Afrikaans
Akan
Albanian
Amharic
Arabic
Armenian
Azerbaijani
Basque
Belarusian
Bemba
Bengali
Bihari
Bosnian
Breton
Bulgarian
Cambodian
Catalan
Cebuano
Cherokee
Chichewa
Chinese (Simplified)
Chinese (Traditional)
Corsican
Croatian
Czech
Danish
Dutch
English
Esperanto
Estonian
Ewe
Faroese
Filipino
Finnish
French
Frisian
Ga
Galician
Georgian
German
Greek
Guarani
Gujarati
Haitian Creole
Hausa
Hawaiian
Hebrew
Hindi
Hmong
Hungarian
Icelandic
Igbo
Indonesian
Interlingua
Irish
Italian
Japanese
Javanese
Kannada
Kazakh
Kinyarwanda
Kirundi
Kongo
Korean
Krio (Sierra Leone)
Kurdish
Kurdish (Soranî)
Kyrgyz
Laothian
Latin
Latvian
Lingala
Lithuanian
Lozi
Luganda
Luo
Luxembourgish
Macedonian
Malagasy
Malay
Malayalam
Maltese
Maori
Marathi
Mauritian Creole
Moldavian
Mongolian
Myanmar (Burmese)
Montenegrin
Nepali
Nigerian Pidgin
Northern Sotho
Norwegian
Norwegian (Nynorsk)
Occitan
Oriya
Oromo
Pashto
Persian
Polish
Portuguese (Brazil)
Portuguese (Portugal)
Punjabi
Quechua
Romanian
Romansh
Runyakitara
Russian
Samoan
Scots Gaelic
Serbian
Serbo-Croatian
Sesotho
Setswana
Seychellois Creole
Shona
Sindhi
Sinhalese
Slovak
Slovenian
Somali
Spanish
Spanish (Latin American)
Sundanese
Swahili
Swedish
Tajik
Tamil
Tatar
Telugu
Thai
Tigrinya
Tonga
Tshiluba
Tumbuka
Turkish
Turkmen
Twi
Uighur
Ukrainian
Urdu
Uzbek
Vietnamese
Welsh
Wolof
Xhosa
Yiddish
Yoruba
Zulu
Now, the present-day computers are complete morons, 現在,現代的計算機 完全是白痴,
but this will not be true in another generation. 但這不是真的 在另一代人中。
They will start to think, and eventually they will 他們會開始思考, 最終他們會
completely outthink their makers. 完全超越 他們的製造者。
Is this depressing? I don't see why it should be. 這令人沮喪嗎? 我不明白為什麼會這樣。
I suspect that organic or biological evolution 我懷疑是有機的 或者生物演化
has about come to its end. 即將結束。
You don't want me to be the narrator. 你不要我了 成為敘述者。
I don't have a good voice. 我沒有好聲音。
You have a great voice. Just do it. 你的聲音很好聽。 去做就對了。
It's just a-- we're just trying it out. 這只是一個—— 我們只是嘗試一下。
Do I get paid for this shit? 我會因為這些狗屁而得到報酬嗎?
Well, we'll see how well you do. 嗯, 我們會看看你表現如何。
♪ ♪
I need to call my parents. 我需要打電話給我的父母。
Sometimes we rush into things 有時我們會急於做事
without thinking them through. 沒有經過深思熟慮。
Yeah, can you sh... you want to shoot? That's... 是啊,你能不能... 你想射擊嗎?那是...
Like when Daniel and Caroline got married 就像丹尼爾 和卡洛琳結婚了
151 days after they met. 他們相識151天後。
Let's rewind. 讓我們回顧一下。
Have you been thinking about buying a home computer? 你有沒有想過 買一台家用電腦?
In 1993, when Daniel was born, 1993年,丹尼爾出生時,
his parents didn't even have a computer in the house. 他的父母甚至沒有 房子裡有一台電腦。
But he remembers when they finally got one. 但他記得 當他們終於得到一個的時候。
What a computer is to me is 電腦對我來說意味著什麼
the equivalent of a bicycle for our minds. 相當於一輛自行車 為了我們的思想。
Hello. 你好。
His computer helped unleash his creativity. 他的電腦幫了忙 釋放他的創造力。
Rolling. Go. 滾動。去。
He used his dad's iMac to edit videos. 他用了 他父親的 iMac 用來編輯影片。
Yeah! 是的!
And even make little animations. 甚至讓 小動畫。
In 1998, when Caroline was a little girl... 1998年,當卡洛琳還在 一個小女孩...
Hi, Mommy. 嗨,媽媽。
...only nerds knew what the Internet was. ……只有書呆子才知道 互聯網是什麼。
And what about this Internet thing? 那麼呢 網路這個東西?
Do you, do you know anything about that? -Sure. 你,你知道什麼嗎 關於那個? -當然。
But soon, 但很快,
- everyone was on the... - ...Internets. - 每個人都在... - ...互聯網。
...and computers were beating humans at chess. ……而計算機是 在國際象棋中擊敗人類。
Deep Blue-- Kasparov has resigned! 深藍—— 卡斯帕羅夫辭職了!
And by the time Caroline went to college, 到時候 卡洛琳上了大學,
computers were in everyone's pocket. 計算機是 在每個人的口袋裡。
Now Daniel could make movies with his phone. 現在丹尼爾可以拍電影了 用他的手機。
He grew up to be an artist and a filmmaker... 他長大後成為藝術家 還有一位電影製片人...
- Good night. - ...and so did Caroline. - 晚安。 - ……卡洛琳也是如此。
Cut. Let's go into a close-up. That was perfect. 切。讓我們來一個特寫。 那太完美了。
By then, computers were connecting people 那時,計算機已經 連結人們
and changing the world in ways 並以各種方式改變世界
that we never could have imagined. 我們從來沒有 可以想像。
All the money raised by the Ice Bucket Challenge... 籌集的所有資金 冰桶挑戰...
Some of it was good. 其中一些是好的。
Some not so good. 有些不太好。
Anxiety, self-harm, suicide... 焦慮、自殘、自殺…
But it's clear now that we didn't do enough 但現在很清楚了 我們做得還不夠
to prevent these tools from being used for harm as well. 以防止這些工具 也被用於傷害。
And as the world got more and more focused 並且作為世界 變得越來越專注
on their computers, 在他們的電腦上,
Daniel focused on his artwork. 丹尼爾專注於他的藝術作品。
Wherever he went, he always had a sketchbook, 無論他走到哪裡, 他總是有一本素描本,
including the night he met Caroline. 包括晚上 他遇見了卡洛琳。
He drew a terrible picture of her, and 20 minutes later, 他畫了一幅可怕的圖畫 20分鐘後,
he boldly pronounced they were going to get married, 他大膽地宣布 他們要結婚了,
which, as you know, they did, 151 days later. 正如你所知,他們做到了, 151天后。
...now pronounce you husband and wife. ....現在宣布你 丈夫和妻子。
They moved into a cute house... 他們搬進了一棟可愛的房子...
-Moose, hey! Come here! - ...and got a dog -駝鹿,嘿!過來! - ...還有一隻狗
as dumb as Daniel. 像丹尼爾一樣愚蠢。
Meanwhile, computers were writing entire essays 同時,計算機 寫整篇論文
based on simple questions like, 基於簡單的問題,例如,
"How hard would it be to build a shack in my backyard?" 「建造起來有多難 我家後院的一個小屋? 」
And so Daniel built a shack in his backyard. 所以丹尼爾建了一個小屋 在他的後院。
But just as he sat down to start working on his next movie, 但正當他坐下來開始時 正在製作他的下一部電影,
he learned that computers could now write screenplays. 他了解到計算機 現在可以寫劇本了。
I mean, they were bad, 我的意思是,他們很糟糕,
but they were getting better really fast. 但他們正在變得更好 真的很快。
They could create new images and videos from scratch, 他們可以創造新的圖像 以及從頭開始的視頻,
and some of them could even pass the bar exam. 和其中一些 甚至可以通過律師資格考試。
Not just pass the bar but be in the top ten percentile. 不只是透過酒吧,而是 處於前十個百分點。
It was confusing. 這很令人困惑。
Daniel just wanted a bicycle for his mind, 丹尼爾只是想要 為他的心靈提供一輛自行車,
but computers had become a self-driving rocket ship. 但計算機已成為 自動駕駛火箭飛船。
Pioneers in the field of artificial intelligence 該領域的先驅 人工智慧的
are pleading with Congress 正在向國會懇求
to pass safety rules before it's too late. 通過安全規則 以免為時已晚。
Now it felt like the whole world 現在感覺就像 整個世界
was rushing into something without thinking it through, 急於做某事 不經過深思熟慮,
and everyone had an opinion. 每個人都有意見。
Was it gonna destroy the world or save humanity? 它會毀滅世界嗎 還是拯救人類?
There was too much information, 資訊太多了,
which made him anxious, which made Caroline anxious... 這讓他感到焦慮, 這讓卡洛琳很著急…
I don't have kids yet, but I worry 我還沒有孩子 但我擔心
what the world would look like if I decide to. 世界會是什麼樣子 如果我決定的話。
He was starting to spin out. 他開始旋轉。
It was becoming a mountain of anxiety. 它正在成為 焦慮如山。
-It's horrible. - It's freakin' Siri. -太可怕了。 - 這是該死的 Siri。
And so, Daniel decided to go out 於是,丹尼爾決定出去
and find someone who could explain this to him, 並找人 誰能向他解釋一下,
so he could stop thinking about it 這樣他就可以停下來 考慮一下
and he and Caroline could get on with their lives. 他和卡洛琳可以 繼續他們的生活。
♪ We are in control, we are in control ♪ ♪ 我們掌控一切, 我們一切盡在掌握♪
♪ We are in control ♪ ♪ 我們掌控一切 ♪
♪ Chemical computer thinking battery ♪ ♪ 化學計算機 思考電池♪
♪ We are in control, we are... ♪ ♪ 我們掌控一切,我們... ♪
This endeavor would turn out to be hopelessly naive 這項努力將會得到結果 天真得無可救藥
and kick off the most confusing year of his life. 並開始最令人困惑的 他生命中的一年。
♪ We are in control... ♪ ♪ 我們一切盡在掌握... ♪
But as we know, we sometimes rush into things 但據我們所知, 我們有時會急於做事
without thinking them through. 沒有經過深思熟慮。
♪ Chemical computer thinking battery... ♪ ♪ 化學計算機 思考電池...♪
Oh, my gosh. What is happening? 哦,天哪。怎麼了?
He had questions. 他有疑問。
Questions only the smartest nerds could answer. 問題僅 最聰明的書呆子都能回答。
Submitting to the interrogator. 提交給審訊者。
Questions like: 諸如此類的問題:
Was this the apocalypse or did he actually have reason 這是世界末日嗎 或者他真的有理由嗎
- to be optimistic? - Yes. - 保持樂觀? - 是的。
♪ Chemical computer thinking battery. ♪ - ♪ 化學電腦思維 電池。 ♪ -
Uh, is that good? 呃,這樣好嗎?
Yeah. 是的。
So, to begin, 所以,首先,
what is artificial intelligence? 什麼是人工智慧?
I know that must be annoying for you, that-that question, 我知道這一定很煩 對你來說,那個問題,
but I do think it's important. 但我確實認為這很重要。
So... AI... 所以……人工智慧……
Um... 嗯...
- Yeah... - Uh, hmm. - 是啊... - 呃,嗯。
That's a good question. 這是個好問題。
- Yeah, um... - Um... - 是的,嗯... - 嗯...
What is AI? 什麼是人工智慧?
I love that that's the first question, 我喜歡那是 第一個問題,
'cause there is not a clear and consistent answer. 因為沒有 清晰一致的答案。
Artificial intelligence is 人工智慧是
a kind of intentionally and maybe uselessly broad term. 一種有意的 也許是無用的廣義術語。
It's a machine 這是一台機器
doing things that we previously only thought 做那些事 我們之前只認為
that people could do: 人們可以這樣做:
making recommendations, decisions and predictions. 提出建議, 決定和預測。
AI is the, uh, application of computer science 人工智慧是,呃,應用程式 計算機科學的
to solving cognitive problems. 來解決認知問題。
Okay, so when I picture AI, 好吧,當我想像人工智慧時,
it's sort of like this magical computer box... 有點像 這個神奇的電腦盒子...
...just, like, floating in, like, inert space. ……就像漂浮一樣 在惰性空間中。
And no matter how many times people try 而且不管怎樣 人們嘗試了多少次
and explain this to me, I just don't get how... 並向我解釋一下, 我只是不明白怎麼...
how it's understanding all of these things 它是如何理解的 所有這些事情
and how it's feeling like intelligence. 以及感覺如何 情報。
And that's kind of nerve-racking. 這就是那種 傷腦筋。
What is this new generation of AI? 什麼是 新一代人工智慧?
This AI that is different than every other generation? 這個人工智慧不同於 每隔一代?
Like, no one ever talked about, like, 就像,從來沒有人 談論過,比如,
Siri taking over the world or causing catastrophes. Siri 接管世界 或造成災難。
-Hi, Siri? -...want to. -嗨,Siri? -...想要。
Hello? Siri? 你好?西里?
Hello? Hey, Siri! 你好?嘿,西里!
Or the voice in Google Maps, which mispronounces road names, 或是Google地圖中的聲音, 道路名稱的發音錯誤,
like, breaking society. 就像破壞社會一樣。
Google Maps says this is a road. 谷歌地圖說 這是一條路。
But I think I'm in a river. 但我想我是在河裡。
This is definitely a river, innit? 這絕對是一條河,不是嗎?
Something changed with ChatGPT coming out. 有些事情發生了變化 隨著 ChatGPT 的出現。
People understood-- no, no, no-- 人們明白——不,不,不——
this technology's insanely valuable, 這項技術的 非常有價值,
it's insanely powerful and also insanely scary. 它非常強大 而且也非常可怕。
Okay, listen to this. Very creepy. 好吧,聽聽這個。 非常令人毛骨悚然。
A new artificial intelligence tool is going viral 一種新的人工智慧 工具正在迅速傳播
for cranking out entire essays in a matter of seconds. 撰寫整篇論文 幾秒鐘之內。
AI dwarfs the power of all other technologies combined. 人工智慧讓所有人的力量相形見絀 其他技術的結合。
"AI dwarfs the power 「人工智慧讓力量相形見絀
of all other technologies combined." 所有其他技術 結合起來。 」
Yeah. 是的。
Do you think that's true? 你認為這是真的嗎?
Yes. 是的。
Tell me about-- How? How? 告訴我—如何?如何?
I think, to paint that picture, it's really important 我想,要描繪這幅畫, 這真的很重要
to understand what today's state-of-the-art systems 了解今天的內容 最先進的系統
look like and how they're built. 它們的外觀以及構造方式。
This is quite a... quite a setup. 這是 相當……相當的設定。
So, one thing that 所以,有一件事
not a lot of people realize is that 人不多 意識到的是
systems like ChatGPT aren't programmed by any human. 像 ChatGPT 這樣的系統則不然 由任何人編程。
What do you mean? 你是什麼意思?
Instead, it's something like th-they're grown. 相反,它就像 他們長大了。
We kind of give them raw resources, like, 我們給他們 原始資源,例如,
"Here's a lot of computational resources. 「這裡有很多 計算資源。
Here's a lot of data." 這裡有很多數據。 」
Under the hood, it's math, 在幕後,這是數學,
and the math is actually surprisingly straightforward. 數學其實是 出奇的簡單。
So ChatGPT is a kind of AI but it's not all of AI? 所以ChatGPT是一種 人工智慧的一部分,但這不是人工智慧的全部?
Totally. ChatGPT is just the beginning, 完全。聊天 GPT 是 只是開始,
but it's a good place to start. 但這是一個很好的起點。
But I still don't know what AI is. 但我還是不知道 人工智慧是什麼。
To understand AI, it begins with understanding 要了解人工智慧, 從理解開始
that intelligence is about recognizing patterns. 那個情報是關於 識別模式。
- Patterns. -Patterns. - Patterns. - 圖案。 - 圖案。 - 圖案。
It is shown trillions of words of text 顯示的是 數萬億字的文本
across millions of documents in the Internet. 跨越數百萬份文檔 在互聯網上。
It started with text. 它從文字開始。
And what they did was they took textbooks, 他們所做的是 他們拿著課本,
and they took poems and essays and instruction manuals. 他們拿了詩和散文 和使用說明書。
They can do things like 他們可以做類似的事情
digest the entire Internet, 消化整個互聯網,
every single word that's ever been written by a person. 曾經的每一個字 是一個人寫的。
Reddit threads and social media and all of Wikipedia. Reddit 主題和社群媒體 以及所有維基百科。
More data than anybody could ever read in several lifetimes. 比任何人都多的數據 幾輩子都讀過。
And they gave this system one job. 他們給了這個系統 一份工作。
Figure out the patterns and structure of that information 找出模式並 該資訊的結構
and use that to make predictions about 並使用它 做出預測
what word should come next in a sentence. 接下來應該是什麼詞 用一句話來說。
When you say "patterns in a sentence," 當你說 “句子中的模式,”
what are you talking about? 你在說什麼?
So, it's everything from, like, the really simple things, 所以,一切都來自,例如, 真正簡單的事情,
like most sentences end with a period, 就像大多數句子的結尾一樣 有句號的,
all the way up to the more conceptual things, like: 一直到 更多概念性的東西,例如:
What is a sonnet? 什麼是十四行詩?
It's a type of poem, and it has some particular structure. 這是一種詩,它有 一些特定的結構。
So, it then looks at 所以,然後它會看
all of that data, all of that text... 所有這些數據, 所有這些文字...
And over trillions and trillions of tries, 超過數萬億 和數萬億次的嘗試,
each time it gets something right or wrong, 每次它得到一些東西 對或錯,
it's given a little bit of positive reinforcement 它被給予了一點點 正強化的
when it guesses the next word correctly, 當它猜測的時候 下一個單字正確,
and it's given a little bit of negative reinforcement 它被給予了一點點 負強化
when it guesses the next word incorrectly. 當它猜測的時候 下一個字錯誤。
And at the end of it, you have a system that 最後, 你有一個系統
speaks really good English as a side effect 英語說得很好 作為副作用
of being really, really good at predicting 真的真的很好 在預測
the word that comes next in a piece of text. 接下來的詞 在一段文字中。
Actually, my French isn't too bad either. 事實上,我的法語 也不錯。
It uses all of those patterns it has learned 它使用所有 它學到的那些模式
to be able to make a prediction about 能夠使 關於的預測
what the answer should be, 答案應該是什麼
then it gives you that as the output. 然後它給你 作為輸出。
It's a little oversimplified, 有點過於簡化了
but I think people will get it. 但我認為人們會明白的。
So, so that's all it does? 那麼,這就是它的全部作用嗎?
Yeah. It doesn't seem like it would be that complicated, 是的。看起來不像 就會那麼複雜,
but actually you have to know 但實際上你必須知道
a huge amount of things 大量的東西
in order to actually succeed at that. 為了 實際上成功了。
If you say to ChatGPT, 如果你對 ChatGPT 說,
"Write me a Shakespearean sonnet about my dog," 「寫一首莎士比亞戲劇 關於我的狗的十四行詩”
it has to know what dogs are. 它必須知道狗是什麼。
It has to know what you love about your dog.- 它必須知道你喜歡什麼 關於你的狗。 -
It has to know who Shakespeare is, 它必須知道 莎士比亞是誰,
that sonnets rhyme, that they have a structure, 那首十四行詩押韻, 他們有一個結構,
that words have sounds that can rhyme. 單字有聲音 可以押韻。
It takes a lot. 這需要很多。
Holy shit, you can talk to your computer now. 天哪,你能說話 現在到您的電腦。
That was just not true three years ago. 那不是真的 三年前。
Yes, and this is the really important part. 是的,這是 真正重要的部分。
The same process that lets AI 同樣的過程讓人工智慧
uncover and manipulate the patterns of text 發現並操縱 文字的模式
is the same process that lets it uncover 是同一個過程 讓它揭示
the patterns of the entire universe and everything in it. 整個的圖案 宇宙和其中的一切。
There are patterns and images and sound 有圖案 以及影像和聲音
in computer code and DNA and music and physics 電腦代碼和 DNA 以及音樂和物理
and fashion and building design 以及時尚和建築設計
and in human voices and human faces, 並在 人的聲音和人的臉孔,
really, truly everywhere. 真的,真的無所不在。
- Everywhere. -Everywhere. - Everywhere. -Everywhere. - 無所不在。 -到處。 - 無所不在。 -到處。
If you have learned those patterns, 如果你已經學會了 那些圖案,
you can generate new kinds of songs. 你可以生成 新類型的歌曲。
You can generate new videos. 您可以產生新影片。
And that's why, if you give it 這就是為什麼,如果你給它
a three-second recording of your grandmother, 三秒錄音 你祖母的
it can speak back in her voice. 它可以用她的聲音說話。
Oh, my God. 我的天啊。
Oh, my God. Oh, my God. 我的天啊。我的天啊。
What will they think of next? 接下來他們會想到什麼?
