AI is coming for us![/caption]AI is coming for us!The year 2017 proved to be a watershed in the development of artificial intelligence (AI). AI achieved historic victories over human players in a range of games, notably poker. So, why should the non-poker players among us care? Why does it matter? The future of poker and everything else is set to change as AI takes a quantum leap forward in capability, with big implications for us all. Here’s a look at some of the areas.AI Destroys Poker ProsIn January 2017 at the Rivers Casino in Pittsburgh, Pa., a poker-playing AI called Libratus achieved a decisive victory against four seasoned poker pros in a 20-day competition billed as “Brains vs. Artificial Intelligence.”Developed by researchers at Carnegie Mellon University, Libratus battled the pros in over 120,000 hands of heads-up no-limit Texas Hold ‘Em, considered the “benchmark challenge for AI in imperfect-information games”. When the dust settled, Libratus finished up with $1,766,250 chips, leaving the pros to split a $200,000 prize purse as consolation for their landmark defeat.Then, in a study published in Science in March 2017, the pros were sent scuttling for cover again by an AI called DeepStack. The AI developed by researchers at the University of Alberta in Canada defeated 10 out of 11 players in the same variant of Texas Hold ‘Em. In a four-week tournament of 3,000 games and 44,000 hands, the AI again scored a decisive victory.[caption id="" align="alignnone" width="1024"]
The ancient Chinese strategy game Go: AI’s latest conquest.[/caption]The ancient Chinese strategy game Go: AI’s latest conquest.All Systems Go for AIIn 2017, another first came when AI conquered the ancient Chinese strategy game Go, a feat considered impossible by experts as recently as the 1990s. But in the last decade, advances in machine learning have led to breakthroughs in Go-playing AI. And by the final days of 2016, a player name “Master” began competing against players on the Asian game server Tygem and beat multiple champions. On January 4, 2017, “Master” was revealed to be Google DeepMind’s AI AlphaGo.Then, in May 2017, the number one ranked world Go champion Ke Jie took on AlphaGo and lost each of the three games in a crushing defeat. But Google wasn’t done. In October 2017, Google announced a vastly improved version called AlphaGo Zero, which learns entirely from playing against itself over and over. Naturally, this led to the release of AlphaZero, which can beat opponents in any game it is programmed to work with including chess and shogi (Japanese Chess), in December 2017.Chess is Different from PokerComputers beating humans at games is not a new phenomenon. In 1997, IBM’s Deep Blue beat world chess champion and grandmaster Garry Kasparov in a series of games, and in the 2000s, machines were regularly dominating their human opponents. So, what’s changed?In chess, the position of each piece on the board is always known. A computer can be taught what to do in each possible situation and then calculate the best move to make given all the possible scenarios. This “brute force” approach that powered Deep Blue was improved upon by Matthew Lai at Imperial College London, who in 2015, developed an improved AI-based chess called “Giraffe” on neural network technology.Poker is Harder for AI So, if human chess players are so easy for a computer to beat, what makes poker special?Unlike in chess where the position of each piece is known, in poker, a player does not know what cards their opponent holds. It’s what experts call an “imperfect information game.” To succeed at poker, a machine needs to learn how to bluff.The conditions of poker are an approximate of our everyday lives in which we need optimize our positions with imperfect information. Often, we’re dealing with an opponent that is trying to do the same. Take buying a new car for example. We need to know how much to offer and on what terms to obtain what we want for the lowest possible cost. The car dealer is trying to obtain the highest possible price.[caption id="" align="alignnone" width="1024"]
AI is set to revolutionize many industries.[/caption]AI is set to revolutionize many industries.Poker is a Great Test for AIPoker is such an enticing prospect for developers of new AI technologies. If they can discover the algorithms that help a computer beat human poker players, then those same algorithms could be applied in countless real-world settings, including security, business negotiation or buying a new car.According to the Libratus creators, Tuomas Sandholm and Noam Brown, “The techniques that we developed are largely domain independent and can thus be applied to other strategic imperfect-information interactions, including non-recreational applications. Due to the ubiquity of hidden information in real-world strategic interactions, we believe the paradigm introduced in Libratus will be critical to the future growth and widespread application of AI.”Is it the rise of the machines? It’s clear that for lovers of poker, playing online against AI-powered bots will likely mean certain defeat. AI will continue to improve with advances in technology, unlike its human counterparts. The challenge for online poker rooms will be to develop ways to keep the bots out and keep the players in. Right now, opinions among professional players on AI are mixed.Of course, people don’t only play poker online. If online poker becomes severely compromised by AI, then we might see a resurgence in good old-fashioned casino poker where humans can look each other in the eye and, at least for now, avoid the AI.Online poker rooms are already exploring new ways to keep players engaged, and emerging technologies like virtual reality hold significant promise. They could even help to establish a middle ground between anonymous online play and in-person play in physical locations.For the rest of us, these recent advances in AI will impact us in ways that we cannot yet conceive. Some like Elon Musk consider AI to be an existential threat to humans. He believes that AI power will become concentrated in the hands of a few companies who will then have “extreme power.” “Maybe there’s a 10 percent chance of us making AI safe,” he said recently.For the poker pros, it might be time to dust off the CV and find a real job. For the rest of us, only time will tell how AI will change the game. Stay tuned.
All Rights Reserved © 2024
