About half of Claude users say artificial intelligence can already handle 50 percent or more of their work, according to a major new survey published by Anthropic as part of its Economic Index research program.

The findings, released on June 27, 2026 in a report the company calls "Cadences," draw on responses from roughly 9,700 users across Claude's Chat, Cowork, and Code products. The data offers one of the largest first-party glimpses to date into how people who actively use a frontier AI assistant perceive its impact on their jobs. The results underscore how quickly workers are folding generative AI into daily tasks, even as debates over productivity gains and job displacement continue. For broader context on how this fits into the wider AI industry coverage, the survey adds hard numbers to a trend many have only described anecdotally.

What the Numbers Show

Anthropic asked users to estimate what share of their concrete, task-based work AI could already perform. The breakdown:

  • About 33 percent said AI was usable for 30 to 60 percent of their tasks.
  • Roughly 14 percent put that figure at 60 to 90 percent.
  • Around 4 percent believed Claude could effectively do their entire job.

Taken together, the company noted, that places approximately half of respondents above the 50 percent threshold. Anthropic was careful to distinguish between discrete tasks and the full fabric of a job, which includes the knowledge transfer and judgment that happen between tasks.

Looking 12 Months Ahead

When asked to project forward one year, expectations climbed. About 26 percent of respondents said they expected AI to take over most of their work within 12 months. Anthropic described these forward-looking expectations as "strikingly consistent" across experience levels, locations, and professions, framing the trend as a "rising tide" in which capabilities are improving broadly rather than in isolated pockets.

Where AI Is Doing the Most Work

The survey also examined which work activities users lean on AI for most, focusing on Claude's Artifacts feature, where the output is a concrete deliverable such as a document, interactive graphic, or query result rather than a simple chat reply. The most heavily used work categories were:

  • Database queries — 82 percent
  • Blog or article writing — 81 percent
  • Marketing content — 80 percent

The concentration in writing-heavy and data-retrieval tasks aligns with broader industry observations that generative AI's clearest near-term value lies in text generation, content drafting, and structured data work.

Early-Career Workers Worry Most; Heavy Users Stay Optimistic

One of the more nuanced findings concerned attitudes rather than usage. Early-career workers reported the highest share of AI-capable tasks in their roles, and they also expressed the most concern about what that means for their job security.

The counterintuitive twist: the heaviest users of Claude were the most optimistic. Respondents who used the tool intensively tended to believe their own skills were becoming more valuable, not less, because AI was amplifying what they could deliver. How people judged their career prospects depended heavily on two factors: how much hands-on AI experience they had, and how far along they were professionally.

A Preference for Partnership Over Replacement

Despite the headline-grabbing finding that some users see AI doing most of their work, the survey found that most respondents do not want to be replaced. Instead, they reported wanting AI to take on routine, repetitive work, and they expressed hope that the resulting productivity gains would be shared widely rather than captured by a narrow few.

That sentiment tracks with the central tension in the report: workers are simultaneously impressed by what AI can already do and anxious about who ultimately benefits. Anthropic, which has a commercial interest in Claude adoption, published the data as part of its broader Economic Index effort to study AI's labor effects.

Why It Matters

The survey is not a neutral academic study, and its respondents are self-selected Claude users, who are likely more AI-fluent than the general workforce. But the sample size, roughly 9,700 users, is substantial, and the consistency of expectations across demographics suggests the attitudes reflect a genuine shift rather than a niche viewpoint.

For employers, the data signals that AI adoption inside knowledge work may be further along than many leadership teams realize, with employees quietly automating significant portions of their output. For workers, the survey reinforces a now-familiar message: familiarity with the tools correlates strongly with optimism about the future, while avoidance correlates with anxiety.

Stay Ahead of AI

The gap between AI's accelerating capabilities and the workforce's readiness to adapt is widening every quarter. Tracking these shifts matters, whether you are an employer planning for change or a professional deciding where to invest your time. Bookmark our homepage to stay ahead of AI developments as they happen.

Read more AI news →