robots that communicate with people.

那些能夠與人交流的機器人。

But this story doesn't start with robotics at all,

不過這個故事不是從機器人說起，

it starts with animation.

而是要從動畫說起。

When I first saw Pixar's "Luxo Jr.,"

當我第一次看到皮克斯的

I was amazed by how much emotion

我驚呆了，

they could put into something

一個如此微不足道的檯燈

as trivial as a desk lamp.

竟能表現如此多的感情。

I mean, look at them -- at the end of this movie,

你看看他們啊！電影結尾的時候，

you actually feel something for two pieces of furniture.

你真的開始喜歡上這兩件小小的家具了。

(Laughter)

（笑聲）

And I said, I have to learn how to do this.

我對自己說，我要學會做這樣的東西。

So I made a really bad career decision.

所以我在職涯上做了一個很差的決定，

And that's what my mom was like when I did it.

我做出這個決定的時候，我媽媽就是這樣的。

(Laughter)

（笑聲）

I left a very cozy tech job in Israel

我辭去了在以色列一個軟體公司的

at a nice software company and I moved to New York

一份舒服的技術工作，而搬到了紐約

to study animation.

去學習動畫。

And there I lived

在那裡

in a collapsing apartment building

我和我的室友住在哈林區

I'm not using this phrase metaphorically,

這並不是比喻，

the ceiling actually collapsed one day

有一天我們客廳的天花板

in our living room.

真的塌下來了，

Whenever they did those news stories

每次新聞報導紐約的違章建築時，

they would put the report in front of our building.

他們都會跑到我們的大樓下進行採訪。

As kind of like a backdrop

就好像讓你看看現場有多糟糕一樣。

Anyway, during the day I went to school and at night

言歸正傳，我白天上學，

I would sit and draw frame by frame

晚上我不停地一幅一幅地用鉛筆畫著畫。

And I learned two surprising lessons --

我學到了兩個令人驚訝的事情 --

one of them was that

其中一個是：

when you want to arouse emotions,

當你想要喚起某些情感時，

it doesn't matter so much how something looks,

外觀並不算太重要，

it's all in the motion -- it's in the timing

關鍵是動作 -- 物體運動的時機。

of how the thing moves.

關鍵是動作 -- 物體運動的時機。

And the second, was something

第二個是一位老師告訴我們的。

He actually did the weasel in Ice Age.

他製作了電影《冰原歷險記》的黃鼠狼。

And he said:

他說：

"As an animator you are not

“作為一個動畫製作者，

So, if you want to find the

所以如果你要為一個角色

don't think about it, go use your body to find it --

不要想，用你的身體找到它，

stand in front of a mirror, act it out

站在鏡子面前，在攝影機前，演出來，

in front of a camera -- whatever you need.

做出你需要做的。

And then put it back in your character.

然後再把這個動作放在你的角色上。

A year later I found myself at MIT

一年以後，我去了麻省理工學的

in the robotic life group, it was one of the first groups

機器人生命小組，

researching the relationships

這是最早開始研究人類和

And I still had this dream to make

我依然有個夢想

an actual, physical Luxo Jr. lamp.

做出真正的、真實的頑皮跳跳燈。

But I found that robots didn't move at all

但是我發現機器人

in this engaging way that I was used to

完全不按照動畫課程中那種

for my animation studies.

引人入勝的方式移動。

Instead, they were all --

相反的，他們都 --

how should I put it, they were all kind of robotic.

該怎麼說呢？他們都有點兒機器人。

(Laughter)

（笑聲）

And I thought, what if I took whatever

我就想，如果我可以把

and used that to design my robotic desk lamp.

應用在設計我的機器人檯燈會怎樣？

So I went and designed frame by frame

因此我一幅一幅的設計，

to try to make this robot

試圖讓這個機器人

as graceful and engaging as possible.

盡量優雅且有吸引力。

And here when you see the robot interacting with me

這裡你可以看到這個桌子上的機器人

on a desktop.

在跟我互動，

And I'm actually redesigning the robot so,

我其實是在重新設計這個機器人，

unbeknownst to itself,

而這個機器人完全不知道，

it's kind of digging its own grave by helping me.

