声明: 本站全部内容源自互联网,不进行任何盈利行为
仅做 整合 / 美化 处理
I have a question.
[AI] 我有个问题。
Can a computer write poetry?
[AI] 电脑能写诗吗？
This is a provocative question.
[AI] 这是一个具有挑衅性的问题。
You think about it for a minute.
[AI] 你考虑一下。
and you suddenly have a bunch of other questions like:
[AI] 你突然有了一大堆其他的问题，比如：
What is a computer?
[AI] 什么是计算机？
What is poetry?
[AI] 什么是诗？
What is creativity?
[AI] 什么是创造力？
But these are questions
[AI] 但这些都是问题
that people spend their entire lifetime trying to answer.
[AI] 人们一生都在试图回答这个问题。
not in a single TED Talk.
[AI] 一次TED演讲都没有。
So we're going to have to try a different approach.
[AI] 所以我们必须尝试一种不同的方法。
So up here. we have two poems.
[AI] 所以在这里。我们有两首诗。
One of them is written by a human.
[AI] 其中一本是人类写的。
and the other one's written by a computer.
[AI] 另一个是电脑写的。
I'm going to ask you to tell me which one's which.
[AI] 我要请你告诉我哪个是哪个。
Have a go:
[AI] 试一试：
Poem 1: Little Fly / Thy summer's play. / My thoughtless hand / Has brush'd away.
[AI] 诗1：小苍蝇/你夏天的游戏我那轻率的手[刷]掉了。
Am I not / A fly like thee? / Or art not thou / A man like me?
[AI] 我不是像你一样的苍蝇吗或者你不是像我这样的人？
Poem 2: We can feel / Activist through your life's / morning /
[AI] 诗2：在你生命的早晨，我们可以感受到你的活力/
Pauses to see. pope I hate the / Non all the night to start a / great otherwise (...)
[AI] 停下来看看。pope我讨厌/不喜欢整晚开始一场/伟大的比赛，否则（…）
Alright. time's up.
[AI] 好吧时间到了。
Hands up if you think Poem 1 was written by a human.
[AI] 如果你认为第一首诗是人类写的，请举手。
OK. most of you.
[AI] 好啊你们大多数人。
Hands up if you think Poem 2 was written by a human.
[AI] 如果你认为第二首诗是人类写的，请举手。
Very brave of you.
[AI] 你真勇敢。
because the first one was written by the human poet William Blake.
[AI] 因为第一本是人类诗人威廉·布莱克写的。
The second one was written by an algorithm
[AI] 第二个是由算法编写的
that took all the language from my Facebook feed on one day
[AI] 这一天就把我Facebook上的所有语言都拿走了
and then regenerated it algorithmically.
[AI] 然后通过算法重新生成。
according to methods that I'll describe a little bit later on.
[AI] 根据我稍后将描述的方法。
So let's try another test.
[AI] 让我们再做一次测试。
Again. you haven't got ages to read this.
[AI] 再一次你没有足够的时间来阅读这篇文章。
so just trust your gut.
[AI] 所以只要相信你的直觉。
Poem 1: A lion roars and a dog barks. It is interesting / and fascinating
[AI] 诗1：狮子吼叫，狗吠叫。这很有趣
that a bird will fly and not / roar or bark. Enthralling stories about animals
[AI] 鸟会飞，不会吼叫或吠叫。关于动物的引人入胜的故事
are in my dreams and I will sing them all if I / am not exhausted or weary.
[AI] 在我的梦中，如果我不疲惫，我会把它们全部唱出来。
Poem 2: Oh! kangaroos. sequins. chocolate sodas! / You are really beautiful!
[AI] 诗2：哦！袋鼠。亮片。巧克力汽水！/你真漂亮！
Pearls. / harmonicas. jujubes. aspirins! All / the stuff they've always talked about (...)
