From Transcripts to Insights: Uncovering Corporate Risks Using Generative AI

We explore the value of generative AI in uncovering corporate risks. We use a Large Language Model (LLM) to construct firm-level measures of political, climate, and AI-related risks. The proposed measures possess significant information content and outperform dictionary-based measures in predicting firm-level volatility and explaining investment decisions. Information in risk assessments dominates that in risk summaries, establishing the value of general AI knowledge. Our measures continue to perform well outside the LLM's training window and at the aggregate level. They are also priced in equity markets. Collectively, generative AI offers an effective solution to the measurement of hard-to-quantify corporate risks.

Keywords: GPT, ChatGPT, large language models, generative AI, risk information, firm-level risk exposure, conference call, political risk, AI risk, climate change risk

JEL Classification: C45, D81, G12, G30, G32, M41

Suggested Citation: Suggested Citation

Kim, Alex G. and Muhn, Maximilian and Nikolaev, Valeri V., From Transcripts to Insights: Uncovering Corporate Risks Using Generative AI (October 26, 2024). Chicago Booth Research Paper No. 23-19, Fama-Miller Working Paper, University of Chicago, Becker Friedman Institute for Economics Working Paper No. 2023-132, Available at SSRN: https://ssrn.com/abstract=4593660 or http://dx.doi.org/10.2139/ssrn.4593660