Toward Human-Like AI: Cognitive Architecture, Common Model of Cognition & Interactive Task Learning

Abstract:

AI is in weird place right now. There is more talk than ever about Artificial General Intelligence, but the emphasis appears to be on slices of System 1: recognition, classification, or reactive action with very little on the internal processing that is characteristic of human cognition. Within those slices, there is human-level or even super-human performance, but these are very thin slices, focused on one problem or phenomenon to the exclusion of developing anything that can work on many different problems, or integrate with other cognitive capabilities. In contrast, humans are flexible and general – they can work on many different problems, switching effortlessly from one task to another, using a wide variety of different cognitive capabilities. Moreover, they can learn new tasks from scratch in real-time from natural language instruction. In this talk, I take us on a journey through attempts to create human-like AI agents using cognitive architectures that are designed to integrate the cognitive capabilities we find in humans. I also explore the Common Model of Cognition, an attempt to unify research results from across cognitive architecture and cognitive science. Finally I discuss our research on building AI agents that integrate cognitive capabilities (natural language processing, planning, perception, motor control, memory, decision making, and learning) to support Interactive Task Learning (ITL). We have developed an AI system that is embodied in a variety of robotic platforms and can learn through natural language over 50 games and puzzles as well as complex mobile robot tasks.

Bio:

John E. Laird is the John L. Tishman Professor of Engineering at the University of Michigan, where he has been since 1986. He received his Ph.D. in Computer Science from Carnegie Mellon University in 1983 working with Allen Newell. From 1984 to 1986, he was a member of research staff at Xerox Palo Alto Research Center. He is one of the original developers of the Soar architecture and leads its continued evolution. He was a founder of Soar Technology, Inc. and he is a Fellow of AAAI, AAAS, ACM, and the Cognitive Science Society. He is the 2018 co-winner of the Herbert A. Simon Award for Advances in Cognitive Systems.