Anthropic’s Economic Index measures how people use Claude—what tasks they assign, how much autonomy they permit, and how often the assistant succeeds. The report, initiated about a year ago, finds that AI reshapes jobs unevenly: it boosts productivity for many highly skilled professionals while some routine roles risk deskilling. Claude’s applicability has increased—44% of jobs can now use AI for at least 25% of tasks (up from 36%)—and Anthropic will open-source the underlying data for researchers and policymakers.
Anthropic’s New Index Shows AI Is Reshaping — Not Simply Replacing — Jobs
Anthropic cofounders Dario, left, and Daniela Amodei.(Courtesy of Anthropic)
New research from Anthropic suggests the answer to “Will AI take my job?” is far from simple. The company’s latest Economic Index analyzes real-world usage of its chatbot Claude—what tasks users assign it, how much autonomy they grant it, and how often it succeeds—to build a data-driven picture of how AI is changing work.
What the Index Found
Anthropic economist Peter McCrory, who helped lead the project launched about a year ago, says the results show uneven effects across occupations. In many highly skilled professions—radiology and therapy are two examples—Claude can shoulder time-consuming tasks, freeing professionals to focus on patient interaction, complex judgment, and higher-value activities. In other roles, such as data entry, some IT tasks, and travel booking, work may be simplified without clear new pathways to higher-skilled responsibilities, a dynamic often described as deskilling.
“AI is a general-purpose technology that will affect every job in some way,” McCrory said, noting both the speed of adoption and the importance of human oversight.
Key Trends and Context
Among the report’s measurable trends: 44% of occupations can now use AI for at least a quarter of their tasks, up from 36% in Anthropic’s previous index. Adoption is accelerating across the U.S.; states that previously lagged appear to be catching up. Importantly, the research predates the release of Anthropic’s Opus 4.5 model and the new Claude "Cowork" app—tools that expand Claude’s ability to act on users’ behalf by opening, reading and analyzing files—so real-world capability and uptake may already be higher.
Human Roles, Delegation, and Managerial Skills
McCrory highlighted that delegation and managerial skills may become more valuable as people learn to assign nuanced tasks to AI. He noted his own practice of delegating research-like tasks to Claude that he would previously have asked a research assistant to handle. That points to a shift in how work is organized: AI augments human expertise most effectively when humans provide oversight, judgment and coordination.
Anthropic plans to open-source the dataset behind the index so economists, policymakers and researchers can better understand the labor-market implications and prepare more informed responses.
Bottom Line
The Economic Index suggests AI is expanding capability and adoption quickly, producing productivity gains in many skilled professions while raising the risk of deskilling in other roles. The consequences will be uneven, and human collaboration and oversight remain essential.
By Sharon Goldman — AI reporter. This story originally appeared on Fortune.com.
Help us improve.
