Brian McClendon
① 他说过的话
7 条
而如果你能成功地把你的问题空间设计成有利于 AI 的方式，你就会很快得到一个人类还没想到的解决方案。
指向原始笔记的链接
And if you can successfully design your problem space in a way that’s conducive to AI, you’re going to get a solution quickly that humans hadn’t thought of yet.
—— Brian McClendon · [27:43]
我认为 AI 在总体上有一个特点，就是那些可以自我检验的问题，AI 可以持续钻研。
指向原始笔记的链接
One of the things that I think AI has in general is that problems that are self-checkable, AIs can grind on.
—— Brian McClendon · [26:40]
你也不应该盲目地同时开三个让它们互相争斗，因为它们很容易让自己全都相信同一个幻觉。
指向原始笔记的链接
And you shouldn’t just blindly turn on three of them fighting each other because they can easily all convince themselves of the same hallucination.
—— Brian McClendon · [37:35]
真实世界不像那个仿真那么清晰锐利。
指向原始笔记的链接
The world is not as sharp and crisp as that simulation.
—— Brian McClendon · [16:12]
我们可以拿 20 年前的照片和视频，做一些 20 年前不可能做到的重建。
指向原始笔记的链接
We can take photos and video from 20 years ago and do reconstructions that were not possible 20 years ago.
—— Brian McClendon · [08:25]
如果数字孪生做得足够好，我们将能够把世界表征到这样的程度：你不会知道自己并不在那里。
指向原始笔记的链接
If the digital twin gets good enough, we will be able to represent the world enough that you won’t know that you’re not there.
—— Brian McClendon · [32:46]
因此我觉得你会发现，我们将不再朝着目标终点随机地开发软件，而是会为那些容易解决的问题做优化，直到 AI 变得足够聪明，你再抬高门槛说：好，现在也许你可以瞄准这么高的目标了。
指向原始笔记的链接
And so I think that you’ll find that we’re going to start developing software not in a random line towards the target of destination, but we are going to optimize for those problems that are easy to solve until AI gets smart enough that you raise the bar and say, okay, now maybe you can go aim at this high.
—— Brian McClendon · [39:39]
② 出现在这些集
1 集
③ 他谈到的
点进去有真内容 —— 本页主要出口
Niantic Spatial · Google Earth · Google Maps · NVIDIA · 四维世界模型 · 视觉定位 · Gaussian Splats · 仿真到真实 · 数字分身 · 自检机制