TED@Merck KGaA, Darmstadt, Germany
Daniel Susskind: 3 myths about the future of work (and why they're not true)
丹尼爾薩斯金: 關於未來工作的三項迷思（以及為什麼它們不是真的）
Filmed:
Readability: 4.5
1,519,249 views
「機器會取代人類嗎？」每個有可能會失去工作的人心中都有這個問題。丹尼爾薩斯金正視這個問題以及我們對自動化未來的三項迷思，他建議我們改問另一個問題：當世界上的工作變少或甚至消失時，我們要如何分配財富？
Daniel Susskind - Economist
Daniel Susskind explores the impact of technology, particularly artificial intelligence, on work and society. Full bio
Daniel Susskind explores the impact of technology, particularly artificial intelligence, on work and society. Full bio
Double-click the English transcript below to play the video.
00:12
Automation anxiety
has been spreading lately,
has been spreading lately,
0
760
3376
近期，自動化焦慮一直在散佈，
00:16
a fear that in the future,
1
4160
2656
它是種恐懼，害怕在未來
00:18
many jobs will be performed by machines
2
6840
2456
許多工作會由機器來進行，
00:21
rather than human beings,
3
9320
1336
而不是人類，
00:22
given the remarkable advances
that are unfolding
that are unfolding
4
10680
2936
因為現在已可以看到在人工智慧
00:25
in artificial intelligence and robotics.
5
13640
2776
和機器人學領域的驚人進步。
00:28
What's clear is that
there will be significant change.
there will be significant change.
6
16440
2816
很清楚的一點是，
將來會有顯著的改變。
將來會有顯著的改變。
00:31
What's less clear
is what that change will look like.
is what that change will look like.
7
19280
3616
比較不那麼清楚的是，
改變會是什麼樣的。
改變會是什麼樣的。
00:34
My research suggests that the future
is both troubling and exciting.
is both troubling and exciting.
8
22920
4936
我的研究指出，未來
既讓人困擾又讓人興奮。
既讓人困擾又讓人興奮。
00:39
The threat of technological
unemployment is real,
unemployment is real,
9
27880
3736
科技造成失業的威脅是真的，
00:43
and yet it's a good problem to have.
10
31640
2056
但，能有這種問題也是件好事。
00:45
And to explain
how I came to that conclusion,
how I came to that conclusion,
11
33720
3216
為了解釋我如何得到這個結論，
00:48
I want to confront three myths
12
36960
2536
我想要來正視三項迷思，
00:51
that I think are currently obscuring
our vision of this automated future.
our vision of this automated future.
13
39520
4280
我認為這些迷思
目前遮掩了我們的視線，
目前遮掩了我們的視線，
讓我們看不清自動化的未來。
00:56
A picture that we see
on our television screens,
on our television screens,
14
44880
2336
我們在電視上、書中、電影中、
00:59
in books, in films, in everyday commentary
15
47240
2216
每天的評論中所看到的描繪，
01:01
is one where an army of robots
descends on the workplace
descends on the workplace
16
49480
3696
通常是機器人大軍湧入工作場所，
01:05
with one goal in mind:
17
53200
1376
心中只有一個目標：
01:06
to displace human beings from their work.
18
54600
2496
在工作上取代人類。
01:09
And I call this the Terminator myth.
19
57120
2696
我稱這個想法為「終結者迷思」。
01:11
Yes, machines displace
human beings from particular tasks,
human beings from particular tasks,
20
59840
3976
是的，在特定的工作任務上，
機器會取代人類，
機器會取代人類，
01:15
but they don't just
substitute for human beings.
substitute for human beings.
21
63840
2256
但它們不會就這樣代替人類。
01:18
They also complement them in other tasks,
22
66120
1976
它們在其他工作任務上會補足人類，
01:20
making that work more valuable
and more important.
and more important.
23
68120
3616
讓工作更有價值、更重要。
01:23
Sometimes they complement
human beings directly,
human beings directly,
24
71760
3336
有時，它們會直接補足人類，
01:27
making them more productive
or more efficient at a particular task.
or more efficient at a particular task.
25
75120
4016
讓人類在特定的工作任務上
能更有生產力或更有效率。
能更有生產力或更有效率。
01:31
So a taxi driver can use a satnav system
to navigate on unfamiliar roads.
to navigate on unfamiliar roads.
26
79160
4616
計程車司機在不熟悉的路上
可以用衛星導航系統來協助導航。
可以用衛星導航系統來協助導航。
01:35
An architect can use
computer-assisted design software
computer-assisted design software
27
83800
3336
建築師可以用電腦輔助的設計軟體
01:39
to design bigger,
more complicated buildings.
more complicated buildings.
28
87160
3096
來設計更大、更複雜的建築物。
01:42
But technological progress doesn't
just complement human beings directly.
just complement human beings directly.
29
90280
3696
但科技進步並不只會直接補足人類。
01:46
It also complements them indirectly,
and it does this in two ways.
and it does this in two ways.
30
94000
3336
它也會用間接方式補足人類，
間接的方式有兩種。
間接的方式有兩種。
01:49
The first is if we think
of the economy as a pie,
of the economy as a pie,
31
97360
3336
第一，如果我們把
經濟想成是一塊派，
經濟想成是一塊派，
01:52
technological progress
makes the pie bigger.
makes the pie bigger.
32
100720
2896
科技進步會讓派變更大。
01:55
As productivity increases,
incomes rise and demand grows.
incomes rise and demand grows.
33
103640
3856
隨著生產力增加，
收入會增加，需求會成長。
收入會增加，需求會成長。
01:59
The British pie, for instance,
34
107520
1776
比如，英國的派
02:01
is more than a hundred times
the size it was 300 years ago.
the size it was 300 years ago.
35
109320
3960
與三百年前相比，現在超過百倍大。
02:05
And so people displaced
from tasks in the old pie
from tasks in the old pie
36
113920
3216
在舊派工作被取代的人，
02:09
could find tasks to do
in the new pie instead.
in the new pie instead.
