Speaker: Xingjian Zheng (Shanghai Advanced Institute of Finance (SAIF))
Time: 2025-12-15 14:00-15:00
Location: 910, SIMIS
Zoom Meeting ID: 844 3630 6474 Passcode: SIMIS
摘要：
Generative AI is increasingly being used as economic agents. However, we know very little about their financial decision-making rules. Exploiting a novel experimental setting, we show that it uses memories to make decisions, even when the memories are not in the same decision domain. When cued with images with positive emotional content, it makes riskier choices, even if it can form perfectly Bayesian beliefs. This mechanism is further causally supported with a supervised fine-tuning technique known as knowledge injection that can edit the language model’s memories. Empirical analysis shows that this memory-driven behavior substantially impacts the AI agent’s investment decisions and return predictability, creating significant upward or downward biases that correspond to the valence of its memories. Finally, we develop a memory-based economic model to explain the investment behavior of GAI agents.
About Speaker
Xingjian Zheng is a sixth-year Ph.D. candidate at the Shanghai Advanced Institute of Finance (SAIF). He was a visiting PhD student at Harvard Business School, sponsored by the China Scholarship Council (CSC). His
working papers on decision-making rules of AI agents have been accepted for presentation at over twenty top international conferences, including the AEA, AFA poster, ABFER poster, CICF, SFS Cavalcade, and the NAWM
Econometric Society. Additionally, his research on deep learning and asset pricing has been published in the Journal of Management Science in China, a top-tier domestic journal. Xingjian has served as a discussant for multiple international conferences and as a reviewer for international journals. He is a recipient of several awards and grants, such as the National Scholarship, Academic Excellence Scholarship, Outstanding Teaching Assistant Award, and an AFA Travel Grant.
简体中文 
English