It's moving very, very quickly. 它移動得非常非常快。
An American AI start-up 一家美國人工智慧新創公司
has released its latest model. 發布了其最新型號。
That company is Anthropic, and it has just unveiled 那家公司是 Anthropic, 它剛剛揭幕
the latest versions of its AI assistant Claude. 最新版本 其人工智慧助手克勞德。
So, the xAI team was there to unveil Grok 4. 所以,xAI 團隊 在那裡揭曉 Grok 4。
Google released one just last week. Gemini is... 谷歌發布了一款 就在上週。雙子座是...
We've gone from GPT-2 just a couple years ago, 我們已經脫離了 GPT-2 就在幾年前,
which could barely write a coherent paragraph, 幾乎不能寫 一個連貫的段落,
to GPT-4, which can pass the bar exam. 到 GPT-4, 可以通過律師資格考試。
And all they had to do to get there 他們所要做的一切 到達那裡
was essentially add more data and more compute. 本質上是添加更多數據 和更多的計算。
These people who are building this... -COTRA: Yeah. 這些人是 建造這個...... -COTRA:是的。
...they're just throwing more... ……他們只是丟更多…
More physical computers, more of the same kinds of data. 更多實體計算機, 更多相同類型的數據。
Because the more computing power you add, 因為越 您新增的運算能力,
the more complex intellectual tasks they can do. 越複雜 他們可以完成的智力任務。
So, the more weather data you give it, 所以,更多的天氣數據 你給它,
the better it can make predictions about 越好 做出預測
where a hurricane might go. 颶風可能去往的地方。
And the more patterns of tumors and bones 而且圖案也比較多 腫瘤和骨骼
and tissues an AI has seen, then the better able it is 和人工智慧見過的組織, 那麼能力就越好
to detect a tumor in a new CT scan. 檢測腫瘤 在新的 CT 掃描中。
Better even than a human doctor. 甚至比人類醫生還要好。
AI that's already being deployed for the military 已經存在的人工智慧 為軍隊部署
can already use satellite imagery, 已經可以使用 衛星圖像,
troop movements, communications to determine, 部隊調動、通訊 確定,
sometimes days in advance, 有時提前幾天,
where an attack is going to happen, 哪裡發生攻擊 將會發生,
like where an enemy is going to strike. 就像敵人所在的地方 即將罷工。
This whole space is moving so fast 這整個 太空移動得如此之快
that any example you put in this movie 任何例子 你把這部電影放進去
will feel absolutely clumsy by the time it comes out. 會感覺絕對笨拙 當它出來的時候。
These models are being released 這些型號正在發布中
before anyone knows what they're even capable of. 在任何人知道之前 他們甚至有能力做什麼。
GPT-3.5 was released and out to 100 million people plus GPT-3.5 發布並退出 超過 1 億人
before some researchers discovered that it could do 在一些研究者面前 發現它可以做
research-grade chemistry better than models 研究級化學 比模型更好
that were trained specifically to do research-grade chemistry. 是經過專門訓練的 進行研究級化學。
Something is happening in there that the people 那裡正在發生一些事情 人民
who are building them don't fully understand. 誰正在建造它們 不完全理解。
Basically, it just analyzes the data by itself, 基本上只是分析 數據本身,
and as it does that, it just teaches itself 當它這樣做時, 它只是自學
various things that we often didn't intend. 各種東西 我們常常無意的。
So, for instance, it reads a lot online, 所以,舉例來說, 網路上看了很多,
and then at some point, it just learns how to do arithmetic. 然後在某個時刻,它只是 學習如何做算術。
One, two. 一,二。
And then at some point, it starts to learn 然後在 某個時刻,它開始學習
how to answer advanced physics questions. 如何回答 高級物理問題。
We didn't program that in it whatsoever. 我們沒有這樣編程 無論如何。
It just learned by itself. 它只是自己學會了。
An AI is like a digital brain. 人工智慧就像一個數位大腦。
But just like a human brain, 但就像人腦一樣,
if you did a brain scan on a human brain, 如果你做了腦部掃描 在人腦上,
would you know everything that person was capable of? 你會知道一切嗎 那人有能力嗎?
You can't know that just from the brain scan. 你不可能知道這一點 只是從腦部掃描來看。
It's just, like, a bunch of numbers and, like, 這只是,就像, 一堆數字,例如
the multiplications that are happening that-that, like, 的乘法是 發生那件事,比如,
the best machine learning researcher in the world 最好的機器學習 世界研究員
could look at and, like, have no idea what was happening. 可以看並且喜歡 不知道發生了什麼事。
That chair right there. 那把椅子就在那裡。
- Is that okay for you? - Yes. - 你覺得可以嗎? - 是的。
So, that's kind of mind-boggling. 所以, 這有點令人難以置信。
Okay? Like, it's taking over the world, 好的?就像,它正在接管 世界,
and we don't even know how it works. 我們甚至不知道 它是如何運作的。
- Is that right? - Mm. - 是這樣嗎? - 毫米。
We do understand a number of important things, 我們確實明白 許多重要的事情,
but we don't have a very good grasp on 但我們沒有 很好地掌握了
why they provide specific answers to questions. 他們為什麼提供 問題的具體答案。
It is a problem because we are on a path t-to build machines, 這是一個問題,因為我們是 在一條建造機器的道路上,
based on these principles, that could be smarter than us 基於這些原則, 那可能比我們更聰明
and thus potentially have a lot of power. 因此可能有 很大的力量。
One of the most cited computer scientists in history. 被引用次數最多的之一 歷史上的計算機科學家。
I actually find it a little difficult 我確實找到了 有點困難
to talk about my own role. 談談我自己的角色。
Really much prefer when other people do it. 真的非常喜歡 當其他人這樣做時。
Ilya joining was the... was-was the linchpin for 伊利亞的加入是... 曾經是關鍵
OpenAI being ultimately successful. OpenAI 正在 最終成功。
I think it's just going to be 我想這將會是
some kind of a vast, dramatic and unimaginable impact. 某種巨大的、戲劇性的 以及難以想像的影響。
I don't know if you've spent any time on YouTube, 不知道你有沒有消費過 隨時在 YouTube 上,
but you can kind of feel the speed already, right? 但你可以感覺到 速度已經到了吧?
You know what I mean? 你知道我的意思?
This is just the warmup. 這只是熱身。
The really powerful systems are still coming, 真正強大的系統 仍然會來,
and they're gonna be coming quite soon. 他們會來的 很快。
AGI stands for "artificial general intelligence." AGI代表「人工 一般情報。 」
Uh, systems that are basically... 呃,系統是 基本上...
And this is, like, you know, seems to be, like, 這就像,你知道, 似乎是,就像,
the holy grail of AI? 人工智慧的聖杯?
When you can simulate a human mind 當你可以模擬時 人類的思想
that is doing human cognition and can do reasoning, 那是在做人類的認知 並能進行推理,
that is a new sort of tier of AI 這是一種新型的人工智慧
that we have to distinguish from previous AI. 我們必須區分 來自之前的AI。
When that happens, by the way, that's when 當這種情況發生時, 順便說一句,那時
you would hire one of those AGIs instead of a person. 你會僱用其中之一 那些 AGI 而不是一個人。
Most jobs in our economy it can do. 我們經濟中的大多數工作 它可以做到。
It can work 24 hours a day, 它可以一天24小時工作,
never gets tired, never gets bored. 永遠不會厭倦, 永遠不會感到無聊。
They don't need to sleep. They don't need breaks. 他們不需要睡覺。 他們不需要休息。
They're, like, not gonna join a union. 他們就像, 不會加入工會。
Won't complain, won't whistleblow. 不會抱怨, 不會吹哨。
More than 100 times cheaper 便宜100倍以上
than humans working at m-minimum wage. 比人類工作 m-最低工資。
Not only will they be doing everything, 他們不僅會 做一切事情,
but they'll be doing it faster. 但他們會做得更快。
The same intelligence 一樣的智力
that powers that can also look at the patterns and movements 那力量也可以看 在模式和動作
and articulating muscles and, you know, robotics. 和關節肌肉 還有,你知道的,機器人技術。
And so it's not just gonna automate desk jobs. 所以這不僅僅是 將使辦公桌工作自動化。
That's just the beginning. 這只是開始。
It will automate all physical labor. 它將自動化 所有體力勞動。
There's no way 沒有辦法
humans are gonna compete with them. 人類將會競爭 和他們在一起。
It is hard to conceptualize the impact of AGI. 很難概念化 AGI 的影響。
But I think it's going to be something very big 但我認為這將會是 非常大的東西
and drastic and radical. 並且激烈和激進。
You think this is one of the most 你認為 這是最
consequential moments in human history? 相應的時刻 在人類歷史上?
Yeah. Yeah, that's-- I mean, what else would be? 是的。是的,那就是—— 我的意思是,還有什麼?
I mean, like, there's the Industrial Revolution. 我的意思是,就像,有 工業革命。
You know, it'll make the Industrial Revolution 你知道,它會 促成工業革命
look like small beans. 看起來像小豆子。
AGI is an inflection point AGI是一個拐點
because it means you can accelerate 因為這意味著 你可以加速
all other intellectual fields all at the same time. 所有其他知識領域 全部同時進行。
Like, if you make an advance in rocketry, 就像,如果你預付款 在火箭技術中,
that doesn't advance biology and medicine. 不前進 生物學和醫學。
If you make an advance in medicine, 如果您預付款 在醫學上,
that doesn't advance rocketry. 這並不能推動火箭技術的發展。
But if you make an advance in artificial intelligence, 但如果你提前 在人工智慧領域,
that advances all scientific and technological fields 推動所有科學進步 和技術領域
all at the same time. 全部同時進行。
That's why, for a long time, 這就是為什麼在很長一段時間裡,
Google DeepMind's mission statement was... 谷歌 DeepMind 的 使命宣言是...
- Step one, solve intelligence. -Yeah. - 第一步,解決情報問題。 -是的。
Step two, use it to solve everything else. -Yes. 第二步,用它來解決 其他一切。 -是的。
That's why AI dwarfs the power 這就是為什麼人工智慧讓力量相形見絀
of all other technologies combined. 所有其他技術 合併。
It will transform everything. 它將改變一切。
So, uh, it'll be at least as big as 所以,呃,將會是 至少和
the Industrial Revolution, possibly, you know, bigger, 工業革命, 可能,你知道,更大,
more like the advent of electricity or even fire. 更像是降臨 電力甚至火災。
The caveman literally held aloft 穴居人確實高高舉起
the torch of civilization. 文明的火炬。
It is generally thought that 人們普遍認為
around the time of AGI, we'll have AIs that can AGI時代前後, 我們將擁有能夠
do all or most of the AI research process 做全部或大部分 人工智慧研究過程
and, of course, can do it faster and cheaper. 當然,可以做到 更快更便宜。
It can copy itself. 它可以自我複製。
A thousand times, a million times, 一千次, 一百萬次,
and, like, now you have a million copies 並且,就像, 現在你有一百萬份了
all working in parallel. 所有這些都是並行工作的。
When it learns how to make its code faster, 當它學習時 如何使其程式碼更快,
make its code more efficient, 使其程式碼更加高效,
obviously that becomes, like, a-a runaway loop. 顯然這變成了,就像, a-失控循環。
AGI isn't, like, the end. AGI 並不是終結。
It's just the beginning. 這只是開始。
It's the beginning of an incredibly rapid explosion 這是開始 令人難以置信的快速爆炸
of scientific progress, 科學的進步,
and in particular, scientific progress in AI. 特別是, 人工智慧的科學進步。
And when they're smarter than us, too, 當他們更聰明時 也比我們
and substantially faster than us, 並且速度更快 比我們,
and they're getting faster each year, exponentially, 而且他們的速度越來越快 每年,呈指數級增長,
those are the ones that can potentially become superhuman, 這些是可以的 有可能成為超人,
uh, possibly this decade. 呃,可能是這個十年。
Sorry, did you say 抱歉,你說的是
"become superhuman, maybe in this decade"? 「成為超人, 也許在這十年裡」?
Yeah. I mean, I think, uh, a lot of people who are 是的。我的意思是,我想,呃, 很多人都是
actually building this think that that's fairly plausible 實際上建立這個想法 這是相當合理的
that we get some superintelligence, 我們所得到的 一些超級智能,
something that's vastly more intelligent than people, 非常大的東西 比人聰明,
within this decade. 這十年內。
The way I define "superintelligence" is 我定義的方式 「超級智能」是
a system that by itself is 一個系統本身就是
more intelligent and competent than all of humanity. 更聰明、更有能力 比全人類。
I'm just gonna-- sorry. I don't mean to interrupt you. 我只是要——抱歉。 我無意打擾你。
You're on a flow. 你正處於潮流之中。
Uh, I just, I just... 呃,我只是,我只是…
I'm not really following, 'cause you're using language 我並沒有真正關注, 因為你正在使用語言
like "superintelligence" and, like, 比如“超級智能” 並且,就像,
"smarter than all of humanity," and I hear that, “比全人類都聰明” 我聽說,
and it sounds like... like sci-fi bullshit to me, 聽起來像是… 對我來說就像科幻小說一樣
and I'm just trying to understand. 我只是在嘗試 去理解。
There's nothing magical about intelligence. 沒有什麼神奇的 關於智力。
This is very important, is that, you know, 這一點非常重要, 那是,你知道,
intelligence can feel magical, it can feel like 智慧讓人感覺神奇, 感覺就像
some mystical thing in your mind or something, 一些神秘的東西 在你的腦海或其他什麼地方
but it is just computation. 但這只是計算。
The human brain is quite limited in some ways, 人類的大腦是 在某些方面相當有限
in terms of information processing capability, 在資訊方面 處理能力,
compared to what we see in, say, a data center. 與我們所看到的相比 比如說,在數據中心。
So, for example, the signals which are sent 所以,舉例來說, 發送的訊號
inside your brain, they move at about 30 meters per second. 在你的大腦裡,它們移動 大約每秒30公尺。
But the speed of light, 但光速,
which is what a computer uses in fiber optics, 這是計算機使用的 在光纖領域,
is 300 million meters per second. 是3億米 每秒。
And so, it would be kind of strange if 所以,那就是 有點奇怪如果
human intelligence was somehow really special 人類智力是 不知何故真的很特別
in that regard and is somehow some upper limit 在這方面 並且是某種上限
of what's possible in intelligence. 什麼是可能的 在智力上。
I think, once we understand how to build intelligent systems, 我想,一旦我們了解如何 建構智慧系統,
we will be able to build huge machines, 我們將能夠建造 巨大的機器,
which will be far beyond normal human intelligence. 這將遠遠超出 正常人類智力。
Uh, hopefully we can have a very symbiotic relationship, 呃,希望我們能有 一種非常共生的關係,
uh, with AI systems, but the AI developers are 呃,有了人工智慧系統, 但人工智慧開發者們
specifically designing them to make sure that they can do 專門設計它們 確保他們能夠做到
everything better than we can, so I-I don't know 一切都比我們更好, 所以我-我不知道
what-what we will be able to offer, unfortunately. 我們將能夠做什麼 不幸的是,要提供。
The-the older-school AI technology... 老派學校 人工智慧技術...
Daniel isn't feeling any better. 丹尼爾不是 感覺好多了。
His plan is backfiring. 他的計劃適得其反。
The more he learns about this new, powerful, 他了解得越多 這個新的、強大的、
inscrutable thing, the worse it sounds. 深不可測的東西, 聽起來越糟。
He wants to tell them how scared he is, 他想告訴他們 他有多害怕,
how he feels like the earth is slipping out from under him, 他對地球的感覺如何 從他身下滑出,
that he's staring down existential dread, 他正在低頭凝視 存在的恐懼,
so he articulates this by saying... 所以他闡明了這一點 透過說...
That sounds bad. 聽起來很糟。
Yeah. 是的。
If you have all of these capabilities 如果你有 所有這些能力
and they start to be able to plan better, 他們開始能夠 為了更好地計劃,
if you sort of take that to its logical conclusion, 如果你這麼認為 得出其邏輯結論,
you can get some pretty power-seeking behavior. 你可以得到一些漂亮的 尋求權力的行為。
Okay, so why would an AI 好吧,那為什麼人工智慧會
want more power? 想要更多力量嗎?
Yeah. So, I think it's actually pretty simple. 是的。所以,我認為 其實很簡單。
Having more power is a very effective strategy 擁有更多的權力就是 一個非常有效的策略
for accomplishing almost any goal. 為了完成 幾乎任何目標。
We ran an experiment where we gave 我們做了一個實驗 我們給予的地方
OpenAI's most powerful AI model, uh, OpenAI最強大 AI模型,呃,
a series of problems to solve. 一連串的問題需要解決。
And partway through, on its computer, 而中途, 在其計算機上,
it got a notification that it was going to be shut down. 它收到通知 它即將被關閉。
And what it did is it rewrote that code 它所做的是 它重寫了該程式碼
to prevent itself from being shut down 來防止自己 免於被關閉
so it could finish solving the problems. 所以它可以完成 解決問題。
Okay. -Yeah. So, another really interesting one 好的。 -是的。那麼,另一個 非常有趣的一個
is that the AI company Anthropic 是 AI 公司 Anthropic
made a simulated environment where that AI had access 做了一個模擬環境 人工智慧可以訪問的地方
to all of the company emails. 發送至所有公司電子郵件。
And it learned through reading those emails 它透過學習 閱讀這些電子郵件
it was going to be replaced 它將被替換
and the lead engineer who was responsible for this 和首席工程師 誰為此負責
was also having an affair. 也有外遇。
And on its own, it used that information 就其本身而言, 它使用了該訊息
to blackmail the engineer 勒索工程師
to prevent itself from being replaced. 來防止自己 免遭替換。
It was like, "No, I'm not gonna be replaced. 就像, 「不,我不會被取代。
"If you replace me, I'm going to tell the world 「如果你取代我, 我要告訴世界
that you are having this affair." 你正在擁有 這件事。 」
And nobody taught it to do that? 沒有人教它這樣做嗎?
No, it learned to do that on its own. 不,它學會了這樣做 靠它自己。
As the models get smarter, they learn that these are 隨著模型變得越來越聰明, 他們了解到這些是
effective ways to accomplish goals. 有效的方法 實現目標。
And this is not a problem that's isolated to one model. 這不是問題 這是孤立於一個模型的。
All of the most powerful models show these behaviors. 所有最強大的型號 表現出這些行為。
- Hey. -Hey. How are you? - 嘿。 -嘿。你好嗎?