37
117160
2720
能在新派中找到工作。
02:12
But technological progress
doesn't just make the pie bigger.
doesn't just make the pie bigger.
38
120800
3936
但科技進步並不只會讓派變大。
02:16
It also changes
the ingredients in the pie.
the ingredients in the pie.
39
124760
2856
它也會改變派的成分原料。
02:19
As time passes, people spend
their income in different ways,
their income in different ways,
40
127640
3456
隨時間演進，人會以
不同的方式花費他們的收入，
不同的方式花費他們的收入，
02:23
changing how they spread it
across existing goods,
across existing goods,
41
131120
2816
改變既有商品花費上的分配，
02:25
and developing tastes
for entirely new goods, too.
for entirely new goods, too.
42
133960
3216
並也會發展出對於全新商品的品味。
02:29
New industries are created,
43
137200
1776
新的產業會被創造出來，
02:31
new tasks have to be done
44
139000
1816
有新的工作任務需要被完成，
02:32
and that means often
new roles have to be filled.
new roles have to be filled.
45
140840
2536
那就意味著有新角色要有人扮演。
02:35
So again, the British pie:
46
143400
1496
所以，再回到英國的派：
02:36
300 years ago,
most people worked on farms,
most people worked on farms,
47
144920
2976
三百年前，大部分的人在農場工作，
02:39
150 years ago, in factories,
48
147920
2336
一百五十年前，在工廠工作，
02:42
and today, most people work in offices.
49
150280
2856
現今，大部分的人在辦公室工作。
02:45
And once again, people displaced
from tasks in the old bit of pie
from tasks in the old bit of pie
50
153160
4056
再提一次，在老派工作被取代的人，
02:49
could tumble into tasks
in the new bit of pie instead.
in the new bit of pie instead.
51
157240
2800
可能會在新派當中
發現可以做的工作任務。
發現可以做的工作任務。
02:52
Economists call these effects
complementarities,
complementarities,
52
160720
3336
經濟學家把這些效應稱為互補性，
02:56
but really that's just a fancy word
to capture the different way
to capture the different way
53
164080
3256
但那只是個很炫的詞，其實意思就是
02:59
that technological progress
helps human beings.
helps human beings.
54
167360
3136
科技進步用不同的方式在協助人類。
03:02
Resolving this Terminator myth
55
170520
2096
解開這個終結者迷思之後，
03:04
shows us that there are
two forces at play:
two forces at play:
56
172640
2336
會發現有兩股力量在運作：
03:07
one, machine substitution
that harms workers,
that harms workers,
57
175000
3536
第一：機器代替，這會傷害到工人，
03:10
but also these complementarities
that do the opposite.
that do the opposite.
58
178560
2880
但也會有第二股力量，
互補性，反而會幫助工人。
互補性，反而會幫助工人。
03:13
Now the second myth,
59
181960
1376
再來，第二項迷思，
03:15
what I call the intelligence myth.
60
183360
2280
我稱之為「智慧迷思」。
03:18
What do the tasks of driving a car,
making a medical diagnosis
making a medical diagnosis
61
186440
4896
以下這些工作任務：
駕駛一台車、做出醫療診斷，
駕駛一台車、做出醫療診斷，
03:23
and identifying a bird
at a fleeting glimpse have in common?
at a fleeting glimpse have in common?
62
191360
2920
及快速一瞥就辨識出
一隻鳥，有何共通性？
一隻鳥，有何共通性？
03:27
Well, these are all tasks
that until very recently,
that until very recently,
63
195280
2976
這些工作任務都是直到最近
03:30
leading economists thought
couldn't readily be automated.
couldn't readily be automated.
64
198280
3336
仍被經濟學家認為不能
自動化的工作任務。
自動化的工作任務。
03:33
And yet today, all of these tasks
can be automated.
can be automated.
65
201640
3176
然而，現今，所有這些
工作任務都能被自動化。
工作任務都能被自動化。
03:36
You know, all major car manufacturers
have driverless car programs.
have driverless car programs.
66
204840
3496
所有大型汽車製造商都有
無人駕駛汽車的計畫。
無人駕駛汽車的計畫。
03:40
There's countless systems out there
that can diagnose medical problems.
that can diagnose medical problems.
67
208360
3976
外面有數不清的系統
都能夠診斷醫療問題。
都能夠診斷醫療問題。
03:44
And there's even an app
that can identify a bird
that can identify a bird
68
212360
2416
甚至有個應用程式能用來辨識鳥類，
03:46
at a fleeting glimpse.
69
214800
1200
只要快速一瞥。
03:48
Now, this wasn't simply a case of bad luck
on the part of economists.
on the part of economists.
70
216920
4376
這並不是經濟學家運氣不好的情況。
03:53
They were wrong,
71
221320
1296
他們錯了，
03:54
and the reason why
they were wrong is very important.
they were wrong is very important.
72
222640
2496
而他們為什麼會錯的原因很重要。
03:57
They've fallen for the intelligence myth,
73
225160
2256
他們陷入了智慧迷思中，
03:59
the belief that machines
have to copy the way
have to copy the way
74
227440
2896
相信機器必須要複製人類
04:02
that human beings think and reason
75
230360
2056
思考和推理的方式，
04:04
in order to outperform them.
76
232440
1776
才能夠表現得比人類好。
04:06
When these economists
were trying to figure out
were trying to figure out
77
234240
2216
當這些經濟學家在試圖想出
04:08
what tasks machines could not do,
78
236480
1856
機器無法勝任哪些工作任務，
04:10
they imagined the only way
to automate a task
to automate a task
79
238360
2136
他們想像，將工作任務自動化的
04:12
was to sit down with a human being,
80
240520
1816
唯一方式就是和人類坐下來，
04:14
get them to explain to you
how it was they performed a task,
how it was they performed a task,
81
242360
3536
讓人類解釋他們如何執行工作任務，
04:17
and then try and capture that explanation
82
245920
2656
再試著分析他們的解釋，
04:20
in a set of instructions
for a machine to follow.
for a machine to follow.
