ABOUT THE SPEAKER
Sougwen Chung - Artist, researcher
Sougwen 愫君 Chung is an artist and researcher whose work explores the dynamics between humans and systems.

Why you should listen
Sougwen Chung's work explores the mark-made-by-hand and the mark-made-by-machine as an approach to understanding the dynamics of humans and systems. Chung is a former research fellow at MIT’s Media Lab and a pioneer in the field of human-machine collaboration. In 2019, she was selected as the Woman of the Year in Monaco for achievement in the Arts & Sciences.
 
In 2018 she was an inaugural E.A.T. Artist in Resident in partnership with New Museum and Bell Labs, and was awarded a commission for her project Omnia per Omnia. In 2016, Chung received Japan Media Art’s Excellence Award in for her project, Drawing Operations. She is a former research fellow at MIT’s Media Lab. She has been awarded Artist in Residence positions at Google, Eyebeam, Japan Media Arts and Pier 9 Autodesk. Her speculative critical practice spans performance, installation and drawings which have been featured in numerous exhibitions at museums and galleries around the world.
More profile about the speaker
Sougwen Chung | Speaker | TED.com
TED@BCG Mumbai

Sougwen Chung: Why I draw with robots

钟愫君: 我为何与机器人共同作画

Filmed:
160,983 views

当人类和机器人一起创造艺术时会发生什么?在这场令人叹为观止的演讲中,艺术家钟愫君(Sougwen Chung)展示了她如何将自己的艺术风格“传授”给一台机器——并在意外发现机器人也会犯错后,分享了他们合作的成果,她说:“人类和机器系统的美妙之一正是它们固有的、共同的不完美。 ”
- Artist, researcher
Sougwen 愫君 Chung is an artist and researcher whose work explores the dynamics between humans and systems. Full bio

Double-click the English transcript below to play the video.

