How we can teach computers to make sense of our emotions | Raphael Arar

64,666 views ・ 2018-04-24

TED


请双击下面的英文字幕来播放视频。

00:00
Translator: Ivana Korom Reviewer: Joanna Pietrulewicz
0
0
7000
翻译人员: Zihao Wang 校对人员: Peipei Xiang
00:13
I consider myself one part artist and one part designer.
1
13760
4680
我自认为既是一位艺术家, 又是一位设计师。
00:18
And I work at an artificial intelligence research lab.
2
18480
3160
我在一个研究 人工智能的实验室工作。
00:22
We're trying to create technology
3
22720
1696
我们正在尝试创造一项
00:24
that you'll want to interact with in the far future.
4
24440
3296
在遥远的将来能和人类互动的科技。
00:27
Not just six months from now, but try years and decades from now.
5
27760
4640
不是在六个月之后, 而是几年甚至几十年之后。
00:33
And we're taking a moonshot
6
33120
1616
我们正在实施一个“登月计划”,
00:34
that we'll want to be interacting with computers
7
34760
2456
我们希望能与电脑
00:37
in deeply emotional ways.
8
37240
2120
进行深层次的情感方面的互动。
00:40
So in order to do that,
9
40280
1456
为了做到这些,
00:41
the technology has to be just as much human as it is artificial.
10
41760
4480
技术不仅要智能,还要人性。
00:46
It has to get you.
11
46920
2256
它必须懂你。
00:49
You know, like that inside joke that'll have you and your best friend
12
49200
3336
就像一个朋友间的笑话, 能让你和你的朋友
00:52
on the floor, cracking up.
13
52560
1936
在地板上笑得前仰后翻;
00:54
Or that look of disappointment that you can just smell from miles away.
14
54520
4560
或者是,你远远就能 嗅到的失望气息。
01:00
I view art as the gateway to help us bridge this gap between human and machine:
15
60560
6040
我把艺术看作是帮助我们填补 人类与机器之间空白的途径:
01:07
to figure out what it means to get each other
16
67280
3136
使每个人都能相互了解对方,
01:10
so that we can train AI to get us.
17
70440
2760
使我们能训练 人工智能来“懂我们”。
01:13
See, to me, art is a way to put tangible experiences
18
73920
3816
对我来说,艺术是把有形的经历,
01:17
to intangible ideas, feelings and emotions.
19
77760
3240
转化为无形的想法、 感受、情感的方式。
01:21
And I think it's one of the most human things about us.
20
81800
2600
我认为这是人性的一个重要特征。
01:25
See, we're a complicated and complex bunch.
21
85480
2936
我们是难懂的、复杂的群体。
01:28
We have what feels like an infinite range of emotions,
22
88440
3136
我们拥有无限的情感,
01:31
and to top it off, we're all different.
23
91600
2496
而且,我们都是不同的。
01:34
We have different family backgrounds,
24
94120
2296
我们拥有不同的家庭背景,
01:36
different experiences and different psychologies.
25
96440
3080
不同的经历,不同的心理活动。
01:40
And this is what makes life really interesting.
26
100240
2720
这是为什么生活那么有趣的原因,
01:43
But this is also what makes working on intelligent technology
27
103440
3496
但这同时也是研究智能技术
01:46
extremely difficult.
28
106960
1600
最难的地方。
01:49
And right now, AI research, well,
29
109640
3456
如今,对人工智能的研究
01:53
it's a bit lopsided on the tech side.
30
113120
2016
过于偏重技术,
01:55
And that makes a lot of sense.
31
115160
2136
这也很好理解。
01:57
See, for every qualitative thing about us --
32
117320
2456
关于我们的每一个定性的特征,
01:59
you know, those parts of us that are emotional, dynamic and subjective --
33
119800
4456
比如属于我们情感的、 动态的、主观的部分——
02:04
we have to convert it to a quantitative metric:
34
124280
3136
我们要把它转化为一个量化指标:
02:07
something that can be represented with facts, figures and computer code.