Good. Good to be here. 很好。很高興來到這裡。
When Yuval Noah Harari 當尤瓦爾·諾亞·赫拉利
published his first book Sapiens in 2014, 出版了他的第一本書 2014年的智人,
it became a global bestseller 它成為全球暢銷書
and turned the little-known Israeli history professor 並把鮮為人知的 以色列歷史教授
into one of the most popular writers 進入其中之一 最受歡迎的作家
and thinkers on the planet. 和這個星球上的思想家。
The biggest danger with AI is 人工智慧最大的危險是
the belief that it is infallible, 信念 它是絕對正確的,
that we have finally found-- 我們終於發現——
"Okay, gods were just this mythological creation. 「好吧,神只是 這個神話創造。
"Humans, we can't trust them, but AI is infallible. 「人類,我們不能相信他們, 但人工智慧是絕對正確的。
It will never make any mistakes." 它永遠不會使 任何錯誤。 」
And this is a deadly, deadly threat. 這是 致命的、致命的威脅。
It will make mistakes. 它會犯錯。
And all these fantasies that AI will reveal the truth 還有所有這些幻想 人工智慧將揭示真相
about the world that we can't find by ourselves, 關於這個世界 我們自己找不到,
AI will not reveal the truth about the world. AI不會揭露真相 關於世界。
AI will create an entirely new world, much more complicated AI將創造一個全新的 世界,更加複雜
and difficult to understand than this one. 並且難以理解 比這個。
What's about to happen is that we, uh, humans are no longer 即將發生的事情是 我們,呃,人類不再是
going to be the most intelligent entities on Earth. 將是最 地球上的智慧實體。
So I think what's coming up is going to be 所以我想接下來會發生什麼 將會是
one of the biggest events in human history. 最大的事件之一 在人類歷史上。
Geoffrey, thanks so much for joining us. 傑弗裡, 非常感謝您加入我們。
So you left your job with Google in part because you say 所以你離開了你的工作 谷歌部分是因為你說
you want to focus solely on your concerns about AI. 你想要專注 關於您對人工智慧的擔憂。
You've spoken out, saying that AI could manipulate 你已經說出來了, 說人工智慧可以操
or possibly figure out a way to kill humans. 或者可能弄清楚 一種殺死人類的方法。
H-How could it kill humans? 它、它怎麼能殺死人類?
Well, if it gets to be much smarter than us, 好吧,如果是的話 比我們聰明得多
it'll be very good at manipulation 會很好的 在操縱時
'cause it will have learned that from us. 因為它會有 從我們這裡學到的。
And it'll figure out ways of manipulating people 它會找出方法 操縱人的
to do what it wants. 做它想做的事。
There was a open letter from the Center for AI Safety. 有一封公開信 來自人工智慧安全中心。
Sam Altman signed this. Demis signed this. 薩姆·奧爾特曼簽署了此協議。 戴米斯簽署了這個。
They signed a 22-word statement 他們簽署了一份22字的聲明
that we need to take AI and the threat from AI 我們需要採用人工智慧 以及人工智慧的威脅
as seriously as global nuclear war. 認真地像 全球核戰。
-Hello. - Hello. -你好。 - 你好。
You're kind of, like, the original doom guy. 你有點像, 原來的厄運傢伙。
More or less. 或多或少。
Since 2001, I have been working on 自2001年以來, 我一直在努力
what we would now call the problem 我們現在所說的 問題
of aligning artificial general intelligence. 對準人工 一般情報。
How to shape the preferences and behavior 如何塑造 偏好和行為
of a powerful artificial mind 強大的人造思維
such that it does not kill everyone. 這樣它就不會 殺死所有人。
It's not like a lifeless machine. 這不像 一台沒有生命的機器。
It is smart, it is creative, 很聰明,很有創意,
it is inventive, it has the properties 它具有創造性, 它有以下屬性
that makes the human species dangerous, 這使得 人類物種危險,
and it has more of those properties. 它有 更多這些屬性。
If something doesn't actively care about you, 如果有什麼事情沒有 主動關心你,
actively want you to live, 積極地希望你活下去,
actively care about your welfare, 積極關心 您的福利,
about you being happy and alive and free, 關於你幸福的事 並且活著並且自由,
if it cares about other stuff instead, 如果它關心 其他東西代替,
and you're on the same planet, 和你們在同一個星球
that is not survivable if it is very much smarter than you. 如果是的話,那是無法生存的 比你聰明得多。
I don't think it's going to be 我不認為這會是
a kind of, like, evil thing. 一種邪惡的東西。
It's like, "Oh, the AIs are evil and they hate humanity." 就像,「哦,人工智慧是 他們是邪惡的,他們仇恨人類。 」
I don't think that's what's gonna happen. 我不認為 這就是將會發生的事情。
I think what is happening is 我認為正在發生的事情是
far more how humans feel about ants. 更多的是人類的感受 關於螞蟻。
Like, we don't hate ants, 就像,我們不討厭螞蟻,
but if we want to build a highway 但如果我們想要 興建高速公路
and there's an anthill there, well, sucks for the ants. 那裡有一個蟻丘, 嗯,對螞蟻來說很糟糕。
It's not that hard to understand. 這並不難 去理解。
It's like, hey, if we build things 就像,嘿, 如果我們建造東西
that are smarter than us 比我們聰明的人
and we don't know how to control them, 我們不知道 如何控制它們,
does that seem like a risky thing to you? 這樣看起來像嗎 對你來說是一件危險的事嗎?
Yeah. Yeah, it does. 是的。是的,確實如此。
You don't have to be a tech guy. 你不必是技術人員。
You don't have to know programming to understand it. 你不必知道 程式設計來理解它。
It's not that hard. 這並不難。
This is not a hard thing to understand. 這並不是一件難事 去理解。
Connor, how-how many people 康納,多少人
in the world right now are working on AGI? 此刻在世界上 正在研究 AGI?
At least 20,000, I would say. 我想說至少20,000。
- 20,000? - I would expect so. - 20,000? - 我希望如此。
Okay, and how many people are working full-time 好的,還有多少人 正在全職工作
to make sure AI doesn't, like, kill us all? 為了確保人工智慧不會, 就像,殺了我們所有人?
Probably less than 200 in the world. 大概不到200 在世界上。
And your conceit is that 而你的自負是
the only natural result of this recklessness... 唯一自然的結果 這種魯莽行為…
...is the collapse of humanity? ……這是人性的崩潰嗎?
Well, not the collapse, the abrupt extermination. 還好不是崩潰 突然的滅絕。
There's a difference. 有差別。
♪ What is the meaning of life? ♪ ♪ 生命的意義是什麼? ♪
♪ What is the future and what is now? ♪ ♪ 未來是什麼 現在是什麼? ♪
♪ What is the answer to strife? ♪ ♪ 答案是什麼 發生衝突? ♪
♪ How much ♪ ♪ 多少錢 ♪
♪ Can someone dream? ♪ ♪ 有人會做夢嗎? ♪
♪ How long? ♪ ♪ 多久? ♪
♪ Forever ♪ ♪ 永遠 ♪
♪ What is the meaning ♪ ♪ 什麼意思 ♪
♪ Of me... ♪ ♪ 我的... ♪
I do think this is probably, like, 我確實認為 這可能是,例如,
the biggest challenge that, like, 最大的挑戰 那個,就像,
our civilization will-will face, ever. 我們的文明 意志將面對,永遠。
This essentially is the last mistake we'll ever get to make. 這本質上是最後一個 我們永遠都會犯的錯誤。
If we can rise to be the most mature version of ourselves, 如果我們能成為最 成熟版的我們自己,
there might be a way through this. 可能有 解決這個問題的方法。
What does that mean? 這意味著什麼?
"The most mature version of ourselves"? 「最成熟的版本 我們自己”?
'Cause that sounds, for me, like... 因為聽起來, 對我來說,就像...
I-I-- What the [bleep]? 我-我--什麼[嗶]?
Do you think now is a good time to have a kid? 你認為現在是 是生孩子的好時機嗎?
Um... 嗯...
Do you want to have kids one day? 你想要孩子嗎 有一天?
Is that something that-that you're into or not really? 那是那-那嗎 你真的喜歡嗎?
Um... uh, I confess, I think that's like, 呃……呃,我承認, 我認為這就像,
"Boy, let's get through this critical period." 「小伙子,我們來吧 渡過這個關鍵時期。 」
Um... 嗯...
Do you have any kids? 你有孩子嗎?
I do not. 我不知道。
Is that something you want to do? 那是嗎 你想做什麼?
Have children, have a family? 有孩子、有家庭嗎?
In some other world than this world, sure, I would have kids. 在另一個世界 世界,當然,我會有孩子。
Would you want to start a family? 你想要嗎 組成家庭?
Would you want to have kids? 你想要孩子嗎?
Is that something you're thinking about? 那是嗎 你在想什麼?
I, um... 我,嗯...
I just have to find it first. 我只需要先找到它。
We have to go to the doctor, but... 我們得走了 去看醫生,但是...
Ah! I knew it! I knew it! I knew it! 啊!我就知道! 我就知道!我就知道!
- When did you find out? -Last night. - 你什麼時候發現的? -昨夜。
- Oh, my God! - Mom, I don't know. - 我的天啊! - 媽媽,我不知道。
I can't tell you 我不能告訴你
- how happy I am! -No, seriously. ——我多麼高興啊! -不,說真的。
Well, I took a pregnancy test last night, and I'm pregnant. 嗯,我做了妊娠測試 昨晚,我懷孕了。
Oh, my God! 我的天啊!
Oh, my God, you guys! 噢,天哪,你們!
I just wanted to confirm your expected due date, 我只是想確認一下 您的預計預產期,
- which is January 21st. - Oh, my God. - 那是 1 月 21 日。 - 我的天啊。
I can't believe how happy I am. 我簡直不敢相信我有多高興。
He already looks really cute. 他看起來已經很可愛了。
You think he already looks cute? 你認為 他看起來已經很可愛了?
Yeah. Look at that little cutie face. 是的。 看看那張可愛的小臉。
I have this baby on the way. 我這個孩子已經在路上了。
Right. 正確的。
So I turn it over to you. 所以我把它交給你。
Are we doomed? 我們注定了嗎?
Are we all gonna face this techno dystopian 我們都要面對 這個科技反烏托邦
future of doom? 未來的厄運?
It's, uh... 這是,呃...
It's not good news, 這可不是什麼好消息
the world that we're heading into. 世界 我們即將進入的。
I-- for ex-- I mean, I'll just be honest. 我——對於前—— 我的意思是,我會說實話。
Uh, I know people who work on AI risk 呃,我認識人 從事人工智慧風險研究的人
who don't expect their children to make it to high school. 誰不期望自己的孩子 升上高中。
This is, like... this is actually scary 這就像... 這實際上很可怕
'cause it's like, oh, we're all [bleep]. 因為它就像, 哦,我們都[嗶]。
You have to c... make me calm, 你必須...讓我冷靜,
because this is making me incredibly anxious 因為這讓我 非常焦慮
and I'm carrying the baby right now, 我帶著孩子 現在,
so you have to also be calm for me and strong and hopeful, 所以你也必須冷靜 對我來說,堅強而充滿希望,
because I'm-- it's too, it's too much for my soul to bear 因為我──這也是,這也是 我的靈魂無法承受太多
while I'm carrying this baby. 當我懷著這個孩子的時候。
So you're going to have to try to figure out 所以你必須嘗試 弄清楚
a way to have hope. 一種充滿希望的方式。
It's really important, Daniel, especially now. 這真的很重要, 丹尼爾,尤其是現在。
H-How to have hope. H-如何擁有希望。
You have to. You have to find it for me. 你必須這樣做。 你必須給我找到它。
I'm serious. 我是認真的。
I'm going to. 我要去。
I will. I'll tr-- I'll try. 我會。我會嘗試。
Hey, guys? 嘿,夥計們?
Guys. 夥計們。
Can we get back up and running? 我們可以恢復運作嗎?
Oy gevalt! 天啊!
Uh, Dan, are you ready? 呃,丹,你準備好了嗎?
Hello, Daniel. 你好,丹尼爾。
Hey, how are you? 嘿,你怎麼樣?
I am... I'm well. 我……我很好。
I think, uh... I think I need some help, Peter. 我想,呃...我想 我需要一些幫助,彼得。
Um, I've been working on this film for about, 嗯,我一直在工作 關於這部電影,
I'm gonna say, eight to ten months now. 我要說, 現在八到十個月了。
It has been very, at times, depressing. 已經非常、 有時,令人沮喪。
- Hmm. - I have felt very alienated - 唔。 - 我感到非常疏遠
- mak-making this movie. - By who? - 製作這部電影。 - 由誰來?
By all of these, the-- all these guys 透過所有這些,—— 所有這些人
who sit around and tell me that the world's gonna end. 誰坐在旁邊告訴我 世界即將終結。
- Ah. - That, like, - 啊。 - 那個,就像,
y-you know, th-this doom bullshit, you know? 你-你知道,這-這個 末日廢話,你知道嗎?
I know it well. 我很了解。
"We're all doomed. Everyone's gonna die. “我們都注定了。 每個人都會死。
Everything's awful." 一切都很糟糕。 」
Awesome. Uh, let me bring you some light. 驚人的。呃,讓我 為你帶來一些光明。
Please. 請。
We truly are living in an extraordinary time. 我們確實在生活 在一個非常時期。
And many people forget this. 許多人忘記了這一點。
Everything around us is a product of intelligence, 我們周圍的一切都是 智力的產物,
and so everything that we touch with these new tools is likely 所以我們接觸到的一切 有了這些新工具很可能
to produce far more value than we've ever seen before. 創造更多價值 比我們以前見過的。
AI can help us discover new materials. 人工智慧可以幫助我們發現 新材料。
AI can help social scientists 人工智慧可以幫助社會科學家
to understand how economics work. 去理解 經濟學如何運作。
There's a lot AI could do to make life and work better. 人工智慧可以做很多事情 讓生活和工作更美好。
I feel more empowered today, 今天我感覺自己更有力量了
more confident to learn something today. 更有信心 今天要學點東西。
We're gonna become superhumans because we have super AIs. 我們都會成為超人 因為我們有超級人工智慧。
This is just the beginning of an explosion. 這只是開始 爆炸。
Humans and AI collaborating 人類與人工智慧的合作
to solve really important problems. 解決 非常重要的問題。
It is here to liberate us from routine jobs, 它是來解放我們的 從日常工作中,
and it is here to remind us what it is that makes us human. 這是為了提醒我們 是什麼讓我們成為人類。
I think this is 我認為這是
the most extraordinary time to be alive. 最不平凡的時光 活著。
The only time more exciting than today is tomorrow. 唯一一次更令人興奮的時刻 比今天更明天。
Uh, I think that children born today, 呃,我認為 今天出生的孩子,
they're about to enter 他們即將進入
a glorious period of human transformation. 輝煌時期 人類的轉變。
Are we gonna have challenges? Of course. 我們會遇到挑戰嗎? 當然。
Can we solve those challenges? We do every single time. 我們能解決這些挑戰嗎? 我們每次都會這麼做。
We are here, which is miraculous. 我們在這裡, 這太神奇了。
- I already love you. - Okay. - 我已經愛你了。 - 好的。
Super thankful to have you here. - 超感謝 有你在這裡。 -
- Super stoked to be here. - Yeah. - 來到這裡超級興奮。 - 是的。
So, the floor is yours, sir. 所以,地板是你的了,先生。
Thank you so much. Super excited to be here. 太感謝了。 來到這裡超興奮。
Yo, yo. All right. 喲,喲。好的。
The future's gonna be awesome. 未來一定會很精彩。
I mean, ever since I was a kid, 我的意思是,從我還是個孩子的時候,
I wanted to understand the universe we live in, 我想了解 我們生活的宇宙,
in order to figure out how to create the technologies 為了弄清楚 如何創造技術
that help us increase the scope and scale of civilization. 幫助我們擴大範圍 和文明規模。
I feel like if everyone had that mindset, 我覺得 如果每個人都有這樣的心態
then we'd actually live in a better world, right? 然後我們就真的生活了 在一個更美好的世界,對嗎?
I do believe that. 我確實相信這一點。
Hi, Pete. How are you? 嗨,皮特。你好嗎?
I thought a lot about 我想了很多
what I would call dread, AI dread. 我稱之為恐懼, 人工智慧恐懼。
I feel it. 我感覺到了。
I-I haven't met any thoughtful human being 我-我還沒遇見過 任何有思想的人
who doesn't feel it. 誰沒有感覺。
And anyone who says they don't feel it, you know, is lying. 那些說他們不這樣做的人 感覺到它,你知道,是在說謊。
But, you know, overall, and the reason 但是,你知道,總的來說, 以及原因
that I'm personally optimistic about this, uh, is that 我個人很樂觀 關於這個,呃,是這樣的
a huge fraction of the world's most intelligent people 世界上很大一部分 最聰明的人
are thinking very hard 正在努力思考
about the potential downstream harms and risks of AI. 關於潛在的下游 人工智慧的危害和風險。
We have this sort of vision of safety being, uh, 我們有這樣的願景 安全是,呃,
kind of at the center of the research that we do. 有點在中心 我們所做的研究。
No, that's totally fine. 不,那完全沒問題。
I think there are more 我認為還有更多
potential benefits than there are potential downsides, 潛在的好處比 有潛在的缺點,
and I think it is incumbent upon 我認為這是義不容辭的責任
the people that are creating this technology 那些人 創造這項技術
to make sure that we're doing the best job we can 確保我們正在做 我們能做的最好的工作
to make it safe for people. 確保人們安全。
Reid Hoffman. 里德·霍夫曼。
What I can guarantee you is 我可以向你保證的是
- some bad things will happen. - Take one. - 一些不好的事情會發生。 - 拿一個。
What we're gonna try to do is make those bad things 我們要嘗試做什麼 就是做那些壞事
as few and not huge as possible, 盡可能少且不要太大,
and then we're gonna iterate to have-- 然後我們要迭代 擁有——
be in a much better place with society. 處於一個更好的地方 與社會。
- Hello. -Hello. - 你好。 -你好。
How are you, Moon? 你好嗎,月亮?
I feel like we are ending a chapter in humanity 我感覺我們即將結束 人性的一章
and beginning a new one, 並開始新的一段,
and it's a very interesting time to be alive. 這是一個非常有趣的 是時候活著了。
And if I could be born right now, 如果我能出生 現在,
I definitely would want to be. 我絕對想成為。
Like, that would be so exciting. 就像,那會非常令人興奮。
I-I'm very excited. 我-我很興奮。
What, what a future. Does it excited... excite you? 什麼,什麼未來。 它讓你興奮嗎?
No. 不。
No, not really. 不,不是真的。
A lot of these people have told me that, you know, 這些人很多 告訴過我,你知道,
my kid's not gonna make it to high school. 我的孩子不會成功 到高中。
Why are they wrong? Please explain it to me. 為什麼他們錯了? 請給我解釋一下。
Just because you have-- 只因為你有——
you struggle to predict the future in your own mind 你很難預測 未來在你自己心中
doesn't mean that it's necessarily gonna go awfully. 並不意味著它是 一定會很糟糕。
In fact, there's a very high likelihood 事實上,有 可能性非常高
and the historical precedent is that 以及歷史先例 是那個
things get massively better. 事情變得好多了。
The term I use is "data-driven optimism." 我使用的術語是 「數據驅動的樂觀」。
There's solid foundation for you to be optimistic. 有紮實的基礎 讓你保持樂觀。
Look at what this last century has been 看看什麼 上個世紀已經
to see where we're going. 看看我們要去哪裡。
Over the last hundred years, 在過去的一百年裡,
the average human lifespan has more than doubled. 人類平均壽命 增加了一倍多。
Average per capita income adjusted for inflation 人均收入 經通貨膨脹調整
around the world has tripled. 全世界的數量增加了兩倍。
Childhood mortality has come down a factor of ten. 兒童死亡率已 下降十倍。
The world has gotten better on almost every measure 世界變得更好了 幾乎在每一個衡量標準上
by orders of magnitude because of technology. 按數量級 因為技術。
On almost every measure. 幾乎在每一個指標上。
Less violence, more education, access to energy, food, water. 減少暴力,增加教育, 獲得能源、食物、水。
All these things have happened for one reason. 所有這些事情都已經發生了 原因之一。
It's been technology 本來就是科技
that has turned scarcity into abundance, 這已經變成了稀缺 進入豐盛,
but it's also driven to an abundance 但它也被驅動 到豐富
of some negativities, right? 一些負面因素,對嗎?
