TEDSummit
Zeynep Tufekci: Machine intelligence makes human morals more important
图费・克奇: 机器智能时代，坚守人类道德更重要
Filmed:
Readability: 4.9
1,648,711 views
人工智能的到来，使我们在很多主观决策上使用它们。但是人工智能的复杂度使人难以去了解，去控制它。在这个演讲中，科技社会学家图费・克奇解释了机器和人类在出错这方面的差异性，并提出这些问题是我们难以预料和准备的。“我们不能把我们该承担的责任转移给机器”，她说，“我们必须更加坚守人类的价值观和伦理。”
Zeynep Tufekci - Techno-sociologist
Techno-sociologist Zeynep Tufekci asks big questions about our societies and our lives, as both algorithms and digital connectivity spread. Full bio
Techno-sociologist Zeynep Tufekci asks big questions about our societies and our lives, as both algorithms and digital connectivity spread. Full bio
Double-click the English transcript below to play the video.
00:12
So, I started my first job
as a computer programmer
as a computer programmer
0
739
4122
我的第一份工作是程序员，
00:16
in my very first year of college --
1
4885
1956
那是在我刚上大学的时候，
00:18
basically, as a teenager.
2
6865
1507
不到二十岁。
00:20
Soon after I started working,
3
8889
1732
我刚开始工作不久，
00:22
writing software in a company,
4
10645
1610
正当在公司写程序，
00:24
a manager who worked at the company
came down to where I was,
came down to where I was,
5
12799
3635
公司的一位经理来到我旁边，
00:28
and he whispered to me,
6
16458
1268
他悄悄的对我说，
00:30
"Can he tell if I'm lying?"
7
18229
2861
“他能看出来我在撒谎吗？”
00:33
There was nobody else in the room.
8
21806
2077
当时屋子里没有别人。
00:37
"Can who tell if you're lying?
And why are we whispering?"
And why are we whispering?"
9
25032
4389
“你是指谁能看出你在撒谎？
还有，我们干嘛要悄悄地说话？”
还有，我们干嘛要悄悄地说话？”
00:42
The manager pointed
at the computer in the room.
at the computer in the room.
10
30266
3107
那个经理指着屋子里的电脑，说：
00:45
"Can he tell if I'm lying?"
11
33397
3096
“他能看出我在撒谎吗？”
00:49
Well, that manager was having
an affair with the receptionist.
an affair with the receptionist.
12
37613
4362
其实，那个经理和前台有一腿。
00:53
(Laughter)
13
41999
1112
（笑声）
00:55
And I was still a teenager.
14
43135
1766
当时我只有十来岁，
00:57
So I whisper-shouted back to him,
15
45447
2019
我低声地回答他，
00:59
"Yes, the computer can tell
if you're lying."
if you're lying."
16
47490
3624
“是的，电脑什么都知道。”
01:03
(Laughter)
17
51138
1806
（笑声）
01:04
Well, I laughed, but actually,
the laugh's on me.
the laugh's on me.
18
52968
2923
我笑了，但其实我是在笑自己，
01:07
Nowadays, there are computational systems
19
55915
3268
现在，计算机系统已经可以
01:11
that can suss out
emotional states and even lying
emotional states and even lying
20
59207
3548
通过分析人脸来辨别人的情绪，
甚至包括是否在撒谎。
01:14
from processing human faces.
21
62779
2044
01:17
Advertisers and even governments
are very interested.
are very interested.
22
65248
4153
广告商，甚至政府
都对此很感兴趣。
都对此很感兴趣。
01:22
I had become a computer programmer
23
70319
1862
我选择成为电脑程序员，
01:24
because I was one of those kids
crazy about math and science.
crazy about math and science.
24
72205
3113
因为我是那种
痴迷于数学和科学孩子。
痴迷于数学和科学孩子。
01:27
But somewhere along the line
I'd learned about nuclear weapons,
I'd learned about nuclear weapons,
25
75942
3108
其间我也学习过核武器，
01:31
and I'd gotten really concerned
with the ethics of science.
with the ethics of science.
26
79074
2952
我也非常关心科学伦理。
01:34
I was troubled.
27
82050
1204
我曾经很困惑。
01:35
However, because of family circumstances,
28
83278
2641
但是，因为家庭原因，
01:37
I also needed to start working
as soon as possible.
as soon as possible.
29
85943
3298
我需要尽快参加工作。
01:41
So I thought to myself, hey,
let me pick a technical field
let me pick a technical field
30
89265
3299
我对自己说，嘿，
选一个容易找工作
选一个容易找工作
01:44
where I can get a job easily
31
92588
1796
的科技领域吧，
01:46
and where I don't have to deal
with any troublesome questions of ethics.
with any troublesome questions of ethics.
32
94408
4018
并且找个不需要
操心伦理问题的。
操心伦理问题的。
01:51
So I picked computers.
33
99022
1529
所以我选了计算机。
01:52
(Laughter)
34
100575
1104
（笑声）
01:53
Well, ha, ha, ha!
All the laughs are on me.
All the laughs are on me.
35
101703
3410
哈哈哈，
我多可笑。
我多可笑。
01:57
Nowadays, computer scientists
are building platforms
are building platforms
36
105137
2754
如今，计算机科学控制着
01:59
that control what a billion
people see every day.
people see every day.
37
107915
4209
十亿人每天能看到的信息，
02:05
They're developing cars
that could decide who to run over.
that could decide who to run over.
38
113052
3822
它们可以控制汽车朝哪里开，
02:09
They're even building machines, weapons,
39
117707
3213
它们可以建造机器、武器，
02:12
that might kill human beings in war.
40
120944
2285
那些在战争中
用于杀人的武器。
用于杀人的武器。
02:15
It's ethics all the way down.
41
123253
2771
说到底，
都是伦理问题。
都是伦理问题。
02:19
Machine intelligence is here.
