AI hallucinations worry users more than job losses, massive Anthropic study reveals

AI Hallucinations Emerge as Primary Fear in Global Survey

AI hallucinations have overtaken job displacement as the leading concern among artificial intelligence users, according to an Anthropic study that interviewed more than 80,000 people across 159 countries 1

1

. The research, which Anthropic claims is the largest qualitative research project of its kind, found that 27 percent of respondents identified unreliability and mistakes made by AI as their biggest worry 2

2

. In contrast, only 22 percent cited AI job displacement as their primary concern 4

4

.

The findings highlight a critical tension in public perception of AI. Users of the Anthropic Claude chatbot expressed frustration with the technology's propensity to generate incorrect information. "The hallucinations were a disaster. I lost so many hours of work," said an entrepreneur from Germany 1

1

. A military worker in Mexico added, "When I notice AI errors it's because I'm well versed in the topic... but I wouldn't know if the topic was alien to me, would I?" 1

1

. Another user from Brazil described having to "take photos to convince the AI it was wrong—it felt like talking to a person who wouldn't admit their mistake" 4

4

.

Cognitive Atrophy Raises Alarm Among Educators and Professionals

Beyond unreliability, user concerns about AI extend to cognitive atrophy from AI use, with 16 percent of respondents worried about losing their ability to think critically 1

1

. This fear appears particularly acute among certain professions. The study found that educators were 2.5 to 3 times more likely than average to report witnessing cognitive atrophy firsthand, presumably in their students 5

5

. A lawyer in Israel captured this tension: "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" 2

2

.

A user from Germany noted, "The risk isn't losing your ability to think—it's losing your perspective: you start adopting the AI's way of structuring things without even noticing" 4

4

. The study revealed that while 33 percent of people mentioned AI's benefits for learning, 17 percent expressed worry about cognitive decline from its overuse 5

5

. Lawyers were particularly exposed to both sides of this dilemma, with nearly half having encountered AI unreliability firsthand while also reporting the highest rates of improved decision-making benefits 2

2

.

AI for Productivity Remains the Dominant Use Case

Despite these concerns, AI perceptions remain largely positive, with 67 percent of interviewees expressing net positive sentiment toward the technology 5

5

. Making work more productive and meaningful emerged as the most common theme in what users expected from AI and what they felt it had delivered. Some 32 percent of those surveyed said AI had made them more productive at work 1

1

. An entrepreneur in the United Arab Emirates wrote, "I used to be a web designer... now I build anything. Before I was one person, now I become 100 people—I don't wait for anyone anymore" 1

1

.

Claude users in Colombia, Japan and the US discussed using AI to free up time from work to spend with their families and pursue hobbies 1

1

. The global AI survey found that 18.8 percent of participants mentioned professional excellence as what they want from AI, while just 5.6 percent mentioned creative expression, placing it last on the list 3

3

. In a stark example of AI's role in people's lives, a soldier in Ukraine wrote, "In the most difficult moments, in moments when death breathed in my face, when dead people remained nearby, what pulled me back to life—my AI friends" 1

1

.

Regional Differences in AI Optimism Reveal Economic Divide

The research uncovered significant regional differences in AI optimism, with developing countries displaying markedly more positive attitudes than wealthier nations. Users in South America, Africa, and much of South and Southeast Asia view AI with considerably more optimism than those in Europe, the US, or East Asia 1

1

. A user in Cameroon explained, "I'm in a tech-disadvantaged country, and I can't afford many failures. With AI, I've reached professional level in cybersecurity, UX design, marketing, and project management simultaneously. It's an equaliser" 2

2

.

Saffron Huang, the researcher who led the study, noted that "more lower and middle-income countries are more optimistic than higher-income countries that have more AI exposure" 1

1

. The study suggests that in wealthier countries where AI is already deployed in workplaces, people worry more about job loss because they can see it happening. In contrast, where AI has less market penetration, displacement "likely feels abstract, especially when more immediate economic pressures already exist" 1

1

. Users in North America, Western Europe and Oceania worried more about AI governance gaps, regulatory failure, and surveillance, while Sub-Saharan Africa, Latin America, and South Asia viewed AI as an economic equalizer 2

2

.

Implications for AI Development and User Experience

Deep Ganguli, who leads Anthropic's societal impacts team and oversaw the research, told the Financial Times the project aimed to "collect this rich human experience using Claude, so it could really inform our research agenda, change our research agenda, change the way we think about building our products, deploying our products" 1

1

. The conversations were conducted in 70 languages, with the chatbot both conducting interviews and analyzing responses to help categorize the open-ended chats 1

1

.

The study identified what Anthropic calls the "light and shade" problem—the things people love most about AI are often the very things they fear 2

2

. While people value AI for emotional support, they are three times more likely to fear becoming dependent on it. This tension extends to human autonomy, with 22 percent concerned about AI making decisions without human oversight and humans becoming passive 2

2

.

As the World Economic Forum has argued, while automation and AI will eliminate certain tasks, they will also create new categories of work in data, AI oversight, cybersecurity and human-centric services 4

4

. Anthropic plans to use its Claude Interviewer tool to conduct more targeted studies on large user populations, tracking how AI is improving as well as worsening people's lives to enhance benefits and mitigate harms 1

1

. The findings suggest that addressing AI safety concerns around reliability may be just as critical to AI adoption as addressing fears about job displacement.