a fear that in the future,

近期，自動化焦慮一直在散佈，

many jobs will be performed by machines

它是種恐懼，害怕在未來

rather than human beings,

許多工作會由機器來進行，

given the remarkable advances that are unfolding

而不是人類，

in artificial intelligence and robotics.

因為現在已可以看到在人工智慧

What's clear is that there will be significant change.

和機器人學領域的驚人進步。

What's less clear is what that change will look like.

很清楚的一點是， 將來會有顯著的改變。

My research suggests that the future is both troubling and exciting.

比較不那麼清楚的是， 改變會是什麼樣的。

The threat of technological unemployment is real,

我的研究指出，未來 既讓人困擾又讓人興奮。

and yet it's a good problem to have.

科技造成失業的威脅是真的，

And to explain how I came to that conclusion,

但，能有這種問題也是件好事。

I want to confront three myths

為了解釋我如何得到這個結論，

that I think are currently obscuring our vision of this automated future.

我想要來正視三項迷思，

A picture that we see on our television screens,

我認為這些迷思 目前遮掩了我們的視線，

in books, in films, in everyday commentary

讓我們看不清自動化的未來。

is one where an army of robots descends on the workplace

我們在電視上、書中、電影中、

with one goal in mind:

每天的評論中所看到的描繪，

to displace human beings from their work.

通常是機器人大軍湧入工作場所，

And I call this the Terminator myth.

心中只有一個目標：

Yes, machines displace human beings from particular tasks,

在工作上取代人類。

but they don't just substitute for human beings.

我稱這個想法為「終結者迷思」。

They also complement them in other tasks,

是的，在特定的工作任務上， 機器會取代人類，

making that work more valuable and more important.

但它們不會就這樣代替人類。

Sometimes they complement human beings directly,

它們在其他工作任務上會補足人類，

making them more productive or more efficient at a particular task.

讓工作更有價值、更重要。

So a taxi driver can use a satnav system to navigate on unfamiliar roads.

有時，它們會直接補足人類，

An architect can use computer-assisted design software

讓人類在特定的工作任務上 能更有生產力或更有效率。

to design bigger, more complicated buildings.

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。

But technological progress doesn't just complement human beings directly.

建築師可以用電腦輔助的設計軟體

It also complements them indirectly, and it does this in two ways.

來設計更大、更複雜的建築物。

The first is if we think of the economy as a pie,

但科技進步並不只會直接補足人類。

technological progress makes the pie bigger.

它也會用間接方式補足人類， 間接的方式有兩種。

As productivity increases, incomes rise and demand grows.

第一，如果我們把 經濟想成是一塊派，

The British pie, for instance,

科技進步會讓派變更大。

is more than a hundred times the size it was 300 years ago.

隨著生產力增加， 收入會增加，需求會成長。

And so people displaced from tasks in the old pie

比如，英國的派

could find tasks to do in the new pie instead.

與三百年前相比，現在超過百倍大。

But technological progress doesn't just make the pie bigger.

在舊派工作被取代的人，

It also changes the ingredients in the pie.

能在新派中找到工作。

As time passes, people spend their income in different ways,

但科技進步並不只會讓派變大。

changing how they spread it across existing goods,

它也會改變派的成分原料。

and developing tastes for entirely new goods, too.

隨時間演進，人會以 不同的方式花費他們的收入，

New industries are created,

改變既有商品花費上的分配，

new tasks have to be done

並也會發展出對於全新商品的品味。

and that means often new roles have to be filled.

新的產業會被創造出來，

So again, the British pie:

有新的工作任務需要被完成，

300 years ago, most people worked on farms,

那就意味著有新角色要有人扮演。

150 years ago, in factories,

所以，再回到英國的派：

and today, most people work in offices.

三百年前，大部分的人在農場工作，

And once again, people displaced from tasks in the old bit of pie

一百五十年前，在工廠工作，

could tumble into tasks in the new bit of pie instead.

現今，大部分的人在辦公室工作。

Economists call these effects complementarities,

再提一次，在老派工作被取代的人，

but really that's just a fancy word to capture the different way

可能會在新派當中 發現可以做的工作任務。

that technological progress helps human beings.

經濟學家把這些效應稱為互補性，

Resolving this Terminator myth

但那只是個很炫的詞，其實意思就是

shows us that there are two forces at play:

科技進步用不同的方式在協助人類。

one, machine substitution that harms workers,

解開這個終結者迷思之後，

but also these complementarities that do the opposite.

會發現有兩股力量在運作：

Now the second myth,

第一：機器代替，這會傷害到工人，

what I call the intelligence myth.

