News Release

在终极游戏理论场景中人工智能系统打败了人类

Peer-Reviewed Publication

American Association for the Advancement of Science (AAAS)

AI System Beats Humans at the Poker Table, in Complex Game

image: Researchers have created an artificial intelligence system that outperforms humans at the ultimate, no-limit version of Texas Hold'em, a complex game involving 10<sup>160</sup> possible moves. This material relates to a paper that appeared in the 2 March, online issue of <i>Science</i>, published by AAAS. The paper, by M. Moravc ?&iacute;k at University of Alberta in Edmonton, AB, Canada, and colleagues was titled, "DeepStack: Expert-level artificial intelligence in heads-up no-limit poker." view more 

Credit: Carla Schaffer / AAAS

研究人员创制了一个人工智能(AI)系统,它在终极、无限版德州扑克(这种复杂的游戏含有10170种可能的玩法)中战胜了人类。在该结果报告的几个星期前,另一AI系统记录下了一个类似的重大突破:它在德州扑克单挑中战胜了人,这种游戏与无限版不同,它的可能玩法较少,为1014个。诸如国际象棋和围棋等游戏相对简单,因为棋手对可能的走法有着等同的信息。让扑克牌戏难得多的原因是每个玩家对可能的玩法有着不对称的信息。在这里,Matej Morav ík研发了一个被称作DeepStack的AI系统,它在每3000次游戏之后具有统计意义地打败了11名职业扑克选手中的10名(它也打败了第11名选手,但其结果不具统计学显著意义)。在游戏的每个时间点,DeepStack会重新计算游戏策略;它并没有将可能玩法计算到最后,它对允许进行的玩法深度和类型做了限制。它将可能的决策点从超过10160个减少至大约1017个,DeepStack的电脑芯片能在5秒钟内完成决策。这种AI系统会令涉及信息不对称的现实世界问题(如捍卫战略资源及做出重要的医疗建议等)的解决取得进展。

###


Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.