42
127183
2058
机器智能来了。
02:21
We're now using computation
to make all sort of decisions,
to make all sort of decisions,
43
129823
3474
我们用计算机来做各种决策，
02:25
but also new kinds of decisions.
44
133321
1886
包括人们面临的新决策。
02:27
We're asking questions to computation
that have no single right answers,
that have no single right answers,
45
135231
5172
我们向计算机询问多解的、
02:32
that are subjective
46
140427
1202
主观的、
02:33
and open-ended and value-laden.
47
141653
2325
开放性的或意义深远的问题。
02:36
We're asking questions like,
48
144002
1758
我们会问，
02:37
"Who should the company hire?"
49
145784
1650
“我们公司应该聘请谁？”
02:40
"Which update from which friend
should you be shown?"
should you be shown?"
50
148096
2759
“你该关注哪个朋友
的哪条状态？”
的哪条状态？”
02:42
"Which convict is more
likely to reoffend?"
likely to reoffend?"
51
150879
2266
“哪种犯罪更容易再犯？”
02:45
"Which news item or movie
should be recommended to people?"
should be recommended to people?"
52
153514
3054
“应该给人们推荐
哪条新闻或是电影？”
哪条新闻或是电影？”
02:48
Look, yes, we've been using
computers for a while,
computers for a while,
53
156592
3372
看，是的，我们使用计算机
已经有一段时间了，
已经有一段时间了，
02:51
but this is different.
54
159988
1517
但现在不一样了。
02:53
This is a historical twist,
55
161529
2067
这是历史性的转折，
02:55
because we cannot anchor computation
for such subjective decisions
for such subjective decisions
56
163620
5337
因为我们在这些主观决策上
无法主导计算机，
无法主导计算机，
03:00
the way we can anchor computation
for flying airplanes, building bridges,
for flying airplanes, building bridges,
57
168981
5420
不像我们在
管理飞机、建造桥梁、
管理飞机、建造桥梁、
登月等问题上，
可以主导它们。
可以主导它们。
03:06
going to the moon.
58
174425
1259
03:08
Are airplanes safer?
Did the bridge sway and fall?
Did the bridge sway and fall?
59
176449
3259
飞机会更安全吗？
桥梁会摇晃或倒塌吗？
桥梁会摇晃或倒塌吗？
03:11
There, we have agreed-upon,
fairly clear benchmarks,
fairly clear benchmarks,
60
179732
4498
在这些问题上，我们
有统一而清晰的判断标准，
有统一而清晰的判断标准，
03:16
and we have laws of nature to guide us.
61
184254
2239
我们有自然定律来指导。
03:18
We have no such anchors and benchmarks
62
186517
3394
但是在复杂的人类事务上，
03:21
for decisions in messy human affairs.
63
189935
3963
我们没有这样的客观标准。
03:25
To make things more complicated,
our software is getting more powerful,
our software is getting more powerful,
64
193922
4237
让问题变得更复杂的，
是我们的软件正越来越强大，
是我们的软件正越来越强大，
03:30
but it's also getting less
transparent and more complex.
transparent and more complex.
65
198183
3773
同时也变得更加不透明，
更加复杂。
更加复杂。
03:34
Recently, in the past decade,
66
202542
2040
最近的几十年，
03:36
complex algorithms
have made great strides.
have made great strides.
67
204606
2729
复杂算法已
取得了长足发展，
取得了长足发展，
03:39
They can recognize human faces.
68
207359
1990
它们可以识别人脸，
03:41
They can decipher handwriting.
69
209985
2055
它们可以破解笔迹，
03:44
They can detect credit card fraud
70
212436
2066
它们可以识别信用卡欺诈，
03:46
and block spam
71
214526
1189
可以屏蔽垃圾信息，
03:47
and they can translate between languages.
72
215739
2037
它们可以翻译语言，
03:49
They can detect tumors in medical imaging.
73
217800
2574
他们可以通过
医学图像识别肿瘤，
医学图像识别肿瘤，
03:52
They can beat humans in chess and Go.
74
220398
2205
它们可以在国际象棋
和围棋上击败人类。
和围棋上击败人类。
03:55
Much of this progress comes
from a method called "machine learning."
from a method called "machine learning."
75
223264
4504
类似的很多发展，
都来自一种叫“机器学习”的方法。
都来自一种叫“机器学习”的方法。
04:00
Machine learning is different
than traditional programming,
than traditional programming,
76
228175
3187
机器学习不像传统程序一样，
04:03
where you give the computer
detailed, exact, painstaking instructions.
detailed, exact, painstaking instructions.
77
231386
3585
需要给计算机详细、
准确的逐条指令。
准确的逐条指令。
04:07
It's more like you take the system
and you feed it lots of data,
and you feed it lots of data,
78
235378
4182
它更像是你给系统
喂了很多数据，
喂了很多数据，
04:11
including unstructured data,
79
239584
1656
包括非结构化数据，
04:13
like the kind we generate
in our digital lives.
in our digital lives.
80
241264
2278
比如我们在数字生活中
产生的数据。
产生的数据。
04:15
And the system learns
by churning through this data.
by churning through this data.
81
243566
2730
系统扎进这些数据中学习，
04:18
And also, crucially,
82
246669
1526
重要的是，
04:20
these systems don't operate
under a single-answer logic.
under a single-answer logic.
83
248219
4380
这些系统不再局限单一答案。
04:24
They don't produce a simple answer;
it's more probabilistic:
it's more probabilistic:
84
252623
2959
他们得出的不是一个
简单的答案，而是概率性的：
简单的答案，而是概率性的：
04:27
"This one is probably more like
what you're looking for."
what you're looking for."
85
255606
3483
“这个更像是你在寻找的。”
04:32
Now, the upside is:
this method is really powerful.
this method is really powerful.
