Name: 誰能在新AI時代生存並蓬勃發展？！ (Who will survive and thrive in the new AI era?)
Uploaded: 2026-02-16T21:36:53.000Z
Duration: 11 min 27 s

程度副詞

Almost all the conflicts in the AI era are between competing groups of people, not between machines versus people.

AI 時代幾乎所有的衝突，都是人與人之間的競爭，而不是機器對抗人類。

David Autor is professor of Economics at MIT and the co director of the Labor Studies Program at the National Bureau of Economics Research.

David Autor 是 MIT 的經濟學教授，也是國家經濟研究局勞工研究計畫的共同主任。

He's also emerged as one of the world's leading experts on the impacts of AI on the American worker.

他也成為全球頂尖專家之一，專門研究 AI 對美國勞工的影響。

He spoke to us from MIT's campus in Cambridge, Massachusetts.

他從麻州劍橋的 MIT 校園跟我們分享。

The interests of the creators may very well not be the interests of the end users.

創作者的利益，可能跟終端使用者的利益完全不一樣。

We see this with social media, for example, all the time.

像是社群媒體，我們就常常看到這種情況。

And so we shouldn't pretend that this is an engineering problem.

所以我們不應該假裝這是一個工程問題。

The problems Otter refers to are at the center of a new series of essays featuring authors like Google's Eric Schmidt and OpenAI CFO Sarah Fryer.

Otter 提到的問題，是新一系列論文的核心，這些論文的作者包括 Google 的 Eric Schmidt 和 OpenAI 的財務長 Sarah Fryer。

Like, over the next couple of months, Stanford professor Eric Brynjolfsson helped organize the project, which they call the Digitalist papers.

像是接下來這幾個月，史丹佛大學的 Eric Brynjolfsson 教授協助組織了這個專案，他們稱之為 Digitalist 論文集。

We made a list of people that we thought would be ideal to write about the economic implications.

我們列出了一些我們認為最適合撰寫經濟影響相關內容的人選。

And we had sort of our A list, our B list, our C list, and

然後我們有 A 級、B 級、C 級名單，

我們從來沒用到 B 級或 C 級名單。

All of these 22 essays address things that could be really profound.

這 22 篇論文都探討了可能非常深遠的議題。

But if you were forced, what are the three or the four that you think really deserve the most attention right now?

但如果非得選，你認為哪三到四篇是現在最值得關注的呢？

Oh, you're asking me to choose between my children, but I'll pick out a few.

噢，你這是要我選我的孩子們，但我還是會挑幾篇。

I mean, in the beginning of the volume, we have some people setting the technological landscape.

我的意思是，在論文集的開頭，我們有一些人描繪了技術的發展藍圖。

I thought Eric Schmidt did a great job describing what he calls the San Francisco Consensus.

我認為 Eric Schmidt 精彩地描述了他所說的「舊金山共識」。

Later in the volume, people like David Otter and Neil Thompson talk about how it's changing expertise and some of the sometimes counterintuitive effects that the same technology may upskill one occupation while deskilling another occupation, depending on how that occupation was configured previously.

書的後半段，David Otter 和 Neil Thompson 等人談到這項技術如何改變專業知識，以及同樣的技術可能會讓一種職業加強技能，同時卻讓另一種職業降低技能，這取決於該職業之前的配置方式，其中有些效果是違反直覺的。

Otter says that artificial intelligence is already reshaping the value of human expertise.

Otter 說，人工智慧已經在重塑人類專業知識的價值。

And in his paper, he lays out a new economic framework to understand which.

在他的論文中，他提出了一個新的經濟框架來理解這一點。

Which jobs will be affected by AI and how it's very natural simply to think about what will be automated.

哪些工作會受到 AI 的影響，以及如何影響，我們很自然地就會想到哪些工作會被自動化。

And, you know, you think, oh, what is exposed?

然後，你會想，哦，哪些東西會暴露出來？

Whatever is exposed to technology will shrivel up and die.

