（补）0322 NBC新闻 中英字幕

现在大众可以接触到的很多人工智能的最新进展
Recent advances in artificial intelligence now available to the masses

让许多美国人着迷和陶醉。
have both fascinated and enthralled many Americans.

但是，在所有对人工智能的赞叹声中，
But amid all the wows over A.I.,

有一些人在说 大家等等
there are some saying wait

包括2位前硅谷内部人士，
including a pair of former Silicon Valley insiders

他们现在警告科技公司，
who are now warning tech companies

一旦把人工智能的精灵放出来 就很难再把它们收回来了
there may be no returning the A.I. genie to the bottle.

在我们系列报道《人工智能革命》 采访了他们
I sat down with them for our series A.I. Revolution.

ChatGPT(人工智能聊天机器人程序)发布才4个月
It's hard to believe it's only been four months

人工智能大比拼就被推向高潮
since ChatGPT launched kicking the A.I. arms race into high gear.

这就像打响了发令枪，现在所有其他公司都说
That was like firing the starting gun that now all the other companies said

好吧，如果我们不跟着一起 我们就会输给微软。
well, if we don't also deploy, we're gonna lose the race to Microsoft.

特里斯坦-哈里斯是谷歌的前设计伦理学家。
Tristan Harris is Google’s former design ethicist.

他与阿扎-拉斯金共同创立了人道技术中心，
He co-founded the Center for Humane Technology with Aza Raskin

两人都在人工智能中看到了可喜的可能性。
both see in A.I. welcome possibilities.

我们想要的是 人工智能 丰富我们生活
What we want is A.I. that enriches our lives

帮助我们治愈癌症，
that is helping us cure cancer

帮助我们找到气候解决方案。
that is helping us find climate solutions.

但是，新的人工智能军备竞赛会带我们到那里吗？
But will the new A.I. arms race take us there

还是它会让我们走上一条更黑暗的道路呢？
or down a darker path?

大家争先恐恐去部署人工智能变成了不计后果地盲目部署
The race to deploy becomes the race to recklessness

因为这么短的时间要把它部署好 就不可能面面俱到
because they can't deploy it that quickly and also get it right.

在2020年Netflix的纪录片《社交困境》中，
In the 2020 Netflix doc The Social Dilemma

他们对社交媒体的危险性敲响了警钟。
they sounded the alarm on the dangers of social media.

我们造了这些东西，
We built these things

我们有责任去改变它。
and we have a responsibility to change it.

但今晚他们对忽视人工智能的危险提出了更可怕的警告。
But tonight they have an even more dire warning about ignoring the perils of artificial intelligence.

这将是人类所犯的所有错误中最糟糕的一个。
It would be the worst of all human mistakes to have ever been made.

而且我们真的不知道它的工作机制
And we literally don't know how it works.

我们不知道它们到底最终会干什么
We don't know all the things that will do.

在我们不能确保它是否一定安全之前
And we're putting it out there

我们就把它放出去了
before we actually know whether it's safe.

拉斯金指出，最近对人工智能研究人员的一项调查显示，
Raskin points to a recent survey of A.I. researchers

近一半的研究者表示，
where nearly half said they believe

他们认为人工智能至少有10%的可能性最终会导致极其糟糕的结果，
there's at least a 10% chance A.I. could eventually result in an extremely bad outcome

如人类灭绝。
like human extinction.

你怎么看这个问题？
What do you come down on that？

我不知道。
I don't know.

这个点在于
The point is

说真的，你就是不知道。
Seriously you don't know.

是的 我不知道
Yeah,

我的意思就是
well, here's the point like it's an,

想象一下，你即将登上一架飞机，
imagine you're about to get on an airplane

造这架飞机的工程师们，他们有一半都说，
and 50% of the engineers that built airplane say

飞机可能会坠毁的机率在10% 机上所有人都会死
there's a 10% chance that the airplane might crash and kill everyone.

不让让我上飞机啊
Leave me at the gate.

对，就是这个意思
Right, exactly.

人工智能类的工具已经可以模仿声音，
A.I. tools can already mimic voices,

让人通过考试，创造艺术和诊断疾病。
ace exams, create art and diagnose diseases.

它们每天都在变得更聪明。
And they're getting smarter every day.

在两年后 即总统选举时，
In two years by the time of the election,

人类将无法区分什么是真实的和什么是假的。
human beings will not be able to tell the difference between what is real and what is fake.

谁在建造护栏呢？
Who's building the guardrails here？

没有人在建造护栏。
No one is building the guardrails.

它的发展速度
And this has moved so much faster

完全超出了我们政府所能理解或意识到的
than our government has been able to understand or appreciate.

大家要注意 那些大的人工智能实验室的首席执行官们
It's important to note the CEOs of the major A.I. labs

他们都说我们确实需要对人工智能进行好好监管
And they've all said we do need to regulate A.I..

总是有这些声音说
There's always that notion that

也许这些公司可以自我监督。
well, maybe these companies can police themselves.

有用吗?
Does that work?

没用
No, no.

自我监督行不通
Self-policing doesn't work.

是的 这真的没用
No, it cannot work.

但是 最终 不是还是会有人来控制它们吗？
But doesn't a person ultimately control it？

我能不能就直接把插头拔掉呢？
Can I simply just pull the plug？

不幸的是，这项技术正被分散到越来越多的人手中，
Unfortunately this is being decentralized into more and more hands

所以这个技术并不只是一家公司在运行，
so that technology isn't just run inside of one company

你不可以说 我想拔掉谷歌的插头。
that you can just say I want to pull the plug on Google.

另外，想想看，
Also think about how hard it would be

要拔掉谷歌的插头
to pull the plug on Google

或拔掉微软的插头会有多难。
or pull the plug on Microsoft.

如果你可以规诫一位硅谷公司的首席执行官 你想说些什么
So what would you tell a CEO of a Silicon Valley company right now

即 虽然知道你不想成为最后一个，但你可以暂停一下吗？
and now you don't wanna be last but can you take a pause？

我的意思是 这么做可行吗？
I mean is that realistic？

你说得对
You know it, you are right.

只要求一家公司这么做 是不可行的
It's not realistic to ask one company.

我们需要做的是让这些公司聚到一起
What we need to do is get those companies,

进行有建设性的积极的交流
have to come together in a constructive positive dialogue about,

就像讨论类似禁止核试验条约一样
think of it like you know the nuclear test ban treaty, right.

我们把所有国家放在一起，
We kinda all the nations together

说我们能不能同意 我们不想在地面上部署核弹。
saying can we agree we don't want to deploy nukes above ground.

他们说 如果控制不好 危害会特别大
The stakes they say are impossibly high.

但是，当我们开启一场军备竞赛
But when we're in an arms race

用最快地速度将人工智能部署到地球上的每一个人身上，
to deploy A.I. to every human being on the planet as fast as possible

测试又不到位
with as little testing as possible

但结果一定会很糟糕
that's not an equation that's going to end well.

我们联系了一些处于人工智能发展前沿的公司，
We reached out to some of the companies in the forefront of A.I. development

微软公司告诉我们人工智能可以帮助解决人类的一些最大问题。
Microsoft told us A.I. has the potential to help solve some of humanity's biggest problems.

但他们认为 一定要让人们用到人工智能 但一定要做好管控
And they believe it's important to make A.I. tools available with guardrails.

谷歌说他们致力于负责任的开发，
Google said they are committed to responsible development

包括严格的测试和道德审查。
including rigorous testing and ethics reviews.