[AI] 珍珠口琴。枣子。阿斯匹林！他们一直在谈论的所有事情（…）
Alright. time's up.
[AI] 好吧时间到了。
So if you think the first poem was written by a human.
[AI] 如果你认为第一首诗是人类写的。
put your hand up.
[AI] 把手举起来。
OK.
[AI] 好啊
And if you think the second poem was written by a human.
[AI] 如果你认为第二首诗是人类写的。
put your hand up.
[AI] 把手举起来。
We have. more or less. a 50/50 split here.
[AI] 我们有。或多或少这里五五分之一。
It was much harder.
[AI] 这要困难得多。
The answer is.
[AI] 答案是。
the first poem was generated by an algorithm called Racter.
[AI] 第一首诗是由一种叫做莱克特的算法生成的。
that was created back in the 1970s.
[AI] 这是在20世纪70年代创建的。
and the second poem was written by a guy called Frank O'Hara.
[AI] 第二首诗是一个叫弗兰克·奥哈拉的人写的。
who happens to be one of my favorite human poets.
[AI] 他碰巧是我最喜欢的人类诗人之一。
(Laughter)
[AI] （众笑）
So what we've just done now is a Turing test for poetry.
[AI] 所以我们现在做的是一个图灵诗歌测试。
The Turing test was first proposed by this guy. Alan Turing. in 1950.
[AI] 图灵测试最初是由这个家伙提出的。艾伦·图灵。1950年。
in order to answer the question.
[AI] 为了回答这个问题。
can computers think?
[AI] 计算机能思考吗？
Alan Turing believed that if a computer was able
[AI] 艾伦·图灵认为如果计算机能够
to have a to have a text-based conversation with a human.
[AI] 要想拥有一个虚拟世界，就必须与人进行基于文本的对话。
with such proficiency such that the human couldn't tell
[AI] 如此熟练以至于人类无法分辨
whether they are talking to a computer or a human.
[AI] 不管他们是在和电脑还是人说话。
then the computer can be said to have intelligence.
[AI] 那么计算机可以说是有智能的。
So in 2013. my friend Benjamin Laird and I.
[AI] 2013年也是如此。我和我的朋友本杰明·莱尔德。
we created a Turing test for poetry online.
[AI] 我们创建了一个在线诗歌图灵测试。
It's called bot or not.
[AI] 这叫不叫机器人。
and you can go and play it for yourselves.
[AI] 你可以自己去玩。
But basically. it's the game we just played.
[AI] 但基本上。这是我们刚才玩的游戏。
You're presented with a poem.
[AI] 你收到了一首诗。
you don't know whether it was written by a human or a computer
[AI] 你不知道它是人类写的还是计算机写的
and you have to guess.
[AI] 你得猜。
So thousands and thousands of people have taken this test online.
[AI] 成千上万的人在网上参加了这项测试。
so we have results.
[AI] 所以我们有了结果。
And what are the results?
[AI] 结果如何？
Well. Turing said that if a computer could fool a human
[AI] 好图灵说如果计算机能愚弄人类
30 percent of the time that it was a human.
[AI] 30%的时间它是人类。
then it passes the Turing test for intelligence.
[AI] 然后它通过了图灵智力测试。
We have poems on the bot or not database
[AI] 我们在机器人上有诗，或者没有数据库
that have fooled 65 percent of human readers into thinking
[AI] 这已经愚弄了65%的人类读者
it was written by a human.
[AI] 这是一个人写的。
So. I think we have an answer to our question.
[AI] 所以我想我们的问题已经有了答案。
According to the logic of the Turing test.
[AI] 根据图灵测试的逻辑。
can a computer write poetry?
[AI] 电脑能写诗吗？
Well. yes. absolutely it can.
[AI] 好对当然可以。
But if you're feeling a little bit uncomfortable
[AI] 但是如果你觉得有点不舒服
with this answer. that's OK.
[AI] 有了这个答案。没关系。
If you're having a bunch of gut reactions to it.
