中国科学院数学与系统科学研究院

数学研究所

数学科学全国重点实验室

学术报告

计算机科学研讨班

Speaker: Professor Weiru Liu（University of Bristol）

Inviter: 李阳阳 副研究员

Language: English

Title: Agents before and after LLM Revolution: comparing symbolic and neural approaches

Time&Venue: 2025年6月30日（星期一）18:00-19:00

&腾讯会议：725-176-435

Abstract: The concept of agent and multi-agent systems in AI traces back over four decades (e.g., Konolige, K., A first-order formalization of knowledge and action for a multi-agent planning system. Tech Note 232, SRI International, Menlo Park, California, 1980). Since then, MASs has remained a core research area in AI. Recently, this subfield has experienced a renewed interest driven by advances in large language models (LLMs), leading to a surge in research and application involving LLM-based agent systems. In this talk, I will compare symbolic/logic-based agent systems versus LLM-based agent systems, focusing on their approaches to planning and belief/knowledge management. By contrasting the two paradigms, we aim to identify synergies that can advance the development of more robust and scalable agent systems in the future.