Anthropic Projects AI Could Double US Economy but Spike Unemployment

Anthropic has released a detailed economic model outlining three possible futures for the United States by 2030. The analysis suggests that while artificial intelligence could drive unprecedented growth, it may simultaneously displace a significant portion of the workforce. The company behind the Claude assistant argues that the impact of AI will depend heavily on how much of daily professional tasks it can automate.

In the most optimistic view, AI acts as a tool that enhances human productivity, similar to the early days of the internet. However, the most extreme scenario presents a stark contrast. Here, AI systems perform the majority of knowledge work independently, leading to an economy that doubles in size every 4.5 years. The trade-off is severe: unemployment among knowledge workers could rise to nearly 18 percent, a level the authors describe as historically unprecedented.

Task-based analysis replaces job-level fears

Rather than asking if AI will take a specific job, the model examines which individual tasks within a role can be automated. Using nursing as an example, the report notes that while AI cannot physically care for patients, it can handle administrative duties like drafting discharge instructions or ordering supplies. This granular approach allows for a more precise prediction of how much of the workforce is affected.

The model also accounts for new roles that may emerge to oversee these systems. For instance, someone will still be needed to verify that AI triage decisions are correct. By multiplying these task-level shifts across the entire economy, Anthropic estimates where displaced workers might end up and how much value shifts from labor to capital ownership.

Three scenarios map the economic spectrum

The mildest scenario assumes AI assists with about 4 percent of work tasks. In this case, economic growth nudges slightly above current rates, and unemployment remains stable. This outcome is comparable to the historical impact of the internet, where technology boosted productivity without causing mass displacement. It represents a steady, manageable transition for most workers.

The intermediate scenario sees AI replacing human effort in about 12 percent of tasks. Growth accelerates to 5.4 percent annually, but unemployment rises as workers are displaced. The most extreme scenario involves AI performing 30 percent of all tasks autonomously. This leads to explosive growth of 15.4 percent per year but also the highest unemployment figures, with nearly one in five knowledge workers losing their jobs.

Wealth concentration shifts toward owners

Across all three scenarios, including the calmest one, a larger share of economic value goes to those who own capital rather than those who provide labor. As AI takes over more tasks, the return on investment in technology and infrastructure outpaces the returns on human work. This structural shift means that even in scenarios with moderate growth, the gap between asset owners and wage earners is likely to widen.

According to GN technics/ai (en-US), the authors warn that the consequences of the high-automation scenario go beyond any event in history. The report serves as a cautionary framework, highlighting that while the potential for wealth creation is immense, the social costs in terms of employment and inequality require careful policy consideration to mitigate.