86
260023
3070
它的优势是：
它真的非常强大。
它真的非常强大。
04:35
The head of Google's AI systems called it,
87
263117
2076
Google 人工智能系统的
负责人称它为：
负责人称它为：
04:37
"the unreasonable effectiveness of data."
88
265217
2197
“不可思议的数据效率”。
04:39
The downside is,
89
267791
1353
缺点在于，
04:41
we don't really understand
what the system learned.
what the system learned.
90
269738
3071
我们无法清楚的了解
系统学到了什么，
系统学到了什么，
04:44
In fact, that's its power.
91
272833
1587
事实上，这也正是
它的强大之处。
它的强大之处。
04:46
This is less like giving
instructions to a computer;
instructions to a computer;
92
274946
3798
不像是给计算机下达指令，
04:51
it's more like training
a puppy-machine-creature
a puppy-machine-creature
93
279200
4064
更像是在训练一个机器狗，
04:55
we don't really understand or control.
94
283288
2371
我们无法精确的
了解和控制它。
了解和控制它。
04:58
So this is our problem.
95
286362
1551
这就是我们遇到的问题。
05:00
It's a problem when this artificial
intelligence system gets things wrong.
intelligence system gets things wrong.
96
288427
4262
人工智能会出错，
这是一个问题。
这是一个问题。
05:04
It's also a problem
when it gets things right,
when it gets things right,
97
292713
3540
但他们得出正确答案，
又是另一种问题。
又是另一种问题。
05:08
because we don't even know which is which
when it's a subjective problem.
when it's a subjective problem.
98
296277
3628
因为我们面对主观问题，
是不应该有答案的。
是不应该有答案的。
05:11
We don't know what this thing is thinking.
99
299929
2339
我们不知道
这些机器在想什么。
这些机器在想什么。
05:15
So, consider a hiring algorithm --
100
303493
3683
所以，考虑一下招聘算法－
05:20
a system used to hire people,
using machine-learning systems.
using machine-learning systems.
101
308123
4311
通过机器学习构建的招聘系统。
05:25
Such a system would have been trained
on previous employees' data
on previous employees' data
102
313052
3579
这样的系统会用员工
现有的数据进行自我培训，
现有的数据进行自我培训，
05:28
and instructed to find and hire
103
316655
2591
参照公司的优秀员工
05:31
people like the existing
high performers in the company.
high performers in the company.
104
319270
3038
来寻找和招聘新人。
05:34
Sounds good.
105
322814
1153
听起来很好。
05:35
I once attended a conference
106
323991
1999
有次我参加了一个会议，
05:38
that brought together
human resources managers and executives,
human resources managers and executives,
107
326014
3125
会上聚集了很多
人力资源部的经理和总监，
人力资源部的经理和总监，
05:41
high-level people,
108
329163
1206
都是高管，
05:42
using such systems in hiring.
109
330393
1559
让他们使用这样的招聘系统。
05:43
They were super excited.
110
331976
1646
他们都非常兴奋，
05:45
They thought that this would make hiring
more objective, less biased,
more objective, less biased,
111
333646
4653
认为这可以让招聘变得
更加客观，从而减少偏见，
更加客观，从而减少偏见，
05:50
and give women
and minorities a better shot
and minorities a better shot
112
338323
3000
给女性和少数族裔
更多的机会，
更多的机会，
05:53
against biased human managers.
113
341347
2188
减少他们自身的偏见。
05:55
And look -- human hiring is biased.
114
343559
2843
你知道的，
招聘是存在偏见的，
招聘是存在偏见的，
05:59
I know.
115
347099
1185
我也很清楚。
06:00
I mean, in one of my early jobs
as a programmer,
as a programmer,
116
348308
3005
在我刚开始做程序员的时候，
06:03
my immediate manager would sometimes
come down to where I was
come down to where I was
117
351337
3868
我的直接主管会来找我，
06:07
really early in the morning
or really late in the afternoon,
or really late in the afternoon,
118
355229
3753
在早晨很早或下午很晚的时候，
06:11
and she'd say, "Zeynep,
let's go to lunch!"
let's go to lunch!"
119
359006
3062
说，“ 图费, 我们去吃午饭！”
06:14
I'd be puzzled by the weird timing.
120
362724
2167
我就被这奇怪的时间
给搞糊涂了，
给搞糊涂了，
06:16
It's 4pm. Lunch?
121
364915
2129
现在是下午4点，吃午饭？
06:19
I was broke, so free lunch. I always went.
122
367068
3094
我当时很穷，所以
不会放过免费的午餐。
不会放过免费的午餐。
06:22
I later realized what was happening.
123
370618
2067
后来我才想明白原因，
06:24
My immediate managers
had not confessed to their higher-ups
had not confessed to their higher-ups
124
372709
4546
我的主管们没有
向他们的上级坦白，
向他们的上级坦白，
06:29
that the programmer they hired
for a serious job was a teen girl
for a serious job was a teen girl
125
377279
3113
他们雇了一个十多岁的小女孩
来做重要的编程工作，
来做重要的编程工作，
06:32
who wore jeans and sneakers to work.
126
380416
3930
一个穿着牛仔裤，
运动鞋工作的女孩。
运动鞋工作的女孩。
06:37
I was doing a good job,
I just looked wrong
I just looked wrong
127
385174
2202
我的工作做得很好，
我只是看起来不合适，
我只是看起来不合适，
06:39
and was the wrong age and gender.
128
387400
1699
年龄和性别也不合适。
06:41
So hiring in a gender- and race-blind way
129
389123
3346
所以，忽略性别和种族的招聘，
06:44
certainly sounds good to me.
