The Turing test: Can a computer pass for a human? - Alex Gendler

翻译人员: Amy H. Fann 校对人员: Jing Zhou

What is consciousness?

什么是意识？

Can an artificial machine really think?

人造的机器真的能思考吗？

Does the mind just consist of neurons in the brain,

人类的大脑到底仅仅是 一个神经元的集体

or is there some intangible spark at its core?

还是是一种神秘意志的存在？

For many, these have been vital considerations

对很多人来说， 这些都是关于

for the future of artificial intelligence.

未来人工智能的 一些重要考虑因素。

But British computer scientist Alan Turing decided to disregard all these questions

但是对于一位英国电脑科学家， 阿兰-图灵，而言，

in favor of a much simpler one:

他关心一个更简单的问题：

can a computer talk like a human?

电脑可以如人类一样地交流吗？

This question led to an idea for measuring aritificial intelligence

这个问题让他有了测量人工智能的想法，

that would famously come to be known as the Turing test.

之后，这便演变为著名的“图灵测验” 。

In the 1950 paper, "Computing Machinery and Intelligence,"

在1950年，图灵在 “计算机器与智能”的报告里

Turing proposed the following game.

提出了一个游戏：

A human judge has a text conversation with unseen players

一个判官和看不到的选手用短信交流，

and evaluates their responses.

然后评价他们的回答。

To pass the test, a computer must be able to replace one of the players

要通过考试， 电脑必须能在没有改变结果的情况下，

without substantially changing the results.

代替其中一位选手。

In other words, a computer would be considered intelligent

换句话说，如果无轻易法分辨 一台电脑与一个人的区别，

if its conversation couldn't be easily distinguished from a human's.

这台电脑就是“聪明”的。

Turing predicted that by the year 2000,

图灵预计在2000年，

machines with 100 megabytes of memory would be able to easily pass his test.

拥有100兆字节内存的机器会 轻易地通过图灵测试。

But he may have jumped the gun.

但是图灵预计错了。

Even though today's computers have far more memory than that,

虽然现代的电脑具备更多的内存，

few have succeeded,

没有几个通过了图灵测试。

and those that have done well

那些通过图灵测试的电脑

focused more on finding clever ways to fool judges

并不是用了压倒性的计算能力，

than using overwhelming computing power.

而是用了巧妙的手段来迷惑判官。

Though it was never subjected to a real test,

虽然没有经过正式的考试，

the first program with some claim to success was called ELIZA.

ELIZA成了历史上第一个有资格成功的程序。

With only a fairly short and simple script,

仅用了一个十分简短的脚本，

it managed to mislead many people by mimicking a psychologist,

ELIZA成功地迷惑了很多人，

encouraging them to talk more

模仿心理专家， 鼓励他们多说话，

and reflecting their own questions back at them.

同时也发问他们的问题。

Another early script PARRY took the opposite approach

另一个早期程式脚本，PARRY, 运用相反的方式，

by imitating a paranoid schizophrenic

模仿了偏执的精神分裂症患者

who kept steering the conversation back to his own preprogrammed obsessions.

一直将话题转移回他自己预设的困扰。

Their success in fooling people highlighted one weakness of the test.

它们玩弄人们的成功 凸显出测试的缺点。

Humans regularly attribute intelligence to a whole range of things

人类常常把很多并不聪明的事物

that are not actually intelligent.

归类于“聪明”。

Nonetheless, annual competitions like the Loebner Prize,

尽管如此，年度竞赛 比如洛伯纳奖，

have made the test more formal

使测试变得更规范，

with judges knowing ahead of time

让判官们事先知道

that some of their conversation partners are machines.

有些对话选手是机器。

But while the quality has improved,

虽然总体质量上升了，

many chatbot programmers have used similar strategies to ELIZA and PARRY.

很多聊天机器人还是局用了 与ELIZA和PARRY类似的技巧。

1997's winner Catherine

1997年的获胜者，凯瑟琳

could carry on amazingly focused and intelligent conversation,

能够进行惊人般的聪慧会话，

but mostly if the judge wanted to talk about Bill Clinton.

但是这是在讨论与 比尔·克林顿 相关的话题。

And the more recent winner Eugene Goostman

最近的获奖者， 尤金·古斯特曼

was given the persona of a 13-year-old Ukrainian boy,

拥有13岁的乌克兰少年的个性，

so judges interpreted its nonsequiturs and awkward grammar

使判官将 他的别扭语法

as language and culture barriers.

理解为语言及文化的阻碍。

Meanwhile, other programs like Cleverbot have taken a different approach

在这同时，其他的程式如 Cleverbot， 采取了另一个方式

by statistically analyzing huge databases of real conversations

通过分析与统计巨大的真实对话数据，

to determine the best responses.

决定最好的回答方式。

Some also store memories of previous conversations

有一些还存储先前对话的记忆，

in order to improve over time.

以便长期改善。

But while Cleverbot's individual responses can sound incredibly human,

但尽管Cleverbot 自己的答复 听起来非常像人类，

its lack of a consistent personality

对始终如一的个性的缺乏

and inability to deal with brand new topics

及无法回答崭新的话题

are a dead giveaway.

则完全暴露了它。

Who in Turing's day could have predicted that today's computers

在图灵的时代里， 谁可能预料到今日的电脑

would be able to pilot spacecraft,

能够驾驶宇宙飞船，

perform delicate surgeries,

能操控精巧的手术，

and solve massive equations,

还能解答大量的方程，

but still struggle with the most basic small talk?

可仍与最基本的对话斗争？

Human language turns out to be an amazingly complex phenomenon

人类语言是样 如何巧妙及复杂的现象，

that can't be captured by even the largest dictionary.

连最大最广泛的字典都 无法记录。

Chatbots can be baffled by simple pauses, like "umm..."

聊天机器人会被简单的停顿， 如“额...”，

or questions with no correct answer.

或被没有正确回答的问题 而弄得团团转。

And a simple conversational sentence,

一个简单的对话句子，

like, "I took the juice out of the fridge and gave it to him,

比如“我从冰箱里拿出了果汁， 然后给了他，

but forgot to check the date,"

但忘了查明日子“，

requires a wealth of underlying knowledge and intuition to parse.

需要丰富的潜在知识 与直觉来解析。

It turns out that simulating a human conversation

事件证明若要模仿人类对话，

takes more than just increasing memory and processing power,

比单纯地增加记忆力和运算量 更加复杂。

and as we get closer to Turing's goal,

在接近图灵的目标的同时，

we may have to deal with all those big questions about consciousness after all.

我们也终究需考虑下 与“意识”相关的重大问题。

翻译：Amy H. Fann