Can artificial intelligence learn to read poker tells and figure out when someone is bluffing? AI engineer Luke Geel (geelai.net) created software designed to do just that, and his work was featured this year on the TV coverage of the World Series of Poker Main Event. I’m the author of some poker tells books (readingpokertells.com), and when I first heard about this program, I wondered: what exactly is his program doing? And how well does it work? In this episode, Luke explains how he built and structured his program. We discuss what his software can actually do, and why teaching machines to read behavior in poker is much harder than it might seem. Along the way, we explore some challenges of machine learning, including how AI finds patterns, why understanding context and personal variation matters, and how researchers can distinguish meaningful behavioral signals from noise. We also talk about Luke’s newer work using AI to analyze sales conversations, with the goal of identifying subtle clues about people’s intentions and discovering which communication strategies are most persuasive. This is a talk about poker, AI, psychology, and the possibilities and limitations of using machines to understand what people are thinking and feeling.

Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.