用 AI 和 AR 赋予人类超能力

每个人，超人的视力、超人的听力、超人的记忆、超人的认知。这就是我要说的。一项极其平等的技术。这就是我们对这些可穿戴设备的愿景。Biz，很高兴见到你。我期待这次录制很久了。我甚至是从酒店房间通过 Zoom 或 Riverside 接入的。那么，我们就直接开始吧。很多人可能已经知道这一点，但我怀疑。你参与了加州的项目，在那里你饲养牲畜，基本上是在农场长大的，顺便说一句，我高中也是在农场上的，所以我非常理解。我很好奇那段经历如何影响了今天的你？

Every person, superhuman vision, superhuman hearing, superhuman memory, superhuman cognition. That's what I'm talking about. A tremendously equalizing technology. That is the vision that we have for these wearables. Biz, it's great to see you. I've been looking forward to this recording for a long time. I'm even zooming in or Riversiding in from a hotel room for this. So, let's just get into it. And many people might already know this, but I doubt it. But you were part of the California program where you raised livestock and essentially grew up on a farm, which by the way, I went to a high school on a farm, so I'm very sympathetic. I'm curious how did that experience influence the person you are today?

是的，人们常常惊讶于我是在农场长大的，我的家族从很久以前就务农，至今仍在务农。但实际上，如果你了解农民，有三件事很重要。第一，他们受时间、日光和季节的支配。他们有固定的日照时间来完成任务，而且必须完成，因为季节在推进。这迫使另外两件事。第一，每个人都是工程师。你现在就得修好那台拖拉机，因为你要完成收割。你今天就得修补篱笆。你没时间把牲畜赶回围栏。所以，他们都是工程师。不是科学家。我是说那种用现有工具解决问题的工程师。第二，他们是企业家。你知道，我家经营马场。马场意味着很多马粪。处理马粪有两种方式。一种是花钱请人清理运走，这要花钱。另一种是把马粪作为肥料销售，还能赚点钱。这是一举两得。所以，我的表亲们还在务农，一切要么是成本，要么是机会，利润很薄。所以，你必须让这些机会发挥作用。因此，至少在我的经验中，这并不像人们想的那么难。但我要特别感谢 4-H 全国组织和加州州立 4-H。我在 4-H 学会了编程。正如我们常说的，它不只是养牛和烹饪。第一个教我编程的是另一位 4-H 成员，就这样让我接触了计算机。所以，这是一个很棒的项目，我仍然在国家层面参与其中。我的表亲、侄子们，他们都还在 4-H。所以，对我来说，这不仅从创业和工程角度，而且直接编程计算机，都是一个很棒的项目。

Yeah, as people are often surprised by the fact that I grew up on a farm and my family's farming from way back and still is farming to this day. But actually, if you know about farmers, three things are important to know about farmers. Number one is they're governed by time, daylight, and seasons. They have x number of daylight hours to get the work done, and they got to get it done because the seasons are moving forward. And that forces two other things. The first one is every one of them's an engineer. You got to fix that tractor now because you got to get that crop done. You got to mend that fence today. You don't have time to be dealing with getting this livestock back in the pen. So, they're all engineers. Not scientists. I mean real like get it done with what you have on hand kind of engineers. And the second one is they're entrepreneurs. You know, my family runs a horse ranch. Well, a horse ranch means you got a lot of manure. Well, you got two ways to handle that. The one is you can pay someone to pick that manure up and haul it off, and that's going to cost you money. The second one is you can market that manure as fertilizer, and you can make a little money. And that's a two-for-one swing. And so, you know, my cousins who are still farming and my who are still farming, it's everything's either a cost or an opportunity and the more you can turn things into opportunities and the margins are slim. So, you got to make those opportunities work. So, in my experience at least it's not as big a stretch as people thought. But I do want to shout out yeah, the 4-H National 4-H organization and California State 4-H. I learned how to program in 4-H. It's not just cows and cooking as we like to say. The first person who taught me to program was a fellow 4-H'er and got me into computers that way. So, it's a great program and I still am involved with it to some degree at the national level. My cousins, my nephews, they're all still in 4-H. So, it was a great program for me not just from an entrepreneurial and engineering standpoint, but also directly programming computers.