83
248600
2776
轉換成一組指令，讓機器照著做。
04:23
This view was popular in artificial
intelligence at one point, too.
intelligence at one point, too.
84
251400
4176
在人工智慧領域，這種觀點
曾在某個時點很流行過。
曾在某個時點很流行過。
04:27
I know this because Richard Susskind,
85
255600
2176
我知道這點，因為理查薩斯金，
04:29
who is my dad and my coauthor,
86
257800
2856
他是我爸爸也是我的共同作者，
04:32
wrote his doctorate in the 1980s
on artificial intelligence and the law
on artificial intelligence and the law
87
260680
4056
在八〇年代，在牛津大學
寫了一篇關於人工智慧
寫了一篇關於人工智慧
04:36
at Oxford University,
88
264760
1416
與法律的博士論文，
04:38
and he was part of the vanguard.
89
266200
1576
他是先鋒部隊之一。
04:39
And with a professor called Phillip Capper
90
267800
2256
和一位名叫菲利普卡波的教授，
04:42
and a legal publisher called Butterworths,
91
270080
2096
以及一間法律出版社
叫做 Butterworths，
叫做 Butterworths，
04:44
they produced the world's first
commercially available
commercially available
92
272200
5896
他們合作製作出了
世界上第一個商業用的
世界上第一個商業用的
04:50
artificial intelligence system in the law.
93
278120
2776
法律人工智慧系統。
04:52
This was the home screen design.
94
280920
2616
這是首頁的畫面設計。
04:55
He assures me this was
a cool screen design at the time.
a cool screen design at the time.
95
283560
2696
他向我保證，在當時
這是很酷的畫面設計。
這是很酷的畫面設計。
04:58
(Laughter)
96
286280
1016
（笑聲）
04:59
I've never been entirely convinced.
97
287320
1696
我從來沒有被說服。
05:01
He published it
in the form of two floppy disks,
in the form of two floppy disks,
98
289040
2616
他用兩張軟碟片的形式將之出版，
05:03
at a time where floppy disks
genuinely were floppy,
genuinely were floppy,
99
291680
3536
在那個時代，軟碟片真的是軟的，
05:07
and his approach was the same
as the economists':
as the economists':
100
295240
2336
而他的方式就和經濟學家一樣：
和一名律師坐下來，
05:09
sit down with a lawyer,
101
297600
1256
讓她向你解釋如何解決法律問題，
05:10
get her to explain to you
how it was she solved a legal problem,
how it was she solved a legal problem,
102
298880
3176
05:14
and then try and capture that explanation
in a set of rules for a machine to follow.
in a set of rules for a machine to follow.
103
302080
5376
接著就試著把她的解釋
轉成一組指令給機器執行。
轉成一組指令給機器執行。
05:19
In economics, if human beings
could explain themselves in this way,
could explain themselves in this way,
104
307480
3616
在經濟上，如果人類能夠用
這種方式解釋自己做的事，
這種方式解釋自己做的事，
05:23
the tasks are called routine,
and they could be automated.
and they could be automated.
105
311120
3296
這種工作任務就叫做例行事務，
是可以被自動化的。
是可以被自動化的。
05:26
But if human beings
can't explain themselves,
can't explain themselves,
106
314440
2336
但如果人類無法解釋出怎麼做，
05:28
the tasks are called non-routine,
and they're thought to be out reach.
and they're thought to be out reach.
107
316800
4256
這種工作任務叫做非例行事務，
應該是不能自動化的。
應該是不能自動化的。
05:33
Today, that routine-nonroutine
distinction is widespread.
distinction is widespread.
108
321080
3296
現今，將事務區別為例行
與非例行是處處可見的。
與非例行是處處可見的。
05:36
Think how often you hear people say to you
109
324400
2056
想想看，你有多常聽到別人對你說
05:38
machines can only perform tasks
that are predictable or repetitive,
that are predictable or repetitive,
110
326480
3256
機器能進行的工作任務
只有可預測的、重覆性的、
只有可預測的、重覆性的、
05:41
rules-based or well-defined.
111
329760
1896
以規則為基礎的，或定義清楚的。
05:43
Those are all just
different words for routine.
different words for routine.
112
331680
2936
那些詞只是例行事務的不同說法。
05:46
And go back to those three cases
that I mentioned at the start.
that I mentioned at the start.
113
334640
3976
回到我一開始提到的三個案例。
05:50
Those are all classic cases
of nonroutine tasks.
of nonroutine tasks.
114
338640
2896
那些案例是典型的非例行事務。
05:53
Ask a doctor, for instance,
how she makes a medical diagnosis,
how she makes a medical diagnosis,
115
341560
2976
比如，去問一位醫生
如何做醫療診斷，
如何做醫療診斷，
05:56
and she might be able
to give you a few rules of thumb,
to give you a few rules of thumb,
116
344560
2656
她可能會給你少數經驗法則，
05:59
but ultimately she'd struggle.
117
347240
1656
但最終，她會很掙扎。
06:00
She'd say it requires things like
creativity and judgment and intuition.
creativity and judgment and intuition.
118
348920
4816
她會說，你還需要創意、
判斷，以及直覺才行。
判斷，以及直覺才行。
06:05
And these things are
very difficult to articulate,
very difficult to articulate,
119
353760
2376
這些東西是很難明確表達的，
06:08
and so it was thought these tasks
would be very hard to automate.
would be very hard to automate.
120
356160
3096
所以這些工作任務就會
被認為很難自動化。
被認為很難自動化。
06:11
If a human being can't explain themselves,
121
359280
2536
如果人類無法解釋他們自己的做法，
06:13
where on earth do we begin
in writing a set of instructions
in writing a set of instructions
122
361840
2896
我們究竟要從何開始寫指令
06:16
for a machine to follow?
