TEDGlobal 2010
Peter Molyneux: Meet Milo, the virtual boy
彼得．莫力納斯展示虛擬男孩「米羅」
Filmed:
Readability: 3.5
852,770 views
彼得．莫力納斯展示運用微軟 Kinect 控制技術，所發展出來的高度互動電玩遊戲「米羅」，他就像一個真的十一歲男孩，他可以看到、聽到和學習，他也認得出你這個人，還可以和你互動。
Peter Molyneux - Game changer
The head of Microsoft's European games division, Peter Molyneux is building an astonishing new "virtual friend" who interacts with you. Full bio
The head of Microsoft's European games division, Peter Molyneux is building an astonishing new "virtual friend" who interacts with you. Full bio
Double-click the English transcript below to play the video.
00:16
When I saw a piece of technology called Kinect --
0
1000
3000
當我注意到有一項叫做 Kinect 的新科技時，
00:19
it was called Natal -- I was inspired,
1
4000
3000
那時它還叫做 Natal，我突然有一個想法，
00:22
and I thought for a moment,
2
7000
2000
我認真地想了一下，
00:24
maybe it's possible
3
9000
2000
或許我們可以
00:26
to address that one problem of storytelling,
4
11000
3000
利用這個科技來說故事，
00:29
to create a character
5
14000
3000
創造一個主角，
00:32
which seemed alive,
6
17000
2000
一個活生生的人物，
00:34
which noticed me,
7
19000
2000
他可以看得到我，
00:36
that could look me in the eyes
8
21000
2000
可以看向我內心深處，
00:38
and feel real,
9
23000
2000
感覺到所有真實的一切，
00:40
and sculpt a story about our relationship.
10
25000
3000
並且和我們建立情感。
00:43
And so a year ago,
11
28000
3000
所以在一年前，
00:46
I showed this off
12
31000
2000
我在 E3 這個電腦展上，
00:48
at a computer show called E3.
13
33000
3000
展示了這項科技，
00:51
And this was a piece of technology
14
36000
2000
那時我們是讓
00:53
with someone called Claire interacting with this boy.
15
38000
3000
Claire 小姐與影片中的男孩互動。
00:56
And there was a huge row online
16
41000
3000
這在網路上引起熱烈討論，
00:59
about, "Hey, this can't be real."
17
44000
3000
大家都說：「嘿，這不是真的吧？」
01:02
And so I waited till now
18
47000
2000
所以我等到現在，
01:04
to have an actual demo
19
49000
2000
才為大家現場展示
01:06
of the real tech.
20
51000
2000
這項嶄新科技。
01:08
Now, this tech incorporates
21
53000
2000
這項科技結合了
01:10
three big elements.
22
55000
2000
三大要素，
01:12
The first is a Kinect camera,
23
57000
2000
第一項是Kinect的攝影機，
01:14
which will be out in November,
24
59000
3000
這個攝影機將在十一月上市；
01:17
some incredible AI
25
62000
2000
還有就是驚人的 AI 人工智慧，
01:19
that was hidden in the dusty vaults,
26
64000
3000
隱藏在保險箱裡，
01:22
collecting dust
27
67000
2000
幫微軟
01:24
in Microsoft,
28
69000
2000
看好錢財；
01:26
plus our quite crude
29
71000
2000
最後再加上
01:28
attempts at AI
30
73000
2000
我們新開發的 AI 人工智慧，
01:30
at a company called Lionhead,
31
75000
3000
由 Lionhead 協助開發。
01:33
mixing all those things together
32
78000
2000
把這三大要素結合起來，
01:35
just to get to this one simple idea:
33
80000
3000
完成了我最初的想法：
01:38
to create a real, living
34
83000
3000
創造一個真實的
01:41
being in a computer.
35
86000
2000
電腦人物出來。
01:43
Now, I'll be honest with you
36
88000
3000
我要誠實地告訴你們，
01:46
and say that most of it
37
91000
2000
大部分的科技
01:48
is just a trick,
38
93000
2000
不過是些小把戲罷了，
01:50
but it's a trick that actually works.
39
95000
3000
但這些小把戲卻能派上用場。
01:53
So why don't we go over and have
40
98000
2000
我們現在就來看看
01:55
a look at the demo now.
41
100000
2000
實際的操作展示吧...
