Ai Will Make Online Poker Extinct Within 5 Years
An artificial intelligence machine performed trillions of calculations against itself in order to perfect bluffing. The AI machine, Pluribus, then significantly beat five top poker players. Researchers at Facebook and Carnegie Mellon created Pluribus, which is a huge leap forward for AI’s capabilities.
Pluribus is the latest Ai machine to beat humans at their own game. In recent history, AI machine Deep Blue beat former world champion, Garry Kasparov, Google’s DeepMind mastered Go, and Libratus, Carnegie Mellon’s first poker machine, bested a poker star in a one-on-one game. Additionally, AI machines have bested humans in checkers, Jeopardy, Dota-2, and so on. Nevertheless, Six-man Texas Hold-Em is more than just the latest in a long line of conquered games.
Pluribus is a step above the rest because of its ability to deal with uncertainty while using less memory. Pluribus ran on 128 gigabytes of memory using only two chips. For reference, DeepMind (chess AI) used 480 custom chips, AlphaGo (Go AI) used 1920 chips, and Libratus (one on one poker AI) used 100 chips. Pluribus developed its winning strategy and bluffing by playing trillions of hands against five clones of itself. Following each round, it analyzed its decisions and if it won the hand, it was more likely to opt for those strategies. The bot realized it could win by making a bet even when it had a weaker hand (bluffing).
Through using its winning strategies, the bot tore through professional opponents. Note that in the game played between professionals and Pluribus there was a standard deviation of 1 big blind per 100 hands and a mean of 0 big blinds lost (or won) per 100 hands. Individual results against the bots ranged anywhere from -.5 to -4 big blinds per hundred hands. In aggregate, the humans lost by 2.3 big blinds per 100 hands. Hence, aggregate humans lost by a statistically significant amount because they were below two standard deviations of the mean. Thus, although technically not every human significantly lost, the bot was vastly better than the aggregate.
Pluribus certainly could be the end for online poker (over a one-billion-dollar industry), but this does not mean it is all bad. Pluribus is a big step for artificial intelligence in the way it deals with uncertainty and deception (after all poker is a game of deception). This kind of AI technology could have applications in defense, financial services, gaming, or health care. However, for now, Pluribus is a gentle reminder that data is king in our world.
Interview with NASA Astronaut Scott Kelly: An American Hero
Why Choose Machine Learning Over A Traditional Financial Advisor?
On Black Holes: Gateway to Another Dimension, or Ghosts of Stars’ Pasts?
Tesla's Augmented Reality & Virtual Reality Founder Tyler Lindell Opens Up
Supersonic Travel: The Future of Aviation
Lockheed Martin Confirms the SR-72 – Son of Blackbird Will Reach Anywhere in the World in One Hour
Shedding Light on Dark Matter: Using Machine Learning to Unravel Physics’ Hardest Questions
Interview with the Inventor of Amazon's Alexa
Aquaponics: How Advanced Technology Grows Vegetables In The Desert
The World Cup Does Not Have a Lasting Positive Impact on Hosting Countries
The Implications of Machine Learning on Condensed Matter Physics & Quantum Computing
Written by Willie Turchetta, Edited by Matthew Durborow & Alexander Fleiss
Sources: