Tutorial 2

Causality and AI Agent

13:30-14:45, Dec. 18, 2024, UTC+8. Seminar Room: Level 4, AESR 4-1.

Abstract
In AI, agent decision making has evolved from control and planning methods to advanced reinforcement learning, now including large language models (LLMs) that leverage expert data, prompt engineering, and human feedback. However, traditional approaches often fall short in dynamic, open-world settings. This tutorial advocates for causality as a critical enhancement for agent decision making, enabling a deeper understanding of environment structures and more reliable interventions. By integrating causal reasoning—through disentanglement, counterfactuals, and targeted policy analysis—LLMs can make adaptive, trustworthy decisions. We conclude by exploring challenges and future directions to further integrate causality into decision-making frameworks for robust AI systems.

Speakers Information

Mengyue Yang
Mengyue Yang
About the Speaker
Mengyue Yang is currently the Lecturer (equals to US assistant professor) in AI at University of Bristol. She studied for her Ph.D. in Computer Science at University College London, under the supervision of Professor Jun Wang. Her main research interests include causal representation learning, multi-agent systems, and reinforcement learning, with a primary focus on decision-making systems based on causal representations. She has published several research papers in top-tier AI conferences and journals such as NeurIPS, CVPR, KDD, SIGIR, WWW, and ACM TOIS. She has been recognized as a Rising Star in AI by KAUST. Additionally, she serves as a PC member or reviewer for conferences and journals including NeurIPS, ICML, ICLR, KDD, AISTATS and TNNLS etc.
Details
Haoxuan Li
Haoxuan Li
About the Speaker
Haoxuan Li is a fourth-year Ph.D. candidate at Peking University, advised by Prof. Xiao-Hua Zhou, coadvised by Prof. Zhi Geng and Prof. Peng Cui. He has more than 30 publications appeared in several top conferences such as ICML, NeurIPS, ICLR, SIGKDD, WWW, SIGIR, AAAI, and IJCAI. His research interests span from causal machine learning theory, counterfactual fairness, recommender system debiasing, out-of-distribution generalization, multi-source data fusion, bioinformatics, and large language models. Moreover, he is supported by the Young Scientists Fund of the National Natural Science Foundation of China (¥300,000), and have served as the AC or SPC/PC-member for top-tier conferences including ICML, NeurIPS, ICLR, SIGKDD, WWW, AAAI, IJCAI, and the invited reviewer for prestigious journals such as TOIS, TPAMI, TKDE, TKDD, TNNLS, JASA, SCIENCE CHINA Information Sciences, and The Innovation.