Haifeng Xu- Towards Agentic Intelligence: Past Insights, Future Directions
The dream of building intelligent agents has driven artificial intelligence (AI) since its inception. From early pioneers designing autonomous agents to play games such as chess and checkers, to building robots, to the study of multi-agent systems for coordinating rational actors (human and machine alike), and now to today’s language model agents accomplishing unprecedentedly complex tasks, the pursuit of agentic intelligence has been central to the field.
In this talk, I will reflect on key insights and lessons from my past few years of research on studying systems with intelligent agents, illustrated through concrete researches from our lab such as: (a) how to employ multi-agent system modeling and optimization to coordinate rational agents’ behaviors towards better welfare; and (b) how to enable strategic communications among agents in foundational planning problems. With the rapid advances of large language models, research on AI agents is now at an especially exciting stage. I will conclude by sharing some personal thoughts on interesting future directions towards agentic intelligence.
Speakers
Haifeng Xu
Haifeng Xu is an assistant professor in computer science and data science at the University of Chicago, and directs the Strategic Intelligence for Machine Agents (SIGMA) research lab which focuses on AI agents and data-driven decision making. His works have been recognized by a few awards, including an AI2050 Early Career Fellow, IJCAI Early Career Spotlight, a WWW best paper and a Google Faculty Research Award.