除了农场和 4-H 经历，显然让你接触了编程和数字世界，还有别的吗？农场中还有其他东西塑造了你对数字和物理世界接口的思考吗？

Was there anything in addition for the farming and the 4-H experience like obviously getting you exposure to you know, kind of programming and the digital world. Was anything else in the kind of farming that also shaped how you think about the interface between the digital and physical world?

嗯，这很有趣。我认为，人类很忙。他们有很多事情要处理，工具要么为他们工作，要么就不值得。你知道，我真的认为，当我们思考如何构建这些工具时，很多时候我们被我们看到的巨大价值所困扰。尤其是在我们的行业，我们看到了它可能有多棒。这并不是新问题。这可以追溯到 Douglas Engelbart，对吧？他发明了鼠标。在某种程度上，他在 SRI 的愿景失败了，因为他的想法太复杂。他想让人类做大量工作来获得足够的技能，以解锁机器的全部力量。据我所知，即使在他 90 年代末退休时，他也感叹我们选择了简单的出路。我们拿走了电脑鼠标并继续使用，只是做点按操作。而他想用和弦键盘取代键盘。他有所有这些想法。我认为这是我们不断吸取的教训，那就是它必须非常简单。就像最后的大价值很棒，但它仍然必须非常非常容易获取，这样你才能引导人们走上那条路，因为如果他们必须参加课程来学习如何使用，他们就不会去做。无论它有多有价值。我认为这适用于——这与任何事无关，只是人性——但就像在木工车间、汽车修理厂一样。你能拿起来就说“是的，我看到了。我懂了。我用了。它有效。”的工具，才是人们一次又一次伸手去拿的工具，而不是超级复杂精细的那个。所以我一直在思考这一点，因为这就是在农场长大的真理：你必须完成它。你没有时间去尝试学习新东西。你必须完成它。

Well, it's interesting. I think you know, this point that humans are busy. They got a lot of things in their plate and the tool either has to work for them or it's just not worth it. You know, I think I really think that when we think about how we build these tools a lot of times we're beset by the tremendous value that we see. Especially in our industry, we see how great it could be. And this isn't new. This goes back Douglas Engelbart, right? Who invented the mouse. To some degree his vision at SRI failed because he had such a complex idea. He wanted humans to do a ton of work to get skilled enough to unlock the full power of the machine. And my understanding is even in his retirement in his late in the '90s, he lamented the fact that we took the easy exit. We took the computer mouse and we ran with it and just did you know, this point and click stuff. Whereas he wanted to replace the keyboard with the chord set. Like he had all these ideas. And I think that's a lesson we keep which is like it has to be so easy. Like the big pot of value at the end is great, but it still has to be so so easy to get at it that you can lead people down that path because if they have to take they have to take a course to learn how to do it, they're not going to do it. It doesn't matter how valuable it is. And I think that's true for that's not nothing to do with anything but just how humans are but that's like that's true in a wood shop, that's true in an auto shop. The tool that you can pick it up and you go yep, I see it. I get it. I use it. It works. That's the tool people reach for time and time again, not the super elaborate complex one. So I think about that all the time because that was the truth growing up on a farm is you just you had to get it done. You didn't have time to be trying to learn a new thing. You had to get it done.

所以我喜欢这样的开场，因为当人们想到技术时，至少最近，他们想得太多的是数字世界，感觉没有什么比在农场工作、用工具修理设备、确保奶牛挤奶等所有现实世界的事情更真实了，但我们正处于这个新时代，这个由可穿戴设备、智能手机，当然还有眼镜组成的新生态系统。那么，你如何看待这个新世界，我们将通过技术、通过数字来导航和参与物理世界，就像我们过去与数字空间互动一样？

So I love that way to start because when people think about technology, at least lately, they think so much about the digital world and it feels like there's nothing more real world than like working on a farm and using tools to fix your equipment and making sure the cows get milked and sort of all that stuff that's in the real world, but we're in this new era, this new ecosystem of device wearables and smartphones and of course glasses. And so how do you think about sort of this new world where we're actually going to navigate and engage with the physical world with technology through technology through digital in a way that we used to interact with digital spaces?

