This page organizes papers related to artificial intelligence published around the world. This page is summarized using Google Gemini and is operated on a non-profit basis. The copyright of the paper belongs to the author and the relevant institution. When sharing, simply cite the source.
Your AI, Not Your View: The Bias of LLMs in Investment Analysis
Created by
Haebom
Author
Hoyoung Lee, Junhyuk Seo, Suhwan Park, Junhyeong Lee, Wonbin Ahn, Chanyeol Choi, Alejandro Lopez-Lira, Yongjae Lee
Outline
This paper addresses the knowledge conflict problem that arises in large-scale language models (LLMs) in the financial sector due to the mismatch between pre-trained knowledge and real-time market data. Specifically, we highlight that in investment services, inherent biases in models can misalign with institutional objectives and lead to unreliable recommendations. To explore the inherent investment biases in LLMs, this study proposes an experimental framework that uses hypothetical scenarios to identify potential model biases and measure their persistence.
Takeaways, Limitations
•
Takeaways:
◦
A framework for quantitatively analyzing model-specific biases in LLM-based investment analysis is presented.
◦
Observation of bias towards technology stocks, large cap stocks, and contrarian strategies.
◦
Confirmation bias occurs when bias is reinforced and initial judgments are fixed.
◦
Provides a public leaderboard that benchmarks the biases of various models.
•
Limitations:
◦
No specific mention of Limitations in the paper (not possible to determine from the abstract alone)