01:57
This is Dimitri.
42
102000
2000
這位是狄米崔，
01:59
Dimitri, just waggle your arm around.
43
104000
3000
狄米崔，試著搖晃你的手。
02:02
Now, you notice he's sitting.
44
107000
2000
各位注意，他現在是坐著的，
02:04
There are no controllers,
45
109000
2000
手上沒有拿任何遙控器，
02:06
no keyboards,
46
111000
2000
沒有鍵盤，
02:08
or mice,
47
113000
2000
沒有滑鼠，
02:10
or joysticks, or joypads.
48
115000
3000
也沒有搖桿這類的東西，
02:13
He is just going to use
49
118000
2000
他只要用自己的雙手、
02:15
his hand, his body and his voice,
50
120000
3000
自己的身體和聲音，
02:18
just like humans interact with their hands, body and voice.
51
123000
3000
就像一般人類的互動一樣，只用手、身體和聲音。
02:21
So let's move forward.
52
126000
2000
讓我們接下去看看，
02:23
You're going to meet Milo for the first time.
53
128000
3000
你將第一次見到米羅。
02:27
We had to give him a problem
54
132000
2000
我們得為他設計一個難題，
02:29
because when we first created Milo,
55
134000
2000
因為在我們初次創造米羅時，
02:31
we realized that he came across as a little bit of a brat,
56
136000
2000
老實說，他看起來是
02:33
to be honest with you.
57
138000
2000
有點頑皮，
02:35
He was quite a know-it-all,
58
140000
2000
以為自己什麼都知道，
02:37
and he wanted to kind of make you laugh.
59
142000
2000
他也會逗別人發笑。
02:39
So the problem we introduced to him was this:
60
144000
2000
我們為他設定的難題是：
02:41
he's just moved house.
61
146000
2000
他剛搬新家。
02:43
He's moved from London
62
148000
2000
他剛從倫敦
02:45
to New England, over in America.
63
150000
3000
搬到美國的新英格蘭，
02:48
His parents are too busy
64
153000
2000
他的父母都太忙，
02:50
to listen to his problems,
65
155000
2000
而忽略了他的問題，
02:52
and that's when he starts almost conjuring you up.
66
157000
3000
那就是故事的背景，魔法要開始了。
02:55
So here he is
67
160000
2000
他現在走在
02:57
walking through the grass.
68
162000
2000
住家的草皮上，
02:59
And you're able to interact with his world.
69
164000
3000
你可以和他互動。
03:02
The cool thing is, what we're doing
70
167000
3000
最酷的是，我們所做的一切，
03:05
is we're changing the mind
71
170000
2000
將持續影響及改變
03:07
of Milo constantly.
72
172000
2000
米羅的想法，
03:09
That means no two people's Milos
73
174000
3000
也就是說，每一個人的米羅
03:12
can be the same.
74
177000
2000
個性都不一樣，
03:14
You're actually sculpting a human being here.
75
179000
3000
你在雕塑另一個人的個性。
03:19
So, he's discovering the garden.
76
184000
2000
他在看花園裡有什麼好玩的，
03:21
You're helping him discover the garden
77
186000
2000
你也可以幫他找找看，
03:23
by just pointing out these snails.
78
188000
3000
幫他指出哪裡有蝸牛，
03:27
Very simple at the start.
79
192000
2000
一開始就是這麼簡單。
03:29
By the way, if you are a boy, it's snails;
80
194000
3000
順道提一下，如果你是男孩，就幫忙找蝸牛，
03:32
if you're a girl, it's butterflies
81
197000
3000
如果是女孩，就幫忙找蝴蝶，
03:35
because what we found was that girls hate snails.
82
200000
3000
因為我們發現女孩子不喜歡蝸牛。
03:38
(Laughter)
83
203000
2000
(笑聲)
03:44
So remember, this is the first time you've met him,
84
209000
3000
請記住，這是你們第一次見面，
03:47
and we really want to draw you in and make you more curious.
85
212000
3000
我們希望能抓住你的注意力，並引發你的好奇心。
03:52
His face, by the way,
86
217000
2000
另外，他的表情是完全由
03:54
is fully AI-driven.