[AI] 如果你对它有一系列的本能反应。
that's also OK because this isn't the end of the story.
[AI] 这也没关系，因为这还不是故事的结尾。
Let's play our third and final test.
[AI] 让我们进行第三次也是最后一次测试。
Again. you're going to have to read
[AI] 再一次你得读书
and tell me which you think is human.
[AI] 告诉我你认为哪个是人类。
Poem 1: Red flags the reason for pretty flags. / And ribbons.
[AI] 诗1：红旗美丽旗帜的原因还有丝带。
Ribbons of flags / And wearing material / Reasons for wearing material. (...)
[AI] 旗带/和耐磨材料/耐磨材料的原因。(...)
Poem 2: A wounded deer leaps highest. / I've heard the daffodil
[AI] 诗2：受伤的鹿跳得最高我听过水仙花的声音
I've heard the flag to-day / I've heard the hunter tell; /
[AI] 今天我听到了旗帜的声音/我听到了猎人的诉说/
'Tis but the ecstasy of death. / And then the brake is almost done (...)
[AI] “这不过是死亡的狂喜。”然后刹车几乎完成了（…）
OK. time is up.
[AI] 好啊时间到了。
So hands up if you think Poem 1 was written by a human.
[AI] 如果你认为第一首诗是人类写的，请举手。
Hands up if you think Poem 2 was written by a human.
[AI] 如果你认为第二首诗是人类写的，请举手。
Whoa. that's a lot more people.
[AI] 哇。那是很多人。
So you'd be surprised to find that Poem 1
[AI] 所以你会惊讶地发现那首诗1
was written by the very human poet Gertrude Stein.
[AI] 是由人类诗人格特鲁德·斯坦写的。
And Poem 2 was generated by an algorithm called RKCP.
[AI] 诗2是由一种叫做RKCP的算法生成的。
Now before we go on. let me describe very quickly and simply.
[AI] 在我们继续之前。让我简单快速地描述一下。
how RKCP works.
[AI] RKCP的工作原理。
So RKCP is an algorithm designed by Ray Kurzweil.
[AI] 所以RKCP是Ray Kurzweil设计的一种算法。
who's a director of engineering at Google
[AI] 谁是谷歌的工程总监
and a firm believer in artificial intelligence.
[AI] 而且是人工智能的坚定信徒。
So. you give RKCP a source text.
[AI] 所以为RKCP提供源文本。
it analyzes the source text in order to find out how it uses language.
[AI] 它分析源文本以了解它是如何使用语言的。
and then it regenerates language that emulates that first text.
[AI] 然后它重新生成模拟第一个文本的语言。
So in the poem we just saw before.
[AI] 所以在我们刚才看到的诗中。
Poem 2. the one that you all thought was human.
[AI] 诗2。那个你们都认为是人类的人。
it was fed a bunch of poems by a poet called Emily Dickinson
[AI] 它由一位名叫艾米丽·狄金森的诗人写了一大堆诗
it looked at the way she used language.
[AI] 这是她使用语言的方式。
learned the model.
[AI] 学习了模型。
and then it regenerated a model according to that same structure.
[AI] 然后根据相同的结构重新生成一个模型。
But the important thing to know about RKCP
[AI] 但是关于RKCP最重要的是要知道
is that it doesn't know the meaning of the words it's using.
[AI] 就是它不知道它所用的词的意思。
The language is just raw material.
[AI] 语言只是原始材料。
it could be Chinese. it could be in Swedish.
[AI] 可能是中国人。可能是瑞典语。
it could be the collected language from your Facebook feed for one day.
[AI] 它可能是从你的Facebook订阅源中收集的一天的语言。
It's just raw material.
[AI] 这只是原材料。
And nevertheless. it's able to create a poem
[AI] 尽管如此。它可以创作一首诗
that seems more human than Gertrude Stein's poem.
