教书育儿做科研分享 http://blog.sciencenet.cn/u/danfouer 以文会友,以理治学,文理交融,教学相长。

博文

AI:诸侠的黄昏,浪子的天堂!

已有 4516 次阅读 2017-1-17 22:10 |个人分类:思考评论|系统分类:海外观察| CMU, 德州扑克, 人机大战

——零距离独家探访CMU德州扑克人机大战

先上几幅图,验明正身,说明我真的是去看了。

美国宾西法利亚州匹斯堡市Rivers 赌场


赌场扑克大厅,本来想拉个360度全景的,被魁梧壮实的黑人保安给制止了,拍了一小半...

宣传彩页和现场实况投影,请大家注意拿着奖杯的这位亚裔小鲜肉


JIMMY CHOU

Jimmy “ForTheSwaRMm” is a leading Heads-Up No-Limit Texas old’em specialist. He is considered one of the top three Heads-Up No-Limit sit-and-go pros and one of the top ten Heads-Up No-Limit cash game pros. He won the 2015 ACOP (Macao) Main Event, and was runner up of the 2014 WCOOP (PokerStars) Heads-Up tournament. He has also made many appearances in high-roller tournaments of buy-ins of $100,000 and beyond. Jimmy currently lives in Las Vegas and can often be found playing the high-limit cash games at the Bellagio poker room.

上面的话简单来说就是说他是德州扑克界的天下三大高手之一,常年混迹于拉斯维加斯的:


不仅如此


人家还很热情,看到我拿着相机过来居然主动站起让我拍照


一聊才知道,人家出生在湖南,跟我这湖北佬算半个老乡。


自己的牌局率先结束了,到隔壁的兄弟那里观战,我顺便拍了拍他们的对手——LIBRATUS

Libratus is an artificial intelligence (AI) program for playing Heads-Up, No-Limit Texas Hold’em. It was developed at Carnegie Mellon University’s School of Computer Science by Prof. Tuomas Sandholm and Ph.D. student Noam Brown. Libratus’s strategy is not based on the experience of expert human players, so its game play could differ markedly from the pros. It uses algorithms to analyze the rules of poker and set its own strategy, based on approximately 15 million core hours of computation at the Pittsburgh Supercomputing Center (PSC). Libratus will continuously sharpen its strategy during the Brains vs. AI competition, performing computations with the PSC’s Bridges computer each night while the pros get some shuteye. During games, Bridges will perform live computations to aid Libratus with its end-game play. The algorithms that created Libratus are not specific to poker. The AI’s ability to reason when faced with incomplete or misleading information have a wide range of possible applications, including business negotiation, medicine, cybersecurity, auctions and more.

上面的话重点如下:

  1. 俺不屑跟你们人类学,俺自己分析规则定策略,俺在匹村超级计算中心的八卦炉中修炼了15,000,000小时;

    注:15,000,000小时=5625,000天=1712.328767123288年

  2. (原来是只千年老妖啊!!!——更准确地说,千头怪兽,估计用了几千个CPU)

  3. 俺边比赛遍修炼,你丫睡觉了俺上网找别人练;

  4. 俺的每一次决策都是实时计算的。


隔壁地兄弟叫做:

JASON LES

Jason “PremiumWhey” started playing poker professionally 10 years ago while completing his degree in Computer Science at UC Irvine. After many years of playing ring poker, he began to specialize in Heads-Up No-Limit Texas Hold’em in 2013. Hard work and dedication paid off as Jason took down opponent after opponent as he rose up the stakes and now sits among the best players in the game. Jason finished in the Top 40 at the 2016 World Series of Poker Main Event. He participated in the 2015 CMU Brains vs. AI challenge to play against the Claudico bot.

Jason 拥有加州大学欧文分校地计算机科学学士学位,曾参加过去年CMU德州扑克的人机大战,在那场比赛中,人类选手击败了Claudico机器人,赢得了总共70万美元的奖金。


下面是我对JIMMY CHOU用英语进行的独家采访,因为担心影响人家比赛,没好意思多问。

:整体情况怎么样?

CHOU:还可以,我们开始落后,但是现在赶上来了。

:机器人做对手跟人类做对手有什么不一样?

CHOU:完全不一样,我们花了几天才适应他们的策略。

:你能给我解释以下屏幕上牌局吗?

CHOU:好的,我同时在进行两副牌局,左边这一副牌局我赢了计算机5000分,右边这一副牌局我输给计算机11000分,所以我总共输了6000分;但是在楼上,我有一个镜像队友,他拿着我这边电脑的牌,而他的电脑拿着我这边的牌,这样可以保证绝对公平。

:喔,就跟桥牌比赛一样,同一幅牌在镜像队友之间交叉。

CHOU:是的,我的队友现在赢了10000分,所以我们现在略微领先。

:你听说最近有一篇文章,说AI在德州扑克上已经击败人类选手了,你知道吗?

CHOU:我知道,但是不知道他们规则是不是像我们这样杜绝了运气的成分,还有就是据我所知,他们的对手并不都是职业选手,有一些,但不是全部。

:最近Master干掉了所有的围棋大师,包括我们中国的。

CHOU:是啊,不完全信息博弈是我们唯一的机会。

我:你们是在代表人类而战,加油!

CHOU:我们会在这里连续比赛二十一天,我们会努力的

:你什么时候开始下一场比赛?

CHOU:我们可以不停的跟电脑比赛,但是我们决定放慢节奏,等Jason他们队比完了,两个队沟通商量一下再同时开始下一场。

:我不耽误你们比赛了,祝你们获胜,再见!

CHOU:谢谢,再见!


从赌场neng'go出来我一直在想一个问题,最能够代表人类智慧的巅峰的,是中国棋院十年如一日培养出来的柯洁、古力,还是这些在赌场上征战多年,时刻用真金白银淬炼自己的赌神? 前者如金庸笔下的大侠——郭靖、萧峰,正大光明,一板一眼,认输也认得干脆;后者如古龙心中的浪子——陆小凤、楚留香,行云流水,自由自在,永不言败。人类可能因为战胜不了AI而放弃围棋,却绝不可能因为玩不过机器而不去赌博。这是否意味着,在浪子的心中,在赌的过程中,包含着人类更为本质的生命元力,而这个,才是人类智慧的根!


而黑客,本质上都是浪子!




https://blog.sciencenet.cn/blog-48882-1028263.html

上一篇:自省打油诗三则
收藏 IP: 155.247.170.*| 热度|

0

该博文允许注册用户评论 请点击登录 评论 (1 个评论)

数据加载中...

Archiver|手机版|科学网 ( 京ICP备07017567号-12 )

GMT+8, 2024-4-26 18:57

Powered by ScienceNet.cn

Copyright © 2007- 中国科学报社

返回顶部