The Anthropic Economic Futures Program: Preparing for an AI-Shaped Economy

Anthropic has launched the Anthropic Economic Futures Program. The initiative aims to promote research on the economic impacts of artificial intelligence and to help understand them comprehensively and develop appropriate measures. It also serves as an extension of the existing Anthropic Economic Index, leveraging its insights into AI use in the world of work.

The Anthropic Economic Index

A fundamental pillar of the Economic Futures Program is the Anthropic Economic Index, which was introduced in February 2025. The initiative aims to understand the impact of AI on labor markets and the economy over a longer period. The first report from the index provides data and analysis based on millions of anonymized conversations with Claude.ai. The data used by the index is made available as open source by Anthropic, so that researchers can build upon it and expand the findings. The index’s methodology focuses on professional tasks rather than entire occupations, because jobs often share common tasks and skills, and AI selectively affects specific tasks. The research uses an internal analysis tool called “Clio” to analyze conversations with Claude.

Here are some findings from the Anthropic Economic Index:

• AI use today is heavily focused on tasks in software development and technical writing, particularly in the computer and mathematics category. Tasks with a high physical component, such as in agriculture, are least affected.

• Although only about 4 percent of occupations use AI in roughly three-quarters of their tasks, more moderate AI use is widespread. 36 percent of occupations use AI in at least a quarter of their tasks. There are no signs that entire occupations are being fully automated; instead, AI is distributed across many tasks, with stronger effects on specific task groups.

• The use of AI tends more towards augmentation, where AI extends and enhances human capabilities, rather than automation, where AI performs tasks directly. This means that in more than half of the cases, AI is not used to replace people, but rather works with them, for example, for validation, learning, or repetitive tasks.

• Interestingly, AI use is most common for tasks in occupations with medium to high salaries, such as computer programmers and data scientists, while it is lower in the lowest- and highest-paid roles. This highlights both the limits of current AI capabilities and practical barriers to using the technology.

Possible Impacts of AI on the Labor Market and Economy

While some in Silicon Valley see new career paths and economic opportunities through generative AI, such as the emergence of new successful startups, others warn that the gains could be distributed unequally and that widespread AI-related job losses are to be expected. Dario Amodei, the CEO of Anthropic, predicted in May 2025 that AI could eliminate half of all entry-level white-collar jobs within the next one to five years and cause unemployment to rise to as much as 20 percent.

However, Sarah Heck, head of policy programs at Anthropic, emphasizes, according to a TechCrunch article, that the main goal of the program is not to achieve biased results, but to understand what is actually happening. She notes that the disruptive changes AI will bring could have both positive and negative aspects. If job losses occur, Heck says, it is important to bring together a group of thought leaders to discuss mitigation measures. It is equally important, she adds, to discuss how to use GDP growth, should it occur. The program therefore aims to examine a much broader range of topics than just the labor market, including the emergence of new occupations and the changing nature of skills.

Early signs of AI’s impact on the world of work are already visible. Research that keeps pace with AI development is necessary to understand in real-time where this technology creates economic opportunities—new jobs, increased productivity, novel forms of value creation—and where it might cause dramatic shifts in labor dynamics.