但也會有第二股力量， 互補性，反而會幫助工人。

What do the tasks of driving a car, making a medical diagnosis

再來，第二項迷思，

and identifying a bird at a fleeting glimpse have in common?

我稱之為「智慧迷思」。

Well, these are all tasks that until very recently,

以下這些工作任務： 駕駛一台車、做出醫療診斷，

leading economists thought couldn't readily be automated.

及快速一瞥就辨識出 一隻鳥，有何共通性？

And yet today, all of these tasks can be automated.

這些工作任務都是直到最近

You know, all major car manufacturers have driverless car programs.

仍被經濟學家認為不能 自動化的工作任務。

There's countless systems out there that can diagnose medical problems.

然而，現今，所有這些 工作任務都能被自動化。

And there's even an app that can identify a bird

所有大型汽車製造商都有 無人駕駛汽車的計畫。

at a fleeting glimpse.

外面有數不清的系統 都能夠診斷醫療問題。

Now, this wasn't simply a case of bad luck on the part of economists.

甚至有個應用程式能用來辨識鳥類，

They were wrong,

只要快速一瞥。

and the reason why they were wrong is very important.

這並不是經濟學家運氣不好的情況。

They've fallen for the intelligence myth,

他們錯了，

the belief that machines have to copy the way

而他們為什麼會錯的原因很重要。

that human beings think and reason

他們陷入了智慧迷思中，

in order to outperform them.

相信機器必須要複製人類

When these economists were trying to figure out

思考和推理的方式，

what tasks machines could not do,

才能夠表現得比人類好。

they imagined the only way to automate a task

當這些經濟學家在試圖想出

was to sit down with a human being,

機器無法勝任哪些工作任務，

get them to explain to you how it was they performed a task,

他們想像，將工作任務自動化的

and then try and capture that explanation

唯一方式就是和人類坐下來，

in a set of instructions for a machine to follow.

讓人類解釋他們如何執行工作任務，

This view was popular in artificial intelligence at one point, too.

再試著分析他們的解釋，

I know this because Richard Susskind,

轉換成一組指令，讓機器照著做。

who is my dad and my coauthor,

在人工智慧領域，這種觀點 曾在某個時點很流行過。

wrote his doctorate in the 1980s on artificial intelligence and the law

我知道這點，因為理查薩斯金，

at Oxford University,

他是我爸爸也是我的共同作者，

and he was part of the vanguard.

在八〇年代，在牛津大學 寫了一篇關於人工智慧

And with a professor called Phillip Capper

與法律的博士論文，

and a legal publisher called Butterworths,

他是先鋒部隊之一。

they produced the world's first commercially available

和一位名叫菲利普卡波的教授，

artificial intelligence system in the law.

以及一間法律出版社 叫做 Butterworths，

This was the home screen design.

他們合作製作出了 世界上第一個商業用的

He assures me this was a cool screen design at the time.

法律人工智慧系統。

(Laughter)

這是首頁的畫面設計。

I've never been entirely convinced.

他向我保證，在當時 這是很酷的畫面設計。

He published it in the form of two floppy disks,

（笑聲）

at a time where floppy disks genuinely were floppy,

我從來沒有被說服。

and his approach was the same as the economists':

他用兩張軟碟片的形式將之出版，

sit down with a lawyer,

在那個時代，軟碟片真的是軟的，

get her to explain to you how it was she solved a legal problem,

而他的方式就和經濟學家一樣：

and then try and capture that explanation in a set of rules for a machine to follow.

和一名律師坐下來，

In economics, if human beings could explain themselves in this way,

讓她向你解釋如何解決法律問題，

the tasks are called routine, and they could be automated.

接著就試著把她的解釋 轉成一組指令給機器執行。

But if human beings can't explain themselves,

在經濟上，如果人類能夠用 這種方式解釋自己做的事，

the tasks are called non-routine, and they're thought to be out reach.

這種工作任務就叫做例行事務， 是可以被自動化的。

Today, that routine-nonroutine distinction is widespread.

但如果人類無法解釋出怎麼做，

Think how often you hear people say to you

這種工作任務叫做非例行事務， 應該是不能自動化的。

machines can only perform tasks that are predictable or repetitive,

現今，將事務區別為例行 與非例行是處處可見的。

rules-based or well-defined.

想想看，你有多常聽到別人對你說

Those are all just different words for routine.

機器能進行的工作任務 只有可預測的、重覆性的、

And go back to those three cases that I mentioned at the start.

以規則為基礎的，或定義清楚的。

Those are all classic cases of nonroutine tasks.