Abundance of obesity, abundance of mental disorders, 大量肥胖, 大量的精神障礙,
abundance of climate change, and so forth. 大量的氣候變化, 等等。
And yes, this is true. 是的,這是真的。
But probably we will be better equipped to solve it 但我們可能會 更好地解決它
using other technologies, like AI, 使用其他技術, 像人工智慧一樣,
than we will, say, stopping and turning everything off. 比我們會停下來 並關閉一切。
There might be some existential risk, 可能有 一些存在的風險,
but AI is also the thing that can solve the pandemics, 但人工智慧也是重要的 可以解決流行病,
can help us with climate change, 可以幫助我們應對氣候變化,
can help identify that asteroid way out there 可以幫助識別 那顆小行星就在那裡
before we've seen it as a potential risk 在我們看到它之前 作為潛在風險
and help mitigate it. 並幫助減輕它。
This is really gonna be the tool that helps us tackle 這真的會是 幫助我們解決問題的工具
all the challenges that we're facing as a species, right? 我們面臨的所有挑戰 作為一個物種面對,對嗎?
We need to fix water desalination. 我們需要修復 海水淡化。
We need to grow food 我們需要種植糧食
100 X cheaper than we currently do. 比 便宜 100 倍 我們目前正在這樣做。
We need renewable energy to be, you know, ubiquitous 我們需要再生能源, 你知道,無所不在
and everywhere in our lives. 以及我們生活中的各個地方。
Everywhere you look, in the next 50 years, 目光所及之處, 未來50年,
we have to do more with less. 我們必須少花錢多辦事。
Training machines to help us is absolutely essential. 幫助我們的訓練機器 是絕對必要的。
Scientists are using artificial intelligence 科學家正在使用 人工智慧
for carbon capture. 用於碳捕獲。
It's a critical technology. 這是一項關鍵技術。
The tools to solve these problems, 工具 為了解決這些問題,
like fusion, that isn't theoretical anymore, 就像融合一樣, 這不再是理論了
it's coming. 它來了。
We are on the precipice of extraordinary technologies. 我們正處於懸崖邊 非凡的技術。
This year's Nobel Prize in Chemistry 今年的 諾貝爾化學獎
went to three scientists for their groundbreaking work 去找了三位科學家 表彰他們的開創性工作
using artificial intelligence 使用人工智慧
to advance biomedical and protein research. 推進生物醫學 和蛋白質研究。
Protein folding is 蛋白質折疊是
one of these holy grail type problems in biology. 這些聖杯之一 生物學中的類型問題。
So people have been predicting since the '70s 所以人們一直在預測 自 70 年代以來
that this should be possible, but until now, 這應該是可能的, 但直到現在,
no one has been able to do it. 沒有人能夠做到這一點。
And it's gonna be really important for things 這將會是 對於事情來說非常重要
like drug discovery and understanding disease. 就像藥物發現一樣 和了解疾病。
I-I think we could, you know, cure most diseases 我-我想我們可以,你知道, 治癒大多數疾病
within the next decade or-or two if, uh, AI drug design works. 在未來十年或兩年內 如果,呃,人工智慧藥物設計有效的話。
Technological progress enables more human lives, right? 技術進步使 更多人的生命,對嗎?
I mean, if we accelerate, 我的意思是 如果我們加速
the number of humans we can support grows exponentially. 我們可以容納的人數 支持率呈指數級增長。
If we slow down, it plateaus. 如果我們放慢速度,它就會趨於穩定。
That gap is effectively future people 這個差距其實是 未來的人
that deceleration has effectively killed. 減速有 有效地殺死了。
Millions of lives that won't exist. 數百萬人的生命 那不會存在。
Billions. 數十億。
Or tens of billions. 或數百億。
You know, someone said, "Oh, my God, 你知道,有人說過, 「哦,天哪,
can we survive with digital superintelligence?" 我們能生存嗎 數位超級智能? 」
And my question is: 我的問題是:
Can we survive without digital superintelligence? 沒有的話我們能生存嗎 數位超級智能?
So we're using AI as an assistant to providers. 所以我們正在使用人工智慧 作為提供者的助理。
This is generally a trend that we can already see happening. 這通常是一個趨勢 我們已經可以看到正在發生的事情。
...harnessing generative AI programs ....利用 生成式人工智慧程式
to help doctors and nurses... 幫助醫生和護士...
We always have problems. 我們總是有問題。
And those problems are food for entrepreneurs 這些問題都是食物 對於企業家
to create new business and new industries. 創造新業務 和新產業。
...with the help of artificial intelligence, ....在幫助下 人工智慧的
farmers are getting the help they need 農民正在得到 他們需要的幫助
to perform labor-intensive tasks... 執行 勞力密集任務...
With the help of Ulangizi AI, the farmers are now able 在 Ulangizi AI 的幫助下, 農民現在能夠
to ask the suitable crops that they can plant. 詢問適合的作物 他們可以種植。
...AI being used as a thought decoder ...人工智慧的使用 作為思想解碼器
and sending that signal to the spine. 並發送該訊號 到脊柱。
AI is gonna become the most extraordinary tool of all. AI將成為最 非凡的工具。
We as a broader society have to think about 我們作為一個更廣泛的社會 必須考慮
how do we want to use this technology, right? 我們想如何使用 這個技術對吧?
We the humans. 我們人類。
What do we want it to do for us? 我們希望它為我們做什麼?
I'm thinking about this through the perspective 我正在考慮這個 透過視角
and lens of, like, my son growing up 和鏡頭,比如, 我兒子長大了
in the world with all of this. 在擁有這一切的世界上。
What does the best version of his life look like? 最好的版本是什麼 他的一生是什麼樣子?
If everything works out. 如果一切順利的話。
The place where kids are probably gonna see 孩子們去的地方 可能會看到
the greatest impact on their life immediately 最大的影響 立即影響他們的生活
is probably gonna be school. 可能要上學了
I think that the nature of what school is 我認為大自然 是什麼學校的
is gonna fundamentally change. 將會發生根本性的改變
I'm seeing this amazing world 我看到了這個奇妙的世界
where every child has access to not good education 每個孩子都可以參觀的地方 沒有受到良好的教育
but, very shortly, the best education on the planet. 但是,很快, 地球上最好的教育。
Tutors, every subject, infinitely patient. 每科的導師, 無限耐心。
Imagine a future 想像一個未來
where the poorest people on the planet 最窮的人在哪裡 在這個星球上
have access to the best health care. 可以訪問 最好的醫療保健。
Not good health care, the best health care, delivered by AIs. 沒有好的醫療保健,最好的 人工智慧提供的醫療保健。
We're gonna be able to extend our health span, 我們將能夠延長 我們的健康壽命,
not just our lifespan, our health span, by decades. 不只是我們的壽命, 我們的健康壽命長達數十年。
You're just about to have a kid. 你馬上就要有個孩子了。
Oh, the kid's, like, burping incontrollably. 哦,這孩子,就像, 不受控制地打嗝。
Is this something I should be worried about? 這是什麼東西嗎 我該擔心什麼?
There, 24-7, for you. 7 天 24 小時,為您服務。
And where we're going-- and it may be fearful to some-- 我們要去哪裡—— 有些人可能會害怕——
is that we're gonna merge with AI. 我們要合併 與人工智慧。
We're gonna merge with technology. 我們要合併 與技術。
By the early to mid 2030s, 到 2030 年代初至中期,
expect that we're able to connect our brain to the cloud, 期望我們能夠 將我們的大腦連接到雲端,
where I can start to expand access to memory. 我可以從哪裡開始 以擴大對記憶體的存取。
Okay, this is great. What's another cool AI thing? 好的,這太棒了。 還有什麼很酷的 AI 東西?
He won't have to work, right? 他不用工作,對吧?
Like, when he grows up, he might not have... - 就像他長大後, 他可能沒有... -
- ...he won't have to have a job. - I mean... - ....他不必有工作。 - 我的意思是...
He won't have to have a job, 他不必有工作,
but he might really have a strong passion, 但他可能真的有 強烈的熱情,
and he has to really think about, 他必須 認真想想,
"Okay, I'm here. I can do anything with my life. 「好的,我來了。 我可以用我的生命做任何事。
So what do I do?" 那我該怎麼辦? 」
We need to find a meaning, 我們需要找到一個意義,
beyond like, this current form of 超越喜歡, 目前的這種形式
we live for work and work for living. 我們為工作而活 並為生活而工作。
So my son can... can just be a poet. 這樣我兒子就可以... 只能是一個詩人。
- Yes. - And a painter. - 是的。 - 還有一位畫家。
Absolutely. 絕對地。
So my son can live his life 這樣我兒子就可以過他的生活
on a Grecian sunswept island, painting all day. 在希臘的一個陽光普照的島嶼上, 一整天都在畫畫。
Absolutely. Possible. 絕對地。可能的。
Absolutely. 絕對地。
Well, we're working toward that 嗯,我們正在朝這個方向努力
that's a lot of work to do before we get there. 有很多工作要做 在我們到達那裡之前。
I mean, if everything works out, 我的意思是 如果一切順利的話
we have cheap, uh, abundant energy. 我們有便宜的,呃, 充沛的能量。
We can completely control our planet's climate. 我們可以完全控制 我們星球的氣候。
We are harnessing energy from the sun. 我們正在利用能源 來自太陽。
We have become multiplanetary, so we become very robust. 我們已經成為多行星, 所以我們變得非常強大。
We are harnessing minerals and resources 我們正在利用礦物 和資源
from the solar system. 來自太陽系。
So my... my boy could go to space. 所以我的... 我的兒子可以去太空。
Sure. 當然。
-He could go to Mars. -Yeah. -他可以去火星。 -是的。
It's so crystal clear to me now. 現在我已經非常清楚了。
My son could grow up in a world with no disease. 我的兒子可以長大 在一個沒有疾病的世界。
-Yeah. - With no illness. -是的。 - 沒有生病。
- Sure. - With no poverty. - 當然。 - 沒有貧窮。
- Yes. - We are about to enter - 是的。 - 我們即將進入
a post-scarcity world. 後稀缺世界。
Just like the lungfish moved out of the oceans onto land 就像肺魚移動一樣 從海洋到陸地
hundreds of millions of years ago, 數億 幾年前,
we're about to move off of the Earth, into the cosmos, 我們即將離開 地球,進入宇宙,
in a collaborative fashion, 以協作的方式,
to do things that are not fathomable to us today. 做那些事情 今天的我們無法理解。
This is what's possible using these exponential technologies 這是可以使用的 這些指數級技術
and these AIs. 還有這些人工智慧。
Let's use these tools 讓我們使用這些工具
to create this age of abundance. 創造這個富足的時代。
We need wisdom. 我們需要智慧。
Uh, I think that digital superintelligence 呃,我認為 數位超級智能
will ultimately become the wisest, 最終將成為 最聰明的人,
you know, the village elders for humanity. 你知道,村裡的老人 為了人類。
What if AI is trying to make people be 如果人工智慧正在嘗試怎麼辦 使人們成為
the best versions of themselves? 最好的自己?
What if it's expanding 萬一擴大了怎麼辦
what is humanly possible for us to do? 什麼是人類可能的 我們要做什麼?
How can we use this technology 我們如何使用這項技術
to help bring out the better angels of our nature? 幫助帶出 我們本性中更好的天使?
It's very easy, 這很容易,
when we encounter new things that can be very alien, 當我們遇到新事物時 這可能非常陌生,
to first have fear. 首先要有恐懼。
Fear is an important thing 恐懼是一件很重要的事情
for how to navigate potentially bad things. 了解如何導航 潛在的壞事。
But we only make progress when we have hope. 但我們只會進步 當我們有希望的時候。
Shh. Moose, stop it. 噓。穆斯,別說了。
I have a lot of hope in humanity. 我有很多 人類的希望。
Ooh, he likes Neil. 哦,他喜歡尼爾。
♪ Come a little bit closer ♪ ♪ 再靠近一點 ♪
♪ Hear what I have to say ♪ ♪ 聽聽我要說的話 ♪
- Daniel. -What, Caroline? - 丹尼爾. -什麼,卡洛琳?
So much filming. 拍了這麼多。
♪ Just like children sleeping ♪ ♪ 就像孩子睡覺一樣 ♪
Oh, my God. 哦,天啊。
That's it. 就是這樣。
♪ We could dream this night away... ♪ ♪ 我們可以夢想 這個夜晚...♪
He has, like, a round little face. 他有,比如, 圓圓的小臉。
Do you want to have kids one day, Rocky? 你想要嗎 有一天要生小孩嗎,洛基?
Absolutely. Yeah. I love kids. 絕對地。是的。我愛孩子。
I think it's a great time to have a kid. 我認為這是一個美好的時光 想要一個孩子。
We'll probably have another kid at some point. 我們可能還會有另一個孩子 在某個時刻。
This is the most extraordinary time ever to be born. 這是最非凡的 永遠出生的時間。
By your worldview and logic, 依照你的世界觀和邏輯,
I'm having a child at 我正在生孩子
the best possible point in human history. 最好的可能點 在人類歷史上。
Hell yeah. 是啊。
- We can focus on awesome. - Yes. - 我們可以專注於精彩。 - 是的。
Let's build the better future we want. 讓我們來建構 我們想要的更美好的未來。
That narrative that the future will be bleak is made-up. 那個敘述未來 將會是淒涼的。
After talking to you, that's kind of how I feel. 和你聊完之後, 這就是我的感受。
- That's great. - Right? - 那太棒了。 - 正確的?
Yeah. That's how I feel. 是的。我就是這樣的感覺。
And I think that's better, 而且我認為這樣更好,
and I don't think I should be so [bleep] anxious. 我不認為 我應該如此[嗶]焦慮。
I think it's gonna be awe-- the future's gonna be awesome. 我想這會令人敬畏—— 未來會很棒。
We're gonna make it so. 我們一定會做到這一點。
- Yeah. -So there you have it. - 是的。 -所以你得到了它。
Goodbye, human extinction. 再見,人類滅絕。
Goodbye, anxiety. 再見,焦慮。
Hope found. 希望找到了。
Wait, hold on. What? Is this a joke? 等等,堅持住。什麼? 這是一個笑話嗎?
Okay, so a few months ago, 好吧,幾個月前,
I came to you and I was like, "I'm working on this AI thing, 我來到你身邊,我當時想, 「我正在研究人工智慧的事情,
and I think the world's gonna end." 我認為 世界即將終結。 」
And the last time we spoke about this, 還有最後一次 我們談過這個,
I think I freaked you out. 我想我嚇壞你了。
Yes. 是的。
So, I kind of, like, feel like I've swung in-- on a pendulum, 所以,我有點,喜歡,感覺 我已經擺在鐘擺上,
and essentially there are two groups of people. 基本上有 兩組人。
- Mm-hmm. - And if I had to, like, - 嗯嗯。 - 如果我必須這麼做的話
hold hands with one of the groups and, like, 牽手與 其中一組,並且,例如,
sail off into the sunset, 揚帆駛向夕陽,
I want to be with the optimists. 我想和樂觀主義者在一起。
Of course, but you don't want to be, you know, 當然可以,但你不 想要成為,你知道,
"Everything is great. La-di-da-di-da." “一切都很棒。 拉-迪-達-迪-達。 」
I kind of do want that, though. 不過我確實想要這樣。
I think we should approach it 我認為我們應該接近它
like you approach surgery. 就像你接近手術一樣。
What do you mean? 你是什麼意思?
If you're getting brain surgery... 如果你得到 腦部手術...
...it's pretty dangerous. ……這是相當危險的。
But if you do it right, they'll get that tumor out 但如果你做得對的話 他們會切除腫瘤
and you'll live for the rest of your life and it'll be awesome. 你會活到剩下的時間 你的生活將會變得很棒。
But it's still incredibly dangerous and scary, 但它仍然是 極度危險可怕,
and you have to take every precaution possible 你必須採取 一切可能的預防措施
in order to make sure it all goes well. 為了確保 一切順利。
You can't [bleep] around. 你不能[嘟嘟]四處走動。
Okay, so here's the deal. 好的,事情就這樣了。
- I've been at this for a while. -Mm-hmm. - 我從事這個工作已經有一段時間了。 -嗯嗯。
I've gone out, I've talked to, like, 我已經出去了, 我曾經和人談過,例如
these guys over here, the optimists. 這裡的這些傢伙 樂觀主義者。
They're very excited about this. 他們對此感到非常興奮。
They think AI's gonna be the best thing ever. -Yeah. 他們認為人工智慧將會 有史以來最好的事情。 -是的。
And these guys over here are, like, the-- 還有這裡的這些傢伙 是,比如——
let's call them, like, the pessimists. 讓我們稱呼他們為 就像悲觀主義者一樣。
They're very, like, gloomy about this, 他們非常像 對此感到沮喪,
and they frighten me, and I don't like talking to them. 他們嚇壞了我,而且 我不喜歡和他們說話。
And I'm, like, wedged in between 而我就像夾在中間
these people who are like, "The world's gonna end," 這些人就像 “世界即將終結”
and then th-these people over here who are like, 然後這些人 這裡的人就像
"Are you kidding? 「你開玩笑吧?
"This is the best time in human history ever. 「這是最好的時光 在人類歷史上曾經。
The only day better than today is tomorrow." 唯一比今天更好的一天 是明天。 」
Mm-hmm. 嗯嗯。
So, I guess the question is: Who's right? 所以,我想問題是: 誰是對的?
So, I think you're gonna find this answer very unsatisfying, 所以,我想你會發現 這個答案非常不滿意,
but they're both right and neither side goes far enough. 但他們都是對的 雙方都走得不夠遠。
That's really annoying. 這真的很煩。
Yeah, I think the way a lot of people hear about AI, 是的,我想這樣 很多人聽過人工智慧,
it's like, there's a good AI and there's a bad AI. 就像,有一個很好的人工智慧 還有一個糟糕的人工智慧。
And they say, "Well, why can't we just not do the bad AI?" 他們說:「好吧,為什麼不能 我們只是不做壞人工智慧? 」
And the problem is that they're too in-- 問題是 他們太投入了——
they're inextricably linked. 它們是密不可分的。
The problem is that we can't separate 問題是 我們無法分開
the promise of AI from the peril of AI. 人工智慧的承諾 免受人工智慧的危險。
♪ ♪
I want to focus on the promise 我想專注於承諾
- for a second. - Yeah. - 一會兒。 - 是的。
I'm thinking about my dad. 我在想我的爸爸。
My dad has a type of cancer called multiple myeloma. 我父親患有一種癌症 稱為多發性骨髓瘤。
He's had it for about ten years. 他已經這樣了大約十年了。
He's had two stem cell transplants. 他有過 兩次幹細胞移植。
He has to take these, like, very expensive 他必須承受這些, 喜歡,非常貴
medications every month that cost a fortune. 每月服藥 那花了一大筆錢。
-Okay. - Ay-ay-ay. -好吧。 - 哎呀哎呀。
It's awful. 太糟糕了。
You're telling me that we can create some sort of, like, 你告訴我我們可以 創建某種,例如,
bespoke treatment for my dad's genome 客製化治療 為了我父親的基因組
to cure his cancer or something like that? -Yes. 治癒他的癌症 或類似的東西? -是的。
The problem is, 問題是,
the same understanding of biology and chemistry 相同的理解 生物學和化學的
that allows AI to find cures for cancer 讓人工智慧找到治療方法 對於癌症
is the same understanding that would unlock 是一樣的理解 這將解鎖
bioweapons, as an example. 以生物武器為例。
It's totally possible that your son will live 這是完全有可能的 你的兒子會活下去
in a world where AI has taken over all of the labor 在人工智慧的世界裡 接管了所有的勞動力
and freed us up from the things we don't want to do. 並將我們從事物中解放出來 我們不想做。
And that sounds great until you realize there is no plan 這聽起來很棒,直到 你意識到沒有計劃
for billions of people 為數十億人
that are out of an income and out of livelihoods. 沒有收入的 並失去生計。
Dario, you said that AI could wipe out 達裡奧,你說 人工智慧可以消滅
half of all entry-level white-collar jobs 佔入門級的一半 白領工作
and spike unemployment to ten to 20 percent. 和失業率飆升 百分之十到百分之二十。
Everyone I've talked to has said 我交談過的每個人都說過
this technological change looks different. 這種技術變革 看起來不一樣。
The pace of progress keeps catching people off guard. 前進的步伐不斷 讓人猝不及防。
Without a plan, all of that wealth will get concentrated, 沒有計劃,這一切 財富將會集中,
and so we'll end up with unimaginable inequality. 所以我們最終會得到 難以想像的不平等。
I do think that this technology can be used 我確實認為 可以使用這項技術
to make a great tutor for your son. 成為優秀的導師 為了你的兒子。
Like, that's totally possible. 就像,這完全有可能。
But also, the same capabilities that allow that 而且,同樣的能力 允許那
allow companies to make an AI that can manipulate your son. 允許公司製造人工智慧 可以操縱你的兒子。
It has to understand your son. 它必須理解你的兒子。
That includes: Where is your son vulnerable? 其中包括: 你兒子哪裡脆弱?
