Anthropic’s new research maps three wildly different futures for the AI economy—and the future of knowledge work

Anthropic's recent research explores three distinct possible futures for the AI economy and its impact on the U.S. workforce. The first scenario presents AI as a helpful sidekick for workers, leading to gradual economic growth akin to the internet's impact. In the second scenario, AI automates half of all knowledge work by 2030, fueling rapid economic expansion twice the usual rate. However, this growth does not translate into higher wages for knowledge workers, while the rest of the economy enjoys gains.

The third scenario is the most drastic, with AI outperforming humans at almost every knowledge-work task, performing most of it autonomously. This leads to a GDP growth of 15% annually, doubling the economy every four and a half years. While society becomes far wealthier, unemployment soars beyond typical recession levels.

Anthropic's interactive tool allows users to input their own assumptions and see the potential outcomes of these scenarios. However, the company emphasizes that the tool does not account for policy responses, business cycles, aggregate demand, financial market disruptions, or catastrophic risks, making it a "stark simplification of a complex reality."

The company's own survey of nearly 11,000 people reveals a split expectation among the public: while some anticipate real productivity gains, others worry about the pain for workers in AI-exposed jobs. Economist Ben Moll and Alex Imas have dismissed predictions of a doubling of GDP by the end of the decade, arguing that five conditions must align perfectly for such explosive economic growth.

Despite the skepticism, Anthropic's research aims to provide economists and policy experts with concrete scenarios to guide discussions about AI's potential impact on the economy. Cofounder Jack Clark acknowledges his uncertainty about which future is most likely, but stresses that AI's adoption rate will significantly influence the outcomes.