123
364760
1200
給機器遵循？
06:18
Thirty years ago, this view was right,
124
366640
2576
三十年前，這個觀點是對的，
06:21
but today it's looking shaky,
125
369240
2136
但現今，它很不穩固，
06:23
and in the future
it's simply going to be wrong.
it's simply going to be wrong.
126
371400
2256
在未來，它將會是錯的。
06:25
Advances in processing power,
in data storage capability
in data storage capability
127
373680
3256
處理能力、資料儲存容量，
以及演算法設計都在進步，
06:28
and in algorithm design
128
376960
1656
06:30
mean that this
routine-nonroutine distinction
routine-nonroutine distinction
129
378640
2496
這就表示例行與非例行事務間的區別
06:33
is diminishingly useful.
130
381160
1736
越來越沒有用了。
06:34
To see this, go back to the case
of making a medical diagnosis.
of making a medical diagnosis.
131
382920
3256
要了解這點，我們
回到醫療診斷的案例。
回到醫療診斷的案例。
06:38
Earlier in the year,
132
386200
1376
今年早些時候，
06:39
a team of researchers at Stanford
announced they'd developed a system
announced they'd developed a system
133
387600
3296
史丹佛的一個研究者團隊
宣佈他們發展出了一個系統，
宣佈他們發展出了一個系統，
06:42
which can tell you
whether or not a freckle is cancerous
whether or not a freckle is cancerous
134
390920
3056
它能告訴你一個斑點是否為惡性的，
06:46
as accurately as leading dermatologists.
135
394000
2680
正確率不輸給頂尖皮膚科醫生。
06:49
How does it work?
136
397280
1256
它怎麼做到的？
06:50
It's not trying to copy the judgment
or the intuition of a doctor.
or the intuition of a doctor.
137
398560
5296
它並不是嘗試複製
醫生的判斷或是直覺。
醫生的判斷或是直覺。
06:55
It knows or understands
nothing about medicine at all.
nothing about medicine at all.
138
403880
3136
它對於醫學是一竅不通。
06:59
Instead, it's running
a pattern recognition algorithm
a pattern recognition algorithm
139
407040
2576
反之，它進行的是模式辨識演算法，
07:01
through 129,450 past cases,
140
409640
4656
在 129,450 個個案當中，
07:06
hunting for similarities
between those cases
between those cases
141
414320
3096
獵尋那些個案與欲探究的損害
07:09
and the particular lesion in question.
142
417440
2080
之間有哪些相似性。
07:12
It's performing these tasks
in an unhuman way,
in an unhuman way,
143
420080
3216
它是用非人類的方式
在進行這些工作任務，
在進行這些工作任務，
07:15
based on the analysis
of more possible cases
of more possible cases
144
423320
2336
且是以大量案例的分析來當依據，
07:17
than any doctor could hope
to review in their lifetime.
to review in their lifetime.
145
425680
2616
案例數多到是醫生
一輩子都看不完的。
一輩子都看不完的。
07:20
It didn't matter that that human being,
146
428320
1896
無所謂人類，也就是醫生，
07:22
that doctor, couldn't explain
how she'd performed the task.
how she'd performed the task.
147
430240
2800
是否能解釋她如何進行此工作任務。
07:25
Now, there are those
who dwell upon that the fact
who dwell upon that the fact
148
433640
2336
有些人老是會想著
這些機器被建立時
沒有依循我們的形象。
沒有依循我們的形象。
07:28
that these machines
aren't built in our image.
aren't built in our image.
149
436000
2296
07:30
As an example, take IBM's Watson,
150
438320
2056
以 IBM 的「華生 」為例，
07:32
the supercomputer that went
on the US quiz show "Jeopardy!" in 2011,
on the US quiz show "Jeopardy!" in 2011,
151
440400
4856
那是台超級電腦，2011 年參加
美國的益智節目《危險邊緣》，
美國的益智節目《危險邊緣》，
07:37
and it beat the two
human champions at "Jeopardy!"
human champions at "Jeopardy!"
152
445280
3016
在節目中它打敗了兩位人類冠軍。
07:40
The day after it won,
153
448320
1696
它獲勝之後的隔天，
07:42
The Wall Street Journal ran a piece
by the philosopher John Searle
by the philosopher John Searle
154
450040
3296
《華爾街日報》刊了一篇
哲學家約翰希爾勒的文章，
哲學家約翰希爾勒的文章，
07:45
with the title "Watson
Doesn't Know It Won on 'Jeopardy!'"
Doesn't Know It Won on 'Jeopardy!'"
155
453360
3376
標題是〈華生不知道
它自己贏了《危險邊緣》 〉。
它自己贏了《危險邊緣》 〉。
07:48
Right, and it's brilliant, and it's true.
156
456760
1976
是的，這篇文章很聰明也沒說錯。
07:50
You know, Watson didn't
let out a cry of excitement.
let out a cry of excitement.
157
458760
2456
華生並沒有興奮地放聲大叫。
它沒有打電話給它的父母
說它的表現多棒。
說它的表現多棒。
07:53
It didn't call up its parents
to say what a good job it had done.
to say what a good job it had done.
158
461240
3096
它沒有去酒吧喝酒慶祝。
07:56
It didn't go down to the pub for a drink.
159
464360
2336
07:58
This system wasn't trying to copy the way
that those human contestants played,
that those human contestants played,
160
466720
4456
這個系統並沒有試圖複製
那些人類參賽者比賽的方式，
那些人類參賽者比賽的方式，
08:03
but it didn't matter.
161
471200
1256
但那無所謂。
08:04
It still outperformed them.
