The battle between humans and Artificial Intelligence has been a long-running one, and one that is transforming all the time. One interesting sub-plot to the advancement of AI is in gaming, and whether AI can ever have the tactical whereabouts to beat human intelligence time and again. AI may have previously beaten world-class humans at games such as backgammon, Othello, Connect Four, drafts and chess, but recent victories have been even more impressive.
In 2016, AlphaGo, a system designed by the Google-owned AI firm DeepMind, defeated South Korean professional Lee Sedol at the notoriously complex Go.
The bots ‘Go’ on
Sedol managed to win just a single game compared to AlphaGo’s four during the five-match series. The robot’s success has been called an AI milestone due to the complexity of the game, which includes more possible moves than atoms in the universe.
The technique of DeepMind differs from the traditional ‘brute force’ power used to win. Its deep learning technology enables it to calculate rules from vast amounts of data. It is then asked to learn from the experience- not dissimilar to a child being taught to communicate.
It has been one year since DeepMind beat the world’s Go champion and close to two decades since the IBM chess-playing computer Deep Blue defeated the world champion of chess, Garry Kasparov, over six matches. This year, AI turned its attention to a game previously considered too difficult for robots to get to grips with.
Bots know when to hold ‘em
Earlier this year, AI proved that it could play poker by defeating professional players at Texas Hold’ Em. Researchers at DeepStack broke their losing streak by combining deep machine learning and new algorithms.
Over four weeks, DeepStack was pitted against 33 poker players, chosen by the International Federation of Poker. The players went up against DeepStack in 44852 games of heads-up no-limit Hold ‘em, a two-person version of the game which allows players to bet an unlimited amount of money; as long as they have it, of course. DeepStack won significantly in the end.
Poker, while certainly containing an element of luck, is still very much a game of strategy, which will always play to the strengths of AI. While the bots may have gotten to grips with poker, other casino-style games are a different proposition. Strategy may be a part of Blackjack, for example, but the game remains mainly about luck. So far those who enjoy playing games such as blackjack or roulette at an online casino at Casino Euro can enjoy their 200% bonus without the threat of having to worry about taking on undefeatable robots; at least for now, anyway. Meanwhile, the bots find other games to dominate.
No-one can defeat Elon Musk
Earlier this year, during the annual Valve Dota 2 tournament, an unannounced segment introduced possibly the world’s best player- a bot from OpenAI- the Elon Musk-backed bot. The nonprofit’s engineers said that the bot was ready to defeat professionals after just 14 days of real-time preparation, claiming it had amassed multiple lifetimes of experiencing during those two weeks. Musk hailed the achievement, the first where AI beat professionals in e-sports competition.
The demonstration may have been limited to just a few gameplay variables, but it was still quite something to see the bot crush crowd favourite Danylo Ishutin during a live 1-on-1. After two defeats, Dendi forfeited the rest of the matches, expressing surprise by how easily a human could be taken apart by a bot.