那些詞只是例行事務的不同說法。

Ask a doctor, for instance, how she makes a medical diagnosis,

回到我一開始提到的三個案例。

and she might be able to give you a few rules of thumb,

那些案例是典型的非例行事務。

but ultimately she'd struggle.

比如，去問一位醫生 如何做醫療診斷，

She'd say it requires things like creativity and judgment and intuition.

她可能會給你少數經驗法則，

And these things are very difficult to articulate,

但最終，她會很掙扎。

and so it was thought these tasks would be very hard to automate.

她會說，你還需要創意、 判斷，以及直覺才行。

If a human being can't explain themselves,

這些東西是很難明確表達的，

where on earth do we begin in writing a set of instructions

所以這些工作任務就會 被認為很難自動化。

for a machine to follow?

如果人類無法解釋他們自己的做法，

Thirty years ago, this view was right,

我們究竟要從何開始寫指令

but today it's looking shaky,

給機器遵循？

and in the future it's simply going to be wrong.

三十年前，這個觀點是對的，

Advances in processing power, in data storage capability

但現今，它很不穩固，

and in algorithm design

在未來，它將會是錯的。

mean that this routine-nonroutine distinction

處理能力、資料儲存容量，

is diminishingly useful.

以及演算法設計都在進步，

To see this, go back to the case of making a medical diagnosis.

這就表示例行與非例行事務間的區別

Earlier in the year,

越來越沒有用了。

a team of researchers at Stanford announced they'd developed a system

要了解這點，我們 回到醫療診斷的案例。

which can tell you whether or not a freckle is cancerous

今年早些時候，

as accurately as leading dermatologists.

史丹佛的一個研究者團隊 宣佈他們發展出了一個系統，

How does it work?

它能告訴你一個斑點是否為惡性的，

It's not trying to copy the judgment or the intuition of a doctor.

正確率不輸給頂尖皮膚科醫生。

It knows or understands nothing about medicine at all.

它怎麼做到的？

Instead, it's running a pattern recognition algorithm

它並不是嘗試複製 醫生的判斷或是直覺。

through 129,450 past cases,

它對於醫學是一竅不通。

hunting for similarities between those cases

反之，它進行的是模式辨識演算法，

and the particular lesion in question.

在 129,450 個個案當中，

It's performing these tasks in an unhuman way,

獵尋那些個案與欲探究的損害

based on the analysis of more possible cases

之間有哪些相似性。

than any doctor could hope to review in their lifetime.

它是用非人類的方式 在進行這些工作任務，

It didn't matter that that human being,

且是以大量案例的分析來當依據，

that doctor, couldn't explain how she'd performed the task.

案例數多到是醫生 一輩子都看不完的。

Now, there are those who dwell upon that the fact

無所謂人類，也就是醫生，

that these machines aren't built in our image.

是否能解釋她如何進行此工作任務。

As an example, take IBM's Watson,

有些人老是會想著

the supercomputer that went on the US quiz show "Jeopardy!" in 2011,

這些機器被建立時 沒有依循我們的形象。

and it beat the two human champions at "Jeopardy!"

以 IBM 的「華生 」為例，

The day after it won,

那是台超級電腦，2011 年參加 美國的益智節目《危險邊緣》，

The Wall Street Journal ran a piece by the philosopher John Searle

在節目中它打敗了兩位人類冠軍。

with the title "Watson Doesn't Know It Won on 'Jeopardy!'"

它獲勝之後的隔天，

Right, and it's brilliant, and it's true.

《華爾街日報》刊了一篇 哲學家約翰希爾勒的文章，

You know, Watson didn't let out a cry of excitement.

標題是〈華生不知道 它自己贏了《危險邊緣》 〉。

It didn't call up its parents to say what a good job it had done.

是的，這篇文章很聰明也沒說錯。

It didn't go down to the pub for a drink.

華生並沒有興奮地放聲大叫。

This system wasn't trying to copy the way that those human contestants played,

它沒有打電話給它的父母 說它的表現多棒。

but it didn't matter.

它沒有去酒吧喝酒慶祝。

It still outperformed them.

這個系統並沒有試圖複製 那些人類參賽者比賽的方式，

Resolving the intelligence myth

但那無所謂。

shows us that our limited understanding about human intelligence,

它仍然表現得比人類好。

about how we think and reason,

解開這個智慧迷思之後，

is far less of a constraint on automation than it was in the past.

看到的是雖然我們對於 人類智慧、對我們如何

What's more, as we've seen,

思考推理的方式了解有限，

when these machines perform tasks differently to human beings,

但這個限制對於自動化的影響 已經遠比過去小很多。

there's no reason to think

此外，如我們所見，

that what human beings are currently capable of doing

當這些機器用和人類不同的 方式來執行工作任務時，

represents any sort of summit

沒有理由認為

in what these machines might be capable of doing in the future.