是的，这就是元宇宙的概念，我认为它被广泛误解了，或者至少不同的人有不同的理解，即数字和物理融合在一起的想法。实际上，让我们继续用农场的事情，因为它很棒。农民们正在开创一些惊人的工作。你知道，自动驾驶还没有农业自动化那么先进。你知道，能够自动耕地的拖拉机，无人机在田间进行侦察，确定哪些田地需要什么样的处理。这实际上是相当先进的技术，它确实非常注重将物理和数字能力融合在一起。所以，我们在互联网和软件方面经历了惊人的爆炸式增长，我认为我们在手机、笔记本电脑的框架内将软件推到了极致。就像我们把它推到了绝对极限。

Yeah, this is the concept of the metaverse, which I think has been pretty broadly misunderstood or at least is understood differently by different people, is this idea of a blending of the digital and the physical together. Actually, let's keep with the farming thing because it's great. Farmers are on the pioneer or pioneering some amazing work. You know, autonomous driving is not as far along as autonomous farming is. You know, the tractors that are able to plow the fields automatically, drones that are out there doing reconnaissance on what fields need what kind of treatment. That is actually pretty advanced technology and it really is very much about blending the physical and the digital capabilities together. And so, we had this amazing explosion with the internet and software, and I think we took software as far as it could go within the construct of a phone, a laptop. Like it's like we just took it to the absolute limit.

现在让我兴奋的是这些真正物理化的体现，通过先进的传感器，无论是视频、音频、机电一体化，还是无人机、机器人、自动化，最终我认为通过可穿戴设备，我们正在达到一个新的硬件平台，可以进一步让软件呼吸和扩展。

And what's exciting me now is these really physical manifestations where through advanced sensors, whether it be video, audio, mechatronics, whether it be drones or robotics, automation, and then ultimately I think through wearable devices, we're getting to a new plateau of hardware that can further allow software to breathe and expand.

AI 是一个很有趣的例子，因为它现在很时髦。如果人们有机会真正使用，忽略所有的炒作，真正尝试在日常生活中使用这些工具，有些领域它有用得令人难以置信。

AI is such a fun example cuz it's so vogue right now. And if people have had a chance to really use, ignore all the hype, and really try to use these tools in their daily lives, there are some areas where it's mind-blowingly useful.

你知道，我在做一个小型家庭自动化项目，我在调试，有一些不常见的物联网设备，它们的 API 没有列出，我需要很长时间才能构建一个模糊测试器来发现这些。天哪，用这些非常有用的 AI，你可以在几分钟内完成。

You know, I'm doing a little home automation project, and I'm debugging, and there's these obscure internet of things devices that have unlisted APIs, and it would take me a long time to build a fuzzer to discover those. Man, you can do it in minutes with these tremendously useful AIs.

当然，有很多事情它们不擅长，但我仍然觉得与它们的交互非常笨拙，我要么去手机，要么去别的地方，无论是语音还是文本，都是一种非常事务性的东西。

Of course, there's a ton of things that they're not good at, but I still find the interface to them very awkward where I'm going to my phone or I'm going, whether it's voice or text, it's this very transactional thing.

有趣的是，我发现自己突然变成了剪切粘贴机器。比如我在做一个编码项目，我就像，好吧，我得把这个结果、调试器输出剪切到这里，然后它给我答案，我再剪切粘贴回那个东西。我就像，为什么我成了剪切粘贴机器？这应该全部集成起来。

And what's so funny, I find myself I'm like the cut and paste machine suddenly. Like I'm doing a coding project here, and I'm like, okay, I got to cut this result, the debugger output into here, and then it gives me the answer, and I'm cutting and pasting back into the thing. And I'm like, why am I What system of events happened where I'm now the cut and paste machine? This should all be integrated.

我们在可穿戴设备内部有演示，而你们通过我们的早期访问计划，使用市场上现有的 Ray-Ban Metas，可以使用这个叫做 Live AI 的工具。30 分钟的会话，直到电池耗尽，因为它是事后附加的。但在这 30 分钟里，它能看到你所看到的，听到你所听到的。

And we have demos internally in our wearables and you with the Ray-Ban Metas that are in market today through our early access program, you can use this tool called Live AI. 30-minute session until the battery runs out cuz it was kind of bolted on after the fact. But for 30 minutes, you can have it can see what you're seeing, it hears what you're hearing.

