Gianluca Demartini 

Date/Time: 12.06.2025, 11:00 

Location: D2.2.094 

Abstract 

The rise in popularity of general-purpose large language models (LLMs) raises questions around bias and fairness. Do these models reflect the biases and stereotypes present in the data they have been pre-trained on? What should we do about that? In this talk, reviewing recent research we conducted at The University of Queensland, we will discuss issues of bias in human data using as an example gender bias in Wikipedia, and issues of bias in AI using as an example political bias in LLMs. We will then discuss how to explore and manage such bias that exists in data and in LLMs, how these models can be used for sensitive tasks, and how users tend to trust and over-rely on AI agents, even for high-risks tasks. 

Bio 

Gianluca Demartini is a Professor in Data Science and an ARC Future Fellow at the School of Electrical Engineering and Computer Science at the University of Queensland, Australia. His main research interests in Data Science include Information Retrieval, Semantic Web, and Responsible Artificial Intelligence. His research is currently funded by the Australian Research Council, the Swiss National Science Foundation, Meta, Google, and the Wikimedia Foundation. He received multiple Best Paper awards at Artificial Intelligence and Information Retrieval conferences. He has published more than 200 scientific papers at major computer science venues such as the ACM Web Conference, ACM SIGIR, VLDB Journal, ISWC, and ACM CHI.