Where Computers Defeat Humans,and Where They Can’t 計算機的過人與不足之處

文/安德魯·麥卡菲 埃里克·布林約爾松 譯/蔣威 By Andrew McAfee& Erik Brynjolfsson

Where Computers Defeat Humans,and Where They Can’t 計算機的過人與不足之處

文/安德魯·麥卡菲 埃里克·布林約爾松 譯/蔣威 By Andrew McAfee1& Erik Brynjolfsson2

AlphaGo, the artificial intelligence system built by the Google subsidiary3subsidiary子公司。DeepMind, has defeated the human champion, Lee Se-dol4韓國著名圍棋棋手，世界頂級圍棋棋手。2016年3月9日起，李世石與谷歌計算機圍棋程序AlphaGo進行圍棋人機大戰。截至3月15日，李世石不敵人工智能AlphaGo，以總比分1∶4落敗。, four games to one in the tournament of the strategy game of Go. Why does this matter? After all, computers surpassed humans in chess in 1997, when IBM’s Deep Blue beat Garry Kasparov5俄羅斯國際象棋棋手，國際象棋特級大師，在22歲時成為世界上最年輕的國際象棋冠軍，是第13位國際象棋世界冠軍，此后又數次衛冕成功。. So why is AlphaGo’s victory significant?

由谷歌旗下DeepMind公司創建的人工智能系統AlphaGo在一場圍棋比賽中以4比1的成績擊敗人類圍棋冠軍李世石。此事為何意義重大？畢竟，早在1997年IBM公司的“深藍”擊敗加里·卡斯帕羅夫之后，計算機在國際象棋領域就已經超越了人類。那么，為何AlphaGo的勝利還不容小覷呢？

[2]圍棋和國際象棋一樣，也是一種極其復雜的講究策略的游戲，不可能靠巧合和運氣取勝。對弈雙方輪番將白色或黑色棋子落于縱橫各19道線的網格棋盤上；若棋子四面被另一色棋子圍住，則需將其從棋盤上提走，最終棋盤上圍占地盤較大、吃子較多的一方獲勝。

[2] Like chess, Go is a hugely complex strategy game in which chance and luck play no role. Two players take turns placing white or black stones on a 19-by-19 grid; when stones are surrounded on all four sides by those of the other color they are removed from the board, and the player with more surrounded territory and captured stone at the game’s end wins.

[3] Unlike the case with chess, however, no human can explain how to play Go at the highest levels. The top players, it turns out, can’t fully access their own knowledge about how they’re able to perform so well. This self-ignorance is common to many human abilities,from driving a car in traffic to recognizing a face. This strange state of affairs was beautifully summarized by the philosopher and scientist Michael Polanyi6英籍匈牙利裔物理化學家和哲學家，著有《個人知識》和《社會、經濟和哲學》等。,who said, “We know more than we can tell.” It’s a phenomenon that has come to be known as “Polanyi’s Paradox”.

[4] Polanyi’s Paradox hasn’t prevented us from using computers to accomplish complicated tasks, like processing payrolls7payroll工資單。, optimizing flight schedules,routing telephone calls and calculating taxes. But as anyone who’s written a traditional computer program can tell you, automating these activities has required painstaking8painstaking耗力費神的。precision to explain exactly what the computer is supposed to do.

[3]然而，與國際象棋不同的是，沒人能解釋水平最高的圍棋要怎么下。事實上，連頂級棋手本人也不完全清楚為何自己下得如此之好。人類對自身的很多能力都同樣缺乏認知，從在車流中駕駛汽車到識別人臉都是如此。哲學家和科學家邁克爾·波蘭尼曾對這一怪象進行了精彩總結：“我們知道的比能夠言說的要多。”這種現象后來被稱為“波蘭尼悖論”。

[4]“波蘭尼悖論”并沒有阻擋我們利用計算機來完成復雜任務，比如處理工資單、優化航班安排、傳遞電話信號和計算稅金。然而，任何一個寫過傳統計算機程序的人都知道，要實現這些事務的自動化，必須極度精確地指示計算機每一步該怎么做。

[5] This approach to programming computers is severely limited; it can’t be used in many domains, like Go,where we know more than we can tell,or other tasks like recognizing common objects in photos, translating between human languages and diagnosing diseases—all tasks where the rules-based approach to programming has failed badly over the years.

[6] Deep Blue achieved its superhuman performance almost by sheer computing power: It sifted through9sift through篩選。millions of possible chess moves to determine the optimal move. The problem is that there are many more possible Go games than there are atoms in the universe, so even the fastest computers can’t simulate10simulate模擬。a meaningful fraction of them.To make matters worse, it’s usually far from clear which possible moves to even start exploring.

[7] What changed? The AlphaGo victories vividly illustrate11illustrate闡明；說明。the power of a new approach in which instead of trying to program smart strategies into a computer, we instead build systems that can learn winning strategies almost entirely on their own, by seeing examples of successes and failures.

[8] Since these systems don’t rely on human knowledge about the task at hand, they’re not limited by the fact that we know more than we can tell.

