SLAI Seminar 37th Session will be discussing the topic on “How AI Agents Are Built: From Language Models to Actionable Research Assistants”,  from 2:30pm-4pm, May 21 (Thursday) at Room B401, online participation is welcome.

 (Tencent Meeting ID: 162-472-398)

报告主题： AI智能体是如何构建的：从语言模型到可操作的研究助手

时间：2026年5月21日（周四）下午2:30-4:00

地点: 深圳河套学院B401教室

线上参与：腾讯会议号162-472-398

主讲嘉宾：董胤蓬博士

主持人：沈力教授

讲者简介About the Speaker:

董胤蓬博士是清华大学人工智能学院助理教授，本科及博士均毕业于清华大学计算机系。董博士的主要研究方向为机器学习与人工智能安全，在TPAMI、IJCV、NeurIPS、ICML、CVPR等顶级会议和期刊上发表论文80余篇，谷歌学术引用超过16000次。他曾担任ICLR、NeurIPS和ICML的领域主席，并获得中国计算机学会（CCF）优秀博士学位论文奖、清华大学杰出博士后研究者、微软学者奖学金、百度奖学金、字节跳动奖学金等荣誉。

Dr. Yinpeng Dong is an Assistant Professor at College of AI, Tsinghua University. He obtained his B.E. and Ph.D. degrees from the Department of Computer Science, Tsinghua University. His research interests are primarily on machine learning and AI safety. He has published over 80 papers in the prestigious conferences and journals, including TPAMI, IJCV, NeurIPS, ICML, CVPR. These papers have amassed more than 16000 citations (Google Scholar). He served as an Area Chair for ICLR, NeurIPS and ICML. He received CCF Outstanding Doctoral Dissertation Award, Tsinghua Oustanding Postdoctoral Researcher, Microsoft Research Asia Fellowship, Baidu Fellowship, ByteDance Scholarship, etc.

报告摘要：

AI智能体（例如OpenClaw）正逐渐成为超越对话功能的智能系统，能够进行规划、使用工具并在数字环境中执行任务。本次演示将介绍AI智能体背后的核心概念，包括任务分解、工具使用、记忆机制、工作流自动化以及人机协同控制。报告将探讨AI智能体如何支持不同领域的研究与开发工作，例如文献综述、数据处理、实验规划、编程辅助、日程安排和知识管理等。

AI agents (e.g., OpenClaw) are emerging as intelligent systems that can go beyond conversation to plan, use tools, and execute tasks across digital environments. This presentation introduces the core ideas behind AI agents, including task decomposition, tool use, memory, workflow automation, and human-in-the-loop control. The talk will explore how AI agents can support research and development in different fields, such as literature review, data processing, experiment planning, coding assistance, scheduling, and knowledge management.