How do we learn to work with intelligent machines? | Matt Beane

譯者: Val Zhang 審譯者: Melody Tang

It’s 6:30 in the morning,

凌晨六點半，

and Kristen is wheeling her prostate patient into the OR.

克莉絲汀推著她的 前列腺病人進入手術室。

She's a resident, a surgeon in training.

她是住院醫師—— 培訓中的外科醫師。

It’s her job to learn.

學習是她的義務。

Today, she’s really hoping to do some of the nerve-sparing,

今天她非常希望能參與 部分的雙側神經保留手術，

extremely delicate dissection that can preserve erectile function.

需要精湛的手術技巧， 讓病人保有勃起的功能。

That'll be up to the attending surgeon, though, but he's not there yet.

不過，這還要看主治醫師的意思， 而他還沒有到場。

She and the team put the patient under,

克莉絲汀和團隊給病人打了麻醉，

and she leads the initial eight-inch incision in the lower abdomen.

她在病人的下腹部開了 第一道 8 英吋的切口。

Once she’s got that clamped back, she tells the nurse to call the attending.

當她把切口夾好， 她請護理師打電話給主治醫師。

He arrives, gowns up,

主治醫師到場，穿上手術服，

And from there on in, their four hands are mostly in that patient --

接著，他們的四隻手 就大都在病人體內，

with him guiding but Kristin leading the way.

在主治醫師的指導下， 由克莉絲汀操作。

When the prostates out (and, yes, he let Kristen do a little nerve sparing),

當病人的前列腺被取出後，

（太好了！他讓她做了 部分神經保留手術。）

he rips off his scrubs.

主治醫師脫掉了手術服，

He starts to do paperwork.

開始填寫資料。

Kristen closes the patient by 8:15,

而克莉絲汀在八點十五分完成了手術，

with a junior resident looking over her shoulder.

一位資淺的住院醫師在旁觀摩學習。

And she lets him do the final line of sutures.

她讓他為病人做最後的縫合。

Kristen feels great.

克莉絲汀感覺好極了！

Patient’s going to be fine,

病人應該很快就會恢復，

and no doubt she’s a better surgeon than she was at 6:30.

無疑地，比起做這個手術前， 她是一位更好的外科醫師。

Now this is extreme work.

這是種極端的工作。

But Kristin’s learning to do her job the way that most of us do:

但克莉絲汀邊做邊學的方式 和我們大多數人無異：

watching an expert for a bit,

觀察專家如何操作，

getting involved in easy, safe parts of the work

從簡單、安全的部分開始著手，

and progressing to riskier and harder tasks

然後在專家的指導和確認合格之下， 接手風險更高、難度更大的工作。

as they guide and decide she’s ready.

我一直對這樣的學習過程感到著迷。

My whole life I’ve been fascinated by this kind of learning.

It feels elemental, part of what makes us human.

我覺得這似乎是人類 之所以為人類的基本要素之一。

It has different names: apprenticeship, coaching, mentorship, on the job training.

人們為這過程賦予不同的名字： 學徒、訓練、師徒制、在職訓練。

In surgery, it’s called “see one, do one, teach one.”

外科稱之為 「看一次、做一遍、教一位」，

But the process is the same,

但過程是一樣的，

and it’s been the main path to skill around the globe for thousands of years.

這也是數千年來， 全球在培養人才時運用的方式。

Right now, we’re handling AI in a way that blocks that path.

現在我們應用人工智慧的方式 阻礙了這條學習路徑。

We’re sacrificing learning in our quest for productivity.

為了更高的生產率， 我們犧牲了在工作中學習的機會。

I found this first in surgery while I was at MIT,

我最初在麻省理工學院的手術 發現了這一個現象，

but now I’ve got evidence it’s happening all over,

但現在，我發現這個情況隨處可見，

in very different industries and with very different kinds of AI.

遍佈各行各業， 應用著各種人工智慧的技術。

If we do nothing, millions of us are going to hit a brick wall

如果我們對此不做出改變，

在我們學著面對人工智慧技術時， 成千上萬的人將受挫。

as we try to learn to deal with AI.

Let’s go back to surgery to see how.

讓我們回到外科手術作為例子，

Fast forward six months.

時間快轉六個月，

It’s 6:30am again, and Kristen is wheeling another prostate patient in,

同樣是凌晨六點半，克莉絲汀推著 另一位前列腺病人進來，

but this time to the robotic OR.

但這一次，病人被推到機器人手術室，

The attending leads attaching

由主治醫師主導著 把一個有著四支手臂、

a four-armed, thousand-pound robot to the patient.

重一千磅的機器人， 連接到病人身上。

They both rip off their scrubs,

他們都脫下了手術服，

head to control consoles 10 or 15 feet away,

來到 10 ~ 15 英尺外的控制台，

and Kristen just watches.

而克莉絲汀只能旁觀。

The robot allows the attending to do the whole procedure himself,

在機器人的幫助下， 主治醫師一個人便可完成手術，

so he basically does.

