生成型人工智能对金融稳定的影响：驯服动物精神

Federal Reserve Board, Washington, D.C.ISSN 1936-2854 (Print)ISSN 2767-3898 (Online) Financial Stability Implications of Generative AI: Taming theAnimal Spirits Anne Lundgaard Hansen, Seung Jung Lee 2025-090 Please cite this paper as:Hansen,Anne L.,and Seung Jung Lee(2025).“FinancialStability Implications ofGenerativeAI:Taming the Animal Spirits,”Finance and Economics Discussion Se-ries2025-090.Washington:Boardof Governors of the Federal Reserve System,https://doi.org/10.17016/FEDS.2025.090. NOTE: Staff working papers in the Finance and Economics Discussion Series (FEDS) are preliminarymaterials circulated to stimulate discussion and critical comment.The analysis and conclusions set forthare those of the authors and do not indicate concurrence by other members of the research staff or theBoard of Governors. References in publications to the Finance and Economics Discussion Series (other thanacknowledgement) should be cleared with the author(s) to protect the tentative character of these papers. Financial Stability Implications ofGenerative AI: Taming the Animal Spirits∗ Anne Lundgaard Hansena,band Seung Jung Leea aBoard of Governors of the Federal Reserve SystembFederal Reserve Bank of Richmond September 25, 2025 This paper investigates the impact of the adoption of generative AI on financial stability. We conductlaboratory-style experiments using large language models to replicate classic studies on herd behaviorin investment decisions. Our results show that AI agents make more rational decisions than humans,relying predominantly on private information over market trends.Increased reliance on AI-poweredinvestment advice could therefore potentially lead to fewer asset price bubbles arising fromanimal spiritsthat trade by following the herd. However, exploring variations in the experimental settings reveals thatAI agents can be induced to herd optimally when explicitly guided to make profit-maximizing decisions.While optimal herding improves market discipline, this behavior still carries potential implications forfinancial stability. In other experimental variations, we show that AI agents are not purely algorithmic,but have inherited some elements of human conditioning and bias. Keywords: Herd behavior, large language models, AI-powered traders, financial markets, financial sta-bility.JEL Codes: C90, D82, G11, G14, G40. ...[T]here is the instability due to the characteristic of human nature that a large proportion of ourpositive activities depend on spontaneous optimism rather than mathematical expectations [...]. Most,probably, of our decisions [...] can only be taken as the result of animal spirits—a spontaneous urgeto action rather than inaction, and not as the outcome of a weighted average of quantitative benefitsmultiplied by quantitative probabilities. — John Maynard Keynes 1. Introduction Human irrationality is a key driver of the build-up of financial vulnerabilities, contributing to asset pricebubbles and banking crises. History offers numerous examples, including Tulip Mania in the 17th cen-tury, the South Sea Bubble, the dot-com boom, the 2008 financial crisis, the 2010 Flash Crash, and theGameStop short squeeze. A well-established body of research highlights the role of psychological andemotional factors, coinedanimal spiritsorirrational exuberance, in these periods of boom and bust (An-geletos et al., 2018; Grauwe, 2012; Shiller, 2005). Understanding the role of animal spirits for financial stability is already a challenge, given the unpre-dictable nature of human behavior. Now, a new and unknown agent has entered the equation: decisionspowered by generative AI. Humans increasingly rely on AI for information gathering and decision mak-ing, whether as a co-pilot or as autonomous agents.1As generative AI is reshaping workflows acrossinstitutions and individuals, the question arises: How might the increased reliance on generative AI im-pact financial stability? Specifically, will generative AI exaggerate or dampen the role of animal spiritsin the build-up of financial vulnerabilities? Two competing hypotheses emerge. On the one hand, AI is fundamentally algorithmic, operating in aset of instructions and grounded in logic and rational decision making.2If AI-guided decisions replace human intuition, the result could be a reduction in the influence of animal spirits, leading to more stablefinancial markets. On the other hand, generative AI models, such as large language models (LLMs), aretrained on vast amounts of data, sourced from both rigorous materials, such as academic research, andthe, at times, chaotic discourse of social media platforms such as Twitter (X) and Reddit. Consequently,generative AI may inherit and even amplify human biases and irrational tendencies (Hayes et al., 2024;Jiang et al., 2023; Koralus & Wang-Maścianica, 2023; Zhu & Griffiths, 2024). Moreover, many AI modelsundergo reinforcement learning from human feedback (see Wang et al., 2024 for a survey), optimiz