Artificial Intelligence Masters the Game of Poker

Artificial Intelligence Masters the Game of Poker

The game of poker is not just about luck. With the right amount of practice, you can get the edge on your opponents. The good news is that there are some new artificial intelligence programs that are looking to help. The most popular ones in the industry include DeepStack and Pluribus. Whether you’re into poker or not, these are some great programs that can benefit you.


Pluribus is a computer program that has reached superhuman levels against professional poker players. It used a process that radically changed the way it analyzed games. It combines information abstraction and real-time search to determine which strategy to use.

The software used a blueprint strategy, a method that requires less computing power than actual game play. Instead of calculating odds to the end of the game, it only looks ahead a few moves at a time. This approach allows the program to improve over time. It also has the flexibility to change strategies based on information it learns.

It also has the ability to bluff without lying. The program does not fear the act of bluffing, as it identifies bluffing as an action that makes the most money.

During the game, it bets in a manner that seems random to the human opponent. It bets a weak hand as a so-so hand, and it folds a decent hand as a bad hand.


Libratus is the brainchild of Carnegie Mellon University scientists. It is an artificial intelligence system designed to play heads-up no-limit Texas hold’em. The algorithm devises an overall strategy based on Nash equilibria.

The algorithm is part of a three-component system, which includes a supercomputer in the background. The algorithms run during tournaments. They sift through hand data and decide on a seemingly crazy strategy. Then they calculate optimal plays based on game theory.

Libratus’ main purpose is not to win the tournament. Instead, its purpose is to find excellent moves that human players are not aware of. It’s also meant to tackle information-imperfect situations.

Libratus’ strategy is essentially the mathematically sound but unorthodox use of bluffing tactics. But the strategy is only the first of many facets to this algorithm’s complex strategies.

Libratus’ algorithms can also be applied to military robotic systems, medical treatment, business negotiations, and cyber security. In fact, the algorithms power many of the things that computers have been developed to do.

Brains vs AI: Upping the Ante challenge

In the “Brains vs AI: Upping the Ante” challenge, a computer played poker against four poker pros. Its victory was not a stroke of luck. But a breakthrough that opens new doors for artificial intelligence.

The Carnegie Mellon University researchers hoped to see if their AI could beat the best poker players in the world. It was the first time an AI had ever played poker against multiple players. The program, called Libratus, played 200 hours of hands over 20 days.

The software was programmed from scratch and has more computation than previous pokerbots. In the end, Libratus won more than $1.7 million in chips.

According to the head of the computer science department at CMU, this was the first time that a machine was able to make a positive play in a game. Its performance suggested that AI might have a chance to take over other fields.

During the competition, Libratus played against four poker pros. The four included Jimmy Chou, Dong Kim, Jason Les and Daniel McAulay. Each team was given one month to build a poker bot.


DeepStack is the first AI program to beat human professionals in the game of poker. This achievement is a mark of progress in machine intelligence.

The DeepStack system combines a fast approximation technique with the continual re-solving of game scenarios. It uses a neural network to determine the best moves. The DeepStack AI learns from experience, analyzing previous games to discover patterns and developing a strategy. It can play no limit Texas Hold’em.

DeepStack was developed by a team of researchers at the University of Alberta and Charles University. It used deep learning techniques to train its neural networks. This allowed it to play in the imperfect information games of no limit poker. It was tested against eleven professional poker players, averaging a 49 big blind win rate per 100 hands. It was also able to outplay the World Poker Tour bracelet winner, Phil Laak.

Unlike popular approaches, DeepStack does not attempt to play the full game of poker. Instead, it takes action based on current situations. Then it recalculates probabilities and strategies when new information is introduced.