摘要: We develop a measure of firm exposure to artificial intelligence (AI) risk using textual analysis of quarterly earnings conference-call transcripts. We validate the measure through several tests, demonstrating that: (1) it captures intuitive variations over time and across firms; (2) it accurately reflects risk exposure, as evidenced by its significant correlation with both realized and options-implied stock return volatility, even after controlling for well-established aggregate risk measures; (3) firms facing higher AI risks have greater cash holdings, file more AI-related patents, and develop more AI-specific investments; (4) during unexpected AI events, such as the AI Beats Humans event in the Reading Comprehension SQuAD, firms with higher ex-ante risk exposure not only suffer adverse outcomes like increased volatility and decreased sales but also intensify their engagement in AI-focused patent applications and investments in response to these events. These results suggest that our text-based measure provides a credible quantification of firm exposure to AI risk and can thus be reliably used by scholars and policymakers to understand the multifaceted impacts of AI on firms and markets.

个人简介：吴志岩，管理学博士，现任浙江大学教授。其研究聚焦于战略管理，擅长运用因果推断方法。相关成果发表于《管理世界》、Academy of Management Journal、Journal of Business Venturing、Journal of Management、Management Science、Organization Science 等期刊。现任Management and Organization Review副主编，并担任Journal of Business Venturing和Journal of Management的编委。研究曾获EBS最佳论文奖，入选INFORMS播客 Resoundingly Human 专题介绍。此外，他曾获Journal of Management最佳审稿人奖。