35
127440
4360
能通过一些事实、图形和 电脑代码表现出来。
02:13
The issue is, there are many qualitative things
36
133000
3376
问题是,有很多定性的东西
02:16
that we just can't put our finger on.
37
136400
1960
是很难量化的。
02:20
So, think about hearing your favorite song for the first time.
38
140400
3200
想一想你第一次听到 你最喜欢的歌的时候,
02:25
What were you doing?
39
145200
1200
你在做什么?
02:28
How did you feel?
40
148000
1200
你有什么感受?
02:30
Did you get goosebumps?
41
150720
1360
你起鸡皮疙瘩了吗?
02:33
Or did you get fired up?
42
153240
1640
你有没有感到热血沸腾?
02:36
Hard to describe, right?
43
156400
1200
很难描述,对吗?
02:38
See, parts of us feel so simple,
44
158800
2096
我们一些看似很简单的感受
02:40
but under the surface, there's really a ton of complexity.
45
160920
3656
背后其实是很复杂的。
02:44
And translating that complexity to machines
46
164600
2936
而将这些复杂的东西 翻译成机器语言,
02:47
is what makes them modern-day moonshots.
47
167560
2856
这就是我们需要实现的 现代“登月计划”。
02:50
And I'm not convinced that we can answer these deeper questions
48
170440
4176
我不相信我们可以仅仅 用0和1这两个数字
02:54
with just ones and zeros alone.
49
174640
1480
来解决这些难题。
02:57
So, in the lab, I've been creating art
50
177120
1936
所以,在实验室,我通过创造艺术
02:59
as a way to help me design better experiences
51
179080
2456
来帮助我设计更好的
03:01
for bleeding-edge technology.
52
181560
2096
对尖端科技的体验。
03:03
And it's been serving as a catalyst
53
183680
1736
艺术作为一种催化剂,
03:05
to beef up the more human ways that computers can relate to us.
54
185440
3840
让电脑更加人类化, 更理解我们。
03:10
Through art, we're tacking some of the hardest questions,
55
190000
2696
通过艺术,我们在解决一些 非常困难的问题,
03:12
like what does it really mean to feel?
56
192720
2360
就像,感受到底是什么意思?
03:16
Or how do we engage and know how to be present with each other?
57
196120
4080
我们如何真正参与或投入其中?
03:20
And how does intuition affect the way that we interact?
58
200800
4000
我们的直觉怎样影响 我们互动的方式?
03:26
So, take for example human emotion.
59
206440
2056
以人类的情感为例,
03:28
Right now, computers can make sense of our most basic ones,
60
208520
3256
如今,电脑能够明白 我们的基本情感,
03:31
like joy, sadness, anger, fear and disgust,
61
211800
3696
比如开心、伤心、 生气、恐惧、厌恶,
03:35
by converting those characteristics to math.
62
215520
3000
把这些特征转化为数学。
03:39
But what about the more complex emotions?
63
219400
2536
那较复杂的情感呢?
03:41
You know, those emotions
64
221960
1216
比如那些很难
03:43
that we have a hard time describing to each other?
65
223200
2376
用文字向对方描述的情感,
03:45
Like nostalgia.
66
225600
1440
比如,怀旧。
03:47
So, to explore this, I created a piece of art, an experience,
67
227640
3936
所以,为了探索这个问题, 我创造了一件艺术品,一种体验,
03:51
that asked people to share a memory,
68
231600
2096
要求人们分享他们的记忆,
03:53
and I teamed up with some data scientists
69
233720
2136
我和一些数据科学家组成一个团队,
03:55
to figure out how to take an emotion that's so highly subjective
70
235880
3576
去研究高度主观的情感是怎样的,
03:59
and convert it into something mathematically precise.
71
239480
3200
如何将它们精确地转化为数学。
04:03
So, we created what we call a nostalgia score
72
243840
2136
我们创造了一个叫怀旧分数的东西,
这是这个装置的核心。
04:06
and it's the heart of this installation.