87
219000
2000
AI人工智慧所控制，
03:56
We have complete control over his blush responses,
88
221000
3000
我們可以控制他臉紅的反應，
03:59
the diameter of his nostrils
89
224000
3000
也可以控制他鼻孔的大小
04:02
to denote stress.
90
227000
2000
來表現他所承受的壓力。
04:04
We actually do something called body matching.
91
229000
2000
我們還運用一種叫做動作比對的技術，
04:06
If you're leaning forward,
92
231000
2000
當你往前傾的時時候，
04:08
he will try and slightly change
93
233000
2000
他會稍微改變
04:10
the neuro-linguistic nature of his face,
94
235000
3000
他臉上的細微表情，
04:13
because we went out with this strong idea:
95
238000
2000
因為我們一直有一種想法：
04:15
how can we make you believe that something's real?
96
240000
3000
要怎麼做才能讓你們相信米羅是真的人？
04:18
Now we've used the hand.
97
243000
2000
現在我們只用手來控制而已，
04:20
The other thing to use is your body.
98
245000
3000
我們還可以運用自己的身體。
04:23
Why not just, instead of pushing left and right
99
248000
3000
除了可以用滑鼠或搖桿
04:26
with a mouse or with a joypad,
100
251000
3000
來控制左右方向之外，
04:29
why not use your body just to lean on the chair --
101
254000
2000
為什麼不直接運用自己的身體呢？
04:31
again, relaxed?
102
256000
2000
還比較輕鬆不是嗎？
04:33
You can lean back,
103
258000
2000
你向後靠，
04:35
but the camera will change its perspective
104
260000
2000
攝影機就會改變視野，
04:37
depending on which way you're looking.
105
262000
3000
讓你看到從你的角度可以看到的東西。
04:41
So Dimitri's now going to use --
106
266000
2000
狄米崔現在是用
04:43
he's used his hand; he's used his body.
107
268000
2000
他的手和他的身體來控制，
04:45
He's now going to use the other thing which is essential,
108
270000
3000
他還可以用另一項他本身具備的工具來控制，
04:48
and that's his voice.
109
273000
2000
也就是他的聲音。
04:50
Now, the thing about voice is,
110
275000
2000
關於聲音，
04:52
our experience with voice recognition
111
277000
2000
我們對人聲辨識系統的印象
04:54
is pretty awful, isn't it?
112
279000
2000
一向不好，對吧？
04:56
It never works.
113
281000
2000
好像從來都不成功，
04:58
You order an airline ticket; you end up in Timbuktu.
114
283000
3000
你想要訂個機票，結果卻飛到非洲的提布圖去了。
05:02
So we've tackled that problem,
115
287000
2000
現在我們已經克服這個問題了，
05:04
and we've come up with a solution, which we'll see in a second.
116
289000
3000
等一下就會看到我們的成果。
05:07
Milo: I could just squish it.
117
292000
2000
米羅：我可以踩扁它。
05:09
Peter Molyneux: What are you going to do, Dimitri?
118
294000
2000
彼得：你想怎麼做？狄米崔？
05:11
Female Voice: Squashing a snail may not seem important,
119
296000
2000
女性聲音：踩扁蝸牛不是什麼了不起的大事，
05:13
but remember, even this choice
120
298000
2000
但請記住，即使是這麼小的決定，
05:15
will affect how Milo develops.
121
300000
3000
也會影響米羅日後的發展。
05:18
Do you want Milo to squash it?
122
303000
2000
你希望米羅踩扁蝸牛嗎？
05:20
When you see the microphone,
123
305000
2000
當你看到麥克風時，
05:22
say ... (PM: Squash.) ... yes to decide.
124
307000
2000
說出...(彼得：踩扁它)..你的決定。
05:24
Dimitri: Go on, Milo. Squash it.
125
309000
3000
狄米崔：去吧，米羅，踩扁它。
05:27
PM: No. That's the wrong thing to do.
126
312000
2000
彼得：不對，這樣做不對，
05:29
Now look at his response.
127
314000
2000
看看他的表情。
05:31
He said, "Go on, Milo. Squash it."
128
316000
3000
狄米崔說：「去吧，米羅，踩扁它。」
05:34
What we're using there is,
129
319000
2000
我們所使用的是
05:36
we're using something, a piece of technology called Tellme.
130
321000
2000
一種新科技，叫做 Tellme，
05:38
It's a company that Microsoft acquired some years ago.
