Anthropic has published the results of its largest economic research effort to date, a survey of 81,000 Claude users that asked, in their own words, how artificial intelligence is reshaping their working lives. The results are striking in their honesty, their contradictions, and their implications for how policymakers, employers, and workers should think about AI’s march through the economy.
Published on 22 April 2026 as part of Anthropic’s ongoing Economic Index, the study connects the company’s internal data on what tasks Claude is being used for with direct testimony from the people doing those tasks. The result is something rare in AI research: quantitative exposure data fused with qualitative human experience.
“Well like anyone who has a white collar job these days I’m 100% concerned, pretty much 24/7 concerned about losing my job eventually to A.I.” — Survey respondent, software engineer
Who’s Worried — and Why Their Instincts Are Right
One fifth of respondents voiced concern about job displacement, and they weren’t wrong to be worried. Anthropic’s data shows a clear statistical relationship: the more of a person’s job tasks that Claude is observed performing, the more anxious that person reports feeling about their economic future. For every ten percentage-point increase in what Anthropic calls “observed exposure,” perceived job threat rose by 1.3 percentage points. Workers in the top quarter of exposure mentioned displacement fears three times as often as those in the bottom quarter.
Elementary school teachers, whose work Claude rarely touches, expressed little personal anxiety. Software engineers — whose coding tasks make up a disproportionate share of Claude’s workload — were far more nervous. “When AI arrived, the project managers started giving harder and harder tickets and bugs to solve,” one developer told researchers. The workload didn’t shrink; it expanded to fill the AI-powered capacity.
The career stage divide is stark: early-career workers are significantly more likely to express displacement anxiety than senior professionals, consistent with earlier Anthropic research suggesting a possible slowdown in hiring of recent graduates in some sectors.
The Productivity Boom — and Who’s Capturing It
The picture isn’t all anxiety. On a seven-point scale measuring self-reported productivity gains, the average respondent scored 5.1 — corresponding to “substantially more productive.” The gains are real, widespread, and in many cases transformative. A website that once would have taken months was built in four to five days. Tasks that took four hours now take two.
High-wage workers — executives, software developers, financial professionals — reported the largest gains. But the data contains a genuinely surprising subplot: some of the lowest-paid workers also reported outsized benefits. A delivery driver was using Claude to launch an e-commerce business. A landscaper was building a music app. For these workers, AI isn’t just speeding up existing tasks — it’s unlocking entirely new ones.
“I’m a non-tech guy but now I’m a full stack developer.” — Survey respondent
Scope Beats Speed: The Most Common Benefit
When respondents described how AI helped them, the most common answer wasn’t “it made me faster” — it was “it let me do things I couldn’t do before.” Forty-eight percent of those who described productivity benefits cited expanded scope; 40% cited speed. This distinction matters enormously. Speed gains compress existing work; scope gains create new value — and potentially new roles.
Quality improvements also featured strongly: lawyers reviewing contracts more thoroughly, developers running deeper code checks, analysts producing more comprehensive reports. The notable outliers were lawyers, who expressed frustration at AI’s tendency to drift from precise instructions — a field where exactitude is everything.
A Paradox at the Heart of the Data
Perhaps the most counterintuitive finding is the “U-shaped” relationship between AI speed-up and job anxiety. Workers who found AI slowed them down — fine artists and writers who found it too rigid for their creative process — were more worried, which makes intuitive sense. But among those who reported genuine speed gains, anxiety grew alongside the gains. The faster AI makes you, the more you wonder whether the role itself remains necessary.
This captures something profound about the current moment. AI’s benefits are not arriving quietly. Workers feel them acutely — and that immediacy makes the longer-term question more, not less, pressing.
Who Captures the Gains?
Among respondents who named a specific beneficiary of their productivity gains, most said the value flowed to themselves — through time saved, capabilities expanded, and scope widened. But a significant 10% said employers and clients were simply demanding more output. The split is also generational: only 60% of early-career workers said they personally benefited, compared to 80% of senior professionals, suggesting the distribution of AI’s gains may mirror existing workplace power dynamics.
Anthropic is candid about the study’s limitations. The survey reached people with personal Claude accounts who chose to respond — a self-selecting group likely to skew positive. Occupation and career stage were inferred from conversational context rather than direct questions. The company calls for structured follow-up surveys to confirm these findings.
The headline conclusion, though, is clear: people’s instincts about AI’s impact on their roles are broadly accurate. And that accuracy — that workers already know what the data shows — is itself something policymakers and employers should not ignore.
Key findings:
- Workers in roles most exposed to AI automation express the greatest concern about job displacement — proving people’s intuitions track the data.
- Early-career workers are significantly more worried about AI replacing them than senior professionals.
- The highest- and lowest-paid workers report the largest productivity gains — primarily from expanding the scope of what they can do, not just working faster.
- Workers experiencing the biggest speed-ups from AI are paradoxically the most anxious about their long-term job security.
Source: Anthropic, “What 81,000 people told us about the economics of AI,” 22 April 2026. Read the full research.