它幫我，其實是在自掘墳墓呢。

(Laughter)

（笑聲）

I wanted it to be less of a mechanical structure

比起把它做成一個照明的機械，

giving me light,

比起把它做成一個照明的機械，

and more of a helpful, kind of quiet apprentice

我更想要一個能幫忙的、安靜的學徒，

that's always there when you need

隨時滿足你的需求卻不打擾你。

And when, for example, I'm looking for a battery

比如，當我要找一個我怎麼也

that I can't find,

找不到的電池時，

in a subtle way, it will show me where the battery is.

它可以巧妙地提醒我電池在哪裡。

So you can see my confusion here.

你看到我的困惑了嗎？

I'm not an actor.

我不是一個演員。

And I want you to notice how the same

我希望你們注意到，

mechanical structure can at one point,

同一個機械如何在前一刻

just by the way it moves seem gentle and caring --

非常溫柔，充滿關懷，

and in the other case, seem

下一刻又顯得非常暴力，有進攻性。

And it's the same structure,

一模一樣的結構，改變的僅僅是動作。

Actor: "You want to know something?

演員：”你想知道嗎？你真的想知道嗎？

He was already dead!

他已經死了！

Just laying there, eyes glazed over!"

他就躺在那裡，目光呆滯！ “

(Laughter)

（笑聲）

But, moving in graceful ways is just one

但是，以一種優雅的方式移動只是這整個

called human-robot interaction.

人類機器人互動結構的一塊基石。

I was at the time doing my Ph.D.,

那時候我正在攻讀博士學位，

I was working on human robot teamwork;

我在一個研究人類機器人的團隊，

teams of humans and robots working together.

也就是人類和機器人一起合作。

I was studying the engineering,

我在學習團隊合作的工程學，

the psychology, the philosophy of teamwork.

心理學和哲學。

And at the same time I found myself

同時，我意識到我自己

in my own kind of teamwork situation

和我的一個好朋友，他今天也在這裡，

with a good friend of mine who is actually here.

碰到了一個團隊合作的情境。

And in that situation we can easily imagine robots

在那個情境中，我們很容易想像

in the near future being there with us.

不久的將來機器人會和我們在一起。

It was after a Passover seder.

那是在一個逾越節家宴結束後，

We were folding up a lot of folding chairs,

我們要收起大量的折疊椅，

and I was amazed at how quickly

我對於我們迅速找到了

Everybody did their own part.

每個人都做了自己的那部分，

We didn't have to divide our tasks.

無需分工，

We didn't have to communicate verbally about this.

無需特意口頭溝通。

It all just happened.

就這樣發生了。

And I thought,

於是我想，

humans and robots don't look at all like this.

人類和機器人的互動卻完全不是這樣。

When humans and robots interact,

當人類和機器人互動的時候，

it's much more like a chess game.

就好像他們在下棋。

The human does a thing,

人類走一步，

the robot analyzes whatever the human did,

機器人對此分析一下，

then the robot decides what to do next,

然後機器人決定接下來怎麼做，

plans it and does it.

計劃好，走下一步。

And then the human waits, until it's their turn again.

這時候人類就等著，

So, it's much more like a chess game

所以，人類和機器人的互動更像下棋，

and that makes sense because chess is great

這很好理解，因為

for mathematicians and computer scientists.

對數學家和電腦科學家來說，

It's all about information analysis,

它們都是關於資料分析、

decision making and planning.

決策和計劃。

But I wanted my robot to be less of a chess player,

但比起下棋，我更希望我的機器人

and more like a doer

是一個行動者，

that just clicks and works together.

可以和人類有默契地一起工作。

So I made my second horrible career choice:

於是我在我的職涯上

I decided to study acting for a semester.

我決定學習一學期的表演課程。

I took off from a Ph.D. I went to acting classes.

我放下了我的博士課程，去上了表演課。

I actually participated in a play,

我還參與了一個戲劇，

I hope theres no video of that around still.

希望現在已經找不到那段影片了。

And I got every book I could find about acting,

我看遍每一本關於表演的書，

including one from the 19th century

其中包括一本從圖書館裡借來的

that I got from the library.

