Can a computer write poetry? | Oscar Schwartz

翻译人员: 易帆 余 校对人员: Ying Fang

I have a question.

我有一个问题，

Can a computer write poetry?

电脑可以写诗吗？

This is a provocative question.

这是个有争议的问题。

You think about it for a minute,

你稍微想一下，

and you suddenly have a bunch of other questions like:

脑海里突然就会浮现出 很多其他的问题：

What is a computer?

例如，什么是电脑？

What is poetry?

什么是诗？

What is creativity?

什么是创造力？

But these are questions

但这些问题，

that people spend their entire lifetime trying to answer,

很多人穷尽一​​生才能试着给出答案，

not in a single TED Talk.

单单一场 TED 演说并不能回答。

So we're going to have to try a different approach.

所以，我们必须用不一样的方法，

So up here, we have two poems.

上面这里有两首诗，

One of them is written by a human,

其中一首是人类写的，

and the other one's written by a computer.

另一首是电脑写的。

I'm going to ask you to tell me which one's which.

我会让各位来分辨哪首是谁写的，

Have a go:

我们开始吧：

Poem 1: Little Fly / Thy summer's play, / My thoughtless hand / Has brush'd away.

１号诗：小苍蝇，你夏天的游戏， 给我的手，无心地抹去。

Am I not / A fly like thee? / Or art not thou / A man like me?

我岂不像你，是一只苍蝇， 你岂不像我，是一个人？

Poem 2: We can feel / Activist through your life's / morning /

２号诗：我们可以感受到， 激进派在妳每日生活的清晨出没

Pauses to see, pope I hate the / Non all the night to start a / great otherwise (...)

暂且停下感受，那我憎恶的教皇 并非每晚都能开始，一个伟大的其他可能...

好的，时间到。

Alright, time's up.

Hands up if you think Poem 1 was written by a human.

认为１号诗是人写的请举手，

好的，你们大部分都是。

OK, most of you.

Hands up if you think Poem 2 was written by a human.

认为２号诗是人写的请举手，

你们很勇敢，

Very brave of you,

because the first one was written by the human poet William Blake.

因为第一首诗是由诗人William Blake所写，

The second one was written by an algorithm

第二首诗是由一个演算法所写出来的，

that took all the language from my Facebook feed on one day

选取了我脸书一天的信息流里的文法,

and then regenerated it algorithmically,

然后，用演算法重新制作出来的，

according to methods that I'll describe a little bit later on.

按照我稍后会提到的方法。

So let's try another test.

我们来做另一个测验，

Again, you haven't got ages to read this,

我再次说明， 你不用花太多时间去读它，

so just trust your gut.

所以，相信你的直觉。

Poem 1: A lion roars and a dog barks. It is interesting / and fascinating

１号诗：狮吼，狗吠， 鸟飞，却不吼也不吠，这真迷人且有趣呐

that a bird will fly and not / roar or bark. Enthralling stories about animals

我梦里有着关于动物的迷人故事

are in my dreams and I will sing them all if I / am not exhausted or weary.

如果我不筋疲力尽或疲惫不堪 我会为他们歌颂。

Poem 2: Oh! kangaroos, sequins, chocolate sodas! / You are really beautiful!

２号诗：喔！袋鼠、亮片、 巧克力苏打！你们真漂亮！

Pearls, / harmonicas, jujubes, aspirins! All / the stuff they've always talked about (...)

珍珠、口琴、枣子、阿斯匹林！ 所有他们进场谈论的素材（...)

Alright, time's up.

好的，时间到。

So if you think the first poem was written by a human,

如果你认为第一首诗是人写的，

put your hand up.

请举手。

OK.

好的。

And if you think the second poem was written by a human,

如果你认为第二首诗是人写的，

put your hand up.

请举手。

We have, more or less, a 50/50 split here.

我们这里大约是50／50比例，

这题比较难一点。

It was much harder.

The answer is,

答案是，

the first poem was generated by an algorithm called Racter,

第一首诗是一个名叫Racter的 电脑演算法

that was created back in the 1970s,

在1970年所创造的，

and the second poem was written by a guy called Frank O'Hara,

第二首诗是一位叫 Frank O'Hara的家伙写的，

who happens to be one of my favorite human poets.