130
392493
1865
听起来很适合我。
06:47
But with these systems,
it is more complicated, and here's why:
it is more complicated, and here's why:
131
395031
3341
但是这样的系统会带来更多问题，
06:50
Currently, computational systems
can infer all sorts of things about you
can infer all sorts of things about you
132
398968
5791
当前，计算机系统
能根据零散的数据，
能根据零散的数据，
06:56
from your digital crumbs,
133
404783
1872
推断出关于你的一切，
06:58
even if you have not
disclosed those things.
disclosed those things.
134
406679
2333
甚至你没有公开的事。
07:01
They can infer your sexual orientation,
135
409506
2927
它们可以推断你的性取向，
07:04
your personality traits,
136
412994
1306
你的性格特点，
07:06
your political leanings.
137
414859
1373
你的政治倾向。
07:08
They have predictive power
with high levels of accuracy.
with high levels of accuracy.
138
416830
3685
它们有高准确度的预测能力，
07:13
Remember -- for things
you haven't even disclosed.
you haven't even disclosed.
139
421362
2578
记住，是你没有公开的事情，
07:15
This is inference.
140
423964
1591
这就是推断。
07:17
I have a friend who developed
such computational systems
such computational systems
141
425579
3261
我有个朋友
就是开发这种系统，
就是开发这种系统，
07:20
to predict the likelihood
of clinical or postpartum depression
of clinical or postpartum depression
142
428864
3641
从社交媒体的数据中，
推断患临床或
产后抑郁症的可能性。
产后抑郁症的可能性。
07:24
from social media data.
143
432529
1416
07:26
The results are impressive.
144
434676
1427
结果令人印象深刻，
07:28
Her system can predict
the likelihood of depression
the likelihood of depression
145
436492
3357
她的系统可以
在症状出现前几个月
在症状出现前几个月
07:31
months before the onset of any symptoms --
146
439873
3903
成功预测到
患抑郁的可能性，
患抑郁的可能性，
07:35
months before.
147
443800
1373
提前几个月。
07:37
No symptoms, there's prediction.
148
445197
2246
在有症状之前，
就可以预测到，
就可以预测到，
07:39
She hopes it will be used
for early intervention. Great!
for early intervention. Great!
149
447467
4812
她希望这可以用于
临床早期干预，这很棒！
临床早期干预，这很棒！
07:44
But now put this in the context of hiring.
150
452911
2040
现在我们把这项技术
放到招聘中来看。
放到招聘中来看。
07:48
So at this human resources
managers conference,
managers conference,
151
456027
3046
在那次人力资源管理会议中，
07:51
I approached a high-level manager
in a very large company,
in a very large company,
152
459097
4709
我接近了一位大公司的高管，
07:55
and I said to her, "Look,
what if, unbeknownst to you,
what if, unbeknownst to you,
153
463830
4578
我对她说，“看，如果这个系统
在不通知你的情况下，
在不通知你的情况下，
08:00
your system is weeding out people
with high future likelihood of depression?
with high future likelihood of depression?
154
468432
6549
就剔除了未来
有可能抑郁的人，怎么办？
有可能抑郁的人，怎么办？
08:07
They're not depressed now,
just maybe in the future, more likely.
just maybe in the future, more likely.
155
475761
3376
他们现在不抑郁，
只是未来有可能。
只是未来有可能。
08:11
What if it's weeding out women
more likely to be pregnant
more likely to be pregnant
156
479923
3406
如果它剔除了
有可能怀孕的女性，怎么办？
有可能怀孕的女性，怎么办？
08:15
in the next year or two
but aren't pregnant now?
but aren't pregnant now?
157
483353
2586
她们现在没怀孕，
但未来一两年有可能。
但未来一两年有可能。
08:18
What if it's hiring aggressive people
because that's your workplace culture?"
because that's your workplace culture?"
158
486844
5636
如果因为你的公司文化，
它只雇佣激进的候选人怎么办？”
它只雇佣激进的候选人怎么办？”
08:25
You can't tell this by looking
at gender breakdowns.
at gender breakdowns.
159
493173
2691
只看性别比例，
你发现不了这些问题，
你发现不了这些问题，
08:27
Those may be balanced.
160
495888
1502
性别比例是可以被调整的。
08:29
And since this is machine learning,
not traditional coding,
not traditional coding,
161
497414
3557
并且因为这是机器学习，
不是传统的代码，
不是传统的代码，
08:32
there is no variable there
labeled "higher risk of depression,"
labeled "higher risk of depression,"
162
500995
4907
不会有一个变量来标识
“高抑郁风险”、
“高抑郁风险”、
08:37
"higher risk of pregnancy,"
163
505926
1833
“高怀孕风险”、
08:39
"aggressive guy scale."
164
507783
1734
“人员的激进程度”。
08:41
Not only do you not know
what your system is selecting on,
what your system is selecting on,
165
509995
3679
你不仅无法了解系统
在选什么样的人，
在选什么样的人，
08:45
you don't even know
where to begin to look.
where to begin to look.
166
513698
2323
你甚至不知道
从哪里入手了解。
从哪里入手了解。
08:48
It's a black box.
167
516045
1246
它是个暗箱。
08:49
It has predictive power,
but you don't understand it.
but you don't understand it.
168
517315
2807
它有预测的能力，
但你不了解它。
但你不了解它。
08:52
"What safeguards," I asked, "do you have
169
520486
2369
我问，“你有什么措施
可以保证，
可以保证，
08:54
to make sure that your black box
isn't doing something shady?"
isn't doing something shady?"
170
522879
3673
你的暗箱没有
在做些见不得人的事？”
在做些见不得人的事？”
09:00
She looked at me as if I had
just stepped on 10 puppy tails.
just stepped on 10 puppy tails.
171
528863
3878
她看着我，就好像
我刚踩了10只小狗的尾巴。
我刚踩了10只小狗的尾巴。
09:04
(Laughter)
172
532765
1248
（笑声）
09:06
She stared at me and she said,
173
534037
2041
她瞪着我说：
09:08
"I don't want to hear
another word about this."
another word about this."