00:12
Many许多 of us here use technology技术
in our day-to-day日复一日.
0
937
3165
在座的各位在日常生活中
都会使用科技,
00:16
And some of us rely依靠
on technology技术 to do our jobs工作.
1
4126
3247
许多人依赖科技来
进行他们的工作。
00:19
For a while, I thought of machines
and the technologies技术 that drive驾驶 them
2
7397
3950
有一段时间,我认为机器和科技
00:23
as perfect完善 tools工具 that could make my work
more efficient高效 and more productive生产的.
3
11371
4505
只是让我的工作更高效、高产的
完美工具。
00:28
But with the rise上升 of automation自动化
across横过 so many许多 different不同 industries行业,
4
16403
3254
但随着自动化技术
在各行各业的崛起,
00:31
it led me to wonder奇迹:
5
19681
1372
让我不禁试想:
00:33
If machines are starting开始
to be able能够 to do the work
6
21077
2341
如果机器能够完成
00:35
traditionally传统 doneDONE by humans人类,
7
23442
1667
原本由人类做的工作,
00:37
what will become成为 of the human人的 hand?
8
25133
2333
那我们人类之手又能做些什么呢?
00:40
How does our desire欲望 for perfection完美,
precision精确 and automation自动化
9
28133
4093
对完美,精确和自动化的追求
00:44
affect影响 our ability能力 to be creative创作的?
10
32250
1922
是如何影响我的创造力?
00:46
In my work as an artist艺术家 and researcher研究员,
I explore探索 AIAI and robotics机器人
11
34553
4087
作为艺术家和研究者,
我研究人工智能和机器人,
00:50
to develop发展 new processes流程
for human人的 creativity创造力.
12
38664
3005
以此来开发人类创造的新途径。
00:54
For the past过去 few少数 years年份,
13
42077
1286
过去几年里,
00:55
I've made制作 work alongside并肩 machines,
data数据 and emerging新兴 technologies技术.
14
43387
4376
我运用机器,数据
和新兴科技进行创作。
01:00
It's part部分 of a lifelong终身 fascination魅力
15
48143
1861
其中一部分永恒的魅力
01:02
about the dynamics动力学
of individuals个人 and systems系统
16
50028
2735
在于人与技术间奇妙的动态,
01:04
and all the messiness杂乱 that that entails限嗣继承.
17
52787
2381
还有其中不可避免的混乱。
01:07
It's how I'm exploring探索 questions问题 about
where AIAI ends结束 and we begin开始
18
55192
4808
我借此来探索人工智能与我们的界限,
01:12
and where I'm developing发展 processes流程
19
60024
1642
以及探索发展
01:13
that investigate调查 potential潜在
sensory感觉的 mixes混合 of the future未来.
20
61690
3326
未来感官融合的可能。
01:17
I think it's where philosophy哲学
and technology技术 intersect相交.
21
65675
2857
我想这是哲学与技术的交汇点。
01:20
Doing this work
has taught me a few少数 things.
22
68992
2239
这项工作教会了我一些事。
01:23
It's taught me how embracing拥抱 imperfection缺陷
23
71642
2824
它教会我拥抱不完美
01:26
can actually其实 teach us
something about ourselves我们自己.
24
74490
2489
可以帮助我们认识自我。
01:29
It's taught me that exploring探索 art艺术
25
77428
2336
它教会我探索艺术
01:31
can actually其实 help shape形状
the technology技术 that shapes形状 us.
26
79788
2931
能够更好的构建科技,
从而塑造自我。
01:35
And it's taught me
that combining结合 AIAI and robotics机器人
27
83148
3261
它教会我将人工智能和机器人
01:38
with traditional传统 forms形式 of creativity创造力 --
visual视觉 arts艺术 in my case案件 --
28
86433
3532
结合到传统的创作中——
以我创作的视觉艺术为例——
01:41
can help us think a little bit more deeply
29
89989
2302
能够帮助我们更深入理解
01:44
about what is human人的
and what is the machine.
30
92315
2897
何为人类,何为机器。
01:47
And it's led me to the realization实现
31
95942
1707
它让我意识到
01:49
that collaboration合作 is the key
to creating创建 the space空间 for both
32
97673
3055
在我们进步的路上,
合作是创造人与机器共存空间的关键。
01:52
as we move移动 forward前锋.
33
100752
1267
01:54
It all started开始 with a simple简单
experiment实验 with machines,
34
102387
2746
这一切都始于
一个简单的机器实验,
01:57
called "Drawing画画 Operations操作
Unit单元: Generation 1."
35
105157
2826
实验机器叫“绘图机器:初代”
(Drawing Operations Unit: Generation 1)。
02:00
I call the machine "D.O.U.G." for short.
36
108434
2516
我把它简称为道格(D.O.U.G.),
02:02
Before I built内置 D.O.U.G,