162
472480
1976
它仍然表現得比人類好。
08:06
Resolving the intelligence myth
163
474480
1576
解開這個智慧迷思之後，
08:08
shows us that our limited understanding
about human intelligence,
about human intelligence,
164
476080
3376
看到的是雖然我們對於
人類智慧、對我們如何
人類智慧、對我們如何
08:11
about how we think and reason,
165
479480
1896
思考推理的方式了解有限，
08:13
is far less of a constraint
on automation than it was in the past.
on automation than it was in the past.
166
481400
3456
但這個限制對於自動化的影響
已經遠比過去小很多。
已經遠比過去小很多。
08:16
What's more, as we've seen,
167
484880
1496
此外，如我們所見，
08:18
when these machines
perform tasks differently to human beings,
perform tasks differently to human beings,
168
486400
3416
當這些機器用和人類不同的
方式來執行工作任務時，
方式來執行工作任務時，
08:21
there's no reason to think
169
489840
1256
沒有理由認為
08:23
that what human beings
are currently capable of doing
are currently capable of doing
170
491120
2536
人類目前能夠做到的事
08:25
represents any sort of summit
171
493680
1456
就代表了一種上限，
08:27
in what these machines
might be capable of doing in the future.
might be capable of doing in the future.
172
495160
3000
在未來機器能夠達成的事
都不可能超過這個上限。
都不可能超過這個上限。
08:31
Now the third myth,
173
499040
1256
第三項迷思，
08:32
what I call the superiority myth.
174
500320
2456
我稱之為優越迷思。
08:34
It's often said that those who forget
175
502800
2216
常見的說法是，有些人會
08:37
about the helpful side
of technological progress,
of technological progress,
176
505040
2456
忘記了科技進步的幫助面，
08:39
those complementarities from before,
177
507520
2496
忘記過去的互補性，
08:42
are committing something
known as the lump of labor fallacy.
known as the lump of labor fallacy.
178
510040
3040
這些人所犯的，就是
所謂的「勞動總合謬誤」。
所謂的「勞動總合謬誤」。
08:45
Now, the problem is
the lump of labor fallacy
the lump of labor fallacy
179
513840
2295
問題是，勞動總合謬誤本身
08:48
is itself a fallacy,
180
516159
1496
就是個謬誤，
08:49
and I call this the lump
of labor fallacy fallacy,
of labor fallacy fallacy,
181
517679
2937
我把它稱為
「勞動總合謬誤的謬誤」，
「勞動總合謬誤的謬誤」，
08:52
or LOLFF, for short.
182
520640
2320
簡寫為「LOLFF」。
08:56
Let me explain.
183
524000
1416
讓我解釋一下。
08:57
The lump of labor fallacy
is a very old idea.
is a very old idea.
184
525440
2136
勞動總合謬誤是個很古老的想法。
08:59
It was a British economist, David Schloss,
who gave it this name in 1892.
who gave it this name in 1892.
185
527600
4216
這個名稱是 1892 年由英國
經濟學家大衛許洛斯取的。
經濟學家大衛許洛斯取的。
09:03
He was puzzled
to come across a dock worker
to come across a dock worker
186
531840
2816
有件事讓他百思不解，
他遇到一個碼頭工人，
他遇到一個碼頭工人，
09:06
who had begun to use
a machine to make washers,
a machine to make washers,
187
534680
2336
這個工人開始用機器來製造墊圈，
09:09
the small metal discs
that fasten on the end of screws.
that fasten on the end of screws.
188
537040
3320
墊圈是小型的金屬圓盤，
固定在螺絲底端。
固定在螺絲底端。
09:13
And this dock worker
felt guilty for being more productive.
felt guilty for being more productive.
189
541000
3760
這個碼頭工人對於自己的
高生產力有罪惡感。
高生產力有罪惡感。
09:17
Now, most of the time,
we expect the opposite,
we expect the opposite,
190
545560
2176
通常，我們預期的是相反的反應，
生產力不高才會讓人感到罪惡，
09:19
that people feel guilty
for being unproductive,
for being unproductive,
191
547760
2216
你知道的，工作時
花太多時間滑臉書或推特。
花太多時間滑臉書或推特。
09:22
you know, a little too much time
on Facebook or Twitter at work.
on Facebook or Twitter at work.
192
550000
3016
但這個工人對於
太有生產力感到罪惡，
太有生產力感到罪惡，
09:25
But this worker felt guilty
for being more productive,
for being more productive,
193
553040
2536
問他原因，他說：「我知道我做錯了。
09:27
and asked why, he said,
"I know I'm doing wrong.
"I know I'm doing wrong.
194
555600
2296
09:29
I'm taking away the work of another man."
195
557920
2040
我搶走了另一個人的工作。」
09:32
In his mind, there was
some fixed lump of work
some fixed lump of work
196
560760
2976
在他的認知中，勞動總合是固定的，
09:35
to be divided up between him and his pals,
197
563760
2136
要由他和他的伙伴來分攤，
09:37
so that if he used
this machine to do more,
this machine to do more,
198
565920
2056
所以如果他用機器多做一點，
09:40
there'd be less left for his pals to do.
199
568000
2016
他伙伴能做的就變少了。
09:42
Schloss saw the mistake.
200
570040
1856
許洛斯看到了這個錯誤。
09:43
The lump of work wasn't fixed.
201
571920
1856
勞動總合並不是固定的。
09:45
As this worker used the machine
and became more productive,
and became more productive,
202
573800
2816
當這個工人用機器提高生產力，
09:48
the price of washers would fall,
demand for washers would rise,
demand for washers would rise,
203
576640
2976
墊圈的價格會下降，
對墊圈的需求會提高，
對墊圈的需求會提高，
09:51
more washers would have to be made,
204
579640
1696
就得要做出更多的墊圈，
09:53
and there'd be more work
for his pals to do.
for his pals to do.
205
581360
2096
他的伙伴反而會有更多要做。
09:55
The lump of work would get bigger.
