How Ethical Should AI Be? How AI Alignment Shapes the Risk Preferences of LLMs

48 Pages Posted: 4 Jun 2024

See all articles by Shumiao Ouyang

Shumiao Ouyang

University of Oxford, Saïd Business School

Hayong Yun

Michigan State University - Department of Finance

Xingjian Zheng

Shanghai Jiao Tong University (SJTU) - Shanghai Advanced Institute of Finance (SAIF)

Date Written: May 01, 2024

Abstract

This study explores the risk preferences of Large Language Models (LLMs) and how the process of aligning them with human ethical standards influences their economic decision-making. By analyzing 30 LLMs, we uncover a broad range of inherent risk profiles ranging from risk-averse to risk-seeking. We then explore how different types of AI alignment, a process that ensures models act according to human values and that focuses on harmlessness, helpfulness, and honesty, alter these base risk preferences. Alignment significantly shifts LLMs towards risk aversion, with models that incorporate all three ethical dimensions exhibiting the most conservative investment behavior. Replicating a prior study that used LLMs to predict corporate investments from company earnings call transcripts, we demonstrate that although some alignment can improve the accuracy of investment forecasts, excessive alignment results in overly cautious predictions. These findings suggest that deploying excessively aligned LLMs in financial decision-making could lead to severe underinvestment. We underline the need for a nuanced approach that carefully balances the degree of ethical alignment with the specific requirements of economic domains when leveraging LLMs within finance.

Keywords: Large Language Models, AI Alignment, Risk Preferences, AI in Finance, Underinvestment JEL Codes: G11, G41, D81, O33, C45, C63, D91, A13

Suggested Citation

Ouyang, Shumiao and Yun, Hayong and Zheng, Xingjian, How Ethical Should AI Be? How AI Alignment Shapes the Risk Preferences of LLMs (May 01, 2024). Available at SSRN: https://ssrn.com/abstract=4851711 or http://dx.doi.org/10.2139/ssrn.4851711

Shumiao Ouyang (Contact Author)

University of Oxford, Saïd Business School ( email )

HOME PAGE: http://https://www.shumiaoouyang.com/

Hayong Yun

Michigan State University - Department of Finance ( email )

667 N Shaw Lane, Room 339
East Lansing, MI 48824-1122
United States
517-884-0549 (Phone)

HOME PAGE: http://https://sites.google.com/view/hayongyun/

Xingjian Zheng

Shanghai Jiao Tong University (SJTU) - Shanghai Advanced Institute of Finance (SAIF) ( email )

Shanghai Jiao Tong University
211 West Huaihai Road
Shanghai, 200030
China

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
34
Abstract Views
179
PlumX Metrics