The Computational Gauntlet of Human-Like Learning

Location: BBB 3725
Zoom: https://umich.zoom.us/j/94393573970
Meeting ID: 943 9357 3970
Passcode: aiseminar

Abstract

In this talk, I pose a challenge for the AI research community: to develop systems that learn like humans. I illustrate this idea with two domains — mathematics and driving — where people are effective learners. I also review briefly the history of machine learning, noting that early work made close contact with cognitive psychology but that this is no longer the case. After this, I identify characteristics of human behavior that can serve as a ‘computational gauntlet’ and that, if reproduced, will offer better ways to acquire expertise than statistical induction over massive training sets. In addition, I review four AI systems — some older and others more recent — that pass most of the gauntlet’s obstacles and thus can serve as role models for future work. In closing, I suggest ways to encourage more research on the challenge of human-like learning.

Langley, P. (2022). The computational gauntlet of human-like learning. Proceedings of the Thirty-Sixth AAAI Conference on Artificial Intelligence ((pp. 12268-12273). Vancouver, BC: AAAI Press. http://www.isle.org/~langley/papers/gauntlet.aaai22.pdf

Bio

Dr. Pat Langley is a Principal Research Scientist at Georgia Tech Research Institute and Director of the Institute for the Study of Learning and Expertise. He has contributed to AI and cognitive science for more than 40 years, publishing over 300 papers and five books on these topics. Dr. Langley developed some of the first computational approaches to scientific knowledge discovery, and he was an early champion of experimental studies of machine learning and its application to real-world problems. He is the founding editor of two journals, Machine Learning in 1986 and Advances in Cognitive Systems in 2012, and he is a Fellow of both AAAI and the Cognitive Science Society. Dr. Langley’s current research focuses on architectures for embodied agents, explainable, normative, and justified agency, and induction of dynamic process models from time series and background knowledge.