Title: Advance AI ethics for large language models

Venue: Room 3425, Wenjin Building, Chang'an Campus, Shaanxi Normal University

Time: 10:00 AM, June 30, 2025

Speaker: Prof Henry Han

Abstract: As AI enters the age of large language models (LLMs), it faces increasingly sophisticated ethical challenges—chiefly, managing unprecedented data complexity while guaranteeing reproducibility and fairness. This talk addresses those challenges in two steps. First, we introduce a rigorous framework for estimating AI reproducibility across both deep-learning systems and LLMs. Second, we present a generative-AI approach that counters fairness-critical biases arising from imbalanced training data in LLMs, and demonstrate it in two high-stakes domains: (i) Finance, where domain-specific LLMs forecast stock returns, and (ii) Law, where Legal-LLMs analyze case holdings and mitigate bias. Together, these examples outline practical pathways to strengthen transparency, fairness, and reliability in next-generation AI. To the best of our knowledge, this is the first work in this new area.

Bio: Dr. Henry Han is the McCollum Family Chair in Data Science and Professor of Computer Science at Baylor University, where he also heads the university’s rapidly expanding data-science and AI research initiatives. Prior to joining Baylor, he was a tenured professor at Fordham University, founding the MS in Cybersecurity program and serving as Associate Chair of the Department of Computer & Information Science. He has published more than 130 high-quality papers in leading journals and top conferences across data science, artificial intelligence, fintech, and informatics. Dr. Han has supervised over 70 graduate students and postdocs, many of whom now hold key positions in major financial firms, leading tech companies, and top academic institutions. His research has attracted substantial support from the NSF, NASA, NIH, and other major agencies.