任何暴露在科技之下都會枯萎死亡。

But, you know, if you think about it for a minute, you know, it can't be quite right because, you know, we all use lots and lots of technology, and

但是，如果你仔細想想，這不可能完全正確，因為我們都使用了大量的科技，而且

we're not primarily threatened by it, right?

我們並沒有主要受到它的威脅，對吧？

We're very happy to have the GPS on our phone, to have our personal computer, and

我們很高興手機上有 GPS，有個人電腦，而且

But of course, we also know there's some risks.

不過當然，我們也知道有些風險。

所以有沒有所謂的「好的」暴露？

And the way we frame that question is in terms of expertise.

而我們提出這個問題的方式，是從專業知識的角度來看。

Some of them are what you're essentially really good at.

You're a great researcher, you're a fantastic interviewer, reporter, and

你是個很棒的研究員，是個很棒的訪談者、記者，而且

then there's lots of other parts of your job that are kind of necessary but not really important or essential.

還有很多工作上其他的部分，雖然是必要的，但並不是那麼重要或不可或缺。

If that supporting work is automated by AI or any other technology, you're happy.

如果這些支援性的工作被 AI 或任何其他科技自動化了，你會很高興。

You're like, oh, great, now I can really focus on what I'm good at.

你會覺得，哦，太好了，現在我可以真正專注在我擅長的事情上了。

It makes your expertise more valuable, allows you to specialize, to exercise your comparative advantage and strip away all the time.

讓你的專業更有價值，讓你能夠專精，發揮你的比較優勢，並且省下所有時間。

On the other hand, if technology can suddenly do the thing that you are most good at, research at, reporting at skilled repair, medical diagnosis, then it's really competing with you.

另一方面，如果科技突然能做好你最擅長、最會研究、最會報導、最會修理、最會診斷的事情，那它就是在跟你競爭了。

Otter says the key lies in what tasks within a job are being automated.

Otter 表示，關鍵在於工作中哪些任務被自動化了。

Are they high expertise or low for lawyers, if AI takes over the relatively low expertise job of drafting routine contracts, it may mean that you need fewer lawyers.

是高專業度還是低專業度的任務？對律師來說，如果 AI 接管了起草例行合約這種相對低專業度的任務，可能意味著你需要更少的律師。

But those that are left get paid more because the high expertise part of what they do is even more valuable.

但留下來的人薪水會更高，因為他們工作中的高專業度部分變得更有價值。

Those who remain will be very specialized, right?

留下來的人會非常專精，對吧？

The language translators who remain working, they will be doing diplomatic documents, legal documents.

留下來的語言翻譯，他們會處理外交文件、法律文件。

As technology encroaches, the work that remains will be more specialized, higher paid, but

隨著科技的滲透，留下來的工作會更專精、薪水更高，但

On the other side of the coin, what happens when the high expertise tasks get automated?

那換個角度想，高度專業的任務被自動化後，會發生什麼事？

The barriers to entry fall, employment goes up, wages go down.

進入門檻降低，就業增加，薪資下降。

Think of how ride hailing software changed the taxi driving industry.

想想看叫車軟體如何改變了計程車行業。

So taxi driving historically had two components to it.

所以計程車駕駛這份工作，過去有兩個主要部分。

One was driving a car, which something most people can do, and

一個是開車，這大部分人都做得到，

the other was knowing your way around, which something most people did not do, could not have.

另一個是熟悉路況，這大部分人做不到，或沒有辦法。

I certainly don't know my way around anywhere, even where I've lived.

我自己就路痴，就算住很久的地方，也認不清路。

And ride hailing, sure, it introduces this interface for calling cars and so on, but

而共乘叫車服務，確實引入了叫車的介面等等，但

Suddenly, to be a taxi or chauffeur driver, you don't need to know your way around.

突然之間，要當計程車司機或駕駛，你不需要熟悉路況。

You can go to any city that you don't live in and drive Uber immediately because the phone tells you what to do.

