• Suraj Srinivasan


May 2024, Paper: "As the latest general-purpose-technology (GPT), we study the value of AI innovations as it diffuses across general and application sectors, using the USPTO’s AI patent dataset. Investors value these innovations more than others, as AI patents exhibit a 9% value premium, and 26% higher forward citations compared to non-AI patents from the same patent classification and industry group. This value premium also increases over time, particularly in firms and industries where occupational tasks are more suited for AI. Due to the specialization in AI innovation across general and application sectors, we further find that policies that facilitate knowledge spillovers are key to increasing the value premium in these innovations. Specifically, we show that the value premium of AI innovations in application sectors, increases by 5% after the AIPA patent publication rule, and by 2% after the open sourcing of TensorFlow. Overall, our analysis illustrates the value of AI innovations to investors and highlights the importance of policies that foster knowledge spillovers for AI innovations."