它对我日常生活的有用程度差异很大。我在做一个胶片冲洗项目。胶片冲洗是一件非常讲究的事情，用什么化学药品、在什么温度下、多长时间。所以我通常会在旁边放一台笔记本电脑，我试着在笔记本电脑上输入，好吧，那是什么？用 Live AI 会话来做这件事太不可思议了。它只是看到你在做什么。它就像，好吧，你还要再做 20 分钟，然后你要做另一件事，你为什么不现在准备这个呢？太惊人了。

And the difference in how useful it is to me as I go about my day in the world, I was doing a film development project. And film developing is a real fussy business with how many what kind of chemical at what temperature for how long. And so you usually I'm doing it with like a laptop next to me and I'm trying to type into the laptop, okay, what's the thing? Doing it with the Live AI session was incredible. It just sees what you're doing. It's like, okay, you got 20 more minutes of doing that and then you're going to do this other thing and why don't you go ahead and prepare this now? It's stunning.

所以对我来说，这是关于在双向融合物理和数字。你必须让 AI 访问你操作的物理环境和你在桌面或手机上操作的数字环境。反过来，你希望能够通过机器人、自动化将这些数字构造带入现实。所以我认为在我们面前有一个非常非常令人兴奋的十年，在这种合成中。

So for me, it is about blending the physical and digital together in both directions. You have to give the AI access to the physical context in which you operate and the digital context in which you operate on your desktop or on your phone. And conversely, you want to be able to bring those digital constructs into reality through robotics, through automation. So I see really a very, very exciting kind of decade ahead of us in that synthesis.

我喜欢。所以让我们更广泛地谈谈 AI。我的意思是，我足够极客，对数字和物理世界的接口以及它如何改变人类和 Homo technicus 的本质。我可以把整个讨论都花在这上面，那会很棒。但也有更广泛的 AI。那么，AI 的元观点是什么，在产品中的使用，在世界中的使用，什么是为人们设计的 AI 哲学，以及即将到来的东西，你知道，其中一件事是我提醒自己，我需要给你一本我新书《Super》的签名本，因为我很确定在和你讨论这个时，我们会有一个很棒的对话，我们实际上以非常相似的方式看待世界。

I love it. So let's broaden out to kind of AI generally. I mean like the I'm enough of a geek and kind of the interface with the digital and the physical world and how that transforms what it is to be human and homo technicus. I could spend the entire discussion on this, which would be amazing. But there's also kind of AI, you know, kind of more broadly. So what's what's kind of the meta view of kind of AI and in, you know, use in the products, use in the world, you know, kind of what is the kind of the AI philosophy for design for people and and kind of what's coming and you know, one of the things I I'm reminding I need to get you a signed copy of my new book Super because I'm quite certain in discussing this with you, uh we'd have a great conversation and we're actually seeing the world in very similar ways.

是的，所以在我看来这有三层。第一层是，你知道，我们正在构建这些非常令人兴奋的模型，我有一个深刻的信念。我发现自己有时实际上与 AI 讨论的双方都在交战。我深信这些是极其重要、有意义的事情，将有意义地推进人类能力。

Yeah, so there's three layers to this in my mind. The first one is, you know, we're building these very exciting models, and I have a profound belief. I find myself actually sometimes at war with both sides of the AI discussion. I have a profound belief that these are hugely important, meaningful things that will meaningfully advance human capability.

我把它比作一个词语计算器，对吧？在 1950 年，计算器是一个人。在 1970 年，它不再是一个人。它是另一种东西。我认为我们所有人，起初他们喜欢禁止它们进入学校并摆脱它们，我成长在“你永远不会随时带着计算器”的时代。我随时带着三个计算器。我的高中老师错了。

I kind of liken it to a word calculator, right? In the year 1950, a calculator was a person. In the year 1970, it was not a person anymore. It was a different thing. And I think all of us and and at first they like ban them from school and get rid of them and they I grew up in the you'll never you won't have a calculator with you at all times. I have three calculators with me at all times. My high school teachers were wrong about that.