206
583480
1696
勞動總合變更大了。
09:57
Schloss called this
"the lump of labor fallacy."
"the lump of labor fallacy."
207
585200
2680
許洛斯稱之為「勞動總合謬誤」。
10:00
And today you hear people talk
about the lump of labor fallacy
about the lump of labor fallacy
208
588560
2936
現今，在思考有各類工作的未來時，
會聽到人們談到勞動總合謬誤。
10:03
to think about the future
of all types of work.
of all types of work.
209
591520
2216
沒有固定的勞動總合
10:05
There's no fixed lump of work
out there to be divided up
out there to be divided up
210
593760
2656
要讓人類與機器瓜分。
10:08
between people and machines.
211
596440
1376
是的，機器會取代人類，
讓原本的勞動總合變少，
讓原本的勞動總合變少，
10:09
Yes, machines substitute for human beings,
making the original lump of work smaller,
making the original lump of work smaller,
212
597840
4656
10:14
but they also complement human beings,
213
602520
1856
但它們也會補足人類，
10:16
and the lump of work
gets bigger and changes.
gets bigger and changes.
214
604400
2096
勞動總合會變更大並且改變。
10:19
But LOLFF.
215
607760
1616
但，LOLFF。
10:21
Here's the mistake:
216
609400
1376
錯誤是這樣的：
10:22
it's right to think
that technological progress
that technological progress
217
610800
2216
認為科技進步會讓
要做的勞動總合變大，
要做的勞動總合變大，
10:25
makes the lump of work to be done bigger.
218
613040
1976
這點是沒錯的。
有些工作任務變得較有價值。
有新工作任務需要完成。
有新工作任務需要完成。
10:27
Some tasks become more valuable.
New tasks have to be done.
New tasks have to be done.
219
615040
3016
10:30
But it's wrong to think that necessarily,
220
618080
2536
錯的地方在於，認為安排人類
10:32
human beings will be best placed
to perform those tasks.
to perform those tasks.
221
620640
3256
來做那些工作任務一定是最好的。
10:35
And this is the superiority myth.
222
623920
1616
這就是優越迷思。
10:37
Yes, the lump of work
might get bigger and change,
might get bigger and change,
223
625560
3416
是的，勞動總量可能
會變大也會改變，
會變大也會改變，
10:41
but as machines become more capable,
224
629000
1976
但隨著機器變得更有能力，
10:43
it's likely that they'll take on
the extra lump of work themselves.
the extra lump of work themselves.
225
631000
3896
很有可能它們會自己去接下
那些額外的勞動總量。
那些額外的勞動總量。
10:46
Technological progress,
rather than complement human beings,
rather than complement human beings,
226
634920
3256
科技進步就不是在補足人類了，
10:50
complements machines instead.
227
638200
1880
反而是補足機器。
10:52
To see this, go back
to the task of driving a car.
to the task of driving a car.
228
640920
3016
可以回頭看駕駛汽車的
工作任務來了解這點。
工作任務來了解這點。
10:55
Today, satnav systems
directly complement human beings.
directly complement human beings.
229
643960
4096
現今，衛星導航系統直接補足人類。
11:00
They make some
human beings better drivers.
human beings better drivers.
230
648080
2280
它讓一些人類變成更好的駕駛。
11:02
But in the future,
231
650920
1256
但在未來，
軟體會取代坐在駕駛座上的人類，
11:04
software is going to displace
human beings from the driving seat,
human beings from the driving seat,
232
652200
3096
這些衛星導航系統
就不是在補足人類了，
就不是在補足人類了，
11:07
and these satnav systems,
rather than complement human beings,
rather than complement human beings,
233
655320
2936
而單純就是在讓這些
無人駕駛汽車更有效率，
無人駕駛汽車更有效率，
11:10
will simply make these
driverless cars more efficient,
driverless cars more efficient,
234
658280
2536
11:12
helping the machines instead.
235
660840
1536
改而協助機器。
11:14
Or go to those indirect complementarities
that I mentioned as well.
that I mentioned as well.
236
662400
4056
或也可以回到
我剛提過的間接互補性。
我剛提過的間接互補性。
11:18
The economic pie may get larger,
237
666480
1776
經濟的派可能會變更大，
11:20
but as machines become more capable,
238
668280
1736
但隨著機器更有能力，
11:22
it's possible that any new demand
will fall on goods that machines,
will fall on goods that machines,
239
670040
3143
有可能所有符合新需求的商品都適合
由機器而不是由人類來製造。
11:25
rather than human beings,
are best placed to produce.
are best placed to produce.
240
673207
2649
11:27
The economic pie may change,
241
675880
1896
經濟的派可能會改變，
11:29
but as machines become more capable,
242
677800
1896
但隨著機器變得更有能力，
11:31
it's possible that they'll be best placed
to do the new tasks that have to be done.
to do the new tasks that have to be done.
243
679720
4856
有可能它們最適合運用在
新工作任務中，那些必須解決的事。
新工作任務中，那些必須解決的事。
11:36
In short, demand for tasks
isn't demand for human labor.
isn't demand for human labor.
244
684600
3696
簡言之，對工作任務的需求
並非對人類勞動力的需求。
並非對人類勞動力的需求。
11:40
Human beings only stand to benefit
245
688320
1936
人類只有在仍然能支配
11:42
if they retain the upper hand
in all these complemented tasks,
in all these complemented tasks,
246
690280
3816
這些補足性工作任務的
情況下才有可能受益，
情況下才有可能受益，
11:46
but as machines become more capable,
that becomes less likely.
that becomes less likely.
247
694120
3720
但隨著機器變得更有能力，
那就更不可能發生。
那就更不可能發生。
11:50
So what do these three myths tell us then?
