Transcript · English → Chinese
Discussion on AI Development and Safety
所以，当你听到David Sacks这么说， 这么说：“这些
这些公司的领导者的动机， 全都是，全都是
监管俘获”，我觉得Beck— —这完全只是想
在某种程度上攫取市场中的 某个特定部分。
J——你也认识Sam，也认识Dario。
显然，Dario昨晚刚和总统共进晚餐， 就在昨晚。
你怎么看这件事？
嗯，呃，不管， 不管他们的意图是什么，嗯，
我们想要
有意或无意地避免监管俘获。
对。
而且这项技术 发展得非常快。
想想看，呃， 我们希望技术加速发展，
而不是反过来。
所以——我可以， 我可以回顾一下过去，
想想这件事。
呃，ChatGPT刚推出时，记得吗， 呃，如果你回想一下——
再次使用它，人们说， “GPT-2 本来应该——
太危险了，不应该
甚至被发布出来。”
嗯哼。
当时，我主张 加快人工智能的发展。
原因如下。第一， 我们解决了幻觉——解决了幻觉——
幻觉问题。 现在它更以事实为依据。
注意，人工智能现在能够使用，嗯， 有用的工具，因此它可以
正确地完成工作。嗯， 我们现在可以设定
人工智能，并提示它， 使它拥有更丰富的语境。
不——所有这些能力 才得以实现。
它现在更值得信赖。它更， 你知道，更实用。
当然，它可能——它更可预—— 可预测得多。
所有这些原因，呃， 所有这些能力
得以实现，都是因为我们加快了 人工智能的发展。
加速人工智能发展和加速人工智能 安全 是一回事
理念。我更喜欢今天的汽车，而不是 那种
100年前的汽车。它有ABS。 它有车道保持功能。
它有好得多的制动系统， 还有安全气囊。
我是说， 所有这些不同的技术
之所以成为可能，是因为
技术的进步。因此，我会，我会主张
今天的人工智能， 当我们谈到进步时，别——
你不需要
将其与能力和安全问题分开思考。
对我来说，这是一回事。
对。
所以我们想加速人工智能的发展， 今天我们要推出
这项新技术，Open Shell 和 Nvidia
Century， 这两项技术让我们得以
实现约束
并进行监控。 但要做到这一点，需要大量技术
才能实现。
End of transcript · 389 words · machine translated
Original transcript (English)
So when you hear David Sacks say this, this, "The motivation of the leaders of these, these firms, it's all in, it's all a regulatory capture," I think that's what Beck- It's all just an effort to, to sort of capture a particular part of the, the market. J- And you know Sam, and you know Dario.
Dario just had dinner with the president, apparently, last night. What do you think of that? Well, uh, w- whatever, whatever their intentions are, um, we wanna avoid regulatory capture advertently or inadvertently.
Right. And the technology is moving really quickly. Just think, uh, we want the technology to accelerate, not the other way around. So-- And I can, I can just go back in time and,
and think about this. Uh, when ChatGPT first came out, remember, uh, if you go back and use it again, people said, "GPT-2 ought to be-- was so dangerous it shouldn't be even released."
Mm-hmm. Um, and at the time, I was advocating for AI to accelerate. The reason for that is this. One, we solved the hallu- solved the hallu- hallucination problem. It's now much grounded on facts. Notice how AI is able to now use, um, useful tools so that it could
do the job properly. Um, we can now condition the AI and prompt the AI so that its context is much richer. No-- All of this capability was made possible. It's more trustworthy now. It's more, you know, functional. Of course, it could-- it more t- much more predictable.
All of those reasons are, are, all those capabilities are made possible because we accelerated the development of AI. Accelerating AI and accelerating AI safety is the same idea. I prefer today's car over the car 100 years ago. It had ABS. It has lane keeping.
It has much better braking systems, airbags. I mean, all of these different technologies are made possible because of the advancement of the technology. And so I would, I would advocate that today's AI, when we think about advancement, don't--
You don't have to think about it separately from capability versus safety. To me, it's the same thing. Yeah. And so we wanna accelerate the advancement of AI, and today we're putting
forth this new technology, Open Shell and Nvidia Century, that these two technologies make it possible for us to contain and to monitor. But it takes a lot of technology to make this possible.