[AI] 这似乎比格特鲁德·斯坦的诗更人性化。
and Gertrude Stein is a human.
[AI] 格特鲁德·斯坦是一个人。
So what we've done here is. more or less. a reverse Turing test.
[AI] 所以我们在这里所做的就是。或多或少反向图灵测试。
So Gertrude Stein. who's a human. is able to write a poem
[AI] 所以格特鲁德·斯坦。谁是人类。他会写一首诗
that fools a majority of human judges into thinking that it was written by a computer.
[AI] 这愚弄了大多数人类法官，使他们认为这是电脑写的。
Therefore. according to the logic of the reverse Turing test.
[AI] 因此根据反向图灵测试的逻辑。
Gertrude Stein is a computer. (Laughter)
[AI] 格特鲁德·斯坦是一台计算机。（众笑）
Feeling confused?
[AI] 感到困惑吗？
I think that's fair enough.
[AI] 我认为这很公平。
So far we've had humans that write like humans.
[AI] 到目前为止，我们已经有了像人类一样写作的人类。
we have computers that write like computers.
[AI] 我们有像计算机一样书写的计算机。
we have computers that write like humans.
[AI] 我们有像人类一样书写的计算机。
but we also have. perhaps most confusingly. humans that write like computers.
[AI] 但我们也有。也许最令人困惑的是。像计算机一样写字的人。
So what do we take from all of this?
[AI] 那么我们从这一切中得到了什么呢？
Do we take that William Blake is somehow more of a human
[AI] 我们是否认为威廉·布莱克更像人类
than Gertrude Stein?
[AI] 而不是格特鲁德·斯坦？
Or that Gertrude Stein is more of a computer than William Blake?
[AI] 还是说格特鲁德·斯坦比威廉·布莱克更像一台电脑？
(Laughter)
[AI] （众笑）
These are questions I've been asking myself
[AI] 这些是我一直在问自己的问题
for around two years now.
[AI] 大约两年了。
and I don't have any answers.
[AI] 我没有任何答案。
But what I do have are a bunch of insights
[AI] 但我确实有很多见解
about our relationship with technology.
[AI] 关于我们与技术的关系。
So my first insight is that. for some reason.
[AI] 所以我的第一个见解是。出于某种原因。
we associate poetry with being human.
[AI] 我们把诗歌与人联系在一起。
So that when we ask. "Can a computer write poetry?"
[AI] 所以当我们问的时候。“电脑能写诗吗？”
we're also asking.
[AI] 我们也在问。
"What does it mean to be human
[AI] “这对人类意味着什么
and how do we put boundaries around this category?
[AI] 我们如何界定这一类别？
How do we say who or what can be part of this category?"
[AI] 我们如何说谁或什么可以属于这一类别？"
This is an essentially philosophical question. I believe.
[AI] 这本质上是一个哲学问题。我相信。
and it can't be answered with a yes or no test. like the Turing test.
[AI] 而且不能用“是”或“否”来回答。比如图灵测试。
I also believe that Alan Turing understood this.
[AI] 我也相信艾伦·图灵明白这一点。
and that when he devised his test back in 1950.
[AI] 当他在1950年设计他的测试时。
he was doing it as a philosophical provocation.
[AI] 他这样做是出于哲学上的挑衅。
So my second insight is that. when we take the Turing test for poetry.
[AI] 所以我的第二个见解是。当我们参加图灵诗歌测试时。
we're not really testing the capacity of the computers
[AI] 我们并不是在真正测试计算机的容量
because poetry-generating algorithms.
[AI] 因为诗歌生成算法。
they're pretty simple and have existed. more or less. since the 1950s.
[AI] 它们非常简单，并且已经存在。或多或少从20世纪50年代开始。
What we are doing with the Turing test for poetry. rather.
[AI] 我们正在做的是图灵诗歌测试。相当地
is collecting opinions about what constitutes humanness.
