Name: 印度AI高峰會落幕，出現小規模的「不服從」！ | DW 新聞 (India AI summit ends with a small show of defiance | DW News)
Uploaded: 2026-02-20T15:43:17.000Z
Duration: 7 min 54 s

基礎被動語態

India's Artificial Intelligence Impact Summit wrapped up today with what was meant to be a show of unity.

印度的人工智慧影響高峰會今天落幕，原本是要展現團結。

Prime Minister Narendra Modi asked the world's top AI bosses to raise their hands together on the stage.

總理莫迪要求全球頂尖的 AI 業者在臺上舉起手。

大家都有舉，只有兩個人例外。

There you can see them standing side by side.

你可以看到他們並肩站在一起。

OpenAI Sam Altman and Anthropic's Dario Emodi raised their fists but didn't hold hands.

OpenAI 的 Sam Altman 和 Anthropic 的 Dario Emodi 舉起了拳頭，但沒有握手。

It's just a small moment capturing the AI power struggle behind the summit's best big promises.

這只是個小小的瞬間，捕捉到了高峰會最美好的大承諾背後，AI 權力鬥爭的縮影。

That moment came as India launched the New Delhi Frontier AI commitments.

這個瞬間發生之際，印度推出了新德里邊境 AI 承諾。

That's a set of voluntary principles, not laws, meant to show that companies building the world's most powerful AI systems can act responsibly and work with governments on safety and oversight.

這是一套自願性原則，不是法律，旨在表明那些開發全球最強大 AI 系統的公司，可以負責任地行事，並與政府合作處理安全和監督事宜。

At the summit, he said the world urgently needs regulation to manage the risks of artificial intelligence.

在高峰會上，他說世界迫切需要法規來管理人工智慧的風險。

For a democratic AI future, it is not enough to just give people tools and wealth.

為了民主 AI 的未來，光是給人們工具和財富是不夠的。

We also need to give them agency and power.

我們也需要賦予他們自主權和力量。

Of course, this is not to suggest that we won't need any regulation or safeguards.

當然，這不是說我們不需要任何法規或保障措施。

We obviously do urgently like we have for other powerful technologies.

我們顯然迫切需要，就像我們對待其他強大技術一樣。

Kathleen Muller is president and co founder of Ally, an independent organization that promotes responsible development, deployment and use of AI.

Kathleen Muller 是 Ally 的總裁兼共同創辦人，Ally 是一個推動 AI 的負責任開發、部署和使用的獨立組織。

She has also advised a number of European institutions on AI and human rights.

她也曾就 AI 與人權向許多歐洲機構提供諮詢。

Kathleen，歡迎來到今天的節目。

Now, this summit saw voluntary commitments at the same time as calls for greater regulation are both needed because firms just cannot be trusted.

現在，這次高峰會看到了自願性承諾，同時也有人呼籲加強法規，兩者都是必要的，因為公司根本無法被信任。

Well, I was kind of surprised to say to see this, this, this voluntary commitments.

嗯，我有點驚訝看到這些自願性承諾。

They're not really commitments towards responsible AI.

這並不是對負責任 AI 的真正承諾。

They are about gathering statistical information, which is obviously interesting and very relevant, informed policymaking.

這些是關於收集統計資訊，這顯然很有趣，而且對知情的政策制定非常有幫助。

And they're about multilingual alums, which is also very good.

而且它們是關於多語言校友，這也非常好。

But that's not a commitment towards responsible AI.

但這並不是對負責任 AI 的承諾。

And as to the request for regulation, to be honest, I find that quite rich because we have a very strong piece of regulation here in, in the EU we have the AI Act.

至於對法規的要求，老實說，我覺得這有點諷刺，因為我們在歐盟這裡有非常嚴格的法規，也就是 AI Act。

Altman, who asked for the regulation again, together with all his technology colleagues, big tech colleague, are now lobbying Europe to water down the regulation that we already have.

他再次要求法規，同時他和所有科技界的同事，這些大型科技公司的同事，現在正在遊說歐洲削弱我們已經有的法規。

So if you on the one hand ask for regulation and the other try to burn it down, I cannot really take that seriously.

所以，如果你一方面要求法規，另一方面又試圖把它搞垮，我真的無法認真看待。

So what kind of regulation do you think Mr. Altman would like to see?

那麼，你認為 Altman 先生會希望看到哪種法規呢？

I would like to ask Mr. Altman to look at what kind of regulation is already out there.

我會請 Altman 先生看看現在已經有哪些法規了。

We have the AI act, which is an example which can be used as a blueprint.

我們有《AI 法案》，這是一個可以當作藍圖的例子。

But there's also an AI treaty that was developed by the Council of Europe and that is open to all signatories.

但歐洲理事會也制定了一份《AI 條約》，開放給所有簽署國。

And in fact, the United States was present in the room, was actually part of the negotiating group of that treaty.

事實上，美國當時也在場，也是這個條約協商小組的一員。

So I mean, Mr. Altman can go to his government, say, look, this is the treaty.

所以，Altman 先生可以去找他的政府，說，看看，這就是這個條約。

It might not be perfect, we might not be there yet, but it is a start.

