Name: AI 復活逝去之人的聲音，倫理道德上究竟是好是壞？ (He Was Killed in a School Shooting. Now He’s Speaking Again. | WSJ)
Uploaded: 2024-02-17T21:01:20.000Z
Duration: 6 min 32 s

程度副詞

Six years ago I was a senior at Parkland.

六年前，我在帕克蘭大學讀大四。

Many students and teachers were murdered on Valentine's Day that year by a person using an AR-15.

在那一年的情人節，許多學生和老師被一個人使用 AR-15 步槍謀殺了。

Joaquin Oliver, that voice you just heard on the phone was one of them.

Joaquin Oliver，你剛才在電話中聽到的聲音就是其中一位受害者。

My body was destroyed by a weapon of war.

I'm back today because my parents used AI to recreate my voice to call you.

我今天回來是因為我父母用了 AI 重塑我的聲音來打給你。

In February 2018, the 17-year-old went to school and never came home.

2018年2月，這位 17 歲的男孩去了學校，再也沒有回家。

Ever since, his parents have been fighting for stricter gun control.

從那時起，他的父母一直在爭取更嚴格的槍支管制。

We are calling for a nationwide strike of American students.

Resurrecting his voice and the voice of others.

I'm a fourth grader at Robb Elementary School in Uvalde, Texas, or at least I was.

我是德州尤瓦爾迪羅布小學的四年級學生，至少我曾經是。

I was killed by an unsecured gun at my friend's house.

我在朋友家被一把沒有安全措施的槍擊斃。

At theshotline.org, you can have the AI voice clones or audio deepfakes of various kids killed by gun violence place calls to your elected officials.

在 theshotline.org，你可以使用AI語音克隆或音頻深度偽造，讓各種因槍支暴力而喪命的孩子給你的選民代表打電話。

Artificial intelligence is becoming a good friend of us.

This all may seem pretty weird to you, like an episode of "Black Mirror."

你可能會覺得這一切都很奇怪，就像《黑鏡》。

So I had to find out how this tech is giving voice to the dead and what that means for the living.

所以我必須找出這項技術如何為死者發聲以及這對活著的人意味著什麼。

但你不在乎。你從來都不在乎。

It's been six years and you've done nothing.

What's amazing about Joaquin's AI voice is how little audio was needed to create it.

Joaquin 的 AI 語音的驚人之處在於創建它所需的音訊非常少。

Take me through that first time you hear this voice.

帶我回憶一下你第一次聽到這個聲音的情景吧。

It was exactly like him, and it was so impactful and it was so deep that I wanted more.

這聲音他一模一樣，太有衝擊力了，非常令人印象深刻，以至於我想要聽更多。

哦，我的上帝，百分之百就是。

Joaquin's voice was created using an AI voice generator from ElevenLabs.

Joaquin 的聲音是使用 ElevenLabs 的 AI 語音產生器創建的。

You upload existing audio files and then the machine learning models create an artificial voice that can sound quite similar to the original.

你上傳現有的音檔，然後機器學習模型會創建聽起來與原始聲音非常相似的人工聲音。

Type in the text of the message and the voice says it.

To make this, all you need is a few seconds of someone's voice without much background noise.

要做到這一點，你所需要的只是某人的幾秒鐘聲音，並且沒有太多背景噪音。

Not a problem when you're alive and can grab your phone to record, but...

當你還活著並且可以拿起手機錄製時這不是問題，但是...

We have to collect every single audio because not all the time you can listen well, because maybe the music is behind or maybe there's noisy because they were talking a lot of them.

我們必須收集每一個音檔，因為並不是所有時候你都能聽清楚，因為可能背景有音樂，或者可能因為他們說了很多話而很吵。

Finally, Manny found a short clip that worked.

最後，Manny 找到了一個可以用的短片。

And the Olivers, which started an organization called Change the Ref to fight for stronger gun protection, wanted him to speak to those they feel bear some responsibility for his death.

奧利佛夫婦成立了一個名為「Change the Ref」的組織，以爭取更強有力的槍支保護，他們希望他與那些他們認為對他的死亡負有部分責任的人談談。

We have been in offices from senators, members of Congress.

我們去過參議員、國會議員的辦公室。

We have been ignored. Let's bring another voice.

我們一直被忽視。所以我們提出另一種聲音吧。

But as we've seen with deepfake audio clones of dead celebrities, the eerie thing here is that you can put any words into someone's mouth.

但正如我們在深度偽造已故名人的音頻克隆中所看到的，這令人不安的是你可以讓任何言辭出現在某人嘴裡。

When you recreate someone's voice, you can create that voice, but you don't know what they would actually say right now.