19世紀的書。

And I was really amazed because my

我震驚地發現我的名字是

the previous name was in 1889. (Laughter)

先前的一個名字是1889年。

And this book was kind of waiting for 100 years

這本書已經躺了100年了，

to be rediscovered for robotics.

只為了藉機器人之名被重新發現。

And this book shows actors

這本書教演員

how to move every muscle in the body

如何調動他們身體上的每塊肌肉

to match every kind of emotion

來表達他們想要表達的情感。

But the real revelation was

真正讓我受到啟示的是

when I learned about method acting.

方法演技。

It became very popular in the 20th century.

它在20世紀的時候非常流行。

And method acting said, you don't have

方法演技指出，你不需要

Instead you have to use your body

相反，你可以用你的身體找到對的動作。

You have to use your sense memory

你應該運用你的感覺記憶，

to reconstruct the emotions and kind of

去重新建構情感，

think with your body to find the right expression.

用你的身體找到對的表情。

Improvise, play off yor scene partner.

即興發揮，根據你的場景搭配。

And this came at the same time

這個時候我也正讀到

in cognitive psychology called embodied cognition.

認知心理學中的體現認知，

Which also talks about the same ideas --

這也談到同樣的觀點 --

We use our bodies to think,

我們用我們的身體思考，

we don't just think with our brains

我們並不是用大腦思考，用身體表現，

but our bodies feed back into our brain

而是我們的身體反饋給大腦

to generate the way that we behave.

並做出相對應的動作，

And it was like a lightning bolt.

這對我好像一道靈光，

I went back to my office.

我馬上回了我的辦公室。

I wrote this paper -- which I never really published

我寫了一篇論文，從來也沒發表過，

called "Acting Lessons for Artificial Intelligence."

叫做《人工智慧的表演課》。

And I even took another month

我甚至花了一個月的時間

to do what was then the first theater play

去做當時第一部由人類和機器人

with a human and a robot acting together.

一起主演的戲劇。

That's what you saw before with the actors.

你之前看到的演員和

And I thought:

當時我就想：

How can we make an artificial intelligence model --

我們怎樣可以做出一個人工智慧 --

computer, computational model --

電腦、程式模型 --

that will model some of these ideas of improvisation,

能夠即興發揮、

of taking risks, of taking chances,

會冒險、會投機

even of making mistakes.

甚至會犯錯。

Maybe it can make for better robotic teammates.

它可能會是更好的機器人隊友。

So I worked for quite a long time on these models

因此我花了很多時間去研究這些模型，

and I implemented them on a number of robots.

我試用在幾個機器人身上。

Here you can see a very early example

這裡你可以看到一個早期的例子，

with the robots trying to use this

這個機器人試圖運用體現人工智慧

to try to match my movements

盡量地模仿我的動作，

sort of like a game.

就好像遊戲一樣。

Let's look at it.

我們來看一下。

You can see when I psych it out, it gets fooled.

你可以看到我可以糊弄它。

And it's a little bit like what you might see actors do

有點像你看到的演員們

when they try to mirror each other

互相模仿對方

to find the right synchrony between them.

只為了找到他們之間的默契。

And then, I did another experiment,

然後，我又做了另外一個實驗，

and I got people off the street

我從大街上拉人來使用這個機器人檯燈，

and try out this idea of embodied

試驗體現人工智慧。

So, I actually used two kinds

其實，同樣的機器人我用了兩種大腦，

The robot is the same lamp that you saw,

機器人就是你看到的這個檯燈，

and I put in it two brains.

我給了它兩種大腦。

For one half of the people,

對一半的人，

I put in a brain that's kind of the traditional,

我放入了一個傳統的、

calculated robotic brain.

機械計算的大腦。

It waits for its turn, it analyzes everything, it plans.

它會等待，會分析，會計劃，

Let's call it the calculated brain.

我們暫且稱它為“會計算的大腦”。

The other got more the stage actor, risk taker brain.

給另一半人則是那個

Let's call it the adventurous brain.