也許它並不完美，我們也許還沒達到那個階段，但這是一個開始。

I want to talk about this image of Sam Altman and Daria Modi not holding hands, which has gone viral since this, since the summit.

我想談談 Sam Altman 和 Daria Modi 沒有牽手的這張照片，這張照片自從這次峰會以來就瘋傳了。

From a governance perspective, does the rivalry between the leading companies make self regulation unrealistic at this point in time?

從治理的角度來看，領先公司之間的競爭是否讓現階段的自我監管變得不切實際？

Well, we have examples of self regulation.

I'm not, I'm not really sure what's behind it.

We have the ethics guidelines for trustworthy, which is the best piece of self regulation I think that we have.

我們有「值得信賴」的道德準則，我認為這是我們擁有的最好的自我監管。

Also from Europe, we have a code of practice for general purpose AI that was signed by OpenAI, by Microsoft, by all the big tech companies and a couple of others.

另外，歐洲方面，我們有一個通用人工智慧的行為準則，由 OpenAI、Microsoft、所有大型科技公司以及其他幾家公司簽署。

And what surprises me a bit is that big tech companies are asking for regulations like they cannot help themselves.

讓我有點驚訝的是，大型科技公司竟然在尋求法規，好像他們無法自拔一樣。

Please regulate us, otherwise we'll burn down the house.

請規範我們，否則我們就會燒毀房子。

I mean, if you, if you don't know how to not burn down the house, you shouldn't be playing with the matches.

我的意思是，如果你們不知道如何不燒毀房子，就不應該玩火柴。

I think democratization of a I was a huge topic as well.

我認為 AI 的民主化也是一個很大的議題。

In your opinion, who should ultimately have agency over these systems?

在你認為，最終應該由誰來掌控這些系統？

The companies, from what I'm hearing you say, shouldn't be it, should it be governments or the public?

從你的話裡聽起來，公司不應該是掌控者，應該是政府或公眾嗎？

And how could that be designed and practice?

It's also a very difficult question because there's a lot of artificial intelligence tools that are very useful and that are not dangerous at all.

這也是個很難的問題，因為有很多 AI 工具非常有用，而且一點也不危險。

So this don't have to be in the hands of governments, but

所以這些不一定要由政府掌握，但是

if you look at the systems that have profound effects on society, profound effects on labor markets, profound effects on our safety, on our fundamental rights, you might want to think about having more oversight on those tools.

如果你看看那些對社會、勞動市場、我們的安全、我們的基本權利有深遠影響的系統，你可能會想對這些工具進行更多監督。

The first step that you need to take is regulation.

你需要採取的第一個步驟是法規。

And regulation needs teeth because self regulation doesn't have teeth.

法規需要有約束力，因為自我約束是沒有約束力的。

Voluntary codes of practice don't have teeth.

And this is why I really want to stress that now in Europe, there's a process of watering down the AI Act.

這也是為什麼我真的很想強調，現在歐洲正在經歷一個稀釋 AI 法案的過程。

So the regulation that has not even entered into force fully yet, and that is a dangerous path.

所以，法規甚至還沒有完全生效，而這是一條危險的道路。

We should try out this regulation, see where it goes, and

我們應該試行這項法規，看看它的走向，然後

then start talking about, okay, maybe there are some tools that we need to bring into the hands of the public or of governments because they are too dangerous.

再開始討論，好吧，也許有些工具我們需要交給公眾或政府掌握，因為它們太危險了。

I want to look before we wrap up again at the role of companies here, because you raise a very interesting point there.

在我們再次總結之前，我想探討一下這些公司扮演的角色，因為你提出了個非常有趣的觀點。

Does Sam Altman, does Dario Amadei, do all of these CEOs of these very powerful AI companies really know the power of the tools they are putting out there and how to rein them in if need be?

Sam Altman、Dario Amadei，這些強大 AI 公司的 CEO 們，他們真的知道自己推出的工具有多強大，以及必要時如何約束它們嗎？

And if they don't, they should not build them, they should not roll them out over society without any proper testing, because that is what's happening right now.

如果他們不知道，就不應該開發它們，不應該在沒有適當測試的情況下就推向社會，因為這正是現在正在發生的事。

It's a bit like we've discovered that cocaine makes you happy and the next day the doctor starts prescribing it for treating depression.

這有點像是我們發現古柯鹼能讓人開心，隔天醫生就開始開立它來治療憂鬱症。

And what we see now is these companies, and

而我們現在看到的是，這些公司，還有

maybe to your earlier questions, rivalry between these two men and maybe between also the other men, this rivalry might make them maybe less open to look at those things and try to fight for, well, being the fastest, being the earliest, being the first mover.

也許是回應你先前的問題，這兩個人之間的競爭，還有其他人的競爭，可能會讓他們比較不願意去看看這些事情，然後爭奪，嗯，成為最快的、最早的、第一個採用的。

And with this technology, that's a dangerous path.

而有了這項技術，這是一條危險的路。

That was Kathleen Moeller, president and co founder of alai.

這是 alai 的總裁兼共同創辦人 Kathleen Moeller。

很高興今天能邀請到你上節目。