[5]這種編程方法具有嚴重的局限性，在很多領域都不適用，比如“我們知道但難以言說”的圍棋，或者對照片中常見物體的識別、人類語言間的轉譯以及疾病的診斷等——多年以來，基于規則的編程方法在這些任務中都慘遭失敗。

[6]“深藍”的超人表現幾乎完全是憑借計算能力來實現的：它通過篩選數百萬種可能走法來確定最佳招數。但問題是，圍棋的走法比宇宙中的原子數還要多，即使是速度最快的電腦，也無法模擬其冰山之一角。更糟糕的是，我們往往連從何處入手都不清楚。

[7] AlphaGo有何不同？在AlphaGo中，我們沒有試圖將巧妙的策略編入計算機程序中，而是創建了一系列系統，使它們能夠在近乎完全自主的情況下，通過觀察勝負實例來學習制勝策略。AlphaGo接二連三的勝利便生動地展現了這一新方法的威力。

[8]由于這些系統并不依賴人類對圍棋的已有知識，因此并不會受到“波蘭尼悖論”的局限。

[9] AlphaGo does use simulations and traditional search algorithms12algorithm算法。to help it decide on some moves, but its real breakthrough is its ability to overcome Polanyi’s Paradox. It did this by figuring out13figure out解決；算出；想出。winning strategies for itself,both by example and from experience.The examples came from huge libraries of Go matches between top players amassed14amass積累。over the game’s 2,500-year history. To understand the strategies that led to victory in these games, the system made use of an approach known as deep learning, which has demonstrated remarkable abilities to tease out15tease out梳理。patterns and understand what’s important in large pools of information.

[9]在某些走法中，AlphaGo的確會使用模擬和傳統搜索算法來幫助決策，但其真正的突破在于有能力克服“波蘭尼悖論”。AlphaGo通過以往案例和自身經驗自行得出制勝策略。這些實例來自2500年圍棋史上高手對決的豐富資源。為理解這些對決中使用的制勝策略，系統采用了一種叫作“深度學習”的方法。這種方法在梳理規律、從大量信息中找出重要信息的驚人能力已得到證實。

[10]人類大腦的學習是一個在神經元間形成和鞏固聯結的過程。深度學習系統采用的方法與此極其相似，以至于這種系統一度被稱為“神經網絡”。系統在軟件中設置了數十億個節點和聯結，利用實例組成的“訓練集”來強化刺激（正在進行的圍棋比賽）與反應（下一步棋）之間的聯結，然后讓系統接收新的刺激，看其會作出何種反應。AlphaGo還和自己進行了數百萬場對決，利用一種叫作“強化學習”的技術來記住管用的招數策略。

[10] Learning in our brains is a process of forming and strengthening connections among neurons16neuron神經元。. Deep learning systems take an analogous17analogous類似的。approach,so much so that they used to be called“neural nets.” They set up billions of nodes18node 節點。節點在程序語言中是XML文件中有效而完整的結構的最小單元。內含標示組的節點，加上必要屬性、屬性值及內容，便可構成一個元素。節點的標志符為<>。and connections in software,use “training sets” of examples to strengthen connections among stimuli19stimuli刺激；刺激物，其單數形式為stimulus。(a Go game in process) and responses(the next move), then expose the system to a new stimulus and see what its response is. AlphaGo also played millions of games against itself, using another technique called reinforcement learning to remember the moves and strategies that worked well.

[11] Deep learning and reinforcement learning have both been around for a while, but until recently it was not at all clear how powerful they were, and how far they could be extended. In fact, it’s still not, but applications are improving at a gallop20at a gallop飛快地。, with no end in sight. And the applications are broad, including speech recognition, credit card fraud detection, and radiology21radiology放射學。and pathology22pathology病理學。.Machines can now recognize faces and drive cars, two of the examples that Polanyi himself noted as areas where we know more than we can tell.

[12] We still have a long way to go,but the implications23implication含蓄；含意。are profound. As when James Watt introduced his steam engine 240 years ago, technology-fueled changes will ripple throughout our economy in the years ahead, but there is no guarantee that everyone will benefit equally. Understanding and addressing the societal challenges brought on by rapid technological progress remain tasks that no machine can do for us.

[11]深度學習和強化學習并非新鮮事物，但直到最近人們才意識到它們的威力以及發展潛能。事實上，人們對其認識依然不充分，但這些技術的應用正在取得飛速進步，而且沒有盡頭。它們的應用范圍非常廣泛，包括語音識別、信用卡欺詐偵測，以及放射學和病理學領域的應用。機器如今可以識別人臉和駕駛汽車——這兩項技術都曾被波蘭尼本人歸為“我們知道但難以言說”的領域。

[12]未來的路還有很長，但是意義深遠。正如240年前詹姆斯·瓦特推出蒸汽機一樣，未來由技術推動的變革將會影響整個人類經濟，但并不能保證每個人都能從中獲得同等的好處。快速的技術進步帶來的社會挑戰，依然需要人類自己去理解和應對，這方面沒有任何機器能為我們代勞。

1麻省理工學院理學學士與碩士，哈佛商學院博士，麻省理工學院數字經濟項目負責人，同時任職于哈佛商學院和哈佛大學伯克曼互聯網與社會研究中心。與埃里克?布林約爾松合著有《與機器賽跑》及紐約時報暢銷書《第二次機器革命》等，同時著有《企業2.0》。

2麻省理工學院數字經濟項目負責人，同時任職于麻省理工大學斯隆商學院，美國國家經濟研究局研究助理，與安德魯·麥卡菲合著有《與機器賽跑》及暢銷書《第二次機器革命》等。

（譯者曾獲第五屆“《英語世界》杯”翻譯大賽二等獎）