Anthropic models three AI economy scenarios through 2030

Anthropic released an interactive economic modeling tool Wednesday that allows users to explore three distinct scenarios for how artificial intelligence could transform the U.S. economy by 2030, with outcomes ranging from modest productivity gains to unprecedented white-collar unemployment.

The tool, developed by Anthropic's economics team, uses a task-based framework that segments the economy into cognitive occupations—management, professional, sales, and office work—versus all other occupations. Users can adjust assumptions about AI capabilities, adoption rates, and productivity improvements to generate projections for GDP, wages, labor's share of income, and unemployment levels.

The three scenarios

In the "modest" scenario, AI handles approximately 4% of economic tasks by 2030. This path produces GDP 1.6% higher than a baseline without AI, with minimal impact on employment or unemployment rates, according to details first reported by Axios.

The middle "substantial" scenario envisions AI performing 12% of tasks, driving GDP roughly 8.3% above the no-AI baseline while pushing unemployment to around 4.6%. Knowledge-sector employment would decline by approximately 4% under this path.

The "extreme" scenario—which Anthropic characterizes as having no historical parallel—projects AI taking over nearly one-third of all economic tasks. Output would surge 32% beyond the no-AI trajectory, but unemployment would climb toward 12% and white-collar employment would drop by more than 20%. Under this scenario, labor's share of national income would slide from 60% to 45%, meaning total wages flowing to workers would be no higher than in a world without AI despite the economy being one-third larger.

Why it matters

The model reveals a consistent pattern across all scenarios: AI redistributes income from labor to capital holders, with the gap widening as AI assumes more productive activity. This dynamic persists even in scenarios where overall economic output grows substantially, highlighting a potential disconnect between aggregate growth and worker prosperity. The framework provides business and policy leaders with a structured way to think through the economic implications of different AI adoption trajectories, though it notably excludes economic downturns, financial market disruptions, government policy responses, catastrophic AI failures, or robotics-enabled physical labor automation.

Public expectations and adoption pace

A separate Anthropic survey of nearly 11,000 U.S. adults found median public expectations align most closely with the substantial scenario, anticipating GDP roughly 10% higher by 2030 and overall unemployment rising to around 5%.

Anthropoc co-founder Jack Clark told NPR he expects continued rapid AI development but slower economic integration than many observers predict. "It will get really, really good. But it will make its way into the economy more slowly," Clark said.

The model is supported by a technical paper co-authored by Anton Korinek, Charles Jones, Szymon Sacher, Tess Cotter, and Peter McCrory. The authors emphasize the scenarios are not predictions with assigned probabilities but tools for making different assumptions comparable.

"These scenarios are not predetermined — it's not like an inexorable march," Anthropic economist Peter McCrory told Axios. "Part of the value of doing scenario modeling is so that you can do scenario planning."

Details of the model and scenarios were first reported by Axios and Quartz.