People′s daily: the man-machine chess battle, the end result still suspense-Li Shishi, AlphaGO, go-IT information
I′m afraid no one would expect, one day the whole world had to go brush screen. A full commercial Li Shishi PK Alpha Dog will go the only niche projects prevalent in East Asia, suddenly becomes the ultimate wisdom duel that human and computers. This scene even reminiscent of science fiction films: highly intelligent robots are about to capture the last bastion of human sent the most intelligent warriors battle, who control the planet′s future in this game ... ...
But reality is always better than fiction plot of wonderful. Before the game, go virtually by unanimous decision, to Alpha Dog 5:0 swept the European Championship Fan Hui showed strength, is insufficient to challenge holds 14 world champion Li Shishi. Alpha Dog is subversive "deep learning", however, go with the overall situation, judgment, abandoning taking thought, can express is far from just a few months. Moreover, before the remains of the man-machine war use computer versions, according to Li Shishi "Alpha Dog and Fan Hui games are not professional" judgment, the victory seems to have tilted the balance.
Because of this, Li Shishi first defeat came a little bit off guard. Black led the weakness he seems to see through computer not good layout, but the Alpha Dog′s ability to exceed expectations. Especially when the Middle when situations are more complex, human proud of oncologists and intuition, eventually arrived, but prone to mood swings the biggest weakness. Li Shishi qifeng treacherous, good looking for flaws or manufacturing flaw, but make no mistake, break the computer opponent, instead, he fell into their "trap".
Old man-machine war, both the IBM computer "deep blue" defeated chess Grandmaster Garry Kasparov, or wave Tissot win over Chinese chess Grandmaster "group of five", the advantages of artificial intelligence also remain in the powerful computing capabilities and no change in mentality. And the key factor in such a decision, is whether computers really can rely on "neural networks" model human thinking. Go in addition to endless variation, thickness, taste, judgment is hard to quantify, such as the shape. Li Shishi PK Alpha Dog′s "landmark" values are for.
Li Shishi first why lose? Just a few hours after the game, has a variety of complex and interpretation. Comprehensive strength of the dominant, is widely seen as the Alpha Dog win this weight. If the trend from chess, Li Shishi had nothing to lose a game of chess, with his strength in the five-game comeback in May. Even if he is not like a computer "to settle", cannot always choose objectively the best hand, but many of the flight of the best weird, is also derived from this irrational, non-conventional "unstable". Once Li Shishi to make good this "double-edged sword", the person who wins or loses the war still in suspense.
But the lost first set, also revealed a stark fact: the rate of progress of artificial intelligence, far more than people expected. Even if the Li Shishi would eventually win, as long as the Alpha Dog Show match game, already shows computer master players of the "keys", beat the humans of that day will not be too far away. For professional players are concerned, this is perhaps an ambivalence. Between computers and chess is bound to move the project caught the attention of wider, and when humanity lost to computer at the competitive level, players of the mystery is this removed?
In fact, such worries don′t have to. Go is a competitive sport, but the culture and unique aesthetic value, is fundamental to the project endured, thousands of years. Even though the professional chess utilitarianism has been criticized, but Wen ping game′s charm and fun, is destined to go does not become a prisoner of computer. As the country and the chess-playing computer defeated, but population growth has been pushed to learn chess and gave birth to the computer self-taught master. Even go out the so-called "last bastion of wisdom", but it is not a science fiction movie, competing for the future initiative, still in their own hands.
人民网：人机围棋大战，最终结果仍存悬念 - 李世石,AlphaGO,围棋 - IT资讯
恐怕没有人会预料到，有一天全世界竟然为围棋刷屏。一场充满商业噱头的李世石PK阿尔法狗，将围棋这个仅盛行于东亚的小众项目，倏忽变成人类与电脑的终极智慧对决。这一幕甚至让人联想起科幻电影：高度智能的机器人即将攻占最后一个堡垒，人类派出最聪明的勇士应战，地球未来由谁控制在此一役……
只不过，现实永远比虚构的情节精彩。赛前，围棋界几乎一致判定，以阿尔法狗5：0横扫欧洲冠军樊麾时展现的棋力，尚不足以挑战手握14个世界冠军头衔的李世石。尽管阿尔法狗具有颠覆性的“深度学习能力”，但围棋所拥有的大局、判断、弃取等思维，远非短短几个月就能速成。况且，这次人机大战采用的仍是之前的电脑版本，按照李世石“阿尔法狗与樊麾的对局并非职业水平”的判断，胜利的天平似乎早已倾斜。
正因如此，李世石首战的失利来得有些猝不及防。执黑先行的他似乎看穿了电脑不擅布局的软肋，但阿尔法狗的应对能力却超过人们的预期。尤其当中盘局面趋于复杂时，人类引以为豪的大局观和直觉判断，终究抵不过易受情绪波动的最大弱点。李世石向来棋风诡谲，擅于寻找破绽甚至制造破绽，但面对不犯错误、步步紧逼的电脑对手，他反而落入自己的“陷阱”。
以往的人机大战，无论是IBM电脑“深蓝”击败国际象棋大师卡斯帕罗夫，还是浪潮天梭完胜中国象棋特级大师“五人组”，人工智能的优势还停留在强大的计算能力和毫无变化的心态。而这一次决定胜负的关键因素，在于电脑是否真能依靠“神经网络”模仿人类思维。围棋除了无穷尽的变化，更有厚薄、味道、形状等很难量化的判断。李世石PK阿尔法狗的“划时代”价值正在于此。
李世石为什么首战会输？赛后短短几小时，已有各种复盘与解读。综合实力的占优，被普遍视为阿尔法狗此役取胜的砝码。如果从棋局走势看，李世石输一盘棋并没什么，以他的实力在五番棋中有翻盘的可能。即便他无法像电脑一样“心如止水”，无法永远选择客观上的最佳手，但许多天马行空的妙手怪招，也正来源于这种非理性、非常规的“不稳定”。一旦李世石使好这把“双刃剑”，人机大战孰胜孰负仍有悬念。
但首盘的告负，也暴露出一个无法回避的事实：人工智能进步的速度，远远超过人类的预期。就算此次李世石能最终取胜，只要阿尔法狗展现出旗鼓相当的对弈能力，已然证明电脑掌握了围棋的“钥匙”，击败人类的那一天不会太远。对于职业棋手而言，这或许是一种矛盾心理。电脑与围棋的较量，势必推动这个项目引起更广范围的关注，而一旦人类彻底在竞技层面输给电脑，围棋的神秘面纱是否就此摘除？
其实，这种担忧大可不必。围棋是一项竞技，但其背后的文化底蕴与独特的美学价值，才是这个项目经久不衰、传承千年的根本。纵然当今职业棋坛的功利色彩备受指摘，但纹枰对弈间的魅力与乐趣，注定围棋不会成为电脑的俘虏。国象和象棋被电脑击败后，反而助推了学棋人口的增长，更催生出依靠电脑自学成材的大师。就算围棋失去了所谓“最后一块智慧堡垒”，但这毕竟不是科幻电影，争夺未来的主动权，仍在人类自己手中。