他基本上也這麼做，

He knows she needs practice.

他知道克莉絲汀需要練習，

He wants to give her control.

他也希望可以讓她主導，

But he also knows she’d be slower and make more mistakes,

但是他同樣清楚她會比較慢， 可能會有失誤，

and his patient comes first.

而他的病人第一。

So Kristin has no hope of getting anywhere near those nerves during this rotation.

在這次手術中，克莉絲汀 沒有任何機會接近那些神經，

She’ll be lucky if she operates more than 15 minutes during a four-hour procedure.

在長達四小時的手術中， 若她能操作十五分鐘就算幸運了。

And she knows that when she slips up,

她知道一旦她有失誤，

he’ll tap a touch screen, and she’ll be watching again,

他只要輕敲螢幕， 她又回到旁觀的角色，

feeling like a kid in the corner with a dunce cap.

感覺像個戴笨蛋高帽的 孩子在角落裡罰站。

Like all the studies of robots and work I’ve done in the last eight years,

就像這八年來我所做的 有關機器人與工作的研究，

我以一個重要且有爭議的問題開始：

I started this one with a big, open question:

How do we learn to work with intelligent machines?

我們如何學習與智慧型機器共事呢？

To find out, I spent two and a half years observing dozens of residents and surgeons

為了找出答案， 我花了兩年半的時間

觀察數十位做傳統與機器人手術的 住院醫師和外科醫師，

doing traditional and robotic surgery, interviewing them

訪問他們，

and in general hanging out with the residents as they tried to learn.

基本上，當他們在學習的時候， 我和他們混在一起。

I covered 18 of the top US teaching hospitals,

我涵蓋了十八所美國頂尖的教學醫院，

and the story was the same.

故事都是相同的。

Most residents were in Kristen's shoes.

絕大多數的住院醫師的 處境跟克莉斯汀一樣。

They got to “see one” plenty,

他們有很多「看一次」的機會，

but the “do one” was barely available.

但「做一遍」的機會少之又少。

So they couldn’t struggle, and they weren’t learning.

他們沒有機會受挫， 也沒能進一步學習。

This was important news for surgeons, but I needed to know how widespread it was:

這對外科醫師來說是很重要的消息， 但我想知道這個情況擴散的程度。

Where else was using AI blocking learning on the job?

哪些產業也在運用人工智慧時， 阻礙了做中學呢？

To find out, I’ve connected with a small but growing group of young researchers

為了找出答案，我聯繫了一群 小而成長中的年輕研究者，

who’ve done boots-on-the-ground studies of work involving AI

他們腳踏實地研究過人工智慧

in very diverse settings like start-ups, policing,

在不同的領域的運用， 包括：新創、警政、

investment banking and online education.

投資銀行和線上教育。

Like me, they spent at least a year and many hundreds of hours observing,

像我一樣，他們花了至少一年， 以及數百個小時

觀察、訪問，並經常和他們 研究的對象肩並肩一起工作。

interviewing and often working side-by-side with the people they studied.

We shared data, and I looked for patterns.

我們分享了數據，而我觀察模式。

No matter the industry, the work, the AI, the story was the same.

不論是哪種產業、工作、人工智慧， 故事都是相同的。

Organizations were trying harder and harder to get results from AI,

所有組織都非常努力地 運用人工智慧以得到更好的成果，

and they were peeling learners away from expert work as they did it.

在這過程中，他們剝奪了 學徒習做專家工作的機會。

Start-up managers were outsourcing their customer contact.

新創公司的經理人 外包他們的客服窗口。

Cops had to learn to deal with crime forecasts without experts support.

警察在沒有專家的協助下， 學著做犯罪預測。

Junior bankers were getting cut out of complex analysis,

資淺的銀行家無法 接觸到複雜的分析，

and professors had to build online courses without help.

而教授得在沒有幫助的狀況下 打造線上課程。

And the effect of all of this was the same as in surgery.

所有這些帶來的影響 跟手術的情形是一樣的。

Learning on the job was getting much harder.

做中學變得越來越困難。

This can’t last.

這情況不能繼續下去。

McKinsey estimates that between half a billion and a billion of us

麥肯錫顧問公司估計：

大約有五～十億人在 2030 年以前，

are going to have to adapt to AI in our daily work by 2030.

必須將人工智慧應用在日常工作中。

我們以為當我們在做的時候， 會有在職訓練。

And we’re assuming that on-the-job learning

will be there for us as we try.

Accenture’s latest workers survey showed that most workers learned key skills

埃森哲顧問公司最新的工作調查顯示：

多數人透過做中學得到工作的 關鍵技能，而非透過正式的訓練。

on the job, not in formal training.

So while we talk a lot about its potential future impact,

當我們高談闊論人工智慧 對未來的潛在衝擊，

the aspect of AI that may matter most right now

此時此刻，人工智慧 對我們最重要的影響是

is that we’re handling it in a way that blocks learning on the job

我們應用人工智慧的方式， 阻礙了人們邊做邊學的機會。

just when we need it most.