37
110974
1326
在我建造道格之前,
02:04
I didn't know anything
about building建造 robots机器人.
38
112324
2365
我对造机器人一无所知,
02:07
I took some open-source开源
robotic机器人 arm designs设计,
39
115220
2897
我参考了一些开源的机器臂设计,
02:10
I hacked砍死 together一起 a system系统
where the robot机器人 would match比赛 my gestures手势
40
118141
3341
编成了一个系统来实现匹配手势,
02:13
and follow跟随 [them] in real真实 time.
41
121506
1639
并实时模仿它们。
02:15
The premise前提 was simple简单:
42
123169
1448
前提很简单:
02:16
I would lead, and it would follow跟随.
43
124641
2200
我画,而它会学我。
02:19
I would draw a line线,
and it would mimic模仿者 my line线.
44
127403
2936
我画一条线,
它也会跟着我画一条线。
02:22
So back in 2015, there we were,
drawing画画 for the first time,
45
130363
3698
回到 2015 年,那是我们第一次
02:26
in front面前 of a small audience听众
in New York纽约 City.
46
134085
2619
在纽约的一小群观众前作画。
02:28
The process处理 was pretty漂亮 sparse --
47
136728
2555
整个过程非常冷清——
02:31
no lights灯火, no sounds声音,
nothing to hide隐藏 behind背后.
48
139307
3487
没有灯光,没有音效,
也没有什么悬念。
02:35
Just my palms手掌 sweating出汗
and the robot's机器人 new servos舵机 heating加热 up.
49
143241
3395
只有手掌冒出的冷汗
和机器臂不断升高的温度。
02:38
(Laughs) Clearly明确地, we were
not built内置 for this.
50
146950
2441
(笑声)显然,
这不是我们想要的效果。
02:41
But something interesting有趣 happened发生,
something I didn't anticipate预料.
51
149820
3233
但有趣的事发生了,
完全出乎意料。
02:45
See, D.O.U.G., in its primitive原始 form形成,
wasn't tracking追踪 my line线 perfectly完美.
52
153077
4802
初代的道格并没有
完美地模仿我画的线条,
02:49
While in the simulation模拟
that happened发生 onscreen在屏幕上
53
157903
2333
在计算器模拟中显示
02:52
it was pixel-perfect像素完美,
54
160260
1357
它的模仿事精确完美的,
02:53
in physical物理 reality现实,
it was a different不同 story故事.
55
161641
2531
但到了现实中,却并非如此。
02:56
It would slip and slide滑动
and punctuate圈点 and falter动摇,
56
164196
2817
它会滑动,会卡顿,会晃动,
02:59
and I would be forced被迫 to respond响应.
57
167037
2068
于是我不得不附和它的线条。
03:01
There was nothing pristine质朴 about it.
58
169525
1778
它的状态不完美,
03:03
And yet然而, somehow不知何故, the mistakes错误
made制作 the work more interesting有趣.
59
171327
3238
而这些失误让作品更加有趣,
03:06
The machine was interpreting解读
my line线 but not perfectly完美.
60
174589
2754
机器在模仿我的线条,
但是并不完美,
03:09
And I was forced被迫 to respond响应.
61
177367
1372
于是变成我在附和机器。
03:10
We were adapting适应
to each other in real真实 time.
62
178763
2709
我们不断地实时熟悉彼此。
03:13
And seeing眼看 this taught me a few少数 things.
63
181496
1937
看到这些,教会了我一些事,
03:15
It showed显示 me that our mistakes错误
actually其实 made制作 the work more interesting有趣.
64
183457
4880
我们的失误,实际上
让我们的作品更加有趣,
03:20
And I realized实现 that, you know,
through通过 the imperfection缺陷 of the machine,
65
188663
4249
我从机器的不完美中意识到,
03:24
our imperfections缺陷 became成为
what was beautiful美丽 about the interaction相互作用.
66
192936
3705
我们的不完美成就了这互动之美。
03:29
And I was excited兴奋,
because it led me to the realization实现
67
197650
3087
而我很兴奋,因为它让我意识到
03:32
that maybe part部分 of the beauty美女
of human人的 and machine systems系统
68
200761
3650
或许人类和机器系统的美妙之一
03:36
is their shared共享 inherent固有 fallibility易错.
69
204435
2738
正是他们共同的、固有的不完美。
03:39
For the second第二 generation of D.O.U.G.,
70
207197
1820
对于第二代的道格,
03:41
I knew知道 I wanted to explore探索 this idea理念.
71
209041
2307
我知道我要探索这个想法,
03:43
But instead代替 of an accident事故 produced生成
by pushing推动 a robotic机器人 arm to its limits范围,
72
211372
4418
我并非打算通过放大机器臂的失误,
03:47
I wanted to design设计 a system系统
that would respond响应 to my drawings图纸
73
215814
2897
而是想要设计一个系统
能够以出其不意的方式