174
536556
4333
“我不想再听你多说一个字。”
09:13
And she turned around and walked away.
175
541458
2034
然后她转身走开了。
09:16
Mind you -- she wasn't rude.
176
544064
1486
其实，
她不是无礼，
她不是无礼，
09:17
It was clearly: what I don't know
isn't my problem, go away, death stare.
isn't my problem, go away, death stare.
177
545574
6308
她想表达的其实是：我不知道，
这不是我的错，走开，不然我瞪死你。
这不是我的错，走开，不然我瞪死你。
09:23
(Laughter)
178
551906
1246
（笑声）
09:25
Look, such a system
may even be less biased
may even be less biased
179
553862
3839
看，这样的系统
可能在某些方面
可能在某些方面
09:29
than human managers in some ways.
180
557725
2103
比人类高管
怀有更少偏见，
怀有更少偏见，
09:31
And it could make monetary sense.
181
559852
2146
而且可以创造经济价值。
09:34
But it could also lead
182
562573
1650
但它也可能
09:36
to a steady but stealthy
shutting out of the job market
shutting out of the job market
183
564247
4748
用一种顽固且隐秘的方式，
把高抑郁风险的人清出职场。
09:41
of people with higher risk of depression.
184
569019
2293
09:43
Is this the kind of society
we want to build,
we want to build,
185
571753
2596
这是我们想要的未来吗？
09:46
without even knowing we've done this,
186
574373
2285
把决策权给予我们
并不完全了解的机器，
并不完全了解的机器，
09:48
because we turned decision-making
to machines we don't totally understand?
to machines we don't totally understand?
187
576682
3964
在我们不知情的状况下
构建一种新的社会？
构建一种新的社会？
09:53
Another problem is this:
188
581265
1458
另一个问题是，
09:55
these systems are often trained
on data generated by our actions,
on data generated by our actions,
189
583314
4452
这些系统通常使用
我们真实的
行为数据来训练。
行为数据来训练。
09:59
human imprints.
190
587790
1816
10:02
Well, they could just be
reflecting our biases,
reflecting our biases,
191
590188
3808
它们可能只是在
反馈我们的偏见，
反馈我们的偏见，
10:06
and these systems
could be picking up on our biases
could be picking up on our biases
192
594020
3593
这些系统会
继承我们的偏见，
继承我们的偏见，
10:09
and amplifying them
193
597637
1313
并把它们放大，
10:10
and showing them back to us,
194
598974
1418
然后反馈给我们。
10:12
while we're telling ourselves,
195
600416
1462
我们骗自己说，
10:13
"We're just doing objective,
neutral computation."
neutral computation."
196
601902
3117
“我们只做客观、
中立的预测。”
中立的预测。”
10:18
Researchers found that on Google,
197
606314
2677
研究者发现，在 Google 上，
10:22
women are less likely than men
to be shown job ads for high-paying jobs.
to be shown job ads for high-paying jobs.
198
610134
5313
高收入工作的广告
更多的被展示给男性用户。
更多的被展示给男性用户。
10:28
And searching for African-American names
199
616463
2530
搜索非裔美国人的名字，
10:31
is more likely to bring up ads
suggesting criminal history,
suggesting criminal history,
200
619017
4706
更可能出现
关于犯罪史的广告，
关于犯罪史的广告，
10:35
even when there is none.
201
623747
1567
即使某些根本不存在。
10:38
Such hidden biases
and black-box algorithms
and black-box algorithms
202
626693
3549
这些潜在的偏见
以及暗箱中的算法，
以及暗箱中的算法，
10:42
that researchers uncover sometimes
but sometimes we don't know,
but sometimes we don't know,
203
630266
3973
有些会被研究者揭露，
有些根本不会被发现，
有些根本不会被发现，
10:46
can have life-altering consequences.
204
634263
2661
它的后果可能是
改变一个人的人生。
改变一个人的人生。
10:49
In Wisconsin, a defendant
was sentenced to six years in prison
was sentenced to six years in prison
205
637958
4159
在威斯康星，一个被告
因逃避警察被判刑六年。
10:54
for evading the police.
206
642141
1355
10:56
You may not know this,
207
644824
1186
你可能不知道，
10:58
but algorithms are increasingly used
in parole and sentencing decisions.
in parole and sentencing decisions.
208
646034
3998
但计算机算法正越来越多的
被应用在假释及量刑裁定上。
被应用在假释及量刑裁定上。
11:02
He wanted to know:
How is this score calculated?
How is this score calculated?
209
650056
2955
他想要弄清楚，这个
得分是怎么算出来的？
得分是怎么算出来的？
11:05
It's a commercial black box.
210
653795
1665
这是个商业暗箱，
11:07
The company refused to have its algorithm
be challenged in open court.
be challenged in open court.
211
655484
4205
这家公司拒绝在公开法庭上
讨论他们的算法。
讨论他们的算法。
11:12
But ProPublica, an investigative
nonprofit, audited that very algorithm
nonprofit, audited that very algorithm
212
660396
5532
但是一家叫 ProPublica
的非盈利机构，
的非盈利机构，
根据公开数据，
对这个算法进行了评估，
对这个算法进行了评估，
11:17
with what public data they could find,
213
665952
2016
11:19
and found that its outcomes were biased
214
667992
2316
他们发现这个算法
的结论是有偏见的，
的结论是有偏见的，
11:22
and its predictive power
was dismal, barely better than chance,
was dismal, barely better than chance,
215
670332
3629
它的预测能力很差，
比碰运气强不了多少，
比碰运气强不了多少，
11:25
and it was wrongly labeling
black defendants as future criminals
black defendants as future criminals
216
673985
4416
并且它错误的把黑人被告
未来犯罪的可能性
未来犯罪的可能性
11:30
at twice the rate of white defendants.