而那是我們最需要學習的時候。

Now across all our sites, a small minority found a way to learn.

在我們的研究對象中， 有一小群人找到一種學習方式。

They did it by breaking and bending rules.

他們透過打破常規來學習。

Approved methods weren’t working, so they bent and broke rules

被准許的作法不可行， 所以他們改變了遊戲規則，

to get hands-on practice with experts.

才得以與專家一同實際操作。

In my setting, residents got involved in robotic surgery in medical school

我看到的是，住院醫師 在醫學院裡為了參與機器人手術，

at the expense of their generalist education.

犧牲了上全科醫師的課為代價，

And they spent hundreds of extra hours with simulators and recordings of surgery,

他們多花了數百個小時 使用模擬器與看手術錄影來學習。

when you were supposed to learn in the OR.

那是他們本來應當 在手術室裡學習的。

And maybe most importantly, they found ways to struggle

或許更重要的，

他們在有限的專家指導下， 找到在實際的手術中練習的機會。

in live procedures with limited expert supervision.

I call all this “shadow learning,” because it bends the rules

我稱之為「在陰影中學習」 因為這違反了規則，

and learner’s do it out of the limelight.

學生得要偷偷摸摸地學習。

And everyone turns a blind eye because it gets results.

大家對此睜一隻眼閉一隻眼， 因為這樣的確有效。

Remember, these are the star pupils of the bunch.

但記住，這些僅是少數的明星學生。

Now, obviously, this is not OK, and it’s not sustainable.

顯然，這樣並不恰當， 這不是長久之計。

No one should have to risk getting fired

沒有人應該冒著被開除的風險，

to learn the skills they need to do their job.

去學習他們工作必要的技巧。

But we do need to learn from these people.

但我們必須從這些人身上學習。

They took serious risks to learn.

他們為了學習而承擔高度風險。

They understood they needed to protect struggle and challenge in their work

他們明白必須保護工作中 受挫與挑戰的機會，

so that they could push themselves to tackle hard problems

以推動他們自己去挑戰

比他們的能力能解決的 更困難的問題。

right near the edge of their capacity.

They also made sure there was an expert nearby

他們也確保會有一個專家在旁，

to offer pointers and to backstop against catastrophe.

提供建議跟收拾殘局， 以防他們搞砸了。

Let’s build this combination of struggle and expert support

讓我們設計 在導入每項人工智慧時

into each AI implementation.

加入學習機會及專家協助的組合。

Here’s one clear example I could get of this on the ground.

這裡有個清楚的案例， 我能在現實中找到。

Before robots,

在機器人出現之前，

if you were a bomb disposal technician, you dealt with an IED by walking up to it.

如果你是個拆彈技術專家 面對簡易爆炸裝置時，你得走近它。

A junior officer was hundreds of feet away,

一個資淺的警官在數百公尺外支援，

so could only watch and help if you decided it was safe

他可以觀察並協助，

直到你確定裝置是安全的 並邀請他們到近距離。

and invited them downrange.

Now you sit side-by-side in a bomb-proof truck.

現在你們肩並肩地坐在防彈車裡。

You both watched the video feed.

你們一同觀看機器人傳來的影片資訊。

They control a distant robot, and you guide the work out loud.

資淺者控制著遠端機器人， 而你大聲引導著作業。

Trainees learn better than they did before robots.

受訓者的學習效果 比機器人出現之前更佳。

We can scale this to surgery, start-ups, policing,

我們可以按比例複製這樣的模式 到手術、新創公司、警政、

investment banking, online education and beyond.

投資銀行、線上教育， 以及更多產業。

The good news is we’ve got new tools to do it.

好消息是我們有新的工具去執行。

The internet and the cloud mean we don’t always need one expert for every trainee,

網路跟雲端代表著我們不再需要 一對一的師徒制，

for them to be physically near each other or even to be in the same organization.

他們不再需要去到同一個空間 甚至在不同的組織單位裡。

And we can build AI to help:

我們可以打造人工智慧來協助。

to coach learners as they struggle, to coach experts as they coach

當學生困惑時教導他們， 當專家指導學生時協助專家，

and to connect those two groups in smart ways.

並以聰明的方式聯繫這兩群人。

There are people at work on systems like this,

有人正在開發這樣的系統，

but they’ve been mostly focused on formal training.

但他們大多專注於正式的訓練。

And the deeper crisis is in on-the-job learning.

但更深的危機是工作做中學的部分。

We must do better.

我們必須做得更好。

Today’s problems demand we do better

今天所面臨的挑戰促使我們要更好地

to create work that takes full advantage of AI’s amazing capabilities

創造應用人工智慧無限潛能的工作，

while enhancing our skills as we do it.

同時讓我們在工作時 也能加強我們的技能。

That’s the kind of future I dreamed of as a kid.

這是我從孩提時代以來 一直有的夢想。

And the time to create it is now.

現在就是創造它的時刻。

Thank you.

謝謝大家。

(Applause)

（掌聲）