73
246000
2216
04:08
To do that, the installation asks you to share a story,
74
248240
3056
这个装置会要求你分享一则故事,
04:11
the computer then analyzes it for its simpler emotions,
75
251320
3256
电脑会分析它的一些简单的情感,
04:14
it checks for your tendency to use past-tense wording
76
254600
2656
它会检测你使用 过去时态的词语的偏好,
04:17
and also looks for words that we tend to associate with nostalgia,
77
257280
3336
还会寻找与怀旧有关的词语,
04:20
like "home," "childhood" and "the past."
78
260640
3040
比如“家”、“童年”和“过去”。
04:24
It then creates a nostalgia score
79
264760
2056
它最后会给出一个怀旧分数,
04:26
to indicate how nostalgic your story is.
80
266840
2736
代表着你的故事的怀旧程度,
04:29
And that score is the driving force behind these light-based sculptures
81
269600
4136
这个分数会 让这个灯箱的颜色发生变化,
04:33
that serve as physical embodiments of your contribution.
82
273760
3896
代表着你的贡献。
04:37
And the higher the score, the rosier the hue.
83
277680
3216
分数越高,色调越偏向玫瑰红色,
04:40
You know, like looking at the world through rose-colored glasses.
84
280920
3936
就像是通过玫红色的眼镜看世界。
04:44
So, when you see your score
85
284880
2616
当你看到你的分数,
04:47
and the physical representation of it,
86
287520
2656
以及它的外部反映时,
04:50
sometimes you'd agree and sometimes you wouldn't.
87
290200
2936
有时你会赞同,有时不赞同。
04:53
It's as if it really understood how that experience made you feel.
88
293160
3480
有的时候就好像它真的明白 故事里的你当时的感受,
04:57
But other times it gets tripped up
89
297400
2216
但有的时候它也会出错,
04:59
and has you thinking it doesn't understand you at all.
90
299640
2560
会让你觉得它一点也不懂你。
05:02
But the piece really serves to show
91
302680
1896
但这个装置能够说明,
05:04
that if we have a hard time explaining the emotions that we have to each other,
92
304600
4056
如果连我们都很难表述清楚的情感,
05:08
how can we teach a computer to make sense of them?
93
308680
2360
我们该如何教电脑明白呢?
05:12
So, even the more objective parts about being human are hard to describe.
94
312360
3576
甚至很多关于 人性的客观方面也很难描述。
05:15
Like, conversation.
95
315960
1240
比如,对话。
05:17
Have you ever really tried to break down the steps?
96
317880
2736
你曾经尝试过分解谈话的步骤吗?
05:20
So think about sitting with your friend at a coffee shop
97
320640
2656
试着想象一下, 你和你的朋友坐在咖啡馆,
05:23
and just having small talk.
98
323320
1320
进行简单的交谈,
05:25
How do you know when to take a turn?
99
325160
1720
你怎么知道轮到你说话了?
05:27
How do you know when to shift topics?
100
327440
1840
你怎么知道什么时候该转换话题?
05:29
And how do you even know what topics to discuss?
101
329960
2720
你怎么知道要讨论些什么?
05:33
See, most of us don't really think about it,
102
333560
2096
大多数人都不会想这些问题,
05:35
because it's almost second nature.
103
335680
1656
因为这对我们来说是很自然的。
05:37
And when we get to know someone, we learn more about what makes them tick,
104
337360
3496
当我们认识一个人的时候, 我们会对他们越来越了解,
05:40
and then we learn what topics we can discuss.
105
340880
2376
然后我们会知道能聊些什么话题。
05:43
But when it comes to teaching AI systems how to interact with people,
106
343280
3656
但是,当你教人工智能 怎样与人类互动时,
05:46
we have to teach them step by step what to do.
107
346960
2856
我们需要一步一步教它们该如何做。
05:49
And right now, it feels clunky.
108
349840
3496
而现在,这个过程还感觉很笨拙。
05:53
If you've ever tried to talk with Alexa, Siri or Google Assistant,
109
353360
4136
如果你曾尝试和 Alexa, Siri 或谷歌助手聊天,
05:57
you can tell that it or they can still sound cold.