217
678425
3895
标记为白人的两倍。
11:35
So, consider this case:
218
683891
1564
看下这个案例：
11:38
This woman was late
picking up her godsister
picking up her godsister
219
686103
3852
这个女人急着去佛罗里达州,
布劳沃德县的一所学校，
布劳沃德县的一所学校，
11:41
from a school in Broward County, Florida,
220
689979
2075
去接她的干妹妹。
11:44
running down the street
with a friend of hers.
with a friend of hers.
221
692757
2356
女人和她的朋友在街上狂奔，
11:47
They spotted an unlocked kid's bike
and a scooter on a porch
and a scooter on a porch
222
695137
4099
她们看到门廊上一辆没上锁的
儿童自行车，和一辆电瓶车，
儿童自行车，和一辆电瓶车，
11:51
and foolishly jumped on it.
223
699260
1632
于是就愚蠢的骑上了车。
11:52
As they were speeding off,
a woman came out and said,
a woman came out and said,
224
700916
2599
正在她们要骑走的时候，
另一个女人出来，喊道：
另一个女人出来，喊道：
11:55
"Hey! That's my kid's bike!"
225
703539
2205
“嘿！那是我孩子的自行车！”
11:57
They dropped it, they walked away,
but they were arrested.
but they were arrested.
226
705768
3294
她们扔掉车走开，
但还是被抓住了。
但还是被抓住了。
12:01
She was wrong, she was foolish,
but she was also just 18.
but she was also just 18.
227
709086
3637
她做错了，她很愚蠢，
但她也才刚满18岁，
但她也才刚满18岁，
12:04
She had a couple of juvenile misdemeanors.
228
712747
2544
她之前有不少
青少年轻罪的记录。
青少年轻罪的记录。
12:07
Meanwhile, that man had been arrested
for shoplifting in Home Depot --
for shoplifting in Home Depot --
229
715808
5185
与此同时，这个男人
在连锁超市偷窃被捕了，
在连锁超市偷窃被捕了，
12:13
85 dollars' worth of stuff,
a similar petty crime.
a similar petty crime.
230
721017
2924
偷了价值85美金的东西，
同样的轻微犯罪，
同样的轻微犯罪，
12:16
But he had two prior
armed robbery convictions.
armed robbery convictions.
231
724766
4559
但他有两次持枪抢劫的案底。
12:21
But the algorithm scored her
as high risk, and not him.
as high risk, and not him.
232
729955
3482
这个程序将这位女性判定为
高风险，而这位男性则不是。
高风险，而这位男性则不是。
12:26
Two years later, ProPublica found
that she had not reoffended.
that she had not reoffended.
233
734746
3874
两年后，ProPublica
发现她没有再次犯罪，
发现她没有再次犯罪，
12:30
It was just hard to get a job
for her with her record.
for her with her record.
234
738644
2550
但这个记录
使她很难找到工作。
使她很难找到工作。
12:33
He, on the other hand, did reoffend
235
741218
2076
而这位男性，却再次犯罪，
12:35
and is now serving an eight-year
prison term for a later crime.
prison term for a later crime.
236
743318
3836
并因此被判八年监禁。
12:40
Clearly, we need to audit our black boxes
237
748088
3369
显然，我们需要
审查这些暗箱，
审查这些暗箱，
12:43
and not have them have
this kind of unchecked power.
this kind of unchecked power.
238
751481
2615
确保它们不再有这样
不加限制的权限。
不加限制的权限。
12:46
(Applause)
239
754120
2879
（掌声）
12:50
Audits are great and important,
but they don't solve all our problems.
but they don't solve all our problems.
240
758087
4242
审查是很重要的，
但不能解决所有的问题。
但不能解决所有的问题。
12:54
Take Facebook's powerful
news feed algorithm --
news feed algorithm --
241
762353
2748
拿 Facebook 的强大的
新闻流算法来说，
新闻流算法来说，
12:57
you know, the one that ranks everything
and decides what to show you
and decides what to show you
242
765125
4843
就是通过你的朋友圈
和你浏览过的页面，
和你浏览过的页面，
决定你的
“推荐内容”的算法。
“推荐内容”的算法。
13:01
from all the friends and pages you follow.
243
769992
2284
13:04
Should you be shown another baby picture?
244
772898
2275
它会决定要不要
再推一张婴儿照片给你，
再推一张婴儿照片给你，
13:07
(Laughter)
245
775197
1196
（笑声）
13:08
A sullen note from an acquaintance?
246
776417
2596
要不要推一条熟人
的沮丧状态？
的沮丧状态？
13:11
An important but difficult news item?
247
779449
1856
要不要推一条重要
但艰涩的新闻？
但艰涩的新闻？
13:13
There's no right answer.
248
781329
1482
这个问题没有正解。
13:14
Facebook optimizes
for engagement on the site:
for engagement on the site:
249
782835
2659
Facebook 会根据
网站的参与度来优化：
网站的参与度来优化：
13:17
likes, shares, comments.
250
785518
1415
喜欢、分享、评论。
13:20
In August of 2014,
251
788168
2696
在2014年8月，
13:22
protests broke out in Ferguson, Missouri,
252
790888
2662
密苏里州弗格森市爆发了游行，
13:25
after the killing of an African-American
teenager by a white police officer,
teenager by a white police officer,
253
793574
4417
一个白人警察在不明状况下
杀害了一位非裔少年。
13:30
under murky circumstances.
254
798015
1570
13:31
The news of the protests was all over
255
799974
2007
关于游行的新闻
13:34
my algorithmically
unfiltered Twitter feed,
unfiltered Twitter feed,
256
802005
2685
在我的未经算法过滤的
Twitter 上大量出现，
Twitter 上大量出现，
13:36
but nowhere on my Facebook.
257
804714
1950
但 Facebook 上却没有。
13:39
Was it my Facebook friends?