03:50
in ways方法 that I didn't expect期望.
74
218735
1833
回应我的画作。
03:52
So, I used a visual视觉 algorithm算法
to extract提取 visual视觉 information信息
75
220592
3849
所以,我运用一个视觉算法
来提取我几十年来的
03:56
from decades几十年 of my digital数字
and analog类似物 drawings图纸.
76
224465
2978
数字和实体绘图中的视觉样本信息,
03:59
I trained熟练 a neural神经 net on these drawings图纸
77
227467
2055
以此我训练了一个神经网络
04:01
in order订购 to generate生成
recurring经常性 patterns模式 in the work
78
229546
2865
优化机器的循环模式,
04:04
that were then fed美联储 through通过 custom习惯 software软件
back into the machine.
79
232435
3476
视觉样本由经专门的
软件处理导入机器。
04:07
I painstakingly精心 collected
as many许多 of my drawings图纸 as I could find --
80
235935
4386
于是我煞费苦心地
收集我的所有的画作——
04:12
finished works作品, unfinished未完成 experiments实验
and random随机 sketches素描 --
81
240345
4215
成品,半成品,随手简笔画——
04:16
and tagged标记 them for the AIAI system系统.
82
244584
1999
把它们标记给人工智能系统。
04:18
And since以来 I'm an artist艺术家,
I've been making制造 work for over 20 years年份.
83
246607
3684
作为一位艺术家,
我作画超过了 20 年,
04:22
Collecting收集 that many许多 drawings图纸 took months个月,
84
250315
2024
所以收集这些画作花了好多个月,
04:24
it was a whole整个 thing.
85
252363
1389
这是个大工程。
04:25
And here's这里的 the thing
about training训练 AIAI systems系统:
86
253776
2595
说到训练人工智能:
04:28
it's actually其实 a lot of hard work.
87
256395
2200
这其实大费功夫。
04:31
A lot of work goes on behind背后 the scenes场景.
88
259022
2191
幕后的工作很多很多,
04:33
But in doing the work,
I realized实现 a little bit more
89
261237
2681
但在其中,我对人工智能的构造
04:35
about how the architecture建筑
of an AIAI is constructed.
90
263942
3421
更深入了解了一些。
04:39
And I realized实现 it's not just made制作
of models楷模 and classifiers分类
91
267387
2947
我意识到它不仅是
神经网络的
04:42
for the neural神经 network网络.
92
270358
1322
模型和分屏器。
04:43
But it's a fundamentally从根本上
malleable可锻铸 and shapable沙布 system系统,
93
271704
3532
它是一个可延展的、可塑的系统,
04:47
one in which哪一个 the human人的 hand
is always present当下.
94
275260
3111
人类的手始终参与其中。
04:50
It's far from the omnipotent无所不能 AIAI
we've我们已经 been told to believe in.
95
278395
4000
它不再是我们认为的
无所不能的人工智能。
04:54
So I collected these drawings图纸
for the neural神经 net.
96
282419
2515
所以,我收集画作以训练神经网络,
04:56
And we realized实现 something
that wasn't previously先前 possible可能.
97
284958
3929
而且我们意识到
前所未有的事情发生了,
05:00
My robot机器人 D.O.U.G. became成为
a real-time即时的 interactive互动 reflection反射
98
288911
4091
我对机器人道格
在实时交互创作中,
05:05
of the work I'd doneDONE
through通过 the course课程 of my life.
99
293026
2627
对我过去人生几十年的作品做出回应。
05:07
The data数据 was personal个人,
but the results结果 were powerful强大.
100
295677
3865
数据源于我个人,
但结果却很有力。
05:11
And I got really excited兴奋,
101
299566
1484
我感到非常兴奋,
05:13
because I started开始 thinking思维 maybe
machines don't need to be just tools工具,
102
301074
4582
因为我开始想或许机器不该只是工具,
05:17
but they can function功能
as nonhuman非人 collaborators合作者.
103
305680
3420
它还可以是非人的合作者。
05:21
And even more than that,
104
309537
1547
再进一步想,
05:23
I thought maybe
the future未来 of human人的 creativity创造力
105
311108
2429
也许未来的人类创作
05:25
isn't in what it makes品牌
106
313561
1524
不在于作品本身,
05:27
but how it comes together一起
to explore探索 new ways方法 of making制造.
107
315109
3436
而在于对艺术诞生新方式的探索。
05:31
So if D.O.U.G._1 was the muscle肌肉,
108
319101
2190
所以,如果道格初代是肌肉,
05:33
and D.O.U.G._2 was the brain,
109
321315
1762
那么道格二代就是大脑,
05:35
then I like to think
of D.O.U.G._3 as the family家庭.
110
323101
2928
然后我想道格三代就是家人。
05:38
I knew知道 I wanted to explore探索 this idea理念