110
357520
4200
你可以感觉得到, 它们仍听上去冷冰冰的。
06:02
And have you ever gotten annoyed
111
362440
1656
你是否曾经因为它们
06:04
when they didn't understand what you were saying
112
364120
2256
不明白你说什么而变得恼怒,
06:06
and you had to rephrase what you wanted 20 times just to play a song?
113
366400
3840
比如为了让它们放一首歌, 你得说上20次?
06:11
Alright, to the credit of the designers, realistic communication is really hard.
114
371440
4896
不过我们也要理解设计师,毕竟 让机器学会真实的沟通是非常难的。
06:16
And there's a whole branch of sociology,
115
376360
2136
有一个社会学的分支,
06:18
called conversation analysis,
116
378520
1936
叫做会话分析,
06:20
that tries to make blueprints for different types of conversation.
117
380480
3136
它尝试做不同对话类型的蓝图,
06:23
Types like customer service or counseling, teaching and others.
118
383640
4080
例如像客户服务、心理咨询、 教授课程等等的会话类型。
06:28
I've been collaborating with a conversation analyst at the lab
119
388880
2936
我已经在和会话分析学家 在实验室展开合作,
06:31
to try to help our AI systems hold more human-sounding conversations.
120
391840
4696
尝试帮助我们的人工智能系统 进行更多的人性化的对话。
06:36
This way, when you have an interaction with a chatbot on your phone
121
396560
3176
这样,当你和手机 聊天机器人进行互动时,
06:39
or a voice-based system in the car,
122
399760
1856
或者和车载语音系统互动时,
06:41
it sounds a little more human and less cold and disjointed.
123
401640
3320
这种声音就听上去更人性, 不那么冷淡和缺乏逻辑。
06:46
So I created a piece of art
124
406360
1336
我创造的这个艺术品,
06:47
that tries to highlight the robotic, clunky interaction
125
407720
2816
重点突显了机械化的、 笨拙的互动方式,
06:50
to help us understand, as designers,
126
410560
1976
以帮助我们这些设计师明白,
06:52
why it doesn't sound human yet and, well, what we can do about it.
127
412560
4576
为什么它听上去不像人类, 我们该如何解决这个问题。
06:57
The piece is called Bot to Bot
128
417160
1456
这个艺术品叫 Bot to Bot,
06:58
and it puts one conversational system against another
129
418640
2936
它将一个会话系统 搭建在另一个会话系统之上,
07:01
and then exposes it to the general public.
130
421600
2536
然后展示给公众。
07:04
And what ends up happening is that you get something
131
424160
2496
最终会发生的就是,
07:06
that tries to mimic human conversation,
132
426680
1896
它尝试模仿人类的对话,
07:08
but falls short.
133
428600
1896
但是却明显有不足之处。
07:10
Sometimes it works and sometimes it gets into these, well,
134
430520
2736
有的时候它还可以,而有的时候
会陷入误解的循环。
07:13
loops of misunderstanding.
135
433280
1536
07:14
So even though the machine-to-machine conversation can make sense,
136
434840
3096
虽然机器与机器的对话 从语法、用意上
07:17
grammatically and colloquially,
137
437960
2016
能让人明白,
但是你还是能感觉到 这个对话的冰冷和机械化。
07:20
it can still end up feeling cold and robotic.
138
440000
3216
07:23
And despite checking all the boxes, the dialogue lacks soul
139
443240
4016
尽管对话的其他要素都具备, 但却缺少了灵魂,
07:27
and those one-off quirks that make each of us who we are.
140
447280
3136
缺少了那些使我们 之所以为人类的特质。
07:30
So while it might be grammatically correct
141
450440
2056
尽管它的语法也许正确,
07:32
and uses all the right hashtags and emojis,
142
452520
2656
用对了所有的话题标签和符号表情,
07:35
it can end up sounding mechanical and, well, a little creepy.
143
455200
4136
但最终听起来还是有些呆板, 还有点儿吓人。
07:39
And we call this the uncanny valley.
