The Value of AI Innovation

Abstract

With Artificial Intelligence (AI) emerging as one of the most consequential technologies, this study investigates how investors in financial markets view the relative value of a few important types of AI. The financial market value are used as a proxy for a value that these new technologies add to society. Using a machine learning approach to cluster all US patents from 1976 to 2025 (USPTO Bulk Database) into 178 labeled categories, we separated the potential AI categories from the rest. The AI patents are then classified into Image, Voice/Natural Language Processing, and Robotics using zero-shot classification. To assess the market value of each patent, this study applies the Kogan, Papanikolaou, Seru, and Stoffman (2017) model that uses the abnormal stock return of a publicly traded firm following the announcement of a patent grant. We document that Natural Language Processing patents are the most valuable, followed by Imaging and Robotics.  

Citations: Leonid Kogan, Dimitris Papanikolaou, Amit Seru, Noah Stoffman, Technological Innovation, Resource Allocation, and Growth, The Quarterly Journal of Economics, Volume 132, Issue 2, May 2017, Pages 665–712, https://doi.org/10.1093/qje/qjw040