The AI company called Anthropic, from anthropos, the Greek for human being, just used artificial intelligence to interview 81,000 people about artificial intelligence. There is something almost vertiginous about it.
An AI interviewer. Asking humans. About AI.
All across 159 countries, in 70 languages.
For a moment, let me set aside the findings and sit with the method. Gallup built an empire on the sample survey a few hundred voices standing in for millions. Ethnographers spent careers living inside communities to understand how people thought. Anthropology, was born from the conviction that you could only understand a society by immersing yourself in it. Now a machine has done something none of them could: hold adaptive, open-ended conversations with 81 000 humans simultaneously, in their own languages and make sense of what they said.
The result is not a statistic. It is something closer to a collective confession. Tens of thousands of people speaking about what they believe this technology might do to all of it. And perhaps because the interviewer was a machine, people were often more honest than they might have been with a human. The Anthropic interviewer asked four core open questions, listened to the answers and adapted its follow-up questions accordingly. Like a conversational anthropologist. The sample who took part was then analyzed by a second layer of AI, which coded and categorized their responses, identified themes and pulled representative quotes.
What the study found, at its core, is that people do not divide neatly into AI optimists and AI pessimists. They divide along the lines of what they need shaped above all by where they live and what their societies have, or have failed, to deliver.
The study’s geographic findings are stark. People in wealthy, AI-saturated regions, Western Europe, North America, Australia want AI to give them time to manage life complexities.
In developing and emerging economies, people want something categorically different. They want access to possibility.
The South East Europe (Western Balkans nested inside it) sits squarely at this second inflection point. Not yet rich enough to be asking AI for convenience, but asking it for something older and more urgent: a way out of economic traps that have resisted every other solution for thirty years.
The Regional Cooperation Council’s Balkan Barometer 2025, contains a single figure worth pausing on: 29% of Western Balkans citizens now use AI up from just 10% in 2024.
Nearly tripled. In one year.
The 29% of people now using AI are disproportionately young, educated and digitally connected precisely the population the region is hemorrhaging to Western Europe.
For them, AI opens two paths: stay home and use those skills to work remotely for foreign companies, earning competitive wages without crossing a border; or use those same skills to become more attractive to European labour markets and leave faster. Which path dominates will define the region’s trajectory over the next decade.
The divergence with the rest of Europe runs deeper than adoption numbers, however. It shows up in what people actually want from the technology, and in what they fear.
Across Western Europe, the dominant vision is life management followed closely by personal transformation and time freedom.
South-East Europe, tells a measurably different story: entrepreneurship and financial indipendence rank significantly higher than the European average, while life management drops.
A respondent in Copenhagen asks AI to clear her inbox so she can think more clearly. A respondent in Skopje or Tirana asks AI to help her build something from nothing a business, a skill set, an income stream that didn’t exist before. Both are asking AI to solve a problem.
The divergence is equally sharp on fears and more revealing.
Western Europe’s standout concern is surveillance and privacy at 17%, well above the global average of 13%, followed by governance gaps at 18–19%. These are institutional fears: anxieties about who controls the technology and how it might be turned against citizens by states or corporations.
South-East Europe, by contrast, indexes most heavily on unreliability and jobs, fears that are immediate, personal, and practical.
Concern about jobs and the economy was the strongest single predictor of overall negative AI sentiment across the entire study and it is precisely in regions like the Western Balkans, where unemployment sits above 10% and 68% of young people are already considering leaving, that this concern bites hardest.
One tension, however, is shared universally: cognitive atrophy
Whether you are a lawyer in Vienna or a student in Sarajevo, the fear of outsourcing your thinking to a machine and slowly losing the capacity to think alone crosses every border, every income level, every language. It is, perhaps, the most human concern of all. The Anthropic study found that educators were 2.5 to 3 times more likely than average to report witnessing declining independent thought in their students as a result of AI use. It is a classroom reality unfolding right now, largely unmonitored.
I guess we all agree. The capabilities that produce AI’s benefits are inseparable from the capabilities that produce its harms.
Western Europe is having a different conversation about AI one about the EU AI Act, regulatory alignment, sovereign AI infrastructure, and digital autonomy.
In the Western Balkans, what that reality shows is a population using AI as their best current tool for economic survival.
AI is studying humans, while humans are still trying to understand themselves.
The question is whether the technology that is giving people new reasons to hope will also give their societies the structural capacity to hold them. That is not a question AI can answer. It is a question for the rest of us.