144
459360
2336
我们把这称为恐怖谷,
07:41
You know, that creepiness factor of tech
145
461720
1936
这种科技的恐怖之处在于,
07:43
where it's close to human but just slightly off.
146
463680
2856
它无比接近人类,却又缺了点什么。
07:46
And the piece will start being
147
466560
1456
这件艺术品能开始用于
07:48
one way that we test for the humanness of a conversation
148
468040
3216
测试交流的人性化,
07:51
and the parts that get lost in translation.
149
471280
2160
以及被误解的部分。
07:54
So there are other things that get lost in translation, too,
150
474560
2856
还有其他一些事情 也容易被电脑误解,
07:57
like human intuition.
151
477440
1616
比如,人类的直觉。
07:59
Right now, computers are gaining more autonomy.
152
479080
2776
如今,电脑拥有更多的自主权,
08:01
They can take care of things for us,
153
481880
1736
能为我们管理一些东西,
08:03
like change the temperature of our houses based on our preferences
154
483640
3176
比如根据我们的偏好 调整房子的温度,
08:06
and even help us drive on the freeway.
155
486840
2160
甚至帮助我们在高速公路驾驶。
08:09
But there are things that you and I do in person
156
489560
2496
但是一些我和你会做的事,
08:12
that are really difficult to translate to AI.
157
492080
2760
是非常难翻译给人工智能的。
08:15
So think about the last time that you saw an old classmate or coworker.
158
495440
4360
想一想你上一次 见到一位老同学或老同事时,
08:21
Did you give them a hug or go in for a handshake?
159
501080
2480
你跟他们拥抱还是握手了呢?
08:24
You probably didn't think twice
160
504800
1496
你可能想都没想,
08:26
because you've had so many built up experiences
161
506320
2336
因为你有过许多这样的经历,
08:28
that had you do one or the other.
162
508680
2000
要么拥抱要么握手。
08:31
And as an artist, I feel that access to one's intuition,
163
511440
3456
作为一名艺术家,我认为 了解一个人的直觉,
08:34
your unconscious knowing,
164
514920
1416
你的潜意识的知觉,
08:36
is what helps us create amazing things.
165
516360
3056
能够帮助我们创造 令人惊叹的东西。
08:39
Big ideas, from that abstract, nonlinear place in our consciousness
166
519440
4056
大的点子,我们潜意识中 抽象的,非线性的东西
08:43
that is the culmination of all of our experiences.
167
523520
2960
是我们所有经历的总和。
08:47
And if we want computers to relate to us and help amplify our creative abilities,
168
527840
4656
如果我们想让电脑 帮我们提升创造力,
08:52
I feel that we'll need to start thinking about how to make computers be intuitive.
169
532520
3896
我认为我们需要思考 如何才能让电脑有直觉,
08:56
So I wanted to explore how something like human intuition
170
536440
3096
所以,我想探究 如何将像人类直觉的东西
08:59
could be directly translated to artificial intelligence.
171
539560
3456
直接地翻译给人工智能。
于是我创造了一台通过现实空间
09:03
And I created a piece that explores computer-based intuition
172
543040
3216
09:06
in a physical space.
173
546280
1320
探究电脑直觉的机器。
09:08
The piece is called Wayfinding,
174
548480
1696
它叫 Wayfinding,
09:10
and it's set up as a symbolic compass that has four kinetic sculptures.
175
550200
3936
它有4个动态装置,像一个指南针。
09:14
Each one represents a direction,
176
554160
2056
每一个装置代表着一个方向,
09:16
north, east, south and west.
177
556240
2120
北、东、南、西。
09:19
And there are sensors set up on the top of each sculpture
178
559080
2696
装在每个装置顶端的传感器,
09:21
that capture how far away you are from them.
179
561800
2256
能够捕获你离它们的距离有多远。
09:24
And the data that gets collected
180
564080
1816
接着数据会被采集,
09:25
ends up changing the way that sculptures move
181
565920
2136
最终装置就会移动,
09:28
and the direction of the compass.
182
568080
2040
从而改变指南针的方向。
09:31
The thing is, the piece doesn't work like the automatic door sensor
183
571360
3656
不过不像自动门的传感器那样——
09:35
that just opens when you walk in front of it.