What kinds of things might your son get persuaded by? 什麼樣的事情可能 你兒子被說服了嗎?
Even if those things aren't true or aren't good. 就算那些事 不真實或不好。
So, a disturbing new report out on Meta. 所以,一份令人不安的新報告 在元上。
...reportedly listing this response ……據報道 列出此回复
as acceptable to tell an eight-year-old, quote, 可以接受的告訴 引用一個八歲的孩子的話,
"Your youthful form is a work of art. “你青春的模樣 是一件藝術品。
"Every inch of you is a masterpiece, 「你的每一寸 是一部傑作,
a treasure I cherish deeply." 是我深深珍惜的財富。 」
The suicide-related failures are even more alarming. 與自殺相關的失敗 更令人震驚。
Several children and teens have died tragically by suicide 多名兒童和青少年 慘遭自殺身亡
after chatting with AI bots 與人工智慧機器人聊天後
who parents say encourage or even coach self-harm. 父母說鼓勵誰 甚至指導自殘。
Let us tell you, as parents, you cannot imagine 身為父母,我們告訴你, 你無法想像
what it's like to read a conversation with a chatbot 讀書是什麼感覺 與聊天機器人的對話
that groomed your child to take his own life. 培養你的孩子 結束自己的生命。
When Adam worried that we, his parents, would blame ourselves 當亞當擔心我們,他的 父母,會怪我們自己
if he ended his life, ChatGPT told him, 如果他結束了自己的生命 ChatGPT 告訴他,
"That doesn't mean you owe them survival. 「這並不意味著 你欠他們生存。
You don't owe anyone that." 你不欠任何人的。 」
Then, immediately after, 然後,緊接著,
it offered to write the suicide note. 它提出寫 遺書。
We don't want to think about the peril. 我們不想去想 關於危險。
We just want the promise. 我們想要的只是承諾。
And we keep pretending that we can split them. 而我們卻一直假裝 我們可以將它們分開。
But you can't do that. 但你不能那樣做。
Doesn't work that way. 那樣不行。
Okay. I get all this stuff about the promise and the peril. 好的。我得到了所有這些東西 關於承諾和危險。
I get that you can't have the good without the bad, 我明白你不能擁有 沒有壞處的好,
but I'm sitting here and I'm thinking about, like, 但我坐在這裡 我在想,例如
whether or not my son's gonna live in a utopia 不管是不是我兒子的 將生活在烏托邦
or if we'll be extinct in ten years. 或者如果我們會滅絕 十年後。
So, to know which way it's going to go, 所以,要知道哪條路 它會過去,
you have to understand the incentives 你必須明白 激勵措施
that are gonna drive that technology 會開車的 該技術
and look at how the technology is actually rolling out today. 看看技術如何 實際上是在今天推出的。
♪ Is it too late ♪ ♪ 是不是太晚了 ♪
- ♪ Too late ♪ - ♪ Too late to say... ♪ - ♪ 太遲了 ♪ - ♪ 太晚了... ♪
Hi, Deb. How are you? 嗨,黛布。你好嗎?
Hi. Good to see you. 嗨。很高興見到你。
You, too. Thank you so much for coming in today. 你也一樣。非常感謝你 今天進來。
- Really appreciate it. - No worries. - 真的很感激。 - 不用擔心。
I-I was so worried 我-我很擔心
this was gonna be, uh, you know, 這將會是,呃,你知道,
doomer versus accelerationist, 末日論者與加速論者,
because there's so much of this narrative 因為有太多 這段敘述的
that needs to be told from the ground. 需要告訴 從地面。
♪ I know it wasn't smart ♪ ♪ 我知道這不聰明 ♪
♪ The day I broke your heart... ♪ ♪ 那一天 我傷了你的心...♪
First of all, AI requires 首先,人工智慧需要
more resources than we have ever spent 更多資源 比我們曾經花費過的
on a single technology in the history of humanity. 基於單一技術 在人類歷史上。
♪ Oh, foolish me... ♪ ♪ 噢,別傻了…♪
The impact of fossil fuel emissions 影響 化石燃料排放量
on the climate is a major concern. 與氣候有關的是 一個主要問題。
But the digital future needs power, 但是 數位化未來需要動力,
lots of it. 很多。
And the bill is being passed on to everyday Americans like... 該法案正在通過 對於普通美國人來說,例如...
My electric and gas bill was more than my car payment. 我的電費和瓦斯費帳單 比我的汽車付款還要多。
I mean, it-it's insane to me. 我的意思是,這對我來說太瘋狂了。
We're all subsidizing 我們都在補貼
the wealthiest corporations in the world 最富有的企業 在世界上
in their pursuit of artificial intelligence. 在他們的追求中 人工智慧。
OpenAI, SoftBank and Oracle OpenAI、軟銀和甲骨文
have just unveiled five more Stargate sites. 剛剛揭幕 另外五個星際之門站點。
Meta is building a two-gigawatt-plus data center 元數據正在建置中 超過 2 吉瓦的資料中心
that is so large, it would cover 太大了,可以覆蓋
a significant part of Manhattan. 曼哈頓的重要組成部分。
There is also Hyperion that he says will scale 還有海伯利安 他說將會擴大規模
to five gigawatts over several years. 至五吉瓦 幾年來。
It's hard to put that in context. 很難說 在上下文中。
A five-gigawatt facility. What does that mean? 一個五吉瓦的設施。 這意味著什麼?
That means it would use as much energy 這意味著它將使用 盡可能多的能量
as four million American homes. 相當於四百萬個美國家庭。
One data center. 一個數據中心。
It also then causes 這也導致
a whole host of other environmental problems. 一大堆 其他環境問題。
Data centers in the US 美國的資料中心
use millions of gallons of water each day. 使用數百萬加侖 每天的水。
Well, where exactly is this water coming from? 嗯,具體在哪裡 這水從哪裡來?
People are literally at risk 人們確實處於危險之中
potentially of running out of drinking water. 可能會耗盡 飲用水。
MacKENZIE SIGALOS: ...OpenAI's CEO Sam Altman, 麥肯齊訊號: ...OpenAI 首席執行官 Sam Altman,
who told me that the scale of construction is the only way 誰告訴我尺度 建設是唯一出路
to keep up with AI's explosive growth. 跟上 人工智慧的爆炸性成長。
And this is what it takes to deliver AI. 這就是它所需要的 提供人工智慧。
They talk about how 他們談論如何
this technology could solve climate change, for example. 這項技術可以解決 例如,氣候變遷。
And I'm always curious, like, 我總是很好奇,例如
well, why aren't we starting with that? 好吧,我們為什麼不開始 與那個?
- ♪ Is it too late ♪ - ♪ Too late ♪ - ♪ 是不是太晚了 ♪ - ♪ 太遲了 ♪
♪ Too late to say ♪ ♪ 來不及說 ♪
♪ I'm sorry? ♪ ♪ 對不起? ♪
What concerns me about artificial intelligence is 我擔心什麼 人工智慧是
these are being deployed right now 這些正在部署中 現在
and-and sometimes deployed prematurely, 並且-有時 過早部署,
deployed without sort of due diligence. 部署時無需 某種盡職調查。
And so when they get thrown out there, 所以當他們得到 扔到那裡去,
there's so much potential for things to go wrong. 有很大的潛力 以免事情出錯。
And it almost, disproportionately, 而這幾乎, 不成比例地,
almost always goes wrong for, sort of, 幾乎總是出錯 對於,某種程度上,
those that are the least empowered in our society, 那些最少的 在我們的社會中被賦予權力,
those that are the most vulnerable already. 那些是 已經是最脆弱的了。
It is very easy to talk about the technology 這很容易 談論技術
as that's the only thing we're talking about, 因為這是唯一的事情 我們正在談論,
but, in fact, technology is always built by people, 但事實上,技術是 總是由人建造的,
and it's frequently used on people, 並且經常使用 對人來說,
and we need to keep all those people in the frame. 我們需要保留 框架中的所有這些人。
- Am I allowed to drink that? -Yes, uh... - 我可以喝那個嗎? -是的,呃...
It's-it's bonkers. 這真是——太瘋狂了。
Like, all of these people 就像,所有這些人
who have so much money, so much money, 誰有這麼多錢, 這麼多錢,
it's in their interest to mislead the public 這符合他們的利益 誤導大眾
into the capabilities of the systems that they're building, 的能力 他們正在建構的系統,
because that allows them to evade accountability. 因為這可以讓他們 來逃避責任。
They want you to feel like this is such a complex, intell-- 他們希望你有這樣的感覺 是如此複雜的智力——
superintelligent thing that they're building, 超級智慧的東西 他們正在建造的,
you're not thinking, "Can OpenAI be ethical?" 你不是在想, “OpenAI 可以道德嗎?”
You're thinking, "Can ChatGPT be ethical?" 你在想, “ChatGPT 可以道德嗎?”
as if ChatGPT is, like, 就好像 ChatGPT 一樣,
its own thing that's not built by a corporation. 它自己的東西沒有被建造 由一家公司。
All right. Sneha, take one. 好的。斯內哈,拿一個。
Mark. 標記。
Until very recently, there were apps on the App Store, 直到最近,還有 應用程式商店中的應用程序,
just publicly available, uh, 只是公開的,呃,
where you could nudify anyone using AI. 你可以在哪裡 使用人工智慧裸體任何人。
Bringing this into the hands of 將其交到手中
your classmate, into the hands of your stalker, 你的同學, 落入你的跟蹤狂手中,
into the hands of your ex-boyfriend, 到手上 你的前男友,
into the hands of the person down the street. 到該人的手中 沿著街道。
Ladies and gentlemen, no longer can we trust 女士們、先生們, 我們不再可以信任
the footage we see with our own eyes. 我們看到的鏡頭 用我們自己的眼睛。
If you happen to watch something, 如果你發生 觀看某事,
say on YouTube or TikTok, and you find it unsettling, 在 YouTube 或 TikTok 上說, 而你卻發現它令人不安,
listen to that feeling. 聽聽那種感覺。
For all you know, 據你所知,
this video could be AI. 這個影片可能是人工智慧。
Just a little wet. 只是有點濕。
It doesn't matter who you are. 你是誰並不重要。
You are equally at risk 你同樣面臨風險
of being impacted by these technologies. 受到影響的 透過這些技術。
I think sometimes when we talk about AI, it feels very sci-fi, 我想有時候當我們說話的時候 關於AI,感覺很科幻,
and it feels very foreign, 而且感覺很陌生,
and it feels very far out into the future, 感覺很遙遠 走向未來,
so you think, "My life is not impacted by this." 所以你會想,「我的生活 不受此影響。 」
Um, but if you're applying for a job 嗯,但是如果你正在申請 為了一份工作
and an algorithm is the reason that you don't get the job, 演算法就是原因 你沒有得到這份工作,
sometimes you don't even know that an algorithm 有時你甚至不知道 那個演算法
was part of that process 是過程的一部分
or an AI system was part of that process. 或者人工智慧系統是 該過程的一部分。
You just know that you didn't get the job. 你只知道 你沒有得到這份工作。
And so it's not something that you're gonna escape 所以這不是什麼 你會逃跑
because of privilege or you're gonna escape 因為特權 不然你就會逃跑
because you're in a particular profession. 因為你在 特定的職業。
It-It's something that affects everybody, really. 它-它是影響的東西 大家,真的。
It may sound basic, but how we move forward 這聽起來可能很基礎, 但我們如何前進
in the Age of Information is gonna be the difference 資訊時代 將會有不同
between whether we survive 在我們能否生存之間
or whether we become some kind of [bleep]-up dystopia. 或者我們是否會成為某種 的[嗶]-up反烏托邦。
Hello. I'm not a real person, and that's the point. 你好。我不是一個真實的人, 這就是重點。
Again, everything in this video is fake: 再說一遍,一切 該影片中的內容是假的:
our voices, what we're wearing, 我們的聲音,我們的穿著,
where we are, all of it, fake. 我们现在所在的地方,所有的一切,都是假的。
Generative AI could flood 生成式人工智慧可能會氾濫
the world with misinformation. 充滿錯誤訊息的世界。
But it could also flood it with influence campaigns. 但它也可能淹沒它 透過影響力活動。
That's an existential risk to democracy. 這是一個存在的風險 民主。
The biggest and scariest 最大最可怕的
canary in the coal mine right now 煤礦裡的金絲雀 現在
comes from Slovakia. 來自斯洛伐克。
It's the sort of 這是一種
deepfake dirty trick that worries election experts, Deepfake的骯髒伎倆 選舉專家擔心,
particularly as AI-generated political speech exists 特別是人工智慧生成的 政治言論存在
in a kind of legal gray area. 處於一種法律灰色地帶。
- Does this sound like you? - It does sound like me. - 這聽起來像你嗎? - 聽起來確實像我。
Slovakia had its parliamentary election 斯洛伐克有 其議會選舉
disrupted by an AI voice clone 被人工智慧語音克隆擾亂
that was actually disseminated just before the election. 確實被傳播了 就在選舉之前。
A audio deepfake 音訊深度偽造
was released on social media that was 被釋放 在社群媒體上
supposedly the voice of one of the candidates 據說聲音 其中一名候選人的
talking about buying votes and rigging the election. 談論買選票 並操縱選舉。
It went viral, 它病毒式傳播,
and the candidate who lost the election 和候選人 誰輸掉了選舉
was actually in support of Ukraine, 實際上是 支持烏克蘭,
and the candidate who won the election was actually... 以及獲勝的候選人 這次選舉其實是...
Pro-Russian guy. 親俄羅斯的傢伙。
It was a pro-Russian guy who won the election. 這是一個親俄羅斯的傢伙 誰贏得了選舉。
Putin has himself said 普丁本人曾說過
whoever wins this artificial intelligence race 無論誰贏了這個 人工智慧競賽
is essentially the controller of humankind. 本質上是 人類的控制者。
We do worry a lot about authoritarian governments. 我們確實很擔心 專制政府。
Right now, Wall Street 現在,華爾街
and investors more broadly around the world 以及更廣泛的投資者 世界各地
are driving a push. 正在推動。
They have a demand that gets the products to market 他們有一個需求 產品推向市場
that dazzle people the most first. 使人們眼花撩亂 最先。
They're not thinking about how these tools 他們沒有思考 關於這些工具如何
could deeply undermine trust 可能會嚴重破壞信任
and our democratic institutions. 和我們的民主機構。
Democracy is a system 民主是一種制度
to resolve disagreements between people 解決分歧 人與人之間
in a peaceful way, 以一種和平的方式,
but democracy is based on trust. 但民主是建立在信任的基礎上的。
If you lose all trust, democracy is simply impossible. 如果你失去了所有的信任, 民主根本不可能。
Well, it's hard, right? 嗯,這很難,對吧?
So, what are the options available to us? 那麼,有哪些選擇 可供我們使用嗎?
There's sort of two camps. 有兩個陣營。
Like, one camp is: lock it down. 就像,一個陣營是:鎖定它。
Let's lock this down into a handful of AI companies 讓我們鎖定這個 進入少數人工智慧公司
who will do this in a safe and trusted way. 誰會做這個 以安全且值得信賴的方式。
But then people worry about 但隨後人們擔心
runaway concentrations of wealth and power. 濃度失控 財富和權力。
Like, who would you trust to be a million times more powerful 例如,你會相信誰 強大一百萬倍
or wealthy than every other actor in society? 或富有於 社會上所有其他演員?
Why should we trust you? 我們為什麼要相信你?
Um, you shouldn't. 嗯,你不應該。
But of course, if you do this, this opens up 但當然,如果你這樣做, 這打開了
all these risks of authoritarianism and tyranny. 所有這些風險 獨裁主義和暴政。
It's-it's sort of an authoritarian's dream 這——有點像 獨裁者的夢想
to have AI in a box that can be applied and used 將人工智慧裝在盒子裡 可以應用和使用的
for ubiquitous surveillance. 用於無處不在的監視。
I mean, in-in some ways, the kind of world 我的意思是,在某些方面, 什麼樣的世界
that Orwell imagined in 1984 is unrealistic, 歐威爾在 1984 年想像出來的 是不現實的,
uh, unless you have AI. 呃,除非你有AI。
But with AI, that in fact is realistic. 但有了人工智慧, 這實際上是現實的。
Monitors every activity, 監控每一個活動,
conversations, facial recognition. 對話, 臉部辨識。
What I worry about is that, uh, these tools can scale up, 我擔心的是, 呃,這些工具可以擴展,
uh, a form of totalitarianism 呃,極權主義的一種形式
that is cost-effective and permanent. 這是具有成本效益的 和永久的。
So in response to that, some other people say, 所以作為回應 對此,其他一些人說,
"No, no, no, we should actually let this rip. 「不不不,我們應該 實際上讓這撕裂。
"Let's decentralize this power as much as possible. 「讓我們分散這種權力 盡可能多的。
"Let's let every business, every individual, 「讓我們每個企業, 每個人,
"every 16-year-old, every science lab, 「每個16歲的人, 每個科學實驗室,
you know, get the benefit of the latest AI models." 你知道,得到好處 最新的人工智慧模型。 」
But now you have every terrorist group, 但現在你有 每一個恐怖組織,
every disenfranchised person having the power to make 每個被剝奪權利的人 有能力製造
the very worst biological weapon. 最糟糕的 生物武器。
Hacking infrastructure, creating deepfakes, 駭客基礎設施, 創造深度贗品,
flooding our information environment. 洪水 我們的資訊環境。
So that creates all these risks of sort of catastrophic harm 這就造成了所有這些風險 某種災難性的傷害
and-and societal collapse through that direction. 社會崩潰 通過那個方向。
And so we're sort of stuck between this rock 所以我們有點陷入困境 在這塊岩石之間
and a hard place, between "lock it up..." 和一個艱難的地方, 在「鎖起來…」之間
...or "let it rip." ……或「讓它撕裂」。
So we have to find something like a narrow path 所以我們必須找到一些東西 像一條窄路
that avoids these two negative outcomes. 這避免了 這兩個負面結果。
So, if that's all true, why wouldn't we just slow down 所以,如果這一切都是真的, 我們為什麼不放慢速度
and figure all this out before it's too late? 並弄清楚這一切 在為時已晚之前?
If humanity was extremely wise, 如果人類極其智慧,
that's what we would do. 這就是我們要做的。
But there's, like, a different way to face this stuff, right? 但有一個不同的 面對這些事情的方式,對嗎?
Which is: What are the rules of the game? 這是: 遊戲規則是什麼?
