Artificial Intelligence defeats humans in poker for the first time

Libratus, an artificial intelligence project developed by Carnegie Mellon
University, made history after defeating poker pros for the first time.

Not surprisingly, machines have beaten humans in other strategy games, such as chess, but this time technology has developed an algorithm that teaches the machine the bluff technique, one of the foundations of this card game.

Poker is known as an unpredictable game, for this reason, a player's ability to bluff may be the line that separates a win from a loss. In previous tournaments, where humans stood face to face with AI, the trick up their sleeves was the possibility that they could bluff their advantage.
The fact that machines could not interpret opponents' behavior was their weakness in this game and therefore a poker game was a challenge immune to AI tactics for lack of information to analyze the game.

According to newspaper news The Telegraph, Libratus' victory was given in Philadelphia, USA, where the machine managed to win, with 1.7 million dollars in chips, four players of the world top, in a marathon that lasted 20 days. Of the total duration of the tournament, the professionals got only five days of victories.

Libratus was created in association with a professor of Computer Science at CMU, Tuomas Sandholm, and Noam Brown, a doctoral student. Tuomas Sandohlm stated that the ability of his fantastic invention to develop strategies in a game, even in situations of imperfect or incomplete information, surpasses that of the professionals of poker.

A computer capable of doing this presents a breakthrough in technology and science that may have many applications in the future.

The biggest challenge for the machines to win in a game like poker was to learn to bluff. As stated by Frank Pfenning, chief of the Department of Computer Science at CMU, the computer would be unable to win in poker if it did not learn to bluff.

The recognition of this AI in the marathon has already aroused the interest of several companies around the world, who seek to apply the algorithm of this machine to solve problems.

Adapting the algorithm, developed by Toamus and Noam, to new situations, it is predicted that it will be possible for machines to solve situations where information may be incomplete. Its future applications include negotiations, military strategies, cyber security and medical treatments.

Dong Kim, one of the professional players present at this event, had previously
possibility of challenging other CMU prototypes, one of them being Claudic, an 2015 project.
The player then stated that the tournament was still not in the middle, and he already knew that they would have no chance of beating Libratus. He further mentions that Claudicus' defeat in the previous tournament was due to the flaws in his program's algorithm, in which players were able to simply bluff until victory, and this time he felt the spell turned against the Sorcerer.

What are the next steps in advancing Artificial Intelligence? This will be the first
development of machines that think for themselves?