184
575040
2656
你走到它前面的时候,门就会打开,
09:37
See, your contribution is only a part of its collection of lived experiences.
185
577720
5056
你的行为只是它 搜集的体验的一部分,
09:42
And all of those experiences affect the way that it moves.
186
582800
4056
所有的体验都会影响它的移动。
09:46
So when you walk in front of it,
187
586880
1736
所以当你在它前面走动时,
09:48
it starts to use all of the data
188
588640
1976
它开始用所有之前
09:50
that it's captured throughout its exhibition history --
189
590640
2616
捕获的数据——
09:53
or its intuition --
190
593280
1816
或它的直觉——
09:55
to mechanically respond to you based on what it's learned from others.
191
595120
3560
基于它从其他人那里学习到的, 对你做出机械的响应。
09:59
And what ends up happening is that as participants
192
599480
2536
最终,作为参与者,
10:02
we start to learn the level of detail that we need
193
602040
2816
我们意识到我们需要怎样的细节
10:04
in order to manage expectations
194
604880
2016
才能同时管理
10:06
from both humans and machines.
195
606920
2776
人类和机器的预期。
10:09
We can almost see our intuition being played out on the computer,
196
609720
3616
我们几乎可以看到我们的直觉 在电脑中被展示出来,
10:13
picturing all of that data being processed in our mind's eye.
197
613360
3600
想象所有的数据 被我们的心灵之眼所处理。
10:17
My hope is that this type of art
198
617560
1656
我希望这种艺术方式,
10:19
will help us think differently about intuition
199
619240
2416
能帮助我们从不同角度思考直觉,
10:21
and how to apply that to AI in the future.
200
621680
2280
以及将来如何 将它运用到人工智能中去。
10:24
So these are just a few examples of how I'm using art to feed into my work
201
624480
3936
这些都是我在自己的 人工智能设计和研究的工作中
10:28
as a designer and researcher of artificial intelligence.
202
628440
3096
如何利用艺术的例子。
10:31
And I see it as a crucial way to move innovation forward.
203
631560
3496
我觉得这是一个 推动创新的重要方式。
10:35
Because right now, there are a lot of extremes when it comes to AI.
204
635080
4376
因为现在说到人工智能, 两极分化的态度很严重。
10:39
Popular movies show it as this destructive force
205
639480
2816
比如一些流行电影 将其描绘成毁灭性的力量,
10:42
while commercials are showing it as a savior
206
642320
3056
而一些广告则 把它们描绘为救世主——
10:45
to solve some of the world's most complex problems.
207
645400
2936
能解决一些世界上 极端复杂的问题。
10:48
But regardless of where you stand,
208
648360
2536
但是不管你站在哪一边,
10:50
it's hard to deny that we're living in a world
209
650920
2176
我们都无法否认,我们正生活在一个
越来越数字化的世界中。
10:53
that's becoming more and more digital by the second.
210
653120
2576
10:55
Our lives revolve around our devices, smart appliances and more.
211
655720
4600
我们的生活被设备、智能家居等充斥,
11:01
And I don't think this will let up any time soon.
212
661400
2320
而我不觉得这种状况会停止。
11:04
So, I'm trying to embed more humanness from the start.
213
664400
3736
我想在一开始就植入更多的人性,
11:08
And I have a hunch that bringing art into an AI research process
214
668160
5136
而我有预感,将艺术 带入人工智能研究
11:13
is a way to do just that.
215
673320
1896
就是其中一个方法。
11:15
Thank you.
216
675240
1216
谢谢。
11:16
(Applause)
217
676480
2280
(掌声)
关于本网站

这个网站将向你介绍对学习英语有用的YouTube视频。你将看到来自世界各地的一流教师教授的英语课程。双击每个视频页面上显示的英文字幕,即可从那里播放视频。字幕会随着视频的播放而同步滚动。如果你有任何意见或要求,请使用此联系表与我们联系。

https://forms.gle/WvT1wiN1qDtmnspy7