他刚好是我最喜欢 的“ 人类诗人 ”其中之一，

(Laughter)

（笑声)

So what we've just done now is a Turing test for poetry.

所以，我们为这首诗 做了「图灵测试」。

The Turing test was first proposed by this guy, Alan Turing, in 1950,

「图灵测试」在1950年， 由Alan Turing做第一次发表，

in order to answer the question,

是为了回答一个问题：

can computers think?

「电脑会思考吗？」

Alan Turing believed that if a computer was able

Alan Turing相信，如果电脑能够

和人类进行一场文字交流，

to have a to have a text-based conversation with a human,

流畅到让人无法分辨

with such proficiency such that the human couldn't tell

whether they are talking to a computer or a human,

对方是人还是一台电脑，

then the computer can be said to have intelligence.

那么这台电脑可以被称呼为 拥有人工智慧。

So in 2013, my friend Benjamin Laird and I,

所以在2013年，我的朋友 Benjamin Laird和我，

we created a Turing test for poetry online.

我们创造了一个 在线的针推诗的图灵测试，

It's called bot or not,

叫做「bot or not」（是不是机器人），

and you can go and play it for yourselves.

你可以上线自己玩玩看。

But basically, it's the game we just played.

但基本上，它就是我们刚刚玩的游戏，

You're presented with a poem,

你会看到一首诗，

you don't know whether it was written by a human or a computer

你不知道它是人写的还是电脑写的，

and you have to guess.

然后你必须猜一猜。

So thousands and thousands of people have taken this test online,

好几千人已经在线上做测验，

so we have results.

所以，我们有一个结论，

And what are the results?

那结论是什么呢？

Well, Turing said that if a computer could fool a human

Turing说如果电脑可以骗过30%的人，

30 percent of the time that it was a human,

那它就可以被当作人，

then it passes the Turing test for intelligence.

它就通过了图灵测试的智力部分。

We have poems on the bot or not database

我们在 bot or not 资料库里的诗集

that have fooled 65 percent of human readers into thinking

已经骗过65% 的人，

让他们认为里面的诗是人写的。

it was written by a human.

So, I think we have an answer to our question.

所以，我认为我们的问题有答案了，

According to the logic of the Turing test,

根据图灵测试的逻辑，

can a computer write poetry?

电脑可以写诗吗？

Well, yes, absolutely it can.

是的，它绝对可以。

But if you're feeling a little bit uncomfortable

但，如果你觉得对这答案 有点让你不太舒服，

with this answer, that's OK.

也没关系，

If you're having a bunch of gut reactions to it,

如果你对此有一些直觉的反应，

that's also OK because this isn't the end of the story.

这也没关系，因为故事还没有结束。

Let's play our third and final test.

我们来玩第三个 最后一个测验，

我再说明一下，你们要读完后，

Again, you're going to have to read

and tell me which you think is human.

告诉我哪一个是人写的。

Poem 1: Red flags the reason for pretty flags. / And ribbons.

１号诗：红旗之所以漂亮 除了红色，还有缎带

Ribbons of flags / And wearing material / Reasons for wearing material. (...)

旗上的缎带及耐磨的材质 耐磨材料之所以（．．．）

Poem 2: A wounded deer leaps highest, / I've heard the daffodil

２号诗：受伤的鹿跳最高， 我听见水仙的诉说， ¶

I've heard the flag to-day / I've heard the hunter tell; /

我今天听旗子的话， 我听到猎人的故事；

'Tis but the ecstasy of death, / And then the brake is almost done (...)

这是对死亡的狂喜， 而伤害几乎已经造成（．．．）

OK, time is up.

好的，时间到。

So hands up if you think Poem 1 was written by a human.

认为１号诗是人写的请举手，

Hands up if you think Poem 2 was written by a human.

认为２号诗是人写的请举手，

Whoa, that's a lot more people.

哇！多很多人！

So you'd be surprised to find that Poem 1

你会很惊讶地发现，

was written by the very human poet Gertrude Stein.

１号诗由一位纯正的人类诗人 Gertrude Stein所写的，

And Poem 2 was generated by an algorithm called RKCP.