131
323000
3000
那是微軟在幾年前購併的一家公司，
05:41
We've got a database of words which we recognize.
132
326000
2000
我們把所能辨認的字彙放進資料庫裡，
05:43
We pick those words out.
133
328000
2000
把那些字挑出來，
05:45
We also reference that
134
330000
2000
然後再參考
05:47
with the tonation database
135
332000
2000
我們為狄米崔
05:49
that we build up of Dimitri's voice,
136
334000
3000
或其他用戶的聲音所建立的
05:52
or the user's voice.
137
337000
2000
音調資料庫。
05:54
Now we need to have a bit more engagement,
138
339000
3000
現在我們要多一點互動了。
05:57
and again, what we can do
139
342000
2000
同樣的，我們可以
05:59
is we can look at the body.
140
344000
2000
運用我們的身體進行互動，
06:01
And we'll do that in a second.
141
346000
2000
等一下就會看到。
06:03
Milo: I wonder how deep it is.
142
348000
2000
米羅：這個池塘有多深啊？
06:13
Deep.
143
358000
2000
很深。
06:16
PM: Okay. So what we're going to do now
144
361000
2000
彼得：好，我們現在要做的是
06:18
is teach Milo to skim stones.
145
363000
3000
教米羅怎麼打水漂兒，
06:21
We're actually teaching him.
146
366000
2000
我們要真的教他打。
06:23
It's very, very interesting
147
368000
2000
有趣的是，
06:25
that men, more than women,
148
370000
2000
男人比女人在這方面，
06:27
tend to be more competitive here.
149
372000
3000
表現得更有企圖心，
06:30
They're fine with teaching Milo for the first few throws,
150
375000
3000
他們一開始會教米羅打幾次水漂兒，
06:33
but then they want to beat Milo,
151
378000
2000
但後來就會想要打敗米羅，
06:35
where women,
152
380000
2000
但是女性就不同了，
06:37
they're more nurturing about this.
153
382000
3000
他們會教導的角度來做這件事。
06:43
Okay, this is skimming stones.
154
388000
2000
好，這是打水漂兒的遊戲，
06:45
How do you skim stones?
155
390000
3000
你要怎麼打水漂兒？
06:48
You stand up,
156
393000
3000
你要先站起來，
06:51
and you skim the stone.
157
396000
2000
然後把石頭丟出去，
06:53
It's that simple.
158
398000
2000
就這麼簡單。
06:55
Just recognizing your body,
159
400000
2000
這項技術會先熟悉你的身體，
06:57
recognizing the body's motions, the tech,
160
402000
2000
瞭解你所做的動作，
06:59
understanding that you've gone
161
404000
2000
它現在知道你已經
07:01
from sitting down to standing up.
162
406000
3000
從坐姿改變成站姿了。
07:04
Again, all of this is done
163
409000
2000
同樣的，我們只要
07:06
in the way us humans do things,
164
411000
2000
做出我們平常所做的動作就可以了，
07:08
and that's crucially important
165
413000
3000
這樣的設計很重要，
07:11
if we want Milo to appear real.
166
416000
2000
因為我們希望米羅像真人一樣。
07:13
Female Voice: See if you can inspire him to do any better.
167
418000
3000
女性聲音：看看你是否能讓米羅打得更好，
07:21
Try hitting the boat.
168
426000
2000
試試看能不能打到那艘船。
07:25
Milo: Ahhh. So close.
169
430000
2000
米羅：啊...很接近了。
07:33
PM: That's Dimitri at his most competitive.
170
438000
2000
彼得：狄米崔現在表現出他強烈的企圖心了，
07:35
Now beaten an 11-year-old child. Well done.
171
440000
3000
他打敗了一個十一歲的孩子，幹得好。
07:39
Milo: Okay.
172
444000
2000
米羅：好。
07:42
PM: So, Milo's being called back in by his parents,
173
447000
3000
彼得：現在米羅被他的父母叫回家，
07:45
giving us time to be alone
174
450000
2000
我們有一點自己的時間，
07:47
and to help him out.
175
452000
2000
我們可以幫他一點忙。
07:49
Basically -- the bit that we missed at the start --
176
454000
2000
其實，在這個故事的一開始，
07:51
his parents had asked him to clean up his room.