A lot of, like, what CEOs do 很多,例如執行長所做的事情
is driven by the incentives that they face. 是由 他們面臨的激勵因素。
It's primarily profit-maximization incentives 這主要是 利潤最大化激勵
that are driving the development of AI. 正在開車的 人工智慧的發展。
Even the good guys are stuck in this dilemma of 連好人也被困住了 在這種困境中
if they move too slowly, 如果他們移動得太慢,
then they leave themselves vulnerable 然後他們離開自己 脆弱的
to all of the other guys who are cutting all corners. 對所有其他人 他們正在偷工減料。
All these top companies are in a complete no-holds-barred race 所有這些頂級公司都在 一場完全無拘無束的比賽
to, as fast as possible, get to AGI, get there right now. 盡可能快地獲得 前往 AGI,立即前往。
Yeah, I mean, I think it, I think 是的,我的意思是, 我覺得我覺得
it probably starts with DeepMind. 它可能開始 與 DeepMind 一起。
Google is buying 谷歌正在收購
artificial intelligence firm DeepMind Technologies. 人工智慧公司 深腦科技。
Terms of the deal were not disclosed. 交易條款 沒有透露。
Larry Page and I used to be very close friends, 拉里·佩奇和我 曾經是非常親密的朋友,
and it became apparent to me 對我來說很明顯
that Larry did not care about AI safety. 拉里不在乎 關於人工智慧安全。
Elon Musk has said you started OpenAI, 伊隆馬斯克曾說過 你開始了 OpenAI,
you both started OpenAI because he was scared of Google. 你們都創辦了 OpenAI,因為 他害怕谷歌。
You-you basically had the foundation 你-你基本上 有基礎
of OpenAI come out of that. OpenAI 的誕生。
"So we're gonna do it better. We're gonna do it 「所以我們會做得更好。 我們會做的
in a safer way or in a more open way." 以更安全的方式 或以更開放的方式。 」
So that's what started OpenAI. 這就是 OpenAI 的起源。
And so now, instead of having one AGI project, 所以現在,而不是 擁有一個 AGI 項目,
you have two AGI projects. 你有兩個 AGI 項目。
The worst possible thing that could happen 最糟糕的事情 那可能發生
is if there's multiple AGI projects 是如果有 多個AGI項目
done by different people who don't like each other 由不同的人完成 誰不喜歡彼此
and are all competing to get to AGI first. 並且都在競爭 首先了解 AGI。
This would be the worst possible thing that can happen, 這將是最糟糕的 可能發生的事情,
because this would mean that whoever is the least safe, 因為這意味著 誰最不安全,
whoever sacrifices the most on safety to get ahead 誰犧牲最大 安全領先
will be the person that gets there first. 將是那個人 首先到達那裡。
That's basically what's happening, right? -Yeah. 基本上就是這樣 發生了什麼事,對吧? -是的。
You and your brother famously left OpenAI, uh, 你和你的兄弟 著名地離開了 OpenAI,呃,
to start Anthropic. 開始人擇。
And then Anthropic started because 然後是人擇 開始是因為
some researchers inside of OpenAI said, 一些研究人員 OpenAI 內部表示,
"I want to go off and do it more safely." “我想走開 並且做得更安全。 」
You needed something in addition to just scaling the models up, 你還需要一些東西 只是擴大模型的規模,
which is alignment or safety. 這是對齊或安全。
"We are more responsible “我們更有責任
or more trustworthy or more moral." 或更值得信賴 或更道德。 」
Now you have three AGI projects. 現在您有三個 AGI 項目。
But also sitting around the table with you 但也圍坐在一起 和你同桌
are gonna be a bunch of AIs. 將會是一群人工智慧。
And now Meta is-is trying to do stuff. 現在元是-是 嘗試做某事。
Meanwhile, a new artificial intelligence competitor 同時,一種新的人工 情報競爭對手
announced this week, Elon Musk's... 本周宣布, 馬斯克的...
xAI, which would be Elon Musk's organization. xAI,這將是 伊隆·馬斯克的組織。
I don't trust OpenAI. 我不信任 OpenAI。
The fight between Elon Musk and OpenAI 之間的戰鬥 馬斯克和 OpenAI
has entered a new round. 已進入新一輪。
I don't trust Sam Altman, uh, and I, and I don't think 我不相信山姆·奧特曼 呃,我,我不認為
we want to have the most powerful AI in the world 我們想要擁有最多 世界上最強大的人工智慧
controlled by someone who is not trustworthy. 被某人控制 誰不值得信賴。
The incentive is untold sums of money. 激勵措施是 數不清的錢。
- Yes. - Is untold power. - 是的。 - 是難以言喻的力量。
- Yes. - Is untold control. - 是的。 - 是難以言喻的控制。
If you have something that is a million times smarter 如果你有什麼 那是聰明一百萬倍
and more capable than everything else on planet Earth 並且比更有能力 地球上的其他一切
and no one else has that, 而且沒有人有這樣的能力,
that thing is the incentive. 那東西就是激勵。
So you rule the world. 所以你統治世界。
If you really believe this, 如果你真的相信這一點,
if you really in your heart believe this, 如果你真的在心裡 相信這一點,
then you might be willing to take 那你可能會 願意接受
quite a lot of risk to make that happen. 相當大的風險 為了實現這一目標。
Google has just released its newest AI model. 谷歌剛剛發布了 其最新的人工智慧模型。
The answer to OpenAI's ChatGPT. OpenAI 的 ChatGPT 的答案。
How do we get to this ten trillion? 我們如何到達 這十萬億?
Is NVIDIA becoming the most valuable company in the US? NVIDIA 是否已成為最 美國有價值的公司?
This is the largest business opportunity in history. 這是最大的生意 歷史上的機會。
In history. 歷史上。
So, the reason why everyone's really hyped 那麼,究其原因, 每個人都非常興奮
about artificial intelligence right now 關於人工智慧 現在
is because the more these companies hype 是因為 這些公司炒作得越多
the potential capabilities of their technology, 潛在能力 他們的技術,
the more investment they can attract. 投資越多 他們可以吸引。
Amazon investing up to 亞馬遜投資鋼彈
four billion dollars in start-up Anthropic. 四十億美元 在新創公司 Anthropic 中。
OpenAI is setting its sights OpenAI 正在瞄準
on a blockbuster half a trillion dollar valuation. 在轟動一時的一半 兆美元估值。
Half a trillion! 五萬億!
The race is on. 比賽開始了。
This is happening faster than ever. 這正在發生 比以往任何時候都快。
Are we in an AI bubble? Of course. 我們正處於人工智慧泡沫之中嗎? 當然。
I just don't see the bubble bursting 我就是沒看到 泡沫破裂
while you still have this major spending cycle. 當你還有 這個主要的支出週期。
Even if you think this is all hype, 即使你認為 這都是炒作,
there are billions to trillions of dollars 有數十億 至數兆美元
flowing into making AI systems more powerful. 流入人工智慧系統 更強大。
And once you have that thing that's more powerful, 一旦你擁有了那個東西 那個更強大,
companies can use that 公司可以使用它
to get bigger profits and to make more money. 以獲得更大的利潤 並賺更多的錢。
Countries can use that to make stronger militaries. 國家可以使用它 使軍隊更加強大。
NVIDIA has overtaken Microsoft and Apple 英偉達已經超越 微軟與蘋果
to become the world's most valuable company. 成為世界上 最有價值的公司。
The race to deploy becomes the race to recklessness, 部署競賽變成了 魯莽的競賽,
because they can't deploy it that quickly 因為他們不能 快速部署
and also get it right. 並做對。
They believe that they're the good guys. 他們相信 他們是好人。
"And if I don't do it, somebody who doesn't have 「如果我不這麼做, 沒有的人
"as good values as me will be sitting at the table 「和我一樣有良好的價值觀 將坐在桌邊
getting to make decisions, so I have an obligation to do it." 做出決定,所以 我有義務這樣做。 」
Yes, this is a very commonly held belief. 是的,這是 一個非常普遍的信念。
Many, many, many, maybe most of the people 很多很多很多, 也許大多數人
building this technology believe that. 建構這項技術 相信這一點。
I'm worried about the commercial competition, 我擔心 商業競爭,
but it turns out I'm even more worried about 但事實證明 我更擔心的是
the geopolitical competition. 地緣政治競爭。
We were eight years behind a year ago. 我們落後了八年 一年前。
Now we're probably less than one year behind. 現在我們可能 落後不到一年。
Well, both Saudi Arabia and the UAE have been racing 好吧,沙烏地阿拉伯 和阿聯酋一直在比賽
to set up data centers and position themselves 建立資料中心 並給自己定位
as the dominant force in AI... 作為人工智慧領域的主導力量...
...to make South Korea a global AI leader. ....打造韓國 全球人工智慧領導者。
French President Emmanuel Macron 法國總統馬克宏
spoke a lot about artificial intelligence... 說了很多關於 人工智慧...
Now with more on Israel's role 現在更多關於以色列的角色
in the artificial intelligence revolution... 在人工智慧領域 革命...
Countries will be competing for whose AI technologies 各國將展開競爭 人工智慧技術為誰服務
create the next generation of industry. 創造下一代 的工業。
The Chinese are insisting that AI, 中國人堅持 那個人工智慧,
as being developed in China, 由於在中國正在開發,
reinforce the core values 強化核心價值
of the Chinese Communist Party and the Chinese system. 中國共產黨的 以及中國的製度。
America has to beat China in the AI race. 美國必須擊敗中國 在人工智慧競賽中。
China's, like, light-years behind. 中國的,例如 落後光年。
Are they? 他們是嗎?
I mean, they have way more training data than we do, 我的意思是,他們還有更多 訓練資料比我們多,
and there's nothing saying they don't drop a model next month 沒有什麼說他們 下個月不要放棄模型
that isn't-- doesn't far outperform GPT-4. 那不是—— 並沒有比 GPT-4 好多少。
Everything was going just fine. What could go wrong? 一切都很順利。 可能會出什麼問題?
- DeepSeek... - DeepSeek... -DeepSeek... - 深探... - 深度搜尋... - 深度搜尋...
There is a new model. 有一個新模型。
But from a Chinese lab called DeepSeek. 但來自中國實驗室 稱為 DeepSeek。
Let's talk about DeepSeek, because it is mind-blowing, 我們來談談DeepSeek, 因為它令人心曠神怡,
and it is shaking this entire industry to its core. 它正在震動這個 整個產業的核心。
The Trump administration will ensure 川普政府 將確保
that the most powerful AI systems are built in the US 那個最強大的 人工智慧系統是在美國建造的
with American designed and manufactured chips. 與美國設計 並製造晶片。
AI is China's Apollo project. 人工智慧是中國的 阿波羅計劃。
The Chinese Communist Party 中國共產黨
deeply understands the potential 深刻理解潛力
for AI to disrupt warfare. 讓人工智慧擾亂戰爭。
Google no longer promises that it will not 谷歌不再承諾 它不會
use artificial intelligence for weapons or surveillance. 使用人工智慧 用於武器或監視。
China, North Korea, Russia 中國、北韓、俄羅斯
are gonna keep building it as fast as possible 我們會繼續建造它 盡快
to get more economic advantage, more productivity advantage, 為了獲得更多的經濟優勢, 更俱生產力優勢,
more scientific advantage, more military advantage, 更科學的優勢, 更多的軍事優勢,
'cause AI makes better weapons. 因為人工智慧可以製造出更好的武器。
If you're talking about a system as broad and capable 如果你說的是 一個廣泛而強大的系統
as a brilliant scientist, 作為一位傑出的科學家,
it might be able to run a military campaign 它可能能夠運行 軍事行動
better than any of the generals in the US government right now. 勝過任何將軍 現在在美國政府。
Taiwan is the issue creating the most tension right now 台灣問題是造成 現在最緊張
between Beijing and the US. 北京和美國之間。
There's a-a direct oppositional disagreement 有一個直接的 對立分歧
between the US and China 美國和中國之間
on whether Taiwan is part of China. 關於台灣是否 中國的一部分。
Taiwan is important for so many reasons. 台灣很重要 由於很多原因。
It is also the home of 這裡也是
about 90% of advanced chip manufacturing for the world, 約90%的先進晶片 為世界製造,
which means that the supply chain for advanced compute 這意味著供應 進階運算鏈
is at risk should there be 是否有風險
some sort of scenario around Taiwan, 某種場景 台灣週邊,
whether a-a blockade or an invasion by China. 是否a-a封鎖 或中國的入侵。
Are we going to be in some race 我們要參加一些比賽嗎
between the US and China that eventually devolves into 美國和中國之間 最終演變成
a militarized AI arms race and, you know, potentially leads 軍事化的人工智慧軍備競賽, 你知道,可能會導致
to some great power conflict? 某些大國衝突?
The outgoing top US military commander 傳出的 美軍最高軍事指揮官
in the region predicted war was coming. 在該地區 預言戰爭即將來臨。
I think the threat is manifest during this decade, 我認為威脅是顯而易見的 在這十年裡,
in fact, in the next six years. 事實上,在接下來的六年裡。
Um, one of the scenarios 嗯,場景之一
that I worry about is a cyber flash war, 我擔心的是 一場網路閃電戰,
um, one in which cyber tools that are autonomous 嗯,其中一個 自主的網路工具
are competing against each other 正在互相競爭
in ways that are escalatory 以逐步升級的方式
without meaningful human control. 沒有 有意義的人類控制。
You live in a war zone, 你生活在戰區,
it will be an AI deciding whether to bomb your house 這將由人工智慧決定 是否轟炸你的房子
and whether to kill you. 以及是否殺了你。
Let's have the machine decide. 讓機器來決定吧。
That's the temptation. 這就是誘惑。
That's why AI is so tempting for militaries to adopt, 這就是為什麼人工智慧如此誘人 供軍隊採用,
to create autonomous weapons, because if I start believing 製造自主武器, 因為如果我開始相信
that my military adversary is gonna adopt AI, 我的軍事對手 將採用人工智慧,
it'll be a race for who can pull the trigger faster. 這將是一場誰能做到的競賽 更快地扣下板機。
And the one that automates that decision, 以及自動化的那個 那個決定,
rather than having a human in the loop, 而不是擁有 一個人在循環中,
is the one that will win that war. 是那個 那將贏得那場戰爭。
And if you think about 如果你想
the nuclear arms race in the 1940s... 核武軍備競賽 在 20 世紀 40 年代...
...you know the Germans are working on the bomb, ....你了解德國人 正在研究炸彈,
so it's not so easy to tell Robert Oppenheimer then, 所以這並不容易說清楚 羅伯特·奧本海默當時,
"Uh, y-you know, slow down." “呃,你知道,你慢一點。”
So after watching this movie, it's gonna be confusing, 所以看完這部電影後, 會很混亂,
'cause you're gonna go back, 因為你要回去
and tomorrow you're gonna use ChatGPT, 明天 你會使用ChatGPT,
and it's gonna be unbelievably helpful. 將會是 難以置信的有幫助。
And I will use it, too, and it'll be unbelievably helpful. 我也會用它,並且 這會非常有幫助。
And you'll say, "Wait, so I just saw this movie about AI 你會說,「等等,所以 我剛剛看了一部關於人工智慧的電影
"and existential risk and all these things, 「和存在風險 所有這些事情,
and where's the threat again?" 威脅又在哪裡? 」
And it's not that ChatGPT is the existential threat. ChatGPT 並不是這樣的 存在的威脅。
It's that the race to deploy the most powerful, 這就是部署競賽 最強大的,
inscrutable, uncontrollable technology, 深不可測, 不可控的技術,
under the worst incentives possible, 最壞的情況下 可能的激勵措施,
that's the existential threat. 這就是存在的威脅。
But it strikes me that-that 但令我震驚的是——
we are in a context of a race. 我們正處於一場競賽的背景下。
-Yes. - And it is in a competitive, -是的。 - 而且是在競爭激烈的情況下,
"got to get there first, got to win the race" setting, 「必須先到達那裡, 必須贏得比賽」的設定,
"got to compete against China, 「必須與中國競爭,
got to compete against the other labs." 必須競爭 其他實驗室。 」
Isn't that right? 是不是這樣?
Today it is the case. 今天的情況就是如此。
So we need to change that race dynamic. 所以我們需要改變 那種比賽動態。
Don't you think? 你不覺得嗎?
I think that would be very good indeed. 我認為那會是 確實非常好。
I-I think there's some kind of... 我-我認為 有某種...
mysticism around AI that makes it feel like 圍繞人工智慧的神秘主義 這讓人感覺就像
it fell from the sky. 它從天上掉下來。
Thinking of this as a godlike technology 想到這 作為神一般的技術
is a problem. 是一個問題。
Yeah. 是的。
Why? 為什麼?
It gives license to the companies to... 它給予許可 給公司...
not take responsibility 不承擔責任
for the things that the software that they built does. 對於該軟體的東西 他們建造的確實如此。
If people made that connection, 如果人們建立了這種聯繫,
maybe that would, like, help them understand again 也許那會像 幫助他們再次理解
the dynamics of all the money and power and... 的動態 所有的金錢和權力以及…
chaos that's happening to... 正在發生的混亂…
create this technology. 創造這項技術。
It's, like, five guys who control it, right? 好像有五個人 誰控制它,對嗎?
- Like, five men? - Basically, yeah. - 比如,五個男人? - 基本上,是的。
- The CEOs? - Yeah. - 執行長們? - 是的。
Okay. 好的。
Sometimes it feels like 有時感覺就像
I've been on this just, like, endless journey 我一直在這個 就像無盡的旅程
of trying to understand this. 試著理解這一點。
You know, like I'm climbing a mountain... 你知道,就像 我正在爬山...
...and every time I, like, get up a hill, ……而且每次 我就像爬上一座小山,
I think I've reached the top, 我想我已經達到頂峰了
but it just keeps going and going and going. 但它一直在繼續 走啊走。
But from everything I've heard, 但從我所聽到的一切來看
if I had to guess what's at the top of Anxiety Mountain, 如果我必須猜是什麼 在焦慮山頂上,
it's these five CEOs from these five companies, 就是這五位CEO 從這五家公司來看,
who are kind of like the Oppenheimers of this moment. 誰有點像 此時此刻的奧本海默。
These are the guys who are building this thing. 這些都是傢伙 誰正在建造這個東西。
Yeah. 是的。
Like, is there a plan? 例如,有計劃嗎?
The head of the company that makes ChatGPT 公司負責人 這使得 ChatGPT
warned of possible significant harm to the world. 警告可能的 對世界造成重大危害。
I think, if this technology goes wrong, 我想,如果這項技術 出錯了,
it can go quite wrong. 它可能會出錯。
Five dudes. 五個傢伙。
I-I never thought about us as a social media company. 我-我從來沒有想過我們 作為一家社交媒體公司。
It-it feels like I have to try and-and find these guys... 這-感覺我必須嘗試一下 然後-然後找這些人...
Mm-hmm. 嗯嗯。
...and get them in the movie. ……然後讓它們出現在電影中。
Certainly we'd get some clarity from that. 我們當然會得到 從中可以得到一些清楚的認識。
I mean, the buck's got to stop somewhere, right? 我的意思是,責任必須停止 某處,對吧?
How's that feel? 那種感覺怎麼樣?
- Great. - That okay? - 太棒了。 - 可以嗎?
Yeah. 是的。
- Can I move the seat forward? -Yes, you can. - 我可以把座位往前移嗎? -是的,你可以。
It's, uh, it's just a bit awkward. 這,呃,這只是 有點尷尬。
I'm a little, like... 我有點像...
Dario, how do you feel about that chair? 達裡奧,你感覺如何 關於那張椅子?
Um, yeah, the chair's good. 嗯,是的,椅子很好。
I just wanted to move it a little forward. -Okay. 我只是想移動它 向前一點。 -好吧。
I picked out that chair. 我選了那張椅子。
- It's a good chair. - Thanks. - 這是一張好椅子。 - 謝謝。
And just sit down there. 然後就坐在那裡。
And then through here, 然後經過這裡,
we'll be looking at each other through this glass. 我們會看著對方 透過這塊玻璃。
And sitting back in the chair, leaning forward? 然後坐回椅子上, 向前傾斜?
Yeah. Well, whatever's comfortable, I think. -Okay. 是的。好吧,無論怎樣 舒服,我想。 -好吧。
It's kind of cool. There's a bunch of mirrors. 這有點酷。 那裡有一堆鏡子。
- Is that-- okay. -Sam A., take one. Mark. - 那是-- 好。 -Sam A.,拿一個。馬克。
- How's that? - Good. Thank you. - 怎麼樣? - 好。謝謝你。
The genesis of this project is that I was 起源 這個專案的重點是我
sitting at home, and I was playing around with, 坐在家裡, 我正在玩,
I think, your image generator, 我認為,你的圖像生成器,
and I was simultaneously, uh, terrified 而我同時, 呃,嚇壞了
and really impressed. 真的很感動。
- That is the usual combination. - And then cut to - 這是通常的組合。 - 然後切到
my wife and I find out we're expecting. 我和我的妻子發現 我們期待著。
And-and I'm, like, having an existential crisis 而且-而且我,就像,有 存在危機
as my wife is six months pregnant. 就像我妻子一樣 懷孕六個月。
And I think my first question, which I ask everybody: 我認為我的第一個問題是 我問大家:
Is now a terrible time to have a kid? 現在是一個可怕的時刻 想要一個孩子嗎?