人類目前能夠做到的事

Now the third myth,

就代表了一種上限，

what I call the superiority myth.

在未來機器能夠達成的事 都不可能超過這個上限。

It's often said that those who forget

第三項迷思，

about the helpful side of technological progress,

我稱之為優越迷思。

those complementarities from before,

常見的說法是，有些人會

are committing something known as the lump of labor fallacy.

忘記了科技進步的幫助面，

Now, the problem is the lump of labor fallacy

忘記過去的互補性，

is itself a fallacy,

這些人所犯的，就是 所謂的「勞動總合謬誤」。

and I call this the lump of labor fallacy fallacy,

問題是，勞動總合謬誤本身

or LOLFF, for short.

就是個謬誤，

Let me explain.

我把它稱為 「勞動總合謬誤的謬誤」，

The lump of labor fallacy is a very old idea.

簡寫為「LOLFF」。

It was a British economist, David Schloss, who gave it this name in 1892.

讓我解釋一下。

He was puzzled to come across a dock worker

勞動總合謬誤是個很古老的想法。

who had begun to use a machine to make washers,

這個名稱是 1892 年由英國 經濟學家大衛許洛斯取的。

the small metal discs that fasten on the end of screws.

有件事讓他百思不解， 他遇到一個碼頭工人，

And this dock worker felt guilty for being more productive.

這個工人開始用機器來製造墊圈，

Now, most of the time, we expect the opposite,

墊圈是小型的金屬圓盤， 固定在螺絲底端。

that people feel guilty for being unproductive,

這個碼頭工人對於自己的 高生產力有罪惡感。

you know, a little too much time on Facebook or Twitter at work.

通常，我們預期的是相反的反應，

But this worker felt guilty for being more productive,

生產力不高才會讓人感到罪惡，

and asked why, he said, "I know I'm doing wrong.

你知道的，工作時 花太多時間滑臉書或推特。

I'm taking away the work of another man."

但這個工人對於 太有生產力感到罪惡，

In his mind, there was some fixed lump of work

問他原因，他說：「我知道我做錯了。

to be divided up between him and his pals,

我搶走了另一個人的工作。」

so that if he used this machine to do more,

在他的認知中，勞動總合是固定的，

there'd be less left for his pals to do.

要由他和他的伙伴來分攤，

Schloss saw the mistake.

所以如果他用機器多做一點，

The lump of work wasn't fixed.

他伙伴能做的就變少了。

As this worker used the machine and became more productive,

許洛斯看到了這個錯誤。

the price of washers would fall, demand for washers would rise,

勞動總合並不是固定的。

more washers would have to be made,

當這個工人用機器提高生產力，

and there'd be more work for his pals to do.

墊圈的價格會下降， 對墊圈的需求會提高，

The lump of work would get bigger.

就得要做出更多的墊圈，

Schloss called this "the lump of labor fallacy."

他的伙伴反而會有更多要做。

And today you hear people talk about the lump of labor fallacy

勞動總合變更大了。

to think about the future of all types of work.

許洛斯稱之為「勞動總合謬誤」。

There's no fixed lump of work out there to be divided up

現今，在思考有各類工作的未來時，

between people and machines.

會聽到人們談到勞動總合謬誤。

Yes, machines substitute for human beings, making the original lump of work smaller,

沒有固定的勞動總合

but they also complement human beings,

要讓人類與機器瓜分。

and the lump of work gets bigger and changes.

是的，機器會取代人類， 讓原本的勞動總合變少，

But LOLFF.

但它們也會補足人類，

Here's the mistake:

勞動總合會變更大並且改變。

it's right to think that technological progress

但，LOLFF。

makes the lump of work to be done bigger.

錯誤是這樣的：

Some tasks become more valuable. New tasks have to be done.

認為科技進步會讓 要做的勞動總合變大，

But it's wrong to think that necessarily,

這點是沒錯的。

human beings will be best placed to perform those tasks.

有些工作任務變得較有價值。 有新工作任務需要完成。

And this is the superiority myth.

錯的地方在於，認為安排人類

Yes, the lump of work might get bigger and change,

來做那些工作任務一定是最好的。

but as machines become more capable,

這就是優越迷思。

it's likely that they'll take on the extra lump of work themselves.

是的，勞動總量可能 會變大也會改變，

Technological progress, rather than complement human beings,

但隨著機器變得更有能力，

complements machines instead.

很有可能它們會自己去接下 那些額外的勞動總量。

To see this, go back to the task of driving a car.

科技進步就不是在補足人類了，