AI Does The Unthinkable In Poker Tournament


There have been many cases in the past where artificial intelligence, or AI as it’s commonly referred to, has been put to the test in games against human players. Chess is a great example of an AI vs human player match up with the AI capable of beating even the world’s finest. Artificial intelligence has been put to the test against humans in poker tournaments this year, with poker posing a more difficult challenge for AI program developers than ever before.

A Unibet spokesman explained, the reason many expected artificial intelligence to struggle against human players in online casino and live poker tournaments is because poker is a game of what is known as “imperfect information.” This means that unlike chess for example, the AI never sees the game in full as each player is unaware of the cards that their opponents at the table possess. This makes things twice as hard for the AI as it has to judge each opponent individually on their play style and to then modify its own approach, while never giving away whether it possesses a good hand or not.

Unbelievably though, AI seems to reign supreme, even against the better human poker players around. At the start of the year, the AI went up against 4 of the world’s finest No Limit Texas Hold ‘Em players in a Pennsylvania casino. Over the course of the 20 day tournament, 120,000 hands were played and the AI, known as Libratus, won with a lead of over $1.7 million in chips. The margin of victory is a clear sign that it was down to the skill of the AI and not luck and that artificial intelligence could strategically out think even the best poker players on the planet.

It certainly was an extraordinary feat to see artificial intelligence prosper against such talented opponents. AI had been victorious in other games in the past, the likes of chess, checkers and scrabble were no match for artificial intelligence but up to the point of the tournament in Pennsylvania, AI had failed to win a game of imperfect information where the cards held by the other players were hidden. To win at Texas Hold ‘Em, whether it maybe on an online casino or at a live table, it requires skill, intuition and a bit of luck; but Libratus seemed to have the edge every time.

What the AI did well was that it knew when to bluff, when to bet big, when to change betting patterns to mislead those at the table. It had the ability to out think the some of the world’s best players, while also keeping them guessing at every turn. Nearly two years ago the creators of Libratus used AI to take on some other top poker players but the AI lost. This time around, the AI won and won comfortably leaving all four of the human players at the table with a negative amount of chips remaining after 120,000 hands had been played. It will certainly be interesting to see what challenges AI take on next and whether the results will be the same.