而２号诗是一个叫 RKCP演算法所创造的，

Now before we go on, let me describe very quickly and simply,

在我们要继续以前， 让我简单快速描述一下

how RKCP works.

RKCP是如何运作的。

So RKCP is an algorithm designed by Ray Kurzweil,

RKCP是Ray Kurzweil 所设计的演算法，

who's a director of engineering at Google

他是一位谷歌的工程师主管，

and a firm believer in artificial intelligence.

也是一位人工智慧的坚定支持者。

So, you give RKCP a source text,

那么，你给 RKCP一个源程序正文，

it analyzes the source text in order to find out how it uses language,

为了找出要如何使用这个语言， 它会分析来源文字，

and then it regenerates language

然后，它会重新创造一段话来模仿源文字。

that emulates that first text.

So in the poem we just saw before,

所以，我们刚刚看到的诗，

Poem 2, the one that you all thought was human,

你们认为是人类写的２号诗，

it was fed a bunch of poems

它被灌入了很多一位名叫 Emily Dickinson诗人的诗，

by a poet called Emily Dickinson

it looked at the way she used language,

它取用了这位诗人的语言，

learned the model,

学习她的模式，

and then it regenerated a model according to that same structure.

然后它依据同样的结构 重制一首诗出来。

但我们对RKCP最需要了解的是，

But the important thing to know about RKCP

is that it doesn't know the meaning of the words it's using.

它不明白它自己用的文字意义，

The language is just raw material,

语言只是它的原料，

it could be Chinese, it could be in Swedish,

它可以是中文，瑞典文，

it could be the collected language from your Facebook feed for one day.

它可以是你脸书上一天的文字。

It's just raw material.

它就只是个原料而已。

And nevertheless, it's able to create a poem

然而，它能够写一首

that seems more human than Gertrude Stein's poem,

比Gertrude Stein写的还要更有人味的诗，

and Gertrude Stein is a human.

但Gertrude Stein才是人啊...

So what we've done here is, more or less, a reverse Turing test.

所以，我们刚刚做的 差不多就是，反向图灵测试。

So Gertrude Stein, who's a human, is able to write a poem

所以Gertrude Stein这位人类，

可以写出让大部分人 误认为是电脑写出来的诗。

that fools a majority of human judges into thinking

that it was written by a computer.

Therefore, according to the logic of the reverse Turing test,

所以，根据图灵测试的反向逻辑，

Gertrude Stein is a computer.

Gertrude Stein这人是个电脑． ． ． （笑声）

(Laughter)

感觉很困惑吗？

Feeling confused?

I think that's fair enough.

我认为这情有可原。

So far we've had humans that write like humans,

目前为止，我们有人可以写出 像是人写出的诗、

we have computers that write like computers,

我们有电脑可以写出 像是电脑写出的诗、

we have computers that write like humans,

我们有电脑可以写出 像是人写出的诗，

but we also have, perhaps most confusingly,

但我们同时也有会让我们最容易混淆的 写诗像写得像电脑写的人。

humans that write like computers.

So what do we take from all of this?

所以，我们从这里面了解到什么呢？

Do we take that William Blake is somehow more of a human

我们会认为William Blake

比Gertrude Stein更像是个人吗？

than Gertrude Stein?

Or that Gertrude Stein is more of a computer than William Blake?

或者Gertrude Stein比 William Blake更像是个电脑？

(Laughter)

（笑声）

These are questions I've been asking myself

这些问题是这两年来，

我一直在问我自己，

for around two years now,

and I don't have any answers.

但我没有任何答案，

But what I do have are a bunch of insights

但我真的有领悟到很多有关于

about our relationship with technology.

我们与科技的关系。

So my first insight is that, for some reason,

所以，我的第一个领悟是，

we associate poetry with being human.

为了一些原因， 我们把人与诗结合一起，

So that when we ask, "Can a computer write poetry?"

所以，当我们问，＂电脑会写诗吗？＂

we're also asking,

我们也在问，

"What does it mean to be human

人的定义是什么？

and how do we put boundaries around this category?

我们要如何在这些类别之间划出界限？

How do we say who or what can be part of this category?"

我们要如何分辨谁或是什么东西 是属于这一类的？ ＂

This is an essentially philosophical question, I believe,

我相信，本质上这是一道哲学的问题，

and it can't be answered with a yes or no test,

而且，这不是像图灵测试这样 “对“或”错”的测试来回答

like the Turing test.

