Who’s most optimistic about AI — and who isn’t, according to Anthropic

People in Sub-Saharan Africa and Asia are more optimistic about artificial intelligence than those in Western Europe and North America, according to a report by Anthropic that surveyed around 81,000 people in 159 countries.

The study, published Wednesday, revealed how economic gains from AI usage formed the main aspiration for most respondents, but analysts also warned that not everyone stands to benefit equally.

Anthropic researchers invited users of its Claude large language model to participate in conversations centered around questions about usage habits, hopes and fears over the development of AI.

These conversations, held using Anthropic Interviewer — a variant of Claude trained to conduct interviews — were subsequently also analyzed with Claude. First to filter out “spammy, unserious, or extremely minimal” responses, then for classifying and tagging responses by sentiment.

Respondents reported having both the highest hopes for AI — and seeing its greatest benefits — in their workplaces.

According to the report, 18.8% of respondents sought “professional excellence” from their use of AI. Similarly, 32% reported that AI was most useful for boosting productivity.

Most productivity gains, according to Anthropic, involved respondents outsourcing more mundane tasks to be able to “focus on strategic, higher-level problems.” Others said AI helped to free them up for pursuits beyond work.

Some analysts were unsurprised by these sentiments, as they said the present stage of AI development suited more menial applications.

“At the moment, AI is best suited to highly repetitive, narrowly focused, goal-oriented use cases … similar to specific tasks on an assembly line,” Lian Jye Su, chief analyst at Omdia wrote in an email to CNBC.

More specifically, these applications often include administrative tasks like “HR, billing, and other backoffice functions,” according to Seema Shah, vice president of insights from the market intelligence firm Sensor Tower in an email to CNBC.

The financial spoils of AI also seemed to favor an entrepreneurial class, as independent workers — which includes entrepreneurs, small business owners, and those with side gigs — experienced more than triple the rates of economic empowerment from AI usage over salaried employees, according to Anthropic.

But recent developments have also shown that ostensibly higher-order work may be vulnerable to many of the same disruptions.

After Anthropic launched Cowork in February — a Claude variant capable of handling more complex tasks like financial modeling and data management — stocks of companies ranging from software to research firms saw a broad selloff as investors were spooked by the implications of these launches.

As companies like Anthropic and Alibaba invest billions into agentic AI, developing models now able to perform actions autonomously with limited user supervision, it may become even harder to tell how professional lives are set to be disrupted.

“These agents are going to do increasingly sophisticated tasks on behalf of people, and that is going to have massive impacts,” said Marc Einstein, research director at Counterpoint Research, in a phone call with CNBC.

Given the uncertainty with which future developments were expected to further transform human work, worries about job displacement surfaced as one of the main areas of concern in Anthropic’s study, with 22.3% of respondents expressing job concerns as their biggest sources of worry.

These displacement worries were “spread fairly evenly across job categories,” according to the report, which Anthropic undertook in December 2025.

“When I am coding now, I am mostly just an observer, not a creator anymore. I can see that even for the observer role, I might not be needed,” an unnamed software engineer from the U.S. was quoted by Anthropic as saying.

Amid the dizzying pace of AI development, analysts are split on who really stands to gain from AI’s promises of economic empowerment.

“I see AI as the great equalizer,” Einstein said. “One of the beautiful things about AI is that in rural Indonesia or Brazil, [people] have access to the same AI as [in] the U.S. or Japan.”

Claude users from emerging economies, like Sub-Saharan Africa and Latin America appeared to express 10-12% lower rates of negative sentiments toward AI than users from Western Europe and North America.

Respondents from Sub-Saharan Africa also expressed greater aspirations for entrepreneurship and financial independence through AI usage than users from North America. Similar divergences emerged when North American users were compared against respondents from Latin America and Asia.

But while these findings may reflect real perceptions of opportunity associated with AI usage, particularly as a mechanism for access or economic mobility, this reading of the data is also undermined by the study’s methods, said Lia Raquel Neves, founder of ethical consultancy EITIC.

While the 80,508 responses that met the researchers’ quality threshold was a large sample by any measure, Anthropic was upfront about the methodological limitations associated with conducting a voluntary study on AI from a pool of existing users.

The pool of respondents “[skewed] toward people who have found enough value in AI to keep using it, and likely toward more positive visions than a general population sample would produce,” Anthropic wrote in its appendix.

Nearly half of all respondents also originated from North America and Western Europe.

“The results should be interpreted as an indicator of how early and active users, in different contexts, are framing their experience[s] with AI, and not as a consolidated picture,” Raquel Neves said in an email to CNBC.

While users from emerging economies seemed most excited by the prospects of economic gain from the use of AI, it remains unclear how evenly the spoils of AI development are likely to be distributed.

In a 2025 report, the United Nations Development Programme warned that future AI development could worsen existing socioeconomic inequalities, as economic benefits tended to get captured disproportionately by societies with greater capacity and access to digital infrastructure — which often means wealthier nations.

“In the absence of adequate conditions, [AI] may amplify existing vulnerabilities, namely through digital exclusion, algorithmic biases or dependence on external systems,” Raquel Neves told CNBC.

Although it may be too early to tell who stands to lose most in the AI race, there is little doubt over who the victors might be.

“Whoever successfully brings the [AI] agents that we’re all going to start using, is absolutely going to win,” Einstein said.

Anthropic has not responded to CNBC’s requests for comment.

— CNBC’s Dylan Butts contributed to this report.