177
456000
3000
米羅的爸媽曾經要他整理他自己的房間，
07:54
And we're going to help him with this now.
178
459000
2000
所以我們現在來幫他這個忙吧。
07:56
But this is going to be an introduction,
179
461000
2000
幫他整理房間可以讓米羅認識你，
07:58
and this is all about the deep psychology that we're trying to use.
180
463000
3000
我們希望引用一些深層心理學的技巧，
08:01
We're trying to introduce you
181
466000
2000
我們現在就要讓你們看看
08:03
to what I believe is the most wonderful part,
182
468000
3000
我們覺得最棒的地方：
08:06
you being able to talk
183
471000
2000
你可以用自己的聲音
08:08
in your natural voice to Milo.
184
473000
3000
和米羅說話。
08:11
Now, to do that, we needed a set up,
185
476000
2000
在這之前，我們要先準備一下，
08:13
like a magician's trick.
186
478000
2000
就像魔術表演一樣，
08:15
And what we did was,
187
480000
2000
我們要做的是，
08:17
we needed to give Milo this big problem.
188
482000
2000
丟給米羅一個問題。
08:19
So as Dimitri
189
484000
3000
當狄米崔
08:22
starts tidying up,
190
487000
2000
在幫忙收拾的時候，
08:24
you can overhear a conversation
191
489000
2000
你會聽到背後米羅在和
08:26
that Milo's having with his parents.
192
491000
2000
他的父母交談。
08:28
Milo's Mom: Oh, you've got gravy all over the floor. (Milo: I didn't mean to!)
193
493000
2000
米羅媽媽：噢，你把肉湯弄得到處都是了！(米羅：我又不是故意的！)
08:30
Milo's Mom: That carpet is brand new.
194
495000
2000
米羅媽媽：地毯是新的耶...
08:32
PM: So he's just spilled
195
497000
2000
彼得：米羅剛才把
08:34
a plate of sausages on the floor,
196
499000
2000
一盤肉湯灑到地上了，
08:36
on the brand-new carpet.
197
501000
2000
還灑在全新的地毯上。
08:38
We've all done it as parents; we've all done it as children.
198
503000
3000
身為父母的我們會有這種反應，但其實我們小時候也會犯錯。
08:41
Now's a chance for Dimitri
199
506000
2000
現在，狄米崔要趁機
08:43
to kind of reassure and calm Milo down.
200
508000
2000
安撫一下米羅，
08:45
It's all been too much for him.
201
510000
2000
因為米羅最近的壓力真的很大，
08:47
He's just moved house. He's got no friends.
202
512000
3000
他才剛到新家，又沒有什麼朋友，
08:50
Now is the time
203
515000
2000
所以在我們
08:52
when we open that portal
204
517000
2000
打開房門之後，
08:54
and allow you to talk to Milo.
205
519000
2000
你可以和米羅說說話。
08:57
Female Voice: Why don't you try saying something encouraging
206
522000
3000
女性聲音：要不要說些鼓勵的話，
09:00
to cheer Milo up.
207
525000
2000
來安慰米羅呢？
09:02
Dimitri: Come on, Milo. You know what parents are like.
208
527000
2000
狄米崔：算了吧，米羅，爸媽都是這樣的，
09:04
They're always getting stressed.
209
529000
3000
他們很容易就會發脾氣的。
09:08
Milo: What do they want to come here for anyway?
210
533000
3000
米羅：他們幹嘛要搬來這裡？
09:12
We don't know anyone.
211
537000
3000
我們又不認識什麼人。
09:15
Dimitri: Well, you've got a new school to go to.
212
540000
2000
狄米崔：嗯，你可以去新的學校啊，
09:17
You're going to meet loads of cool, new friends.
213
542000
3000
你會認識一堆很酷的新朋友哦！
09:22
Milo: I just really miss my old house, that's all.
214
547000
3000
米羅：我就是很想念舊家嘛！
09:28
Dimitri: Well, this is a pretty awesome house, Milo.
215
553000
2000
狄米崔：嗯，這個房子也很棒啊，米羅，
09:30
You've got a cool garden to play in and a pond.
216
555000
3000
有一座大花園，還有一個池塘。
09:37
Milo: It was good skimming stones.
217
562000
2000
米羅：打水漂兒是很好玩啦...
09:47
This looks nice.