of human-nonhuman人-非人类 collaboration合作 at scale规模.
111
326482
4793
我知道我想要将对
人类和非人类合作的想法放大。
05:43
So over the past过去 few少数 months个月,
112
331299
1373
于是再过去的几个月里,
05:44
I worked工作 with my team球队
to develop发展 20 custom习惯 robots机器人
113
332696
3135
我和团队造出了 20 个定制的机器人
05:47
that could work with me as a collective集体.
114
335855
1960
与我集体创作。
05:49
They would work as a group,
115
337839
1293
它们像团队一样工作,
05:51
and together一起, we would collaborate合作
with all of New York纽约 City.
116
339156
2889
我们共同与整个纽约市携手合作,
05:54
I was really inspired启发
by Stanford斯坦福 researcher研究员 Fei-Fei菲菲 Li,
117
342069
2944
斯坦福大学的研究员李飞飞
激发了我对灵感,
05:57
who said, "if we want to teach
machines how to think,
118
345037
2515
她说,"若像教机器如何思考,
05:59
we need to first teach them how to see."
119
347576
1984
先要教它们如何看见。"
06:01
It made制作 me think of the past过去 decade
of my life in New York纽约,
120
349584
2785
这让我想起了过去
十年的纽约生活,
06:04
and how I'd been all watched看着 over by these
surveillance监控 cameras相机 around the city.
121
352393
3993
城市上空的监控摄像头监视着我,
06:08
And I thought it would be
really interesting有趣
122
356410
2056
如果我用它们来训练
06:10
if I could use them
to teach my robots机器人 to see.
123
358490
2405
我的机器人的视觉,
那会非常有趣。
06:12
So with this project项目,
124
360919
1888
所以在这个项目中,
06:14
I thought about the gaze凝视 of the machine,
125
362831
1967
我思考机器对我们的凝视,
06:16
and I began开始 to think about vision视力
as multidimensional多维,
126
364822
3226
于是我开始将视觉看成多元化的,
06:20
as views意见 from somewhere某处.
127
368072
1600
视作来自某处的视点。
06:22
We collected video视频
128
370151
1834
我们收集视频,
06:24
from publicly公然 available可得到
camera相机 feeds供稿 on the internet互联网
129
372009
3063
从网络上公共摄像头的影片
06:27
of people walking步行 on the sidewalks人行道,
130
375096
1690
到行人在路上走的片段,
06:28
cars汽车 and taxis出租车 on the road,
131
376810
1712
道路上的汽车和出租,
06:30
all kinds of urban城市的 movement运动.
132
378546
1817
城市中各种车水马龙的片段。
06:33
We trained熟练 a vision视力 algorithm算法
on those feeds供稿
133
381188
2603
基于一种“光流技术”,
06:35
based基于 on a technique技术
called "optical光纤 flow,"
134
383815
2286
我们训练了一种视觉算法,
06:38
to analyze分析 the collective集体 density密度,
135
386125
1977
来分析收集到的人流密度,
06:40
direction方向, dwell and velocity速度 states状态
of urban城市的 movement运动.
136
388126
3637
城市流动的方向,
速度状态以及居住方式。
06:44
Our system系统 extracted提取 those states状态
from the feeds供稿 as positional位置 data数据
137
392178
4269
我们的系统从海量的
位置数据中提取这些信息,
06:48
and became成为 pads for my
robotic机器人 units单位 to draw on.
138
396471
3373
我们的机器人依靠这些信息来作画,
06:51
Instead代替 of a collaboration合作 of one-to-one一到一个,
139
399868
2534
与之前的一对一合作不同,
06:54
we made制作 a collaboration合作 of many-to-many许多一对多.
140
402426
3024
我们实现了多对多的合作。
06:57
By combining结合 the vision视力 of human人的
and machine in the city,
141
405474
3587
通过结合城市中人类与机器的视角,
07:01
we reimagined重新想象 what
a landscape景观 painting绘画 could be.
142
409085
2794
我们重构了一个景观绘图可能的样子。
07:03
Throughout始终 all of my
experiments实验 with D.O.U.G.,
143
411903
2218
在我和道格所有的实验中,
07:06
no two performances演出
have ever been the same相同.
144
414145
2717
没有哪两次的呈现是相同的,
07:08
And through通过 collaboration合作,
145
416886
1382
而且通过合作,
07:10
we create创建 something that neither也不 of us
could have doneDONE alone单独:
146
418292
2864
我们创作了我们
无法独自实现的事情,
07:13
we explore探索 the boundaries边界
of our creativity创造力,
147
421180
2611
我们共同探索了创造力的边界,
07:15
human人的 and nonhuman非人 working加工 in parallel平行.
148
423815
2892
人类和非人类并肩工作。
07:19
