Artificial intelligence holds the potential to dramatically reshape the United States economy, driving significant GDP growth by 2030, but also posing substantial challenges for employment and wage distribution. A new economic scenario model developed by Anthropic's Economics team explores these complex outcomes, suggesting a future where a wealthier nation doesn't automatically translate to better incomes or job security for all.
Anthropic's Economic Scenarios
Anthropic emphasizes that its model presents simulations, not forecasts, designed to illustrate how varying assumptions about AI's development could lead to different economic realities. The model analyzes how AI capability, adoption rates, autonomy, and productivity might influence GDP, employment levels, wages, and the division of economic gains between labor and capital.
Three AI Futures
The model outlines three distinct scenarios for AI's economic integration:
- Modest Scenario: In this future, AI's economic impact is comparable to that of the internet. Productivity sees gradual improvements, and GDP growth remains within historical norms for major technological advancements.
- Substantial Scenario: This represents a more significant disruption. By 2030, AI can perform roughly half of all knowledge-work tasks, many autonomously, though adoption isn't universal. Economic growth effectively doubles its usual rate, with knowledge workers experiencing minimal wage increases, while other occupations benefit from increased demand.
- Extreme Scenario: This radical transformation sees AI surpassing human productivity across most knowledge-work tasks, automating nearly all of them, and creating few new knowledge-work roles for humans. This outcome would likely require recursively self-improving AI and very rapid adoption.
Projected GDP Growth by 2030
Under all three scenarios, Anthropic's model estimates a notable increase in US GDP by 2030, calculated using 2025 price levels:
- Modest: GDP reaches approximately $34.1 trillion, a 1.6% increase compared to an economy without AI.
- Substantial: GDP climbs to about $36.3 trillion, an 8.3% increase.
- Extreme: GDP soars to roughly $44.4 trillion, a significant 32.4% higher.
While the economy expands dramatically in the more extreme scenarios, Anthropic highlights a critical distinction: a larger economy doesn't necessarily guarantee proportionally higher incomes for workers.
AI's Impact on Employment and Wages
The model views jobs as collections of individual tasks, rather than assuming entire occupations will disappear. AI can augment existing tasks, automate them, or create entirely new ones. For example, a nurse might use AI for patient monitoring or documentation, while still performing tasks requiring human interaction, or even take on new responsibilities like reviewing AI-generated care plans.
However, in more transformative scenarios, certain occupations could see a substantial number of tasks automated. Professions like coding and customer service, for instance, might face pressure for workers to transition into less automation-exposed fields, a shift that could be challenging due to retraining and re-employment needs.
Anthropic's findings indicate that while average wages may increase across all scenarios, these gains are uneven. In the substantial scenario, knowledge workers' wages remain largely flat. In the extreme scenario, their wages could fall by over 10% by 2030.
Conversely, wages in occupations less susceptible to AI automation could rise. For example, AI-assisted design and permitting could accelerate construction, increasing demand for construction workers. This creates a unique possibility: AI could make the economy much richer while simultaneously making some highly skilled workers financially worse off.
Shifting Gains: Labor vs. Capital
Perhaps the most profound finding concerns the distribution of economic gains between labor and capital. Starting from an economy where labor receives approximately 60% of output and capital 40%, the model shows a significant shift as AI automates more tasks:
- Modest Scenario: Labor's share slightly decreases to 59.4%, with capital rising to 40.6%.
- Substantial Scenario: Labor's share falls more noticeably to 56.1%, while capital's share increases to 43.9%.
- Extreme Scenario: The split changes dramatically, with labor receiving only 45.2% and capital capturing 54.8%. This suggests that companies and individuals owning AI systems, computing infrastructure, and other productive capital could secure an increasingly larger portion of the economic benefits.
Public Expectations and Model Limitations
A survey of over 10,000 Americans conducted by Anthropic revealed that typical respondents' expectations align broadly with the substantial-change scenario. This implies an economy roughly 10% larger by 2030 than it would otherwise be, accompanied by a rise in unemployment to around 5%. About 10% of respondents' answers were consistent with the extreme scenario.
Anthropic cautions that its model is intentionally simplified. It does not account for various important factors, including potential policy responses, business cycles, aggregate demand effects, financial market disruptions, catastrophic risks, or the future development of highly capable robots.