Artificial Intelligence Developed to See Through the “Poker Face”
30 April 2018
PokerBrave (2828 articles)
Share

When players are at the poker table, there is plenty of deception that goes on. One of the biggest things that players attempt to hide is any expression on their faces of how good or bad their hands are – the proverbial “poker face.” According to scientists, they have now developed an artificial intelligence that can see through anyone’s attempts to hide their emotions and feelings.

In a recent TED Talk, Dolby Laboratories scientist Poppy Crum explained how the company came up with the software to be able to break down what someone’s disposition is. “It’s the end of the poker face,” she said during the TED discussion. “We broadcast our emotions…we will know more about each other than we ever have.”

Using a combination of sensors and monitors along with a sophisticated interpretation program, Crum states that micromovements of the face, a rise in skin temperature and other indicators can be read and interpreted for a variety of emotional reactions. Such things as lying, infatuation, excitement, preparing to fight and other reactions can be identified through the analysis of all this information in a split second by the AI.

The potential implications at the poker table could be significant. If the AI can read – and, perhaps more importantly, interpret the information – correctly, such things as an increase in pulse rate or breathing, the dilation of a player’s pupils, changes in body temperature and several other bodily functions could be examined and a pattern predicted by the computers. If a player were able to get these things in real time, it would be a massive benefit to a player on the felt in sizing up his opponents.

Like most technologies, this breakthrough in artificial intelligence isn’t limited to its usage at the poker tables. The technology could be used to better identify how a human is acting and determine a path forward in communications, negotiations, and other areas. It could also be used for less than altruistic purposes when someone wants to manipulate a person or their actions.

The advent of computers has led to some fascinating breakthroughs in the activity of games. First computers were unleashed against the finite game of chess, with computers being trained in the game to the point that they were able to take on grandmasters of the game. In 1996, a creation of IBM called Deep Blue took on legendary Grandmaster and former World Champion Gary Kasparov, with Kasparov able to handily beat the computer. Fast forward one year and Deep Blue, with another year of development and improvement to its calculations, defeated the chess legend.

With chess out of the way, other computer programmers and scientists began to look at other games they could conquer. The professors and students at the University of Alberta began working on AI that took on the game of Heads Up Limit Texas Hold’em. The reason for the parameters of the game – the heads-up nature and the limit format – kept the number of decisions down for their artificial intelligence.

In 2008, the team at the University of Alberta came up with Polaris, who lost to professional poker players Ali Eslami and Phil Laak in a much-ballyhooed “man vs. machine” battle. After making some critical changes and expanding the AI’s database, the Alberta braniacs brought Polaris back for another round with Eslami and Laak, with the final score coming out in favor of Polaris, 3-2-1.

Not satisfied with their work with Polaris, the University of Alberta experts set about building a poker AI that would play the game of Limit Hold’em perfectly. In 2014, that work brought about Cepheus, which taught itself the entirety of mathematical challenges faced in Limit Hold’em in 70 days of play. In 2015, the University of Alberta stated that Cepheus had “perfected” the game of Limit Hold’em; players can take on Cepheus and learn for themselves how good the program is by playing against Cepheus here.

Alberta’s greatest minds then shifted their focus to NO LIMIT Texas Hold’em, with similar success. In 2015, the professors and scientists presented Claudico, the first heads up No Limit Hold’em playing AI and presented it for battle against four poker professionals including Doug Polk and Bjorn Li. Over the span of 80,000 hands, the AI held its own as there wasn’t enough of a statistical difference to determine a winner. In 2017, Alberta presented Libratus for a second “Brains vs. Artificial Intelligence” battle, with Libratus destroying Jason Les, Daniel McAulay, Jimmy Chou and Dong Kim to the tune of $1.7 million.

With the rise of Libratus and the ability to decipher a poker face, is it long until the computers take over the game of poker? Well, right now yes. Because of the complexities of No Limit Texas Hold’em, AI has only been created for the heads-up game. It also hasn’t been created for games like Stud, Omaha or other variants. Thus, while the news may be in favor of the development of AI and “poker face” decipherers, it should be some time before sentient beings are taking humans money on the real-life poker tables.