Professor Lee Yongjae’s Team at UNIST Achieves 'Most Papers' at the World’s Largest Financial AI Conference

Top Performer Among Global Research Groups with 10 Papers Presented at ACM ICAIF
Highly Praised for LLM Investment Bias Analysis
Global Financial Institutions Seek Collaboration

UNIST has elevated Korea's status at the world's largest conference in the field of financial artificial intelligence, often referred to as the "Olympics" of the industry.

The research team led by Professor Lee Yongjae from the Department of Industrial Engineering at UNIST (President: Chong Rae Park) raised the profile of Korean financial AI by presenting the highest number of papers as a single research group at the world’s largest financial AI conference, 'ACM ICAIF 2025'.

Professor Lee Yongjae (third from the left in the top row) and his research team attending the 'ACM ICAIF 2025' academic conference are posing for a commemorative photo. Provided by UNIST

The International Conference on AI in Finance (ICAIF), organized by the Association for Computing Machinery (ACM), is one of the largest international AI conferences in the financial sector. It brings together global financial institutions such as JP Morgan, BlackRock, and Bloomberg, as well as academic researchers from around the world.

This year’s 'ACM ICAIF 2025' was held in Singapore from November 15 to 18. A total of 349 papers were submitted, and only 111 papers, or 31.8%, were accepted. Of these, 54 were selected for oral presentations and 57 for poster presentations.

Professor Lee’s team presented a total of eight papers-three oral presentations and five poster presentations.

This is the highest number among all regular papers presented this year. The team stood alongside world-class research groups such as Professor Mihai Cucuringu’s team from the University of California, Los Angeles (UCLA) with six papers, and Dr. Dhagash Mehta’s team from BlackRock with five papers.

The quality of the research was also notable. The study titled 'Your AI, Not Your View: The Bias of LLMs in Investment Analysis', which analyzed investment decision biases in large language models (LLMs), received particular attention.

Invited seminars were held at the Bank of Korea and Korea Investment Corporation, and there were additional presentation requests from UBS (Hong Kong), Bloomberg (New York), and global hedge funds.

This paper empirically demonstrated that major large language models (LLMs) such as ChatGPT, Gemini, Llama, and DeepSeek exhibit different 'investment biases' depending on industry sector, company size, and investment strategy. This means that even when asked about the same stock, each model may provide a different assessment.

The degree to which each model changed its stance when presented with evidence contradicting its bias also varied significantly. This suggests that tendencies developed during training may not be flexibly adjusted in response to actual market changes.

Professor Lee emphasized, "Financial institutions seeking to introduce LLM-based financial services must thoroughly examine model-specific biases." The research team has released the validation results in the form of a leaderboard and announced plans to continuously update it as new models are introduced.

The team also participated in organizing three workshops at the conference: ▲ LLMs and Generative AI for Finance ▲ Financial Time Series Analysis ▲ Emerging Market Financial Risk Modeling.

In addition, they contributed to the qualitative growth of the conference by hosting the 'Agent Search for Financial Documents' competition and the 'Decision-Centric Learning for Financial Optimization' tutorial.

Korean research achievements made significant strides at this year’s conference. Based on corresponding authors, 14% of the papers accepted at ICAIF were from Korean researchers, ranking third globally after the United States (31%) and the United Kingdom (27%). The academic community has commented, "The outstanding performance of the UNIST research team has firmly established the presence of Korean financial AI on the international stage."

Professor Lee has steadily expanded his international academic activities. Since submitting his first paper in 2022, he served as a workshop chair and organizing committee member last year.

He stated, "It is deeply moving to become the group with the most papers at the conference I first attended three years ago. I am profoundly grateful to the passionate students and colleagues who have worked hard on this research. I hope that as the foundation for AI research in Korea’s financial sector broadens, we will be able to host such world-class conferences in Seoul in the future."

ICAIF is scheduled to be held in Milan in 2026, in the United States in 2027, and preparations are underway for an Asian venue in 2028.

Professor Lee Yongjae is currently serving as a member of the Presidential National AI Strategy Committee and as a member of the Financial AI Council under the Financial Services Commission. He also acts as an advisory professor for the global financial AI company LinqAlpha, bridging research outcomes with real-world applications.

This content was produced with the assistance of AI translation services.