258
807182
1734
是因为我的 Facebook 好友
不关注这事吗？
不关注这事吗？
13:40
I disabled Facebook's algorithm,
259
808940
2032
我禁用了 Facebook 的算法，
13:43
which is hard because Facebook
keeps wanting to make you
keeps wanting to make you
260
811472
2848
这是很麻烦的一键事，
因为 Facebook 希望
因为 Facebook 希望
13:46
come under the algorithm's control,
261
814344
2036
你一直在它的算法
控制下使用,
控制下使用,
13:48
and saw that my friends
were talking about it.
were talking about it.
262
816404
2238
希望我的朋友持续
地谈论这件事。
地谈论这件事。
13:50
It's just that the algorithm
wasn't showing it to me.
wasn't showing it to me.
263
818666
2509
只是算法没法
给我这些信息。
给我这些信息。
13:53
I researched this and found
this was a widespread problem.
this was a widespread problem.
264
821199
3042
我研究了这个现象，
发现这是个普遍的问题。
发现这是个普遍的问题。
13:56
The story of Ferguson
wasn't algorithm-friendly.
wasn't algorithm-friendly.
265
824265
3813
弗格森事件
对算法是不适用的，
对算法是不适用的，
它不是值得“赞”的新闻，
14:00
It's not "likable."
266
828102
1171
14:01
Who's going to click on "like?"
267
829297
1552
谁会在这样
的文章下点“赞”呢？
的文章下点“赞”呢？
14:03
It's not even easy to comment on.
268
831500
2206
甚至这新闻都不好被评论。
14:05
Without likes and comments,
269
833730
1371
因为没有“赞”和评论，
14:07
the algorithm was likely showing it
to even fewer people,
to even fewer people,
270
835125
3292
算法会减少
这些新闻的曝光，
这些新闻的曝光，
14:10
so we didn't get to see this.
271
838441
1542
所以我们无法看到。
14:12
Instead, that week,
272
840946
1228
相反的，在同一周，
14:14
Facebook's algorithm highlighted this,
273
842198
2298
Facebook 的算法热推了
14:16
which is the ALS Ice Bucket Challenge.
274
844520
2226
ALS 冰桶挑战的信息。
14:18
Worthy cause; dump ice water,
donate to charity, fine.
donate to charity, fine.
275
846770
3742
这很有意义，倒冰水，
为慈善捐款，很好。
为慈善捐款，很好。
14:22
But it was super algorithm-friendly.
276
850536
1904
这个事件对算法是很适用的，
14:25
The machine made this decision for us.
277
853219
2613
机器帮我们做了这个决定。
14:27
A very important
but difficult conversation
but difficult conversation
278
855856
3497
非常重要但艰涩的新闻事件
14:31
might have been smothered,
279
859377
1555
可能会被埋没掉，
14:32
had Facebook been the only channel.
280
860956
2696
因为 Facebook 已经成为
主要的信息来源。
主要的信息来源。
14:36
Now, finally, these systems
can also be wrong
can also be wrong
281
864117
3797
最后，这些系统
也可能会在一些
也可能会在一些
14:39
in ways that don't resemble human systems.
282
867938
2736
不同于人力系统
的那些事情上搞错。
的那些事情上搞错。
14:42
Do you guys remember Watson,
IBM's machine-intelligence system
IBM's machine-intelligence system
283
870698
2922
你们记得 Watson 吧，
那个在智力竞赛《危险边缘》中
那个在智力竞赛《危险边缘》中
14:45
that wiped the floor
with human contestants on Jeopardy?
with human contestants on Jeopardy?
284
873644
3128
横扫人类选手的
IBM 机器智能系统，
IBM 机器智能系统，
14:49
It was a great player.
285
877131
1428
它是个很厉害的选手。
14:50
But then, for Final Jeopardy,
Watson was asked this question:
Watson was asked this question:
286
878583
3569
但是，在最后一轮比赛中，
Watson 被问道：
Watson 被问道：
14:54
"Its largest airport is named
for a World War II hero,
for a World War II hero,
287
882659
2932
“它最大的机场是以
二战英雄命名的，
二战英雄命名的，
14:57
its second-largest
for a World War II battle."
for a World War II battle."
288
885615
2252
它第二大机场是以
二战战场命名的。”
二战战场命名的。”
14:59
(Hums Final Jeopardy music)
289
887891
1378
（哼唱《危险边缘》插曲）
15:01
Chicago.
290
889582
1182
芝加哥。
15:02
The two humans got it right.
291
890788
1370
两位人类选手答对了，
15:04
Watson, on the other hand,
answered "Toronto" --
answered "Toronto" --
292
892697
4348
但 Watson 答的是，
“多伦多”，
“多伦多”，
15:09
for a US city category!
293
897069
1818
这是个猜美国城市的环节！
15:11
The impressive system also made an error
294
899596
2901
这个厉害的系统也会犯
15:14
that a human would never make,
a second-grader wouldn't make.
a second-grader wouldn't make.
295
902521
3651
人类都不会犯的，二年级
小孩都不会犯的错误。
小孩都不会犯的错误。
15:18
Our machine intelligence can fail
296
906823
3109
我们的机器智能系统,
15:21
in ways that don't fit
error patterns of humans,
error patterns of humans,
297
909956
3100
会在一些不符合人类
出错模式的问题上出错，
出错模式的问题上出错，
15:25
in ways we won't expect
and be prepared for.
and be prepared for.
298
913080
2950
这些问题都是我们
无法预料和准备的。
无法预料和准备的。
15:28
It'd be lousy not to get a job
one is qualified for,
one is qualified for,
299
916054
3638
丢失一份完全有能力胜任
的工作时，人们会感到很糟，
的工作时，人们会感到很糟，
15:31
but it would triple suck
if it was because of stack overflow
if it was because of stack overflow
300
919716
3727
但如果是因为机器
子程序的过度堆积，
子程序的过度堆积，
15:35
in some subroutine.
