Intelligent decision-making is widely present in critical societal scenarios such as public safety, financial trading, and medical resource allocation. As technology has advanced, this field has undergone three profound paradigm shifts. In the Classical Algorithm Era (before 2016), represented by dynamic programming and integer programming, the advantages lay in model interpretability and mature theoretical guarantees, but scaling to large real-world environments remained difficult. In the Reinforcement Learning Era (2016–2023), agents learned strategies through continuous interaction with environments, driving large-scale deployment in real-world domains. Today, the Agentic Era centered on large language models (LLMs) has arrived.  LLMs endow agents with world knowledge, reasoning and planning capabilities, and tool-calling abilities, extending decision systems all the way to the execution layer. This talk will systematically review these three paradigm shifts, presenting our latest results in RL-based industrial deployment (smart agricultural efficiency, medical resource allocation, wildlife protection, etc.) and frontier LLM Agent exploration (financial intelligence frameworks, AI value alignment), and will look ahead to the future landscape of human-AI collaboration.

Guojun Xiong is an Associate Professor at the School of Computer Science, Shanghai Jiao Tong University. He was selected for the 2025 Science Fund Program for Excellent Young Scientists (Overseas). He received his Ph.D. from the State University of New York at Stony Brook in 2024, followed by postdoctoral research at the Department of Computer Science of Harvard.His research focuses on AI intelligent network decision-making and structured reinforcement learning, with primary directions including online sequential decision-making under uncertainty, stochastic optimization and control, and distributed optimization and multi-agent reinforcement learning for networked systems. He is dedicated to applying structured reinforcement learning methods to real-world complex systems, with particular attention to their potential impact in public health and social good. He has published over 20 first-author papers at top venues including NeurIPS, ICML, ICLR, AAAI, AAMAS, IEEE INFOCOM, ACM MobiHoc, IEEE/ACM TON, and IEEE TSP, and his work received a Best Paper Nomination at AAMAS 2026. He has also received the 2022 Chinese National Outstanding Overseas Student Award and the 2024 SUNY Outstanding Research Award.