你可以去任何你不住的城市，立刻開 Uber，因為手機會告訴你該怎麼做。

So this created all kinds of new competition.

It commodified the expertise of knowing your way around the place.

它將熟悉路況的專業知識商品化了。

I think the evidence suggests that it lowered wages of people in the taxi driving industry.

我認為證據顯示，這降低了計程車駕駛業者的薪資。

Incumbents did not like it and understandably so.

現有的業者當然不喜歡，這可以理解。

So with this analysis, what is the most likely consequence for the labor market overall?

所以，有了這個分析，對整個勞動市場最可能的影響是什麼？

There are many, many signs that we have long term labor shortage, so we're not going to run out of jobs.

有很多跡象顯示，我們有長期的勞動力短缺，所以工作不會消失。

But that doesn't mean there's nothing to worry about.

A world in which everyone is working as a crossing guard is very different from a world in which everyone is working as a medical doctor.

一個每個人都當交通管理員的世界，跟一個每個人都當醫生的世界，是非常不一樣的。

In the first case, you know, people are relatively interchangeable.

在第一種情況下，你知道，大家基本上是可以互相替代的。

Most people can do that work and so that's not specialized expertise.

大多數人都能做這份工作，所以這不是什麼專業技能。

There might be a lot of people doing it, but, you know, employers won't have much incentive to treat them well.

可能會有很多人做這份工作，但你知道，僱主不太會有動力去好好對待他們。

On the other hand, if everyone isn't, you know, doing specialized work where they, you know, what they do is, you know, high stakes and it really adds a lot of value.

另一方面，如果大家做的不是專業工作，你知道，他們做的事情，你知道，風險很高，而且真的能創造很多價值。

When you talk about what automation has done thus far, sort of pushing people to the extremes, you know, yeah, higher, higher pay, fewer people or lower, lower pay, lots of people.

談到到目前為止自動化做了什麼，有點像是把人們推向極端，你知道，對，更高、更高的薪水，更少的人；或者更低、更低的薪水，更多的人。

Is it indeterminate what I could do to inequality?

One is it's, it's determinate, but I just don't know.

一種是，它是確定的，只是我不知道。

I think a better answer, the one that I believe in, is we have a lot of choice about this.

我認為一個更好的答案，我所相信的，是我們有很多選擇。

Choice is a theme that comes up often in the digitalist papers.

選擇是 Digitalist 論文集中經常出現的主題。

And Stanford's Brynjolfsson says making the right choice depends on having the right information.

史丹佛大學的 Brynjolfsson 說，做出正確的選擇取決於擁有正確的資訊。

Given your sophisticated understanding of transformative AI, are you more confident about the upside from AI today than you were a year ago?

鑑於你對轉型性 AI 的精闢理解，你對今日 AI 的潛在效益是否比一年前更有信心？

I'm confident that the capabilities are breathtaking and they're going to get better.

我相信這些能力令人驚嘆，而且會越來越好。

I'm confident that the productivity benefits are there to be had, but

我相信生產力效益是存在的，但是

I'm worried that we're not going to do it in a way that create widely shared prosperity.

我擔心我們無法以創造廣泛共享繁榮的方式來實現。

But ultimately there's no predetermined future.

There are many, many ways for us to use AI.

我們有很多很多方法可以運用 AI。

It's an incredibly flexible, malleable plastic technology.

它是一項極其靈活、可塑的技術。

You could use it to try to automate people out of existence.

你可以用它來試圖自動化，讓人們消失。

You could also use it to collaborate with people to make them more effective.

你也可以用它來與人們合作，讓他們更有效率。

But I also think that it depends on how we invest, how we build out those technologies.

但我也認為這取決於我們如何投資，如何建構這些技術。

China runs the world's largest surveillance state, the world's largest real time censorship apparatus, and

中國經營著世界上最大的監控國家、世界上最大的即時審查機制，

字幕列表影片播放

誰能在新AI時代生存並蓬勃發展？！ (Who will survive and thrive in the new AI era?)

scenario

opportunity

alternative

progress