If John Connor was a gambler we would all be dead now.
Researchers at Carnegie Mellon University (Pittsburgh, USA) developed a Poker playing AI software called Libratus that won the Brains vs Artificial Intelligence event held at Rivers Casino. The computer software won the 20-day competition consisting of 120,000 hands of Heads-up No-Limit Texas Hold’em against 4 pro Poker players bringing home to the motherboard over $US1.7 million worth of chips.
Unlike many other strategy games, Poker has additional layers of uncertainty, such as the randomness of cards dealt, and the decision making/cunningness of competitors.
So what did Libratus do? It learnt how to bluff, and how to read its opponent’s bluffs. And apparently quite well.
Unlike other Poker playing programs that analyse the strategies of expert players, Libratus developed its own strategy based on 15 million core hours of computation.
Although this won’t spell the end of Poker tournaments or the rise of Skynet (probably), this AI breakthrough has the potential to give various computing platforms more strategic and negotiating ability. Basically games will now have a new difficulty level.
And you thought those NPCs were already difficult to negotiate with now.
If you’re worried about your gambling or someone else’s, Gambling Help Online is a free 24/7 online service for people who are affected by gambling.