This dissertation examines how the commercialization of generative artificial intelligence influences new firm creation, leveraging the release of ChatGPT as an exogenous shock to AI accessibility. Chapter 1 motivates the study by reviewing the economic literature on the impact of AI, which has so far focused largely on its effects on labor markets and incumbent firms while leaving open the question of whether AI also reshapes the rate and composition of entrepreneurial entry. Chapter 2 develops a novel industry-level measure of AI compatibility—the AI Assistance Score—that captures the extent to which AI tools can support early-stage entrepreneurial tasks. The measure is built through a multi-layer pipeline that applies large language models to a corpus of entrepreneurship handbooks, business guides, and case studies, and it is externally validated against an original survey of 582 new entrepreneurs in the United States. Chapter 3 applies this measure in a difference-in-differences design and finds that industries more compatible with AI experienced an increase of over 10% in firm formation following ChatGPT's release. To investigate why, the chapter examines three mechanisms: reduced experimentation costs, capital reallocation, and skill gap bridging. While I find no evidence supporting the latter two, the increase in firm formation is concentrated in industries where AI can effectively assist and experimentation is critical—evidence consistent with a reduction in the cost of experimentation as the key driver. Moreover, new firms created after ChatGPT's release in AI-compatible industries are more likely to survive, grow faster, and attract more educated workers. Taken together, the mechanism and quality results suggest that AI lowers the cost of experimentation in a way that disproportionately benefits high-ability entrepreneurs, leading to both greater and higher-quality firm formation.