218
572000
2000
我的房間變整齊了！
09:50
You cleaned up my room.
219
575000
2000
你幫我整理了房間啊...
09:52
Thanks.
220
577000
2000
謝謝！
09:54
PM: So after three-quarters of an hour,
221
579000
2000
彼得：大約在45分鐘之後，
09:56
he recognizes you.
222
581000
2000
他就能認得你了，
09:58
And I promise you, if you're sitting in front of this screen,
223
583000
3000
而且我告訴你，如果你坐在螢幕前面，
10:01
that is a truly wonderful moment.
224
586000
3000
你將會體會到很棒的事情。
10:04
And we're ready now
225
589000
2000
我們現在已經準備好
10:06
to tell a story about his childhood and his life,
226
591000
2000
要把他的童年和他的一生都告訴你，
10:08
and it goes on,
227
593000
2000
而且還會持續下去，
10:10
and he has, you know, many adventures.
228
595000
2000
他會有很多冒險的旅程，
10:12
Some of those adventures are a little bit dark or on the darker side.
229
597000
3000
有些冒險是去做壞事，
10:15
Some of those adventures are wonderfully encouraging --
230
600000
2000
有些則是能鼓動人心，
10:17
he's got to go to school.
231
602000
2000
像是他會去學校上學這件事。
10:19
The cool thing is
232
604000
2000
最酷的是，
10:21
that we're doing as well:
233
606000
2000
我們也會參與其中，
10:23
as you interact with him,
234
608000
2000
當你和他互動的時候，
10:25
you're able to put things into his world; he recognizes objects.
235
610000
3000
你可以把東西放進他的世界，而他會認得這些東西，
10:28
His mind is based in a cloud.
236
613000
3000
他的心智
10:31
That means Milo's mind,
237
616000
2000
仍像一張純潔的白紙，
10:33
as millions of people use it,
238
618000
2000
只要愈多人使用這個軟體，
10:35
will get smarter and cleverer.
239
620000
2000
米羅就會變得更聰明，
10:37
He'll recognize more objects
240
622000
2000
他會認得更多東西，
10:39
and thus understand more words.
241
624000
2000
也會學會更多字彙。
10:41
But for me,
242
626000
2000
但對我來說，
10:43
this is a wonderful opportunity
243
628000
2000
這是一個絕佳的機會，
10:45
where technology, at last, can be connected with,
244
630000
3000
讓我們終於可以把科技和人類連結在一起了，
10:48
where I am no longer restrained
245
633000
2000
我們不必再被自己的
10:50
by the finger I hold in my hand --
246
635000
2000
手指限制住了，
10:52
as far as a computer game's concerned --
247
637000
2000
因為這是目前的電玩的限制；
10:54
or by the blandness of not being noticed
248
639000
2000
我們也不會再有看電影或閱讀
10:56
if you're watching a film or a book.
249
641000
3000
所缺乏的雙向交流。
10:59
And I love those revolutions,
250
644000
3000
我喜歡這種改變，
11:02
and I love the future that Milo brings.
251
647000
2000
我喜歡有米羅的未來。
11:04
Thank you very much indeed.
252
649000
2000
真心謝謝各位！
11:06
(Applause)
253
651000
3000
(掌聲)
ABOUT THE SPEAKER
Peter Molyneux - Game changerThe head of Microsoft's European games division, Peter Molyneux is building an astonishing new "virtual friend" who interacts with you.
Why you should listen
Game geeks have been buzzing about Project Natal for, seemingly, ever -- Microsoft's now-in-development gaming device that goes the Wii one better by allowing your whole body to become the joystick, using touchless, visual-recognition interface technology. What kind of game will be played on this immersive new equipment? Peter Molyneux is working on one that redefines the whole notion of "game" -- Milo.
Milo is a little boy (there's also to be a little girl, Kate) who quite simply acts like a little boy, learning your name, talking with you, going on walks around the lake. It's an immersive, slightly uncanny experience, as seen in the single year-old demo that's still blowing minds on YouTube. Molyneux himself is a game-industry legend who has created many titles. In 1997 he founded Lionhead Studios, makers of Fable, a role-playing game. Microsoft bought Lionhead in 2006, and will release the much-anticipated Fable III in the fall.
Peter Molyneux | Speaker | TED.com