I think this is just the beginning开始.
149
427823
2334
我想这才是开始,
07:22
This year, I've launched推出 Scilicet西利塞特,
150
430569
2183
今年,我创办了 Scilicet,
07:24
my new lab实验室 exploring探索 human人的
and interhuman人间 collaboration合作.
151
432776
4245
这个新实验室旨在探索
人类和非人类间的合作,
07:29
We're really interested有兴趣
in the feedback反馈 loop循环
152
437339
2120
我们对个体,人工和生态系统
07:31
between之间 individual个人, artificial人造
and ecological生态 systems系统.
153
439483
4230
之间的反馈关系非常感兴趣。
07:36
We're connecting human人的 and machine output产量
154
444276
2269
我们将人类和机器与
07:38
to biometrics生物识别技术 and other kinds
of environmental环境的 data数据.
155
446569
2984
生物特征识别和其他环境数据相结合。
07:41
We're inviting诱人的 anyone任何人 who's谁是 interested有兴趣
in the future未来 of work, systems系统
156
449577
4079
我们邀请任何对未来的作品、系统
07:45
and interhuman人间 collaboration合作
157
453680
1595
和人际间合作感兴趣的人
07:47
to explore探索 with us.
158
455299
1550
和我们共同探索。
07:48
We know it's not just technologists技术专家
that have to do this work
159
456873
3405
我们知道不仅是科技工作者肩负使命,
07:52
and that we all have a role角色 to play.
160
460302
2103
所有人都可以参与其中。
07:54
We believe that by teaching教学 machines
161
462429
2243
我们坚信通过教授机器
07:56
how to do the work
traditionally传统 doneDONE by humans人类,
162
464696
2730
如何去完成人类的传统工作,
07:59
we can explore探索 and evolve发展 our criteria标准
163
467450
2953
我们就能不断探索和创新
08:02
of what's made制作 possible可能 by the human人的 hand.
164
470427
2443
超越人类之手所能达到的可能。
08:04
And part部分 of that journey旅程
is embracing拥抱 the imperfections缺陷
165
472894
3493
这段旅程之一便是拥抱不完美,
08:08
and recognizing认识 the fallibility易错
of both human人的 and machine,
166
476411
3690
发现人类和机器共有的缺憾,
08:12
in order订购 to expand扩大 the potential潜在 of both.
167
480125
2405
才能更好的拓展我们共同的潜能。
08:14
Today今天, I'm still in pursuit追求
of finding发现 the beauty美女
168
482919
2301
今天,我仍在追寻人类和
08:17
in human人的 and nonhuman非人 creativity创造力.
169
485244
2276
非人类协作的美妙之处。
08:19
In the future未来, I have no idea理念
what that will look like,
170
487865
2829
在未来,我不知道会怎样,
08:23
but I'm pretty漂亮 curious好奇 to find out.
171
491627
2024
但是我满怀好奇去寻找答案。
08:25
Thank you.
172
493675
1151
谢谢。
08:26
(Applause掌声)
173
494850
1884
(掌声)
Translated by Yanyan Hong
Reviewed by Cissy Yun

▲Back to top

ABOUT THE SPEAKER
Sougwen Chung - Artist, researcher
Sougwen 愫君 Chung is an artist and researcher whose work explores the dynamics between humans and systems.

Why you should listen
Sougwen Chung's work explores the mark-made-by-hand and the mark-made-by-machine as an approach to understanding the dynamics of humans and systems. Chung is a former research fellow at MIT’s Media Lab and a pioneer in the field of human-machine collaboration. In 2019, she was selected as the Woman of the Year in Monaco for achievement in the Arts & Sciences.
 
In 2018 she was an inaugural E.A.T. Artist in Resident in partnership with New Museum and Bell Labs, and was awarded a commission for her project Omnia per Omnia. In 2016, Chung received Japan Media Art’s Excellence Award in for her project, Drawing Operations. She is a former research fellow at MIT’s Media Lab. She has been awarded Artist in Residence positions at Google, Eyebeam, Japan Media Arts and Pier 9 Autodesk. Her speculative critical practice spans performance, installation and drawings which have been featured in numerous exhibitions at museums and galleries around the world.
More profile about the speaker
Sougwen Chung | Speaker | TED.com