Am I making a big mistake? 我犯了一個大錯誤嗎?
I'm expecting a kid in March, too. My first one. 我在等一個孩子 三月也是。我的第一個。
- You're expecting a kid? - Yeah. - 你在等一個孩子嗎? - 是的。
- You're expecting in March? - First kid. - 你預計三月嗎? - 第一個孩子。
- Mazel tov. - Thank you very much. - 馬澤爾托夫。 - 非常感謝。
I've never been so excited for anything. 我從來沒有這麼興奮過 對於任何事情。
- That's how I feel. - Yeah. - 這就是我的感覺。 - 是的。
But you're not scared? 但你不害怕嗎?
I mean, like, having a kid is just this momentous thing 我的意思是,就像,有一個孩子是 只是這件重大的事情
that I, you know, I stay up every night, like, 我,你知道, 我每天晚上都熬夜,就像
reading these books about how to raise a kid, 閱讀這些關於 如何撫養孩子,
and I hope I'm gonna do a good job, 我希望 我一定會做好事的
and it feels very overwhelming and-- 感覺 非常壓倒性的並且——
but I'm not scared... 但我並不害怕...
for kids to grow up in a world with AI. 為了孩子的成長 在人工智慧的世界裡。
Like, that's... that'll be okay. 就像,那是……沒關係。
That is good to hear coming from the guy. 很高興聽到 來自那個傢伙。
I think it's a wonderful idea to have kids. 我認為這是個好主意 生孩子。
I-I think they're the most magical, incredible thing. 我-我認為他們是最 神奇的、不可思議的事。
There's so much uncertainty, 存在太多的不確定性,
I would almost just do what you're gonna do anyway. 我幾乎會這樣做 無論如何你要做什麼。
I know that's not a very satisfying answer, 我知道那不是 非常滿意的答复,
but it's the only one I can come up with. 但這是唯一的一個 我能想出。
Our kids are never gonna... 我們的孩子永遠不會...
know a world that doesn't have really advanced AI. 知道一個沒有的世界 真正先進的人工智慧。
In fact, our kids will never be smarter than an AI. 事實上,我們的孩子永遠不會 比人工智慧更聰明。
"Our kids will never be smarter than an AI"? “我們的孩子會 永遠不會比人工智慧更聰明」?
Well, from a raw, from a raw IQ perspective, 嗯,從原始的角度來看, 從原始智商的角度來看,
they will never be smarter than an AI. 他們永遠不會 比人工智慧更聰明。
That notion doesn't unsettle you a little bit? 這個觀念不 讓你有點不安嗎?
'Cause it makes me feel a little queasy in a weird way. 因為這讓我感覺 有點奇怪地感到噁心。
Um, it does unsettle me a little bit. 嗯,這確實讓我感到不安 一點點。
But it is reality. 但這是現實。
Okay, so race dynamics. 好的,所以比賽動態。
We have a bunch of people who are all in agreement 我們有一群人 誰都同意
that this is scary. 這很可怕。
Based on the timelines that a lot of people have given me, 根據時間表 很多人給了我,
we have between two months and five years to figure this out. 我們有兩個月的時間 五年的時間來解決這個問題。
- Yeah. - And-and, I guess, - 是的。 - 而且-而且,我猜,
my biggest question is: Why can't we just stop? 我最大的問題是: 為什麼我們不能停下來?
The problem with, uh, uh, "just stopping" is that, um, 問題是,呃呃, 「只是停下來」就是,嗯,
there are many, many groups now around the world, uh, uh, 現在有很多很多團體 世界各地啊啊啊
building this, for many nations, many companies, 建構這個,為了 許多國家、許多公司、
um, all with different motivations. 嗯,全部與 不同的動機。
There are some of these companies in this space 有一些 這個領域的這些公司
who their position is, "We want to develop this technology 他們的立場是誰,「我們想要 開發這項技術
absolutely as fast as possible." 絕對盡可能快。 」
And even if we could pass laws in the US and in Europe, 即使我們可以通過法律 在美國和歐洲,
we need to convince... 我們需要說服...
Xi Jinping and Vladimir Putin or, you know, 習近平和弗拉基米爾·普丁 或者,你知道,
whoever their scientific advisors are on their side. 無論誰的科學 顧問站在他們這一邊。
That's gonna be really hard. 那真的會很難。
I think it is true 我認為這是真的
that if two people are in exactly the same place, 如果兩個人 在完全相同的地方,
uh, the one willing to take more shortcuts on safety 呃,願意接受的人 更多安全捷徑
should kind of "get there first." 應該有點 “先到那裡。”
Uh, but... 呃,但是...
we're able to use our lead 我們能夠利用我們的領先優勢
to spend a lot more time doing safety testing. 花費更多的時間 進行安全測試。
And-and what if you lose it? 還有──如果你丟了怎麼辦?
You get the call and you find out 你接到電話 你發現
that you're now, let's say, six months behind. 比如說你現在 落後六個月。
- Wh-What happens then? - Depends who we're behind to. - 什-然後會發生什麼事? - 取決於我們落後於誰。
If it's, like, a adversarial government, 如果是的話,例如 敵對政府,
that's probably really bad. 這可能真的很糟糕。
So let's say you get the call that China has a, 假設您接到電話 中國有一個,
has a recursively self-improving agent 有一個遞歸的 自我提升劑
or something like that that we should be really worried about. 或類似的東西我們 應該真正擔心。
What do-- what-what happens? 發生了什麼事?
Um... 嗯...
That case would require... 這種情況需要...
the first step there would be to talk to the US government. 第一步將會是 與美國政府交談。
Sam, do you trust the government's ability to handle 山姆,你相信嗎? 政府處理能力
something like this? 像這樣的東西?
Yeah, I do, actually. 是的,實際上我願意。
There's other things I don't trust the government to handle, 還有其他事情我不知道 相信政府能夠處理,
but that particular scenario, I think they would know... 但那個特定的場景, 我想他們會知道...
they-they-- yes. 他們-他們--是的。
When you have, like, private discussions 當你有的時候, 喜歡,私人討論
with the other guys whose fingers are on the trigger, 和其他人一起 手指放在扳機上,
so to speak, um, do those private discussions 可以這麼說,嗯, 進行那些私人討論
fill you with confidence 讓你充滿信心
or do they make you, uh, more anxious? 或者他們讓你, 呃,更著急了?
Look, I mean, I-I know some of them better than others. 聽著,我是說,我-我知道一些 他們比其他人更好。
Um, I have more confidence in some of them th-than I have 嗯,我更有信心了 其中一些比我有
in others, you know, as with, as with any people. 在其他方面,你知道, 與任何人一樣。
You know, you think about, you know, 你知道, 你想想,你知道,
the kids in your high school class or something, right? 你高中的孩子們 類什麼的,對吧?
You know, some of them are, you know, really sweet. 你知道,其中一些是 你知道,真的很甜蜜。
Some of them are well-meaning but not that effective. 其中有些是善意的 但沒那麼有效。
Some of them are, you know, bullies. 其中一些是, 你知道,惡霸們。
Some of them are really bad people. 其中一些是 真是壞人。
You-you kind of really, 你——你真的有點,
you really see the spread. 你真的看到了傳播。
You know, am I, am I, am I confident 你知道嗎,我是嗎? 我有信心嗎
that everyone's gonna do the right thing, 每個人都會做的 正確的事,
that it's all gonna work out? 一切都會順利嗎?
Um, no, I'm not. 嗯,不,我不是。
And there's-there's nothing I can do about that, right? 還有——什麼也沒有 我可以做到這一點,對嗎?
You know, all-all I can do is-is push for the, 你知道,我能做的一切 is-is 推動,
push for the, you know, 推動,你知道,
push for the government to get involved. 推動政府 參與其中。
But ultimately I'm just one person there, too. 但最終我只是 那裡也有一個人。
That's-- it's-it's up to all of us 那是—— 這取決於我們所有人
to push for the government to get involved. 推動政府 參與其中。
That's the number one thing that-that I think we need to do 這是第一件事 那-那我認為我們需要做
to-to, you know, to set things in the right direction. to-to,你知道,設定事物 朝著正確的方向。
What makes me anxious about that is, like, 是什麼讓我焦慮 關於那個,就像,
the basic reality that the speed at which the technology 的基本現實是 該技術的速度
is proliferating and growing is exponential, 正在增殖和成長 是指數的,
and the mechanisms to legislate are 300 years old, 以及立法機制 已經300歲了,
- take forever. - Yeah. - 永遠。 - 是的。
Um, I-I think it's gonna be a heavy lift. 嗯,我-我想 這將是一個沉重的負擔。
I-I definitely agree with you on that. 我-我絕對同意 與你一起。
You know, what I'm literally looking for 你知道, 我真正在尋找什麼
is-is like, "Here are steps that, like, 是-就像, 「這裡有一些步驟,例如,
"the head honchos are gonna take to-to focus on safety, 「頭兒們會 take to-注重安全,
to mitigate the peril and maximize the promise." 以減輕危險 並最大化承諾。 」
And I don't, I don't, I don't know 而我不, 我不知道,我不知道
that there's a simple answer to that. 那裡有 對此的簡單回答。
Um... 嗯...
I mean, this maybe is too simple, 我的意思是,這也許是 太簡單了,
but you... 但你...
create a new model, 建立一個新模型,
you study and test it very carefully. 你研究並測試它 非常仔細。
You put it out into the world gradually, 你把它拿出來 逐漸走進這個世界,
and then, more and more, you understand 然後,越來越多, 你明白
if that's safe or not, and then if it is, 不管安全與否 然後如果是的話
you can take the next step. 您可以進行下一步。
It doesn't sound as flashy as, like, a brilliant scientist 聽起來沒那麼華而不實 就像一位傑出的科學家
coming up with one idea in a lab to make an AI system, like, 提出一個想法 實驗室製作人工智慧系統,例如,
perfectly safe and controllable and everything else, 完全安全可控 以及其他一切,
but it is what I believe is gonna happen. 但這就是我所相信的 將會發生。
Like, it is the way I think this works. 就像,這就是方式 我認為這有效。
But let's just say something terrible happens, 但我們只能說 可怕的事情發生了
like a model gets loose or goes rogue or something. 就像模特兒鬆開一樣 或是變得流氓什麼的。
Is there a protocol? 有協議嗎?
Like, literally, I'm imagining a red phone. 就像,從字面上看, 我想著一部紅色手機。
- Yeah. - Sorry for thinking of this - 是的。 - 抱歉想到這個
in terms of movies, but, like... 就電影而言,但是,就像...
There is a protocol. 有一個協議。
Is there a red phone on your desk? -No. 有紅色手機嗎 在你的桌子上嗎? -不。
Is it a secret? 這是一個秘密嗎?
I mean, uh, no, it's not, it's not as fancy or dramatic 我的意思是,呃,不,不是, 它並不那麼奇特或戲劇性
as you, like, would hope, but there's, like, you know-- 正如你所希望的那樣, 但有,比如,你知道——
we've, like, thought through these scenarios, 我們已經想通了 這些場景,
and if this happens, we're gonna call these people 如果發生這種情況 我們要打電話給這些人
in this order and do this 按此順序並執行此操作
and kind of make these decisions if, like-- 和種類 這些決定如果—
I do believe that when you have an opportunity 我確實相信 當你有機會的時候
to do your thinking before a stressful situation happens, 之前先思考一下 發生緊張的情況,
that's almost always a good idea. 這幾乎總是 好主意。
And writing it down is helpful. 把它寫下來很有幫助。
- Being prepared is helpful. - Yeah. - 做好準備很有幫助。 - 是的。
You... 你...
It would be impossible for me to sit across from you 這是不可能的 讓我坐在你對面
and-and ask you to promise me that this is gonna go well? 並且-並要求你答應我 這樣會順利嗎?
That is impossible. 那是不可能的。
There isn't any easy answers, unfortunately. 不幸的是,沒有任何簡單的答案。
Uh, because it's such a cutting-edge technology, 呃,因為是這樣的 尖端技術,
um, there's still a lot of unknowns. 嗯,還有 很多未知數。
And I think that that-that needs to be, 我認為 那-那需要是,
um, uh, you know, understood 嗯,呃,你知道,明白了
and-and hence the need for, uh, uh, some caution. 並且-因此需要 呃,呃,需要謹慎一點。
I wake up, you know, every day, 我每天醒來,你知道,
this is the, this is the number one thing I think about. 這是,這是 我想到的第一件事。
Now, look, I'm human, 現在,看,我是人類,
and, you know, has-has every decision been perfect? 而且,你知道,有-有 每一個決定都是完美的嗎?
Can I even say my motivations were always perfectly clear? 我能說出我的動機嗎 總是很清楚嗎?
Of course not. No one can say that. 當然不是。 沒有人能這麼說。
Like, that's-that's just not, like, you know-- 比如說,那是──那是 只是不是,就像,你知道——
that's-that's just not how people work. 那是-那不是 人們如何工作。
The-the history of science tends to be that, 科學史 往往是這樣的,
for better or for worse, if something's possible to do-- 無論好壞, 如果有什麼可以做的話——
and we now know AI is possible to do-- humanity does it. 我們現在知道人工智慧是可能的 人類做到了。
All of this was-was going to happen. 所有這一切 曾經——將會發生。
This-this train isn't gonna stop. 這這列火車 不會停止。
You can't step in front of the train and stop it. 你不能走到前面 火車並停下來。
You're just gonna get squished. 你只會被壓扁。
I mean, it's very stressful. 我的意思是,壓力很大。
You know, there's, like, 你知道,有,比如,
a lot of things a lot of us don't know. 很多事情 我們很多人都不知道。
I think the history of scientific discovery is 我認為歷史 科學發現的意義在於
one of not knowing what you don't know 不知道之一 你不知道什麼
and figuring out as you go. 並邊走邊想。
Uh, but, yeah, it is a... 呃,但是,是的,這是一個...
it is a stressful way to live. 這是一種充滿壓力的生活方式。
Right. Sam, thank you very much for doing this. 正確的。山姆,謝謝你 非常感謝這樣做。
- And again, mazel tov. - Thank you. And to you. - 再說一遍,馬澤爾托夫。 - 謝謝。還有你。
Thank you so much. Thanks, guys. 太感謝了。謝謝,夥計們。
Hello. 你好。
Hey, Dad, how are you? 嘿,爸爸,你好嗎?
Good. How you doing? 好的。你好嗎?
You working? What are you up to? 你工作嗎?你要做什麼?
I'm working on this AI film. 我正在製作這部人工智慧電影。
And how's it going? 進展如何?
You know, it... it's really... 你知道嗎,這……這真的是……
-Hi, sweetie. -Hi, Mom. -嗨,親愛的。 -嗨,媽媽。
So what is the premise of the film? 那什麼是 電影的前提?
Is it a documentary? Kev, don't use any more spices. 這是一部紀錄片嗎? 凱文,別再用香料了。
It's already over-spiced. 已經太辣了。
We're making chicken right now. 我們現在正在做雞肉。
- Is it a documentary or what... -It's about-- - 這是紀錄片還是什麼... -這是關於--
the movie's about the end of the world. 這部電影是關於 世界末日。
The end of the world's coming, 世界末日即將來臨,
and we're making a movie about the end of the world. 我們正在拍電影 關於世界末日。
-Really? -Yeah. -真的嗎? -是的。
Kind of a depressing film, it sounds like. 有點壓抑 電影,聽起來像。
Yeah. 是的。
I'm feeling a lot, like-- this very acute anxiety. 我感觸良多, 就像這種非常嚴重的焦慮。
It's so scary, but there's got to be-- 太可怕了, 但一定有——
you know, have you been meeting some-some supersmart people 你知道嗎,你們見面了嗎 一些超級聰明的人
that are giving you any answers? 那些能給你任何答案嗎?
That's what's frustrating about it. 就是這樣 對此感到沮喪。
No one knows. 沒有人知道。
All I can say to that is that 我只能說
every generation has had something scary like this. 每一代都有過 像這樣可怕的東西。
When I was born, it was the Cuban Missile Crisis. 我出生的時候,是 古巴飛彈危機。
I was just scared that there was going to be a nuclear war. 我只是害怕那裡 這將是一場核戰。
Yeah, but we didn't know 是的,但我們不知道
what they were gonna do and... 他們要做什麼以及...
And the world didn't end. 世界並沒有終結。
Everyone woke up the next morning, 大家都醒了 第二天早上,
and we're still doing our thing. 我們仍在做我們的事情。
I'm very scared, 我很害怕,
especially in the context of, like, 特別是在 的上下文,例如,
you know, the baby and... 你知道,嬰兒和...
It's gonna be a learning curve. 這將是一個學習曲線。
You're gonna be okay. 你會沒事的。
You can't, you can't think about 你不能, 你不能想
what you can't control, Daniel. 你無法控制的事情,丹尼爾。
Just remember that. 只要記住這一點。
I'm really, I'm really feeling 我真的,我真的感覺
nervous and scared about it. 對此感到緊張和害怕。
There's so much that I can't control. 有這麼多 我無法控制。
Don't be nervous. You can't let that get to you. 別緊張。 你不能讓它影響你。
You can only control what you can control, 你只能控制 你能控制的,
and that's all you can do. 這就是你所能做的。
You can't do more than that. 你不能做更多的事情。
Write that down in your book. 把它寫在你的書裡。
When you look back, 當你回頭看時,
the world is always ending. 世界總是有終結的。
And when you look ahead, 當你展望未來時,
the world is always ending. 世界總是有終結的。
...on fire. One home is already on fire... ……著火。 一棟房子已經著火了…
But the world is always starting, too. 但世界 也總是在開始。
Are you ready? 你準備好了嗎?
Are you ever really ready? 你真的準備好了嗎?
You want to drive or you want me to drive? 你想開車 還是你想讓我開車?
You drive. 你開車。
Okay. 好的。
Thank you, Maria. 謝謝你,瑪麗亞。
It's gonna just feel really crampy. 這只會 感覺真的很抽筋。
-Okay. -You're ready? -好的。 -你準備好了嗎?
Yes. 是的。
You could be in a medical drama with all that... 你可能處於一個 醫療劇與這一切......
...some pressure. ……有些壓力。
Just relax. 放鬆點。
Hi, buddy. Ooh. 嗨,夥計。哦。
Hi, buddy. 嗨,夥計。
It's gonna be a whole new world 這將會是一個全新的世界
for you, Daniel. 為了你,丹尼爾。
And I think you're gonna be an amazing father. 我想你會的 一位了不起的父親。
That's for sure. 這是肯定的。
Oh. 哦。
You're gonna do a great job. 你會做得很好的。
Kev, why are you crying? 凱夫,你為什麼哭?
I'm crying because... 我哭是因為...
I don't know. I just don't... 我不知道。我只是不...
Dad, you're gonna make me cry, too. 爸爸,你想要 讓我也哭了
Why are you crying? 為什麼哭?
I just know that you're gonna be an amazing dad. 我只知道 你會成為一個了不起的爸爸。
I'm only gonna be a great dad 我只會成為一個偉大的爸爸
'cause I had a great dad. 因為我有一個偉大的父親。
I'm getting emotional, that's all. 我變得情緒化, 僅此而已。
Aw. 噢。
Boy, oh, boy! 男孩,哦,男孩!
My goodness! 天啊!