248
698760
2016
所以，這三項迷思告訴我們什麼？
11:52
Well, resolving the Terminator myth
249
700800
1696
解開終結者迷思之後，
11:54
shows us that the future of work depends
upon this balance between two forces:
upon this balance between two forces:
250
702520
3696
我們知道工作的未來還要
仰賴兩股力量間的平衡：
仰賴兩股力量間的平衡：
11:58
one, machine substitution
that harms workers
that harms workers
251
706240
3136
第一：機器代替，這會傷害到工人，
12:01
but also those complementarities
that do the opposite.
that do the opposite.
252
709400
2576
但也會有第二股力量，
互補性，反而會幫助工人。
互補性，反而會幫助工人。
12:04
And until now, this balance
has fallen in favor of human beings.
has fallen in favor of human beings.
253
712000
4040
直到目前，這平衡是對人類有利的。
12:09
But resolving the intelligence myth
254
717120
1736
但解開了智慧迷思之後，
12:10
shows us that that first force,
machine substitution,
machine substitution,
255
718880
2496
我們知道，第一股力量，機器代替，
12:13
is gathering strength.
256
721400
1296
正在聚集實力。
12:14
Machines, of course, can't do everything,
257
722720
1976
當然，機器並非什麼都能做，
12:16
but they can do far more,
258
724720
1256
但它們能做的很多，
12:18
encroaching ever deeper into the realm
of tasks performed by human beings.
of tasks performed by human beings.
259
726000
4576
能更深進入到人類所進行之
工作任務的領域中。
工作任務的領域中。
12:22
What's more, there's no reason to think
260
730600
1896
此外，沒有理由去認為
12:24
that what human beings
are currently capable of
are currently capable of
261
732520
2216
人類目前已經能做到的事，
12:26
represents any sort of finishing line,
262
734760
1856
就表示是某種終點線，
12:28
that machines are going
to draw to a polite stop
to draw to a polite stop
263
736640
2256
等到機器和我們一樣有能力時
就會禮貌地在終點線前停下來。
12:30
once they're as capable as us.
264
738920
1816
12:32
Now, none of this matters
265
740760
1536
這些都無所謂，
12:34
so long as those helpful
winds of complementarity
winds of complementarity
266
742320
2816
只要機器和人類在工作上
能相得益彰就好。
能相得益彰就好。
12:37
blow firmly enough,
267
745160
1736
12:38
but resolving the superiority myth
268
746920
1936
但解開了優越迷思之後，
12:40
shows us that that process
of task encroachment
of task encroachment
269
748880
3096
我們了解到，工作任務侵佔的過程
12:44
not only strengthens
the force of machine substitution,
the force of machine substitution,
270
752000
3936
不僅是強化了機器代替的那股力量，
12:47
but it wears down
those helpful complementarities too.
those helpful complementarities too.
271
755960
3336
也會耗損那些有助益的互補性。
12:51
Bring these three myths together
272
759320
1936
把這三項迷思結合起來，
12:53
and I think we can capture a glimpse
of that troubling future.
of that troubling future.
273
761280
2936
我想，我們就能對
讓人困擾的未來有點概念。
讓人困擾的未來有點概念。
12:56
Machines continue to become more capable,
274
764240
2016
機器持續變得更有能力，
12:58
encroaching ever deeper
on tasks performed by human beings,
on tasks performed by human beings,
275
766280
3656
比以前更深入人類進行的工作任務，
13:01
strengthening the force
of machine substitution,
of machine substitution,
276
769960
2576
強化機器代替的那股力量，
13:04
weakening the force
of machine complementarity.
of machine complementarity.
277
772560
3616
弱化機器互補性的那股力量。
13:08
And at some point, that balance
falls in favor of machines
falls in favor of machines
278
776200
4296
在某個時點，那平衡
會變得對機器有利，
會變得對機器有利，
13:12
rather than human beings.
279
780520
2056
而非人類。
13:14
This is the path we're currently on.
280
782600
1736
我們目前就在這條路上。
13:16
I say "path" deliberately,
because I don't think we're there yet,
because I don't think we're there yet,
281
784360
3176
我刻意用「路」這個字，
因為我們還沒有到達那裡，
因為我們還沒有到達那裡，
13:19
but it is hard to avoid the conclusion
that this is our direction of travel.
that this is our direction of travel.
282
787560
3640
但無可避免，結論會是：
這就是我們行進的方向。
這就是我們行進的方向。
13:24
That's the troubling part.
283
792640
1456
那是讓人困擾的部分。
13:26
Let me say now why I think actually
this is a good problem to have.
this is a good problem to have.
284
794120
3520
現在讓我說明為什麼我認為
有這個問題是件好事。
有這個問題是件好事。
13:30
For most of human history,
one economic problem has dominated:
one economic problem has dominated:
285
798520
3536
大部分的人類歷史中，
主導的都是這一個經濟問題：
主導的都是這一個經濟問題：
13:34
how to make the economic pie
large enough for everyone to live on.
large enough for everyone to live on.
286
802080
4056
如何讓經濟的派夠大，
確保每個人都得以維生。
確保每個人都得以維生。
13:38
Go back to the turn
of the first century AD,
of the first century AD,
287
806160
2176
回到西元一世紀，
13:40
and if you took the global economic pie
288
808360
2096
如果用全球的派當作例子，
13:42
and divided it up into equal slices
for everyone in the world,
for everyone in the world,
289
810480
3296
將它切成相同的等分，
分給全世界的人，
分給全世界的人，
13:45
everyone would get a few hundred dollars.
290
813800
2136
每個人可能得到幾百美元。
13:47
Almost everyone lived
on or around the poverty line.
on or around the poverty line.
291
815960
2760
幾乎每個人都是在
貧窮水平線上下過生活。
貧窮水平線上下過生活。
13:51
And if you roll forward a thousand years,
292
819320
2176
如果你再向前轉一千年，
13:53
roughly the same is true.
293
821520
1240
大致上也是一樣的。
13:55
But in the last few hundred years,
economic growth has taken off.
economic growth has taken off.