我认为我们正在构建的 AI 是词语计算器。我是认真的。词语、图像，就像视觉，它们是非常复杂的计算器，已经超越了数学的简单符号空间，进入了这个更高阶的空间。我也不认为它是我们所理解的人类智能、能动性、意识和思维的那种东西。

I think of the AIs that we are building as word calculators. I mean I really mean that. Word, image, like visual, like they're really complex calculators that have moved beyond the simple symbolic space of mathematics into this higher order space. I also don't think it is even the kind of thing that is human intelligence as we understand intelligence and agency and consciousness and thought.

所以我发现自己必须与双方斗争。就像 AI 既是一件大事，也不是那种大事。所以这就像我对它的第一个信念。这给了我极大的信心去使用这个工具。

And so I find myself I have to fight both sides. Like AI is both a huge deal and also not that kind of a huge deal. So that's like my first kind of belief about it. Which gives me tremendous confidence in how I use the tool.

第二层是我们正在遇到它的信息论极限。你知道，我们谈过向它投入算力和缩放定律。但如果你一直追溯到诺伯特·维纳和他的控制论，信息论的最初构造，这个想法是我们能从某物中提取多少可泛化的比特，足够可泛化的比特。我们发现对于所有人类媒体语料库，它不够。不够。

The second one is we're running into the information theoretic limits of it. You know, and and we've talked about throwing computer at it and scaling laws. But if you go all the way back to like Norbert Wiener and his cybernetics of the first, you know, constructs of information theory, this idea of how many bits can we pull out of something that are generalizable, that are sufficiently generalizable bits. And we're finding out that for all the corpus of human media ever produced, it's not enough. It's not enough.

我们在机器人技术中也发现了这一点。你知道，机器人技术是我们最近在 Meta 内部启动的一项努力，作为我们 Lama 计划的附属。当你无论有多少视频显示某人抓取咖啡杯，你实际上没有得到你需要的数据，因为你不知道施加了多少力的本体感觉，以及我们如何检测到，好吧，这是一个塑料杯，它会偏转到某个点。而且上面有冷凝水，所以我需要施加更多的力来抵消我经历的摩擦损失。我们自动地做这些。我们甚至没有——当我们做这些事情时，我们脑海中没有一个有意识的念头。

We're finding that in robotics. You know, robotics is an effort that we kicked off recently inside of Meta in partnership kind of was as a kind of an adjunct to our Lama program. And when you no matter how many videos you have of somebody grabbing a coffee cup, there's you're actually not getting the data you need because you don't know the proprioception of how much force is applied and how we detected, okay, this is a plastic cup, it's going to deflect to a certain point. And there's condensation on it, so I need to apply a little bit more force to counter the loss of friction that I'm experiencing. We do that autonomically. We don't even There's not a single conscious thought in our head when we're doing those things.

Arya，当你从口袋里拿出手机时，你不知道你第二个手指的角度，或者你用拇指施加了多少力来避免碰到钥匙，在某种程度上，我们认为的智能，我们谈论的是人类大脑的高阶功能。

Arya, when you're taking your phone out of your pocket, you don't know what the angle of your second digit is or how much force you're applying with your thumb to avoid getting the keys to some degree, the things that we think of as intelligence, we're talking about, you know, the higher-order functions of the human brain.

可以说，智能中不那么令人印象深刻的部分——深层大脑、哺乳动物脑、杏仁核，那种蜥蜴脑智能——才是我们在现代最难捕捉的。所以，尽管我对“词语计算器”感到兴奋，我真心相信 Yann LeCun 的愿景：你必须做这种开创性工作，才能突破到一个具有常识、能以更实质的方式理解因果关系的世界模型——不是统计式的“大杂烩”，而是基于模型的。

That's arguably the less impressive part of intelligence—the deep brain, the mammalian, the amygdala, that lizard brain intelligence. That is wildly hard for us to capture in the modern era. So, as much as I'm excited about the word calculator, I really do believe in Yann LeCun's vision that you have to do this pioneering work to break through to a world model that has common sense, that understands causality in a more substantial way—not in a statistical kind of soup way, but in a model-based way.

