2021年1月23日发(作者：公众人物)
A good many technical people become irate when you call a computer a giant brain. They insist t
hat a computer does only what thinking humans have planned to have it do.
假如你把计算机称
为巨人脑，< br>许多技术人员会感到气愤。
他们坚持认为，
计算机只不过是做有思维的人类安排
它做的事情罢了

Yet one authority states categorically,
de, and choose; it can perform reasonable operations with information. A machine, therefore, ca
n think. Famed mathematician Norbert Wiener, of MIT, envisions a machine that can learn and wi
ll
us.
不过，
一位权威人士明确地 说：
“机器能处理信息，
能进行计算、做出决定和选择，能利用信息从事各种合理的工作，因此 ，机器能够思维。
”
麻省理工学院著名的数学家诺伯特·
威纳做出展望认为，
机器能学习，
它“决不会被迫做出
我们必须作的决定，或是迎合我们。
”显然，他认为 机器能够思维

There is a popular anecdote about a computer programmer who, just for a lark, spent days settin
g up the machine to destroy itself, then watched delightedly as the computer dutifully proceeded
to commit suicide. If that machine could have thought, would it not have circumvented him?
一
个有关一位计算机程序设计员的轶事广为流传。
这位程序设计员只是为了 取乐，
花了几天的
时间，
装了一台毁灭机器自身的机器，
然后兴高采烈地观看 这台计算机忠实地自杀，
那台计
算机假如能够思维的话，难道不会用谋略去战胜他吗

To me the controversy boils down to a definition of the word think . There is no area in physiology
that is less understood than the human brain . Practically all that is known is that the brain contai
ns some ten billion tiny cells called neurons.
对我来说，
这场争论归根结底就是对
“思维”
一词
如何下定义。
从生理学的角度看，
没有哪个领域比人脑了解得更 少的了。
实际上我们所知道
的只是大脑约含有
100
亿个称为神经原的微小细 胞

Apparently neurons are elementary memory units, capable of storing the same kind of informatio
n
extraordinary network of nerve-communication lines, comprise all that is to the brain. But such a
n explanation fails to account for how the brain originates thought.
很显然，神经原是基本的记
忆单位，能储存的信息“块”种类与机器是相等的。彻底唯物的观点是，大脑 是由神经原与
一个特殊的神经通讯网共同组成的。但这一解释并没有说明大脑是如何产生思维的

Probably the clearest differentiation between man and machine is a quantitative one. The brain h
as roughly a million times as many components as the best computer. On the other hand, the diff
erence may lie in a spiritual factor, embraced by religion. At any rate, a machine cannot exercise f
ree will or originate anything-not yet. Whether it ever will is still an open argument.
人
和
机
器
之
间最明显的区别 大概是量的不同。
人脑的组成元件大约是最好的计算机的
100
万倍。
从另一
个方面来说，
区别也许在于宗教所能接受的某精神因素。
不管怎样，
机器不能 行使自由意志
或创造任何东西——现在还不能。将来是否可以尚存争议

Computers can already do a lot of surprising things, which include predicting the weather. The m
achine is able to make forecasts by assimilating vast quantities of data, but this, as well as most of
the other tasks now performed by the thinking machines, is routine, requiring thinking of a very l
ow order. Let us see what happens when we go beyond this step.
计算机已经能做许多 令人吃
惊的事，其中包括预报天气。它可以接收大量的数据、并以此做出预报，但与思维机器现在
所完成的大多数别的工作一样，
这只是它的日常工作，
只需要低思维层次。
我们跨越 这一步
后，再看看会出现什么情况吧

A great many outstanding men are preoccupied with computers that do nonnumerical work: that
is, machines that
ple, whether inanimate devices can exercise judgment, make choices, give birth to ideas, and play
games intelligently. At the far limit of possibility, they wish to know whether, at least on paper, m
achines can reproduce themselves. In other words, are we really certain that a machine can do o
nly what its programmer wills it to do? Already there are in existence a number of machines that
approach these powers. The simplest type is the computer that understands logic.
许多杰出的人
才在潜心研制从事“非数字”工作的计算机。也就是说，所有的数字输入后，机器能够“以
此为基点继 续干下去”
。比方说，这些人想看看无生命的装置是否能进行判断、做出选择、
产生思想、伶俐 地玩游戏，他们想知道，至少在理论上，机器是否可以再生，再生的可能性
究竟有多大。换句话说，我们 真的确信机器只能做其程序设计员要它做的事吗
?
一些差不多
具有这些能力的机器现在已经有了。最简单的一种是懂逻辑的计算机

The mathematics of logic, upon which such machines work, was originated a century ago by an o
bscure Englishman named George Boole.
这种机 器赖以工作的逻辑数学是一个世纪以前一位
毫无名气的叫乔治·
布尔的英国人创造的。
现在的布尔代数，
在电话工程师设计自动交换机
的过程中已经证实了其强大的功能

Boolean algebra, as it is now called, has proved to be of great service to telephone engineers in d
esigning automatic switching apparatus. Boolean algebra caught the eye of two Harvard students
some years ago, and they built a machine called the Logical Truth Calculator. It created quite a stir
in Cambridge, because it worked uncannily well at solving rather simple logical problems. Soon
more sophisticated machines were developed - machines that could play games. The mechanical
game-players fall into a number of classes. The simplest has a dictionary of rales in its memory, in
cluding all possible moves to be made in answer to an opponent. A second class plays only accord
ing to a rigid formula, and can never lose.
好多年前，
布尔代数引起了哈佛大学两位学生的注意，他们造了一台叫做“逻辑真值计算机”的机器。在剑桥引起了轰动，因为它在解决较简单的
逻辑问题 的过程中表现得异常出色。
不久，
更复杂的机器——能够下棋的机器——研制出来
了。
这种机械棋手被分成了几类，
在最简单的一类的储存器中有一部规则辞典，
其中包含了
机器所能走的所有对应招法。第二类只按照严格的程式下，决不会出现失误

A third category applies the principles of approximations, choosing the probable best move by
hinking
machines can play include checkers, chess, bridge, poker, and many other card games. They can' t
always win, because an approximately right move isn't necessarily the best one, but few human
opponents can do better.
第三类采用近似值原理，对所有可以想象得出的招数加以“考虑”
，
并根据某一标准进行试验，
然后选择可能是最佳的一着。
这些机器能玩的游戏包括西洋跳棋、< br>象棋、
桥牌、扑克和其他许多纸牌游戏。
它们不能每回都赢，
因为近似正确的一 着不一定是
最好的一着，但人类与它们对垒很少能比它们做得更好

Dr. Claude Shannon of Bell Telephone laboratories, and his colleague, E. F. Moore, once built a ma
chine to play the game known as Hex, using an analog computer working on electrical voltages. It
beat them about 70 per cent of the time.
nts,
贝尔电话实验所的克劳
德·香农博士和他的同事
E
·F
·穆尔曾采用在电压位环境中工作的模拟计算机制造过一台
名为“巫婆”的游戏机，他们 十回有七回要被它打败。香衣评论说，
“它常常选择一些貌似
奇怪的招法，而分析证明这些招法 都是正确的。这常使它的设计者感到惊讶。

Each of these mechanical game-player is said to
he Boolean algebra pattern, discovering by sad experience that one thing or another doesn't wor