294
823680
3576
但在過去幾百年間，經濟成長起飛。
13:59
Those economic pies have exploded in size.
295
827280
2376
這些經濟的派在尺寸上都爆增。
14:01
Global GDP per head,
296
829680
2056
全球的人均生產總值，
14:03
the value of those individual
slices of the pie today,
slices of the pie today,
297
831760
3376
也就是現今每個人分到的那片派，
14:07
they're about 10,150 dollars.
298
835160
2816
價值約 10,150 美元。
14:10
If economic growth continues
at two percent,
at two percent,
299
838000
2696
如果經濟成長率維持 2%，
14:12
our children will be twice as rich as us.
300
840720
2056
我們的孩子會比我們富有兩倍。
14:14
If it continues
at a more measly one percent,
at a more measly one percent,
301
842800
2296
如果成長率低一點，維持在 1%，
14:17
our grandchildren
will be twice as rich as us.
will be twice as rich as us.
302
845120
2656
我們的孫子會比我們富有兩倍。
14:19
By and large, we've solved
that traditional economic problem.
that traditional economic problem.
303
847800
3680
總的來說，我們解決了
傳統的經濟問題。
傳統的經濟問題。
14:24
Now, technological unemployment,
if it does happen,
if it does happen,
304
852200
3016
如果真的因為科技進步而造成失業，
14:27
in a strange way will be
a symptom of that success,
a symptom of that success,
305
855240
3216
從一種奇怪的角度來看，
那會是一種成功的象徵，
那會是一種成功的象徵，
14:30
will have solved one problem --
how to make the pie bigger --
how to make the pie bigger --
306
858480
3856
它能夠解決一個問題
──如何讓派變大──
──如何讓派變大──
14:34
but replaced it with another --
307
862360
1816
但卻用另一個問題取代它──
14:36
how to make sure
that everyone gets a slice.
that everyone gets a slice.
308
864200
2760
如何確保每個人得到一片派。
14:39
As other economists have noted,
solving this problem won't be easy.
solving this problem won't be easy.
309
867840
3496
如其他經濟學家注意到的，
解決這個問題並不容易。
解決這個問題並不容易。
14:43
Today, for most people,
310
871360
1656
現今，對大部分人而言，
14:45
their job is their seat
at the economic dinner table,
at the economic dinner table,
311
873040
2496
他們的工作就是在
經濟晚餐餐桌上的席位，
經濟晚餐餐桌上的席位，
14:47
and in a world with less work
or even without work,
or even without work,
312
875560
2416
在一個更少或甚至沒工作的世界裡，
14:50
it won't be clear
how they get their slice.
how they get their slice.
313
878000
2056
沒人知道他們如何得到自己的那片派。
14:52
There's a great deal
of discussion, for instance,
of discussion, for instance,
314
880080
2336
比如，有很多的討論都是
14:54
about various forms
of universal basic income
of universal basic income
315
882440
2696
關於全體基本收入的各種形式，
14:57
as one possible approach,
316
885160
1216
這是種可能的方式，
14:58
and there's trials underway
317
886400
1616
且在美國、芬蘭，
15:00
in the United States
and in Finland and in Kenya.
and in Finland and in Kenya.
318
888040
2400
及肯亞都有試驗正在進行中。
15:03
And this is the collective challenge
that's right in front of us,
that's right in front of us,
319
891000
3176
這是我們要面臨的集體挑戰，
15:06
to figure out how this material prosperity
generated by our economic system
generated by our economic system
320
894200
5056
要想出我們的經濟體制
所產生出的物質繁榮要如何
所產生出的物質繁榮要如何
15:11
can be enjoyed by everyone
321
899280
1976
讓每個人都享受到，
15:13
in a world in which
our traditional mechanism
our traditional mechanism
322
901280
2416
而且在這個世界中，
我們的傳統切派機制，
15:15
for slicing up the pie,
323
903720
1856
15:17
the work that people do,
324
905600
1936
瓜分人們所做的工作的機制，
15:19
withers away and perhaps disappears.
325
907560
2160
在衰弱且也許在消失中。
15:22
Solving this problem is going to require
us to think in very different ways.
us to think in very different ways.
326
910280
4360
若要解決這個問題，我們
得要用很不同的方式思考。
得要用很不同的方式思考。
15:27
There's going to be a lot of disagreement
about what ought to be done,
about what ought to be done,
327
915400
4176
對於該做什麼事，
必定會有很多異議，
必定會有很多異議，
15:31
but it's important to remember
that this is a far better problem to have
that this is a far better problem to have
328
919600
3416
但很重要的是要記住，
有這個問題其實算好事，
有這個問題其實算好事，
15:35
than the one that haunted
our ancestors for centuries:
our ancestors for centuries:
329
923040
2816
比我們的祖先煩惱了
幾世紀的問題要好多了，
幾世紀的問題要好多了，
15:37
how to make that pie
big enough in the first place.
big enough in the first place.
330
925880
3376
他們煩惱的是：
一開始要如何讓派變大。
一開始要如何讓派變大。
15:41
Thank you very much.
331
929280
1256
非常謝謝各位。
15:42
(Applause)
332
930560
3840
（掌聲）
ABOUT THE SPEAKER
Daniel Susskind - EconomistDaniel Susskind explores the impact of technology, particularly artificial intelligence, on work and society.
Why you should listen
Daniel Susskind is the co-author, with Richard Susskind, of the best-selling book, The Future of the Professions, and a Fellow in Economics at Balliol College, Oxford University. He is currently finishing his latest book, on the future of work. Previously, he worked in the British Government -- as a policy adviser in the Prime Minister's Strategy Unit, as a policy analyst in the Policy Unit in 10 Downing Street, and as a senior policy adviser in the Cabinet Office. Susskind received a doctorate in economics from Oxford University and was a Kennedy Scholar at Harvard University.
Daniel Susskind | Speaker | TED.com