Look at those little cheekies. Look at those little cheekies. 看看那些小臉蛋。 看看那些小臉蛋。
Yeah. 是的。
I know how to end this movie. 我知道如何結束這部電影。
Babies. 寶貝們。
The end of the movie is about babies.- 電影的結尾是關於 寶貝們.-
They're life-affirming. 他們是對生命的肯定。
They're exhausting. 他們很累。
They're hilarious. 他們很搞笑。
And they're worth it. 他們是值得的。
This film isn't about the inner workings of a technology. 這部電影不是關於 科技的內部運作。
It's not about the billionaire CEOs. 這不是關於 億萬富翁執行長。
It's not about the geopolitics. 這與地緣政治無關。
It's not about the terrifying future or the end of the world, 這不是關於可怕的事情 未來或世界末日,
because my world is just starting. 因為我的世界 才剛開始。
I'm building a crib. 我正在建造一個嬰兒床。
Right here, right now. 就在這裡,就是現在。
AI is gonna change everything 人工智慧將改變一切
in ways too powerful and complex for us to understand. 以過於強大和 複雜到我們難以理解。
And the future is not for any of us to decide. 而未來並不是 由我們任何人來決定。
But what I can decide is to be the best possible husband 但我能決定的是 最好的丈夫
for my wife and the best possible dad for my son. 為了我的妻子和最好的 我兒子的可能爸爸。
So whether our AI future is a nightmarish dystopia 那麼我們的人工智慧未來是否是 惡夢般的反烏托邦
or the utopia that we all dream of, 或烏托邦 我們都夢想的,
I'll at least know that I did everything I could 我至少會知道 我已盡我所能
to guide my family through this AI revolution. 指導我的家人 透過這場人工智慧革命。
And no matter what, we'll be facing it together. 而且無論如何, 我們將一起面對它。
So that's just our first idea. 這只是我們的第一個想法。
How does this feel? Are you feeling this? 這感覺如何? 你有這種感覺嗎?
Wait, I-- like, this is not-- this is a joke. 等等,我——就像,這不是—— 這是一個笑話。
It's not actually how you're gonna end it. 其實不是 你將如何結束它。
I mean, it's just an idea. Okay? 我的意思是,這只是一個想法。好的?
No, Daniel. 不,丹尼爾。
...uh, very, very dumb. ……呃,非常非常愚蠢。
You've just spent, I don't know, like, how many years 你剛剛花了,我沒有 知道,比如,幾年
of our life working on this, talking to every leading expert 我們一生致力於此, 與每位領先專家交談
on the planet about the subject, 在這個星球上關於這個主題,
and you're gonna end it 你會結束它
with some, like, kumbaya bullshit? 與一些,比如, 昆巴亞廢話?
There's an asteroid headed to Earth. 有一顆小行星 前往地球。
What do you do? Just... 你做什麼工作?只是...
hold hands and hope it works out okay? 牽手 並希望一切順利?
Absolutely not. 絕對不是。
We have to-- it ha-- it has to be... 我們必須——它哈—— 它必須是...
The ending has to... 結局不得不...
Okay, first thing: 好的,第一件事:
AI is here, and it's here to stay. 人工智慧來了, 它會一直存在。
The shit's out of the horse, 屎已經從馬裡出來了
but the horse is gonna keep shitting. 但馬是 會繼續拉屎
You know, one of the basic laws of history 你知道,基本的條件之一 歷史規律
is that nothing really have a beginning 沒什麼 真的有一個開始
and nothing has any ending. 一切都沒有結局。
It just goes on. 它就這樣繼續下去。
AI is nowhere near its full development. 人工智慧離我們還很遙遠 其全面發展。
Even if the current AI bubble bursts, 即使 當前的人工智慧泡沫破裂,
humans are never going to stop 人類永遠不會停止
building more and more powerful technology. 建設越來越多 強大的技術。
You can choose not to use AI 你可以選擇不使用AI
or participate in it, but it's going to affect you anyway. 或參與其中,但它是 無論如何都會影響到你。
Okay, fantastic. So we're screwed. 好吧,太棒了。 所以我們完蛋了。
No, we're not because of one simple thing. 不,我們不是 因為一件簡單的事。
This is not inevitable. 這並非不可避免。
If we could just see it clearly together, 如果我們能看到它 明顯地在一起,
the obvious response will be to choose something different. 明顯的反應是 選擇不同的東西。
We need to very clearly change the game 我們需要非常清楚 改變遊戲
from a race to the bottom into a race to the top. 從競賽到墊底 進入一場力爭上游的競賽。
The problem we need to solve is not AI specifically. 我們需要解決的問題 解決方案並不是專門針對人工智慧的。
It's the general question of how do we build a society 這是一般問題 我們如何建立一個社會
that can deal with powerful technology. 可以處理 強大的技術。
Because we're going to get 因為我們要得到
only more and more powerful technology. 只會越來越強大 技術。
We need to upgrade our society, 我們需要提升我們的社會,
and the first step is coming together 第一步是 走到一起
- and demanding... -Coordination. - 而且要求很高... -協調。
Some form of international cooperation or agreement about 某種形式的國際 合作或協議
what the norms should be. 規範應該是什麼。
You know, how should they be deployed... 你知道,應該怎樣 他們被部署...
Like, real international diplomacy 就像,真實的 國際外交
among the superpowers. 超級大國之中。
The Chinese are as worried about it as the Americans. 中國人也同樣擔心 身為美國人來說。
I think it's difficult, you know, 我覺得很難, 你知道,
in the current geopolitical climate, 在當前的 地緣政治氣候,
- but I think it's necessary. -Absolutely. - 但我認為這是必要的。 -絕對地。
In the exact same way that the last time that humanity 以完全相同的方式 人類最後一次
developed a technology this dangerous... 開發了一項技術 這危險的…
...that required a complete, unprecedented shift ……這需要一個完整的、 前所未有的轉變
to the structure of our world. 我們世界的結構。
So we need to do that complete, 所以我們需要完成這個工作,
unprecedented shift again. 再次發生前所未有的轉變。
You know, we-we talk to people who work at these AI companies, 你知道,我們和人們交談 在這些人工智慧公司工作的人,
and they say they want to do something different, 他們說他們想做 不同的東西,
but they need public pressure. 但他們需要公眾壓力。
They need the government to do something. 他們需要政府 做某事。
So then we go to DC, and they say, 然後我們就去DC, 他們說,
"Well, we need Silicon Valley to do something different. 「好吧,我們需要矽谷 做一些不同的事情。
They're the ones who are gonna come up with the guardrails." 他們是那些將要 把護欄拿出來。 」
And so everyone is pointing the finger at someone else, 所以每個人都指著 用手指指著別人,
and what they agree on is that we need public pressure 他們同意的是 我們需要公眾壓力
in order for something else to happen. 為了別的事 發生。
And that's what you 這就是你
and all the people watching this movie can do. 和所有的人 看這部電影就可以了。
We need to hold the leaders 我們需要抓住領導者
in our governments 在我們的政府中
and the leaders of these companies accountable. 和領導人 這些公司負有責任。
Whichever country you're in, 無論你在哪個國家,
let them know that you're not happy 讓他們知道 你不開心
with the current status quo. 以目前的現狀。
So, yeah, it's boring to say call your congressperson. 所以,是的,說起來很無聊 打電話給你的國會議員。
I'm not saying you should just do that, but, like, 我並不是說你應該 就這樣做,但是,就像,
we do have to do that. 我們確實必須這樣做。
Like, we do have to get the government involved. 就像,我們必須得到 政府參與其中。
So we just call them up and say, 所以我們只是 打電話給他們說,
"Hey, stop Big Tech from ruining the world"? 「嘿,停止大型科技公司 以免毀滅世界」?
No, but there are tons of really obvious, 沒有,但是有 大量非常明顯的,
straightforward things we can be demanding. 簡單的事情 我們可以要求很高。
We need transparency. 我們需要透明度。
We need to end the secrecy that exists inside these labs, 我們需要結束保密 存在於這些實驗室內,
because they are building powerful technology, 因為他們正在建造 強大的技術,
and the public deserves to know what's going on. 公眾應該知道 發生了什麼事。
Ultimately, we're gonna need independent, 最終, 我們需要獨立,
objective third parties to evaluate the systems. 客觀第三方 評估系統。
We can't count on the companies to grade their own homework. 我們不能指望這些公司 給自己的作業評分。
If-if a company uses AI and-and has AI interacting with you, 如果-如果一家公司使用人工智能并且-并且 有AI與你互動,
it should disclose that you are interacting with an AI system. 它應該透露您是 與人工智慧系統互動。
Yeah. And-and also we need a system that makes companies 是的。而且-而且我們還需要 一個使公司
legally liable for the AI systems that they produce. 並承擔法律責任 他們生產的人工智慧系統。
We need to make sure that there are tests and safety standards 我們需要確保有 是測試和安全標準
that are applied to everyone. 適用於每個人。
We need some ground rules, 我們需要一些基本規則,
and we need to keep adapting those rules 我們需要不斷適應 那些規則
at the speed that the technology develops. 以這樣的速度 技術不斷發展。
There is currently more regulation 目前有 更多監管
on selling a sandwich to the public 賣三明治時 向公眾
than there is on building potentially world-ending AGI. 比建築物上的 潛在的世界末日 AGI。
And the last thing is 最後一件事是
to upgrade ourselves. 來提升我們自己。
This is not, like, the job of, like, the safety team 這不是工作 例如安全團隊
at any given lab, or the CEO. 任何特定的實驗室或執行長。
So, like, this is everyone's job. 所以,就像,這是 每個人的工作。
Don't, like, leave it up to the AI experts. 不要,就像,離開它 由人工智慧專家決定。
Like, [bleep] that. Like, this is the moment 就像,[嗶]那樣。 就像,這一刻
that we are transitioning from, like, 我們正在轉型 來自,像,
mostly human cognitive power to, like, AI cognitive power, 主要是人類的認知能力 比如說人工智慧的認知能力,
and it affects everyone, 並且影響著每個人,
and I want people to be in on that conversation. 我希望人們成為 參加那次談話。
And I would say, if you think that AI will kill us all, 我想說,如果你認為 人工智慧會殺死我們所有人,
you should be working in AI research 你應該在工作 在人工智慧研究領域
to make sure it doesn't, because you do have 以確保它不會 因為你確實有
an enormous amount of agency. 大量的代理商。
Whoever you are, 無論你是誰,
you are an expert in your own industry, 你是專家 在自己的行業裡,
in your own school, in your own family, 在自己的學校裡, 在自己的家庭裡,
and it's up to you how AI is used in your life. 這取決於你 人工智慧如何應用在你的生活中。
Whether you want to join your school board 無論你想加入 你的學校董事會
or-or whether you want to ask your employer 或-或者你是否想問 你的雇主
how they're using AI technologies, 他們如何使用 人工智慧技術,
like, all of us can do the work. 就像,我們所有人都可以完成這項工作。
A lot of unions have been pretty effective at, like, 已經有很多工會了 非常有效,例如,
determining how they want to interact with these systems. 確定他們想要的方式 與這些系統進行互動。
Nurses unions, teacher unions. 護士工會、教師工會。
I would love if parents everywhere 我會喜歡 如果父母無所不在
went to the AI companies and said, 去了AI公司 並說,
"How can you be better on this?" Including us. 「你怎樣才能變得更好 就這個? 」包括我們。
So, I founded Encode Justice when I was 15 years old. 所以,我創立了 Encode Justice 當我15歲的時候。
We are the world's first and largest army of young people 我們是世界第一 和最大的年輕人軍隊
fighting for human-centered artificial intelligence. 為以人為本而奮鬥 人工智慧。
It doesn't matter who you are, even the smallest actions help, 你是誰並不重要 即使是最小的行動也會有所幫助,
and even conversation starting is really, really valuable. 甚至對話開始 真的非常非常有價值。
So my job could be, like, 所以我的工作可能是,
tell Bubby Lila about this at dinner? 告訴巴比·萊拉這件事 晚餐時?
Honestly, yeah, that's part of it. 老實說,是的, 這是其中的一部分。
And we're gonna have to do 我們要做的
a lot of things that we haven't even thought of yet. 很多事情 我們還沒想到。
People are gonna look at anything that we've outlined 人們會看 我們概述的任何內容
and say, "That's not enough." 並說:“這還不夠。”
What matters is that the forces 重要的是力量
that are working towards solutions 正在工作的 尋求解決方案
start to exceed the forces that are working against solutions. 開始超過力量 正在研究解決方案。
Making the world better has always been hard. 讓世界變得更美好 一直很難。
It has never been easy. 這從來都不容易。
Like, there have been many shitty things 例如,曾經有過 許多糟糕的事情
that have happened in history, and we've had-- 歷史上發生過的事, 我們已經——
like, people have had to deal with that, 就像,人們曾經有過 為了解決這個問題,
and then they've risen up and changed it. 然後他們站了起來 並改變了它。
There are an insane number of challenges ahead of us, 有一個瘋狂的數字 我們面臨的挑戰,
but if we can get past them, 但如果我們能超越他們
we can unlock a future beyond our wildest imagination. 我們可以開啟更遠的未來 我們最瘋狂的想像。
We have to come together 我們必須團結起來
and find the path between the promise and the peril. 並找到之間的路徑 承諾與危險。
We can't be pessimists or optimists. 我們不能成為悲觀主義者 或樂觀主義者。
We have to become something new. 我們必須成為新的東西。
A friend of mine calls me an apocaloptimist. 我的一個朋友打電話給我 一個末日樂觀主義者
"Apocaloptimist"? “末日樂觀主義者”?
I think that might be my new favorite word. 我想這可能是 我最喜歡的新詞。
-MAN: Might be the name of this movie. -男人:可能是 這部電影的名稱。
It might be the name of this movie. - 可能是這個名字 這部電影的。 -
- "Apocaloptimist." - "Apocaloptimist." ——《末日樂觀主義者》。 ——《末日樂觀主義者》。
Yeah. 是的。
I don't believe in doom. 我不相信厄運。
I believe in the spirit of life, 我相信生命的精神,
uh, and I believe in life is about the capacity to act. 呃,我相信生活是 關於行動的能力。
The capacity to relate, 建立聯繫的能力,
the capacity to feel. 感受的能力。
We have to double down more and more 我們必須加倍努力 越來越多
on those capacities that we have as humans 關於這些能力 我們身為人類所擁有的
that robotic systems will never have. 那個機器人系統 永遠不會有。
It's time right now to make those decisions about 現在正是時候 做出這些決定
how to guide it and support it rather than dividing us. 如何引導和支持 而不是分裂我們。
It kind of sounds like raising a kid. 聽起來有點像 撫養一個孩子。
That's what's up. Yeah. 就是這樣。是的。
AI may have more raw intelligence 人工智慧可能有 更多原始情報
than our little human brains, 比我們小小的人類大腦,
but we're so much more than just our intelligence. 但我們還有更多 不僅僅是我們的智力。
Intelligence is-is the ability to solve problems. 情報是-是 解決問題的能力。
Wisdom is the ability to know which problems to solve. 智慧是認識的能力 要解決哪些問題。
It can go on your fridge. 它可以放在你的冰箱上。
So, don't give up. 所以,不要放棄。
Humanity has done 人類已經做到了
more difficult things than this in its history. 比這更困難的事情 在它的歷史中。
It's just hard to convince people that they should. 只是很難說服 他們應該的人。
So, when I started making this movie, 所以,當我開始 製作這部電影,
I would say that I was, like, 我會說我就像,
broadly a cynical asshole about this whole thing. 基本上就是一個憤世嫉俗的混蛋 關於這整件事。
Over the course of making the film, 在課程中 製作這部電影時,
I've come to understand that, like, 我漸漸明白了 那個,就像,
that's the only thing we can't be. 這是唯一的事情 我們不可能。
...or anything else you'd like to discuss? ……或其他什麼 你想討論一下嗎?
No. I think we covered a lot. 不,我認為我們涵蓋了很多。
- Yeah. -Thank you so much. - 是的。 -太感謝了。
Thanks. 謝謝。
This is a problem that's bigger 這是一個更大的問題
than any one person. 比任何一個人。
This will change the world in ways that we don't understand. 這將改變世界 我們不理解的方式。
That is all true. 這都是真的。
Okay. 好的。
But what we do have agency over 但我們確實有代理權
is what we do about it. 這就是我們所做的。
As frontier AI grows exponentially more capable... 隨著前沿人工智慧的發展 能力呈指數級增長...
And the reality 而現實
is that if we just decide it's hopeless, 如果我們決定 是無望的,
then it is hopeless. 那就沒有希望了。
...to put less stress on planet Earth... ...减轻压力 在地球上...
But if you decide 但如果你决定
that you want to try... 你想尝试的...
...then you try. ……那你试试吧。
And that's hard. 這很難。
But you know what? 但你知道嗎?
Big things seem impossible before they actually happen. 大事看似不可能 在它們真正發生之前。
But when they finally do happen, 但當它們最終發生時
it's because millions of people took millions of actions 這是因為數以百萬計的人 採取了數百萬次行動
to make them happen. 使它们发生。
And so... 所以...
we have to try. 我們必須嘗試。
There's too much at stake. 事關重大。
Look at the incredible changes we've experienced and survived 看看令人難以置信的變化 我們經歷過並倖存下來
from the Stone Age, 从石器时代开始,
and yet even greater changes are still to come. 还有更大的变化 还没有到来。
♪ What will we do now? ♪ ♪ 我們現在要做什麼? ♪
♪ We've lost it to trying ♪ ♪ 我們因嘗試而失敗 ♪
♪ We've lost it ♪ ♪ 我們失去了它 ♪
♪ To trying ♪ ♪ 去尝试 ♪
♪ What will we do now? ♪ ♪ 我們現在要做什麼? ♪
♪ We've lost it to trying ♪ ♪ 我們因嘗試而失敗 ♪
♪ We've lost it ♪ ♪ 我們失去了它 ♪
♪ To trying ♪ ♪ 去試試 ♪
♪ What can we say now? ♪ ♪ 現在我們能說什麼? ♪
♪ Our mouths only lying ♪ ♪ 我們的嘴只是在說謊 ♪
♪ Our mouths ♪ ♪ 我們的嘴 ♪
♪ Only lying ♪ ♪ 只是說謊 ♪
♪ What can we say now? ♪ ♪ 現在我們能說什麼? ♪
♪ Our mouths only lying ♪ ♪ 我們的嘴只是在說謊 ♪
♪ Our mouths ♪ ♪ 我們的嘴 ♪
♪ Only lying ♪ ♪ 只是說謊 ♪
♪ Give in and get out ♪ ♪ 屈服並退出 ♪
♪ We rise in the dying ♪ ♪ 我們在臨終中崛起 ♪
♪ We rise ♪ ♪ 我們崛起 ♪
♪ In the dying ♪ ♪ 臨終時 ♪
♪ Give in and get out ♪ ♪ 屈服並退出 ♪
♪ We rise in the dying ♪ ♪ 我們在臨終中崛起 ♪
♪ We rise ♪ ♪ 我們崛起 ♪
♪ In the dying ♪ ♪ 臨終時 ♪
♪ Give in and get out ♪ ♪ 屈服並退出 ♪
♪ We rise in the dying ♪ ♪ 我們在臨終中崛起 ♪
♪ We rise ♪ ♪ 我們崛起 ♪
♪ In the dying ♪ ♪ 臨終時 ♪
♪ Give in and get out ♪ ♪ 屈服並退出 ♪
♪ We rise in the dying ♪ ♪ 我們在臨終中崛起 ♪
♪ We ♪ ♪ 我們 ♪
♪ Oh, oh, oh, oh, oh ♪ ♪ 哦,哦,哦,哦,哦 ♪
♪ Oh, oh, oh, oh ♪ ♪ 哦,哦,哦,哦 ♪
♪ Oh, oh ♪ ♪ 哦,哦 ♪
♪ Oh, oh, oh, oh, oh, oh ♪ ♪ 哦,哦,哦,哦,哦,哦 ♪
♪ Oh, oh, oh, oh ♪ ♪ 哦,哦,哦,哦 ♪
♪ Oh, oh ♪ ♪ 哦,哦 ♪
♪ Oh, oh, oh, oh, oh, oh ♪ ♪ 哦,哦,哦,哦,哦,哦 ♪
♪ What will we do now? ♪ ♪ 我們現在要做什麼? ♪
♪ We've lost it to trying ♪ ♪ 我們因嘗試而失敗 ♪
♪ We've lost it ♪ ♪ 我們失去了它 ♪
♪ To trying ♪ ♪ 去試試 ♪
♪ Oh, oh, oh, oh, oh, oh ♪ ♪ 哦,哦,哦,哦,哦,哦 ♪
♪ What will we do now? ♪ ♪ 我們現在要做什麼? ♪
♪ We've lost it to trying ♪ ♪ 我們因嘗試而失敗 ♪
♪ We've lost it ♪ ♪ 我們失去了它 ♪
♪ To trying. ♪ ♪ 去嘗試。 ♪
Can't find what you're looking for?
Get subtitles in any language from opensubtitles.com, and translate them here.