White, look over here! Li Shishi and complex Pan-human-computer chess world war AlphaGo war, AlphaGo, Li Shishi, go-IT information
Beijing time on March 9, Li Shishi AlphaGo chess five times in the first innings with Google Korea Seasons Hotel end . Li Shishi black 186 Middle lost to AlphaGo, losing five first game of chess.
Figure a, Li Shishi AlphaGo black VS Google. From the sequence, Li Shishi 1 obviously come in cold, or want to use AlphaGo no change to affect it in the database. Black 21 you can choose white flat 22 breaking, the purpose is to make the game into the open and complex situation, but in the actual combat is not favorable for black, Li Shishi intentional but it has paid a price.
Figure II, AlphaGo and judicious. Real AlphaGo does not take too much fighting, but 8-12, respectively 3 chess, although in the top 18-20 sell some variation of suspected, but by and large this complex battle situations, AlphaGo and judicious. But actual 24-26 of cross off some questions may be regarded as unnecessary exchange of stereotypes.
Figure III, AlphaGo loss. Real AlphaGo1 bit length is not necessary, from combat to see Li Shishi is relatively simple, but is the black 16 cover will undoubtedly make some, to hang white 17-right angle is slightly favourable situation can be thought of as Li Shishi.
Figure four, turn to the Thunderball. Yu bin thought, black belly down with help of the unthinkable, however should turn to win the upper hand to grab 3, this game also threatens the safety of White Dragon in its entirety, while black-right business, is overwhelming.
Figure five, chaotic battles. For the upper-left corner, Yu bin, the Li Shishi should take positive moves, from 3 bits out as early as possible, so as to give the white cause a lot of trouble. Real Li Shishi might have the advantage, however, instead of just simple.
Figure VI, Li Shishi technology variants. Li Shishi 1 hanging in the lower-left corner, AlphaGo judge risk on the left, off to fill weaknesses. Subsequently formed under the left shuangfeiyan characters, Li Shishi even went 2 hands but white has the upper hand, I completely distortion. Subsequently, the AlphaGo 16-24 can be seen as a series of coherent ideas, white 24 directly hit the black key, where Li Shishi has not worried about the place. From the perspective of this series contains detailed explanations, AlphaGo has a great depth of the big picture overall judgment and the ability to calculate.
Figure seven, missed the Exchange, if the Li Shishi when stereotypes found in a left-right issue, ready to go to point 1 in barbed exchanges, so can white scored 13 pointed right, black shape perfect.
Figure eight, Li Shishi did great damage on the ground. Combat compared Li Shishi in the right 1-3 should be forced, 3 on the right was cut out by white suffered heavy losses. Ke Jie believe that this is a major failure of Li Shishi.
Figure IX, Li Shishi to lose your cool. Figuration in the right, Li Shishi triangle not cool, should be 5 tip hold the lower right field, real AlphaGo13 mistakes, not 20 offensive after another, caused Li Shishi 18-20 tempo breaks 1 son, losses will not be small. But from a global view, when AlphaGo left corner holding 33 o′clock, Li Shishi has been a disadvantage.
For this Council, AlphaGo′s performance has far exceeded expectations in many complex where students know when. Global was notable 2 mistakes, but overall, Li Shishi deformation more of today′s technology, may want to computers, but play a disorder. Machine was cold, which is to its advantage. Said Li Shishi was lost today to complain.
小白看过来！李世石与AlphaGo大战复盘 - 人机围棋大战,AlphaGo,李世石,围棋 - IT资讯
北京时间3月9日，李世石与谷歌AlphaGo的五番棋第一局在韩国四季酒店结束。李世石执黑186手中盘不敌AlphaGo，输掉了五番棋的首局。
图一，李世石执黑VS谷歌AlphaGo。从序盘看，李世石1位明显走在一步冷招，还是希望用AlphaGo数据库中没有的变化来影响它。黑21面对可以平缓进行的局面选择让白22分断，目的在于让棋局进入开放而复杂的局面，但从实战来看对黑棋并不有利，李世石有意为之但也付出一定代价。
图二，AlphaGo进退有度。实战AlphaGo并没有采取过于激烈的战斗，而是8-12先后分别处理了3块棋，虽然在上边18-20有些卖掉变化的嫌疑，但是总体来说在这个复杂的战斗局面下，AlphaGo的处理可谓进退有度。但是实战24-26的跨断有些疑问，可视为不必要的定型交换。
图三，AlphaGo亏损。实战AlphaGo1位长其实没有必要，从实战来看李世石处理比较简单，但进行到黑16罩时还是无疑赚到一些，至白17右下挂角时，可以看做李世石稍稍有利的局面。
图四，雷霆万钧的一拐。俞斌认为，黑棋中腹落下后手匪夷所思，无论如何应当争到先手抢3位的拐，这手棋还威胁着白棋整条大龙的安危，同时有利于右下黑棋模样的经营，是雷霆万钧的一手。
图五，混乱的战斗。对于左上角，俞斌认为李世石应当采取积极的下法，及早从3位出动，这样可以给白棋造成很大的麻烦。然而实战李世石可能有了优势意识，反而下得简单了。
图六，李世石技术变形。左下角李世石1位挂角，AlphaGo判断左上危险，脱先补住弱点。随后左下形成双飞燕的性状，李世石连走2手最后却让白棋先手做活，技术完全变形。随后，AlphaGo的16-24可以看成是一系列的连贯思路，白24直接击中黑棋要害，这里也是李世石未曾担心的地方。从这一系列招法看，AlphaGo已经具备了非常深度的大局部整体判断和计算的能力。
图七，漏掉的交换，如果李世石在左下定型时发现右边的问题，随时可以走到1位点刺的交换，这样便可以在白棋打入右边时走13位尖，黑棋形状无懈可击。
图八，李世石实地大损。实战中李世石右边1-3的应法比较勉强，右上3子被白棋割掉损失惨重。柯洁认为这个是李世石主要的败因。
图九，李世石失去冷静。右下的定型中，李世石三角位压不冷静，应当在5位尖守住右下角实地，实战中AlphaGo13也出现失误，没有在20位先手接一子，导致被李世石18-20先手断下1子，目数损失不小。但是从全局看，当AlphaGo守住左上角的三三时，李世石就已经呈劣势了。
结合本局来看，AlphaGo的表现已经大大超出人们的预期，在很多复杂的地方懂得进退。全局出现了比较明显的2处失误，但是整体来看，李世石今天的技术变形更大，或许太想针对电脑，反而造成发挥失常。机器是冰冷的，这原本就是它的优势。可以说李世石今天是输得心服口服。