Rutgers computer scientist Michael Littman had a front-row seat to four hot rounds of Texas Hold’em poker, except he was neither in Las Vegas nor did he have his own cards or money on the table. Littman was in Vancouver, B.C., at the world’s leading artificial intelligence conference, serving as official arbiter of a poker match between a computer and two expert human contenders. Only at the end of the last round late on the night of Tuesday, July 24, 2007 was Littman able to declare the human contestants narrow victors.

“I was expecting a draw,” Littman told reporters covering the event, noting that in the first three rounds that began early Monday, the competitors played their first round to a draw and traded victories in the next two. The match pitted two top-notch players from Los Angeles against a program named Polaris designed by University of Alberta computer scientists. The program’s lead architect was himself a one-time professional poker player.

“The poker stars were really impressed with the level of play, and they worked hard to build and hold a lead – it was exciting and interesting to watch,” Littman said.

Michael Littman
For years, computers have upstaged humans at games such as checkers and chess, in which logical principles and clear-cut rules govern play. But poker is a psychological game that involves deception. So even though Polaris lost in the end, attendees billed it a milestone for artificial intelligence.

Littman and his Rutgers colleagues were hands-down winners in another competition at the meeting, the 22nd Conference on Artificial Intelligence sponsored by the Association for the Advancement of Artificial Intelligence. In the conference’s first-ever “AI Video” competition, he brought home a “Shakey” award for best short video, showing the lab’s Sony robotic dog responding to cues to perform simple tasks. Competition organizers named their video awards after the first robot to reason about its actions, dubbed Shakey by its Stanford University inventors four decades ago. The videos illustrate progress in AI and show how researchers can have fun pursuing new challenges.