[AI] 正在收集关于什么是人性的观点。
So. what I've figured out.
[AI] 所以我发现了什么。
we've seen this when earlier today.
[AI] 今天早些时候我们已经看到了这一点。
we say that William Blake is more of a human
[AI] 我们说威廉·布莱克更像一个人
than Gertrude Stein.
[AI] 而不是格特鲁德·斯坦。
Of course. this doesn't mean that William Blake
[AI] 当然这并不意味着威廉·布莱克
was actually more human
[AI] 实际上更人性化
or that Gertrude Stein was more of a computer.
[AI] 或者说格特鲁德·斯坦更像一台电脑。
It simply means that the category of the human is unstable.
[AI] 这仅仅意味着人类的范畴是不稳定的。
This has led me to understand
[AI] 这让我明白了
that the human is not a cold. hard fact.
[AI] 人不是感冒。确凿的事实。
Rather. it is something that's constructed with our opinions
[AI] 相当地这是由我们的意见构成的
and something that changes over time.
[AI] 随着时间的推移会发生变化。
So my final insight is that the computer. more or less.
[AI] 所以我最后的见解是计算机。或多或少
works like a mirror that reflects any idea of a human
[AI] 它就像一面镜子，可以反映人类的任何想法
that we show it.
[AI] 我们展示它。
We show it Emily Dickinson.
[AI] 我们给它看艾米丽·迪金森。
it gives Emily Dickinson back to us.
[AI] 它把艾米丽·迪金森还给了我们。
We show it William Blake.
[AI] 我们给它看威廉·布莱克。
that's what it reflects back to us.
[AI] 这就是它对我们的反映。
We show it Gertrude Stein.
[AI] 我们给它看格特鲁德·斯坦。
what we get back is Gertrude Stein.
[AI] 我们得到的是格特鲁德·斯坦。
More than any other bit of technology.
[AI] 比任何其他技术都重要。
the computer is a mirror that reflects any idea of the human we teach it.
[AI] 计算机是一面镜子，反映了我们教它的人的任何想法。
So I'm sure a lot of you have been hearing
[AI] 所以我相信你们很多人都听说了
a lot about artificial intelligence recently.
[AI] 最近有很多关于人工智能的事情。
And much of the conversation is.
[AI] 大部分对话都是这样的。
can we build it?
[AI] 我们能建造它吗？
Can we build an intelligent computer?
[AI] 我们能造一台智能计算机吗？
Can we build a creative computer?
[AI] 我们能制造一台有创意的电脑吗？
What we seem to be asking over and over
[AI] 我们一遍又一遍地问
is can we build a human-like computer?
[AI] 我们能造一台像人类一样的计算机吗？
But what we've seen just now
[AI] 但是我们刚才看到的
is that the human is not a scientific fact.
[AI] 人类不是一个科学事实。
that it's an ever-shifting. concatenating idea
[AI] 这是一个不断变化的过程。连接思想
and one that changes over time.
[AI] 而且随着时间的推移会发生变化。
So that when we begin to grapple with the ideas
[AI] 所以当我们开始与这些想法搏斗时
of artificial intelligence in the future.
[AI] 未来的人工智能。
we shouldn't only be asking ourselves.
[AI] 我们不应该只问自己。
"Can we build it?"
[AI] “我们能建造它吗？”
But we should also be asking ourselves.
[AI] 但我们也应该问问自己。
"What idea of the human do we want to have reflected back to us?"
[AI] “我们想把什么样的人类观念反映给我们？”
This is an essentially philosophical idea.
[AI] 这本质上是一种哲学思想。
and it's one that can't be answered with software alone.
[AI] 这是单靠软件无法解决的问题。
but I think requires a moment of species-wide. existential reflection.
[AI] 但我认为这需要在物种范围内进行片刻的研究。存在主义反思。
Thank you.
[AI] 非常感谢。
(Applause)
[AI] （掌声）