301
923467
1432
就简直糟透了。
15:36
(Laughter)
302
924923
1579
（笑声）
15:38
In May of 2010,
303
926526
2786
在2010年五月，
15:41
a flash crash on Wall Street
fueled by a feedback loop
fueled by a feedback loop
304
929336
4044
华尔街出现一次
股票闪电崩盘，
股票闪电崩盘，
原因是“卖出”算法
的反馈回路导致，
的反馈回路导致，
15:45
in Wall Street's "sell" algorithm
305
933404
3028
15:48
wiped a trillion dollars
of value in 36 minutes.
of value in 36 minutes.
306
936456
4184
在36分钟内
损失了几十亿美金。
损失了几十亿美金。
15:53
I don't even want to think
what "error" means
what "error" means
307
941722
2187
我甚至不敢想，
致命的自动化武器
致命的自动化武器
15:55
in the context of lethal
autonomous weapons.
autonomous weapons.
308
943933
3589
发生“错误”会是什么后果。
16:01
So yes, humans have always made biases.
309
949894
3790
是的，人类总是会有偏见，
16:05
Decision makers and gatekeepers,
310
953708
2176
法庭上、新闻机构、战争中的，
16:07
in courts, in news, in war ...
311
955908
3493
决策者、看门人…
16:11
they make mistakes;
but that's exactly my point.
but that's exactly my point.
312
959425
3038
他们都会犯错，
但这恰恰是我要说的。
但这恰恰是我要说的。
16:14
We cannot escape
these difficult questions.
these difficult questions.
313
962487
3521
我们无法抛开
这些困难的问题，
这些困难的问题，
我们不能把我们自身
该承担的责任推给机器。
该承担的责任推给机器。
16:18
We cannot outsource
our responsibilities to machines.
our responsibilities to machines.
314
966596
3516
16:22
(Applause)
315
970676
4208
（掌声）
16:29
Artificial intelligence does not give us
a "Get out of ethics free" card.
a "Get out of ethics free" card.
316
977089
4447
人工智能不会给我们
一张“伦理免责卡”。
一张“伦理免责卡”。
16:34
Data scientist Fred Benenson
calls this math-washing.
calls this math-washing.
317
982742
3381
数据科学家 Fred Benenson
称之为“数学粉饰”。
称之为“数学粉饰”。
16:38
We need the opposite.
318
986147
1389
我们需要是相反的东西。
16:39
We need to cultivate algorithm suspicion,
scrutiny and investigation.
scrutiny and investigation.
319
987560
5388
我们需要培养算法的
怀疑、复查和调研能力。
怀疑、复查和调研能力。
16:45
We need to make sure we have
algorithmic accountability,
algorithmic accountability,
320
993380
3198
我们需要确保
有人为算法负责，
有人为算法负责，
16:48
auditing and meaningful transparency.
321
996602
2445
为算法审查，
并切实的公开透明。
并切实的公开透明。
16:51
We need to accept
that bringing math and computation
that bringing math and computation
322
999380
3234
我们必须认识到，
把数学和计算引入
把数学和计算引入
16:54
to messy, value-laden human affairs
323
1002638
2970
解决复杂的、高价值
的人类事务中，
的人类事务中，
16:57
does not bring objectivity;
324
1005632
2384
并不能带来客观性，
17:00
rather, the complexity of human affairs
invades the algorithms.
invades the algorithms.
325
1008040
3633
相反，人类事务
的复杂性会扰乱算法。
的复杂性会扰乱算法。
17:04
Yes, we can and we should use computation
326
1012148
3487
是的，我们可以
并且需要使用计算机
并且需要使用计算机
17:07
to help us make better decisions.
327
1015659
2014
来帮助我们做更好的决策，
17:09
But we have to own up
to our moral responsibility to judgment,
to our moral responsibility to judgment,
328
1017697
5332
但我们也需要在判断中
加入道德义务，
加入道德义务，
17:15
and use algorithms within that framework,
329
1023053
2818
在这个框架下使用算法，
17:17
not as a means to abdicate
and outsource our responsibilities
and outsource our responsibilities
330
1025895
4935
而不是像人与人
之间相互推卸那样，
之间相互推卸那样，
就把责任转移给机器。
17:22
to one another as human to human.
331
1030854
2454
17:25
Machine intelligence is here.
332
1033807
2609
人工智能到来了，
17:28
That means we must hold on ever tighter
333
1036440
3421
这意味着
我们要格外坚守
我们要格外坚守
人类的价值观和伦理。
17:31
to human values and human ethics.
334
1039885
2147
17:34
Thank you.
335
1042056
1154
谢谢。
17:35
(Applause)
336
1043234
5020
（掌声）
ABOUT THE SPEAKER
Zeynep Tufekci - Techno-sociologistTechno-sociologist Zeynep Tufekci asks big questions about our societies and our lives, as both algorithms and digital connectivity spread.
Why you should listen
We've entered an era of digital connectivity and machine intelligence. Complex algorithms are increasingly used to make consequential decisions about us. Many of these decisions are subjective and have no right answer: who should be hired, fired or promoted; what news should be shown to whom; which of your friends do you see updates from; which convict should be paroled. With increasing use of machine learning in these systems, we often don't even understand how exactly they are making these decisions. Zeynep Tufekci studies what this historic transition means for culture, markets, politics and personal life.
Tufekci is a contributing opinion writer at the New York Times, an associate professor at the School of Information and Library Science at University of North Carolina, Chapel Hill, and a faculty associate at Harvard's Berkman Klein Center for Internet and Society.
Her book, Twitter and Tear Gas: The Power and Fragility of Networked Protest, was published in 2017 by Yale University Press. Her next book, from Penguin Random House, will be about algorithms that watch, judge and nudge us.
Zeynep Tufekci | Speaker | TED.com