I also believe that Alan Turing understood this,

我也相信，

Alan Turing在1950年发明这个理论时， 也了解这一点，

and that when he devised his test back in 1950,

he was doing it as a philosophical provocation.

他当时引发了一个哲学上的争议。

So my second insight is that, when we take the Turing test for poetry,

我的第二个领悟是， 当我们在为诗做图灵测试时，

we're not really testing the capacity of the computers

我们并不是真的在测试电脑的能力，

because poetry-generating algorithms,

因为用演算法作诗相当简单，

they're pretty simple and have existed, more or less, since the 1950s.

而且它们大约在1950年代 早就已经存在了。

我们现在为诗做的图灵测试，

What we are doing with the Turing test for poetry, rather,

反而，比较像是在收集关于 什么是构成人性的条件的看法。

is collecting opinions about what constitutes humanness.

So, what I've figured out,

所以，我发现，

we've seen this when earlier today,

稍早我们今天看到的，

we say that William Blake is more of a human

我们说William Blake

than Gertrude Stein.

比Gertrude Stein更像个人，

Of course, this doesn't mean that William Blake

当然，这并不代表

was actually more human

William Blake比较有人性

or that Gertrude Stein was more of a computer.

或者Gertrude Stein比较像电脑。

It simply means that the category of the human is unstable.

这只能单纯的说明， 对人类的界定是不稳定的。

This has led me to understand

这让我明白了一件事，

that the human is not a cold, hard fact.

就是人性不是冷的、死板的事实，

Rather, it is something that's constructed with our opinions

反倒是一种由我们 的意见所构成的东西，

and something that changes over time.

而这个东西会随着时间而改变。

So my final insight is that the computer, more or less,

所以我最后的领悟是，

电脑，或多或少只是

works like a mirror that reflects any idea of a human

一面反映出我们展示给它的人类思想的镜子。

that we show it.

We show it Emily Dickinson,

我们向它展示Emily Dickinson，

it gives Emily Dickinson back to us.

它就展示Emily Dickinson给我们，

We show it William Blake,

我们向它展示William Blake，

that's what it reflects back to us.

它同样也会显示给我们，

We show it Gertrude Stein,

我们向它展示Gertrude Stein，

what we get back is Gertrude Stein.

我们得到的回应仅是Gertrude Stein。

还有其他更多的科技也是，

More than any other bit of technology,

the computer is a mirror that reflects any idea of the human we teach it.

电脑只是一面镜子， 它只是展示我们教给他的任何东西。

所以，我确定你们大部分人都曾听过

So I'm sure a lot of you have been hearing

a lot about artificial intelligence recently.

很多有关人工智慧的事情。

And much of the conversation is,

而大部分的对话就类似：

「我们该建造它吗？」

can we build it?

Can we build an intelligent computer?

「我们可以建立一个智慧型电脑吗？」

Can we build a creative computer?

「我们可以建立一个创造型电脑吗？」

What we seem to be asking over and over

我们一次又一次的被问到，

is can we build a human-like computer?

我们可以建立一个 类似人类的电脑吗？

But what we've seen just now

但就我们刚刚看到的，

is that the human is not a scientific fact,

人类不是一个科学事实，

that it's an ever-shifting, concatenating idea

人类是一个会不断地变化、串联想法、

and one that changes over time.

随时间改变的物种。

So that when we begin to grapple with the ideas

所以，当我们开始要努力克服

of artificial intelligence in the future,

未来人工智慧的这个想法时，

we shouldn't only be asking ourselves,

我们不应该只问我们自己，

"Can we build it?"

我们可以建造它吗？」

But we should also be asking ourselves,

我们还得问我们自己，

"What idea of the human do we want to have reflected back to us?"

「我们希望可以得到什么样的人性回应？」

This is an essentially philosophical idea,

这绝对是个哲学想法，

and it's one that can't be answered with software alone,

而且不是单靠软件就可以回答出来的，

but I think requires a moment of species-wide, existential reflection.

但我认为，这需要一个各类物种 共存的反应时刻，

谢谢各位。

Thank you.

(Applause)

（掌声）