我們暫且稱它為“愛冒險的大腦”，

It sometimes acts without knowing

有的時候它在並不知道

It sometimes makes mistakes and corrects them.

有的時候它會犯錯然後去糾正。

And I had them do this very tedious task

我讓他們完成一項無比乏味的任務，

that took almost 20 minutes

這個任務要花近20分鐘，

and they had to work together.

他們必須一起合作完成，

Somehow simulating like a factory job

有點類似在工廠工作，

of repetitively doing the same thing.

機械地重複一件事情。

And what I found was that people actually loved

我發現人們非常喜歡

the adventurous robot.

那個“愛冒險的機器人”。

And they thought it was more intelligent,

他們覺得它非常聰明，

more committed, a better member of the team,

更投入，是一個很好的團隊成員，

contributed to the success of the team more.

一起幫助團隊成功。

They even called it 'he' and 'she,'

他們甚至稱它為“他”和“她”，

whereas people with the calculated brain called it 'it.'

而另外那些人稱那個

And nobody ever called it 'he' or 'she'.

沒有人稱它為“他”或“她”。

When they talked about it after the task

任務完成後，

with the adventurous brain,

那些與“會冒險的大腦”互動的人說：

they said, "By the end, we were good

“最後，我們成了好朋友，

Whatever that means.

不管那是啥意思

(Laughter) Sounds painful.

（笑聲）

Whereas the people with the calculated brain

然而，那些與“會計算的大腦”互動的人

said it was just like a lazy apprentice.

則說“它就像一個懶徒弟，

It only did what it was supposed

只會做最基本的。 “

Which is almost what people expect robots to do,

這基本上和同人對機器人期待一樣，

so I was surprised that people

所以我有些驚訝，比起那些機器人專家，

of robots, than what anybody in robotics

人們居然對機器人有更高的期望。

And in a way, I thought, maybe it's time --

但從另一個角度，我想該是時候了 --

just like method acting changed the way

就像方法演技改變了

people thought about acting in the 19th century,

19世紀人們思考表演的方式一樣，

from going from the very calculated,

改變這種通過精確計算的

planned way of behaving,

行為方式，

to a more intuitive, risk-taking,

而轉向一種更直覺的、冒險的、

Maybe it's time for robots

也許應該是

to have the same kind of revolution.

機器人革命的時刻了。

A few years later,

幾年後，

I was at my next research job

我在亞特蘭大的喬治理工大學做研究，

and I was working in a group

我在一個研究機器人音樂家的

dealing with robotic musicians.

小組工作。

And I thought, music, that's the perfect place

我想，音樂是正好是

to look at teamwork, coordination,

研究團隊合作、協調、

timing, improvisation --

時機和即興表演的領域，

and we just got this robot playing marimba.

我們有這個玩馬林巴的機器人。

Marimba, for everybody who was like me,

和我一樣對樂器不在行的朋友，馬林巴是

it was this huge, wooden xylophone.

一個巨大的木琴。

And, when I was looking at this,

當我看著這個，

I looked at other works in

又看了那些人類和機器人的即興互動，

yes, there are other works in

沒錯，還有人和機器人的即興互動

and they were also a little bit like a chess game.

也都差不多像是一個下棋的互動。

The human would play,

人類走一步，

the robot would analyze what was played,

機器人對此分析，

would improvise their own part.

然後決定下一步。

So, this is what musicians called

音樂家們稱其為

a call and response interaction,

呼叫和應答互動，

and it also fits very well, robots

這對機器人和人工智慧很合適。

But I thought, if I use the same ideas I used

但是我想，如果我可以運用

in the theater play and in the teamwork studies,

戲劇表演和團隊合作中的研究發現，

maybe I can make the robots jam together

也許我可以讓這些機器人

like a band.

組成一個樂隊，

Everybody's riffing off each other,

每個人都在即興發揮，

And so, I tried to do the same

於是這次我嘗試用音樂做試驗，

where the robot doesn't really know

機器人並不知道

what it's about to play.

它會演奏什麼，

It just sort of moves its body

它就這樣移動它的身體，

and uses opportunities to play,

找機會演奏，