October 13, 2021
Artificial Intelligence (AI) finds use in many areas of our lives. In the gaming industry, games with pre-defined rules such as chess and checkers were the first to experience AI technology. Significant strides in the AI field have recently made a system that beats humans in Texas Hold’em, a game that requires strategy and intuition. The article will introduce you to a new era of gambling through the use of AI.
Many moves are needed to win in poker. In a game like Texas Hold’em, a player must anticipate the opponent’s moves, call bluffs, and play with an ever-changing strategy to win. For a long time, AI could only learn games that had pre-defined moves. The ability to teach a system to make decisions based on intuition and strategy has researchers very excited. The reason being, the technology can find many uses outside poker.
The steps to get to a working AI system started long ago. Before the discovery of Pluribus, other systems existed. Here is a brief description of each:
DeepStack - DeepStack was developed by a team from the University of Alberta in Canada. Through a blend of deep learning algorithms, the system could win in two-player Texas Hold’em. According to the developers, DeepStack makes decisions based on neural networks. As proof of concept, that system played 44,852 poker games. The results were ten times the wins professional players would expect in such a situation.
Libratus - Two researchers from Carnegie Mellon University joined efforts to build Libratus, an unbeatable system in two-person poker. To showcase the ability of this fantastic system, the team put it up against four experienced players in a 20-day poker tournament. In the end, the AI system was up $1.8 million chips.
Pluribus - Pluribus is by far the most successful AI project in poker. It was developed through collaborative efforts from Facebook and researchers from Carnegie Mellon University. The AI system played in 10,000 poker games where the opponents were 5 of the best earners in poker. On average, the wins recorded by Pluribus were $480 for every 100 games. Such an impressive winning streak can only be achieved by very few professional poker players.
The above systems paint a clear picture of what AI technology can achieve. Poker leaners can become skilled by practicing against AI Systems. That way, they don’t have to lose money by playing trial and error games like in the olden days.
The algorithms used by the AI systems to beat players in poker can be used to make decisions in other areas. For example, self-driving cars need to make the best decision from numerous possibilities in a split second.
Researchers have barely scratched the surface of Artificial Intelligence. The possibilities in the field are limitless. Hopefully, AI systems will learn even more games in the future. As you can tell from the sections above, advancement in one field opens discovers techniques that find use in other areas of our lives.