Uploaded May 2025 | Updated September 2026, 2 weeks ago
Columbia Engineering's Lecture Series in AI explores cutting-edge topics in artificial intelligence and brings to campus thinkers and leaders shaping tomorrow’s technology landscape in a wide variety of fields. Join us to unravel the complexities and possibilities of AI in today's rapidly evolving world.
The success of large language models (LLMs) like OpenAI’s GPTs has fascinated us, with their astounding capabilities in a wide range of tasks such as genius-level standard test performance, IMO-level math reasoning, encyclopedic-like question-answering, and human-like ability in conversation and writing. However, LLMs suffer inherent limitations in their capabilities for embodied reasoning, physical reasoning, social reasoning, and strategic planning thereupon, in the real world.
This talk features Eric Xing, Mohamed bin Zayed University of Artificial Intelligence. He introduces “PAN”. a new architecture for General and Purposeful Reasoning toward such capabilities, which includes a new World Model capable of instructible and steerable simulation of an actionable universe (of any next-world resultant from agentic action), and simulative reasoning (through experiments over alternative next-worlds) in such a virtual universe; and a new Agent Model capable of autonomously acting, planning, strategizing, learning, and adapting in this universe toward complex objectives and over diverse platforms and tasks.
Columbia Engineering's Lecture Series in AI explores cutting-edge topics in artificial intelligence and brings to campus thinkers and leaders shaping tomorrow’s technology landscape in a wide variety of fields. Join us to unravel the complexities and possibilities of AI in today's rapidly evolving world.
The success of large language models (LLMs) like OpenAI’s GPTs has fascinated us, with their astounding capabilities in a wide range of tasks such as genius-level standard test performance, IMO-level math reasoning, encyclopedic-like question-answering, and human-like ability in conversation and writing. However, LLMs suffer inherent limitations in their capabilities for embodied reasoning, physical reasoning, social reasoning, and strategic planning thereupon, in the real world.
This talk features Eric Xing, Mohamed bin Zayed University of Artificial Intelligence. He introduces “PAN”. a new architecture for General and Purposeful Reasoning toward such capabilities, which includes a new World Model capable of instructible and steerable simulation of an actionable universe (of any next-world resultant from agentic action), and simulative reasoning (through experiments over alternative next-worlds) in such a virtual universe; and a new Agent Model capable of autonomously acting, planning, strategizing, learning, and adapting in this universe toward complex objectives and over diverse platforms and tasks.
