Anthropic has released a new economic model that seeks to address a question on everyone's mind: What happens if artificial intelligence continues to advance? The model focuses on AI's impact on employment, wages, and the US economy through 2030, presenting three scenarios—mild, substantial, and extreme—by adjusting variables such as AI capabilities, corporate adoption rates, autonomy levels, productivity gains, and the difficulty of retraining displaced workers.
Anthropic's mild scenario aligns with past technological change experiences: AI boosts productivity and drives economic growth, but the overall impact remains within historical norms. The substantial scenario, however, is markedly different—it assumes that by 2030, AI will be able to perform half of all knowledge work tasks and complete most of them autonomously. In this scenario, the US GDP would reach $36.3 trillion, 8.3% higher than the baseline without AI, and economic growth would be roughly double the normal rate. Yet, under this scenario, the wages of knowledge workers would largely remain stagnant.
AI's Impact on Knowledge Work Will Significantly Alter the Future Job Market
The concept of knowledge workers was introduced by management expert Peter Drucker in 1959, referring to a group of professionals whose primary capital is specialized knowledge and critical thinking. Anthropic's model does not view jobs as a simple binary of "AI can replace or cannot replace," but rather breaks them down into specific tasks: some AI can enhance, some can automate, and some must still be done by humans. For example, in healthcare, AI can draft discharge instructions, assist in patient monitoring, and schedule care plans, but it cannot bathe patients. AI reduces the demand for some digital roles while the productivity it creates spurs new demand in other areas—for instance, AI shortens the design and approval cycles for infrastructure, making more projects feasible and thereby increasing the demand and wages for construction workers.
The extreme scenario is the most alarming. This version assumes that AI becomes more efficient than humans in the vast majority of knowledge work domains, operates almost entirely autonomously, and fails to create enough new knowledge work to compensate for the lost jobs. Achieving this scenario might require recursively self-improving AI coupled with extremely rapid corporate adoption. Under these conditions, the US GDP would reach $44.4 trillion by 2030, expanding the economy by 32.4% compared to the no-AI baseline; however, the wages of knowledge workers would fall by more than 10%, and the unemployment rate would rise to levels typically associated with economic recession.
The economy is growing, but the people creating the value are not sharing in the benefits.
Anthropic is not making predictions, and the model has clear limitations: it does not consider government policy responses, business cycles, financial market fluctuations, changes in consumer demand, or catastrophic AI risks, nor does it simulate the impact of highly capable robots on physical labor. In August 2026, Anthropic surveyed 10,980 Americans about their expectations for AI and fed the responses into the model. It found that the typical respondent was closer to the "substantial" scenario—the economy would be about 10% larger than the baseline, with an unemployment rate of around 5%. But even this middle ground implies that the way millions of people work will change substantially.