Artificial intelligence's economic trajectory raises a pivotal question: will it spark unprecedented growth, trigger widespread job displacement, or chart a course toward a future we have yet to imagine?

The economics team at Anthropic has released a new set of economic scenario models designed to quantify AI's potential impact on employment, growth, and wages. These models, accessible through an interactive public tool, allow users to project different economic outcomes based on their own assumptions about AI's capabilities.

The framework explores three distinct scenarios, ranging from a moderate, gradual path to a transformative scenario where economic growth doubles, and culminating in an extreme case with annual GDP growth of 15% but a sharp decline in knowledge-based jobs. Across most scenarios, unemployment and wage fluctuations remain within historical norms. However, in the extreme case, knowledge workers face wage reductions and prolonged joblessness, while capital's share of economic output rises significantly.

Anthropic intends for this model, along with its broader research, to inform labor market studies it funds and to underpin policy recommendations, with the goal of ensuring AI-driven economic benefits are distributed broadly across society. The current version, v1.0, released in September 2026, acknowledges certain limitations and will undergo continuous refinement.

Redefining the Economy: Every Job Is a Bundle of Tasks

Understanding AI's economic influence begins with rethinking the notion of a "job." In Anthropic's model, all work in the economy is not a singular activity but a collection of "task bundles." Consider a nurse's typical day: checking patient rooms, drawing blood, triaging patients, recording vital signs, and ordering supplies. How might AI reshape this bundle? Some tasks remain exclusively human, such as bathing patients, where AI cannot assist. Other tasks are enhanced by AI, helping nurses draft discharge instructions more quickly, monitor patients remotely, or optimize shift schedules. Certain tasks become fully automated, like recording vital signs or ordering supplies. Additionally, new tasks emerge, as history shows technology creates new roles, such as nurses reviewing AI-generated triage results or evaluating AI-proposed care plans.

The outcome? As AI integrates into daily work, nurses become more productive, freeing time for patient interaction. When this task-level transformation occurs millions of times daily across industries, it forms the foundation of a future economy valued at over $30 trillion. GDP, labor market dynamics, and workers' wage shares will hinge on the speed and breadth of AI-driven task enhancement and automation.

Three Scenarios for the Road Ahead

Future economic directions depend on the pace of AI capability growth and adoption across sectors. Anthropic's report outlines three distinctly different macro scenarios. The first, a moderate scenario, envisions gradual gains. Here, AI's economic impact resembles the internet's rollout, driving real growth but within historical norms for new technologies, progressing incrementally. In macro data, AI's immediate shock may even be hard to detect. The second, a major scenario, involves a revolution in cognitive work. By 2030, AI can handle half of knowledge-based tasks, mostly autonomously, though not fully adopted. Economic growth doubles, but notably, knowledge workers' wages stagnate while other worker types see gains. Anthropic's August survey of over 10,000 Americans shows most people's expectations align with this scenario: GDP 10% higher than without AI by 2030, with unemployment rising to around 5%. The third, an extreme scenario, brings profound economic restructuring. Here, AI surpasses human productivity in most knowledge tasks, operating almost entirely autonomously while creating few new knowledge roles. This typically requires recursive self-improvement in AI, rapidly adopted. The results are striking: annual GDP growth hits 15%, doubling the economy every 4.5 years. Society becomes wealthier than ever, but at the cost of a sharp decline in knowledge workers and unemployment surging beyond typical recession levels.

Four Hard Truths About 2030

From these models, Anthropic draws four key findings worth serious consideration. First, the economic pie grows, but may bring historic unemployment. In all scenarios, AI boosts GDP and overall societal wealth. Yet, in the extreme scenario, unemployment could reach record highs due to recursive self-improvement and rapid adoption. Second, a painful occupational reshuffle looms. In the major and extreme scenarios, knowledge workers face severe automation displacement. On an individual level, programmers and call center agents may need to pivot to less AI-affected roles like electricians or nurses. But cross-industry transitions are tough: workers may resist change, require new skills, and find few openings. The more transitions demanded, the longer unemployment spells become. Third, wages diverge, with knowledge workers facing cuts. Models show that while average wages rise, gains concentrate in non-knowledge work. Reduced demand for human cognitive labor puts downward pressure on knowledge worker salaries. Conversely, AI boosts efficiency in cognitive tasks, like faster infrastructure design and permitting, increasing demand for physical labor and pushing up blue-collar wages. In the major scenario, knowledge worker wages stay flat; in the extreme scenario, they decline over 10% by 2030. Fourth, capital gains while labor's share shrinks. Today, each dollar of output splits roughly 60 cents to labor and 40 cents to capital. As AI takes over more tasks, capital's tools and resources become more valuable, raising their price. Studies indicate labor's share declines notably in the major and extreme scenarios, with capital's share rising. In the extreme case, despite a booming economy, total labor income barely changes by 2030. Most knowledge workers face pay cuts or job loss, leaving the working class with a smaller slice of a larger pie.

A Final Note: The Future Is Not Predetermined

Facing this somewhat stark forecast, Anthropic stresses that 2030's economic landscape is not fixed destiny. It depends on AI's actual capabilities, how businesses and workers implement it, and crucially, how the financial gains from this technology are distributed. The team admits the v1.0 economic model simplifies complex realities, omitting factors like policy responses, economic cycles, or super-robots. Yet, the core value of this explorer lies in its warning: under extreme AI prosperity, our primary challenge shifts from achieving growth to ensuring benefits are widely shared, not unevenly burdening specific groups, especially knowledge workers. The AI era's train has departed. We must not only watch its speed but also ensure every passenger has a seat.