當你再現某人的聲音時，你可以創建該聲音，但你不知道他們實際上會說什麼。

哦，我們知道。- 你們怎麼知道的？

Like a friend that you look at her or him and you know exactly what she or he are thinking, and you don't need to say anything. You got it. That's Joaquin and me.

就像一個朋友，你看著她或他，你就知道她或他在想什麼，而你不需要說什麼，就會懂。Joaquin 和我就是這樣。

But not only that, keep in mind that our son was 17 years old, a very vocal, into social media, communicating the whole time loud and clear.

但不僅如此，請記住，我們的兒子當時 17 歲，在社群媒體上非常活躍，自始至終都大聲且清晰地進行交流。

We're not pulling out phrases or statements that wouldn't belong to Joaquin.

我們不會刪除不屬於 Joaquin 的話語。

Manny and Patricia said they drew on Joaquin's previous social media posts when writing the message.

Manny 和 Patricia 表示，他們在撰寫這則訊息時參考了 Joaquin 之前的社群媒體貼文。

When you combine these new AI tools with someone's digital trail, you can create a digital legacy that would've seemed unimaginable even a few years ago.

當你將這些新的 AI 工具與某人的數位軌跡結合時，你可以創建幾年前似乎難以想像的數位遺產。

There are some that will say, "This is creepy bringing back a voice of a child who died."

有些人會說：「讓逝去孩子的聲音重現太嚇人了。」

是的，我的意思是，當然，這讓人毛骨悚然。

Not worse than a 19-year-old holding an AR-15 inside a school.

不會比一名在學校持有 AR-15 的 19 歲青少年更糟。

This is a United States problem, and we have not been able to fix it.

這是美國的問題，我們一直沒能解決。

If we need to use creepy stuff to fix it, welcome to creepy.

如果我們需要使用令人毛骨悚然的東西來修復它，歡迎使用令人毛骨悚然的東西。

New AI tools have stunned us, but they also strike fear in us that they could be inevitably used for evil.

新的 AI 工具讓我們震驚，但也讓我們擔心它們可能不可避免地被用於邪惡的事情。

What if Joaquin's voice was used to spread misinformation or hate speech?

如果 Joaquin 的聲音被用來傳播錯誤訊息或仇恨言論怎麼辦？

See, you have to take the risk. We have to take the risk.

你必須冒這個險。 我們必須冒這個險。

You will see other families' loved ones' voices too.

你還會看到其他家庭的親人的聲音。

So this is done in a very responsible way.

Like we want this to happen. We want this voice out there.

我們希望這一切發生。我們希望有這樣的聲音。

I mean, we are sad and miserable most of the time, but sometimes you want like a small victory and keep it, and this is a small victory.

我們大部分時間都很悲傷和痛苦，但有時，你也會想取得一個小小的勝利，並將它保持下去，這是一個小小的勝利。

All of the messages for the campaign state that the voices were created with AI.

活動的所有資訊都說明這些聲音是 AI 創造的。

The rise of new generative AI tools like ChatGPT and ElevenLabs make it more accessible for anyone to create a bot of their dead loved one.

ChatGPT 和 ElevenLabs 等新型生成式 AI 工具的興起，讓任何人都可以更輕鬆地為已故親人創建機器人。

But that's not the intent with Joaquin's voice here.

但這並不是 Joaquin 聲音的本意。

That technology will keep increasing and bringing more and more things to life that maybe they can recreate Joaquin exactly the way it is.

這項技術將不斷發展，讓越來越多的東西變得栩栩如生，也許他們可以按照原樣再現 Joaquin。

I don't want that. That's not gonna bring Joaquin back.

我不想那樣。那不會讓 Joaquin 回來的。

Instead, the goal is to use the technology to build support for a cause and to help a 17-year-old boy who had a dream that he never got to fulfill.

相反，我們的目標是利用該技術為一項事業提供支持，並幫助一名 17 歲的男孩，他有一個從未實現的夢想。

He randomly out of nowhere, asked me, "Papi, how do you wanna be remembered?"

他有天突然問我：「爸爸，你希望人們如何記住你？」

So I asked him back, "How do you wanna be remembered?"

於是我反問他：「你想讓別人怎麼記住你？」

He said, "I wanna be remembered big, like Lennon, like Mandela, like Ali."

他說：「我想像藍、曼德拉，阿里一樣被世人銘記。」

We can let more people know about Joaquin because of technology.

我們可以讓更多人因為科技瞭解 Joaquin。

我們可以接觸到越來越多的人。

And yes, the tech also had another benefit.

是的，這項技術還有另一個好處。

That was more a moment of wow, I got a chance to listen to his voice again, will say that it filled my heart voices and it brought me a smile.

那真是一個令人驚嘆的時刻，我有機會再次聽到他的聲音，我會說它充滿了我的心聲，它給我帶來了微笑。

現在就行動吧，我們會找到願意的人。