他总用这个例子：我 4 岁的女儿——她现在 7 岁了——4 岁时就能开高尔夫球车。如果路上有只火鸡（在卡梅尔有时会发生），她能绕过去。而 Waymo 只会停下来。就像：我不知道——我没有火鸡过滤器。我没有关于火鸡的数据。我只能停下。我放弃。

He always uses this example: my 4-year-old daughter—she's now seven—when she was four, could drive a golf cart. And if there was a turkey in the road, which would happen in Carmel sometimes, she could drive around it. Whereas Waymo would just stop. It's like, I don't know—I don't have a turkey filter. I don't have the data on the turkey. I've got to stop. I give up.

然后我的第三层是：我希望它是具身的。我真的觉得这几乎可以追溯到 J.C.R. Licklider，他是最早坐在终端前进行实时编程的计算机科学家之一，他相信那个愿景。他后来资助了后来成为 SRI 的机构，资助了 Engelbart，资助了一堆 DARPA 项目。当时是在 DARPA 的 IPTO。我觉得我们正处在那个时代。我们处在 AI 的终端时代，但它不想只是那样。它想无处不在。它想无所不在。它想充分了解你的生活背景，以及你是谁、你的生活的历史。

And then my third layer of this stuff is I want it to be embodied. I really think it almost goes back to J.C.R. Licklider, who was one of the first computer scientists to sit down at a terminal and do live programming, and he believed in that vision. He was really the one who then funded what would become SRI, funded Engelbart, funded a bunch of DARPA. It was at the IPTO at DARPA. I feel like we're in that era. We're in the terminal era of AI, and it doesn't want to be like that. It wants to be everywhere. It wants to be ubiquitous. It wants to be in full context of what your life is and with history of who you are and what your life is.

我想到 Douglas Hofstadter 的书《我是个怪圈》，我们大脑里都运行着彼此意识的小型版本、彼此意识的模拟，这让我们能有效协作。而我的 AI 显然没有这些。它不知道我想要什么、我是什么样的人。它无法从上下文推断任何东西。所以对我来说，我喜欢我们现在的状态。我是坚定的信仰者。我也希望我们投资这些世界模型，并且我想把它从终端中解放出来。

I think about Douglas Hofstadter's book I Am a Strange Loop and how we all have little mini versions of each other's consciousness, simulations of each other's consciousness running in our brains that allow us to collaborate effectively. And my AI obviously doesn't have that. It has no idea of what I want, what I'm about. It can't infer anything from context. So for me, I love where we are. I'm a huge believer. I also want us to invest in these world models and I want to free it from the terminal.

那么世界模型和从终端中解放出来——这些是与规模 Transformer 根本不同的新技术吗？是一种混合组合吗？就像透过玻璃模糊地思考这会如何实现？

And is the world models and the free from the terminal—are those fundamentally new different technologies from the kind of scale transformer? Is that a mixed combination? Like what's the kind of looking through a glass darkly thinking about how this would be done?

是的，我真的认为世界模型是一种新的不同的东西。因此，我认为它的时间线是不可知的。你知道，当我思考世界模型时，我有点想起我们在机器学习方面的处境。当我还是本科生，2004 年从哈佛毕业，我教了一门人工智能导论课。实际上我就是在那里遇到 Mark 的。他是我那门课的学生。事实上，我们当时教的是神经网络曾一度有前途，但现在已知是死技术。

Yeah, I think I really think that the world model is a new different thing. And as a consequence I think it's an unknowable timeline. You know, I'm reminded a little bit when I think about the world model of where we are with machine learning. When I was an undergraduate and I graduated Harvard in 2004 and I taught a class introduction to artificial intelligence. It's actually where I met Mark. He was a student of mine in that class. We taught as a matter of fact that neural networks were a once promising now known to be dead technology.

每个人都得构建一个手写识别神经网络，而且它成功了。然后他们说，是的，你只能用它做这个。就这样。恭喜，这是你的学位。而现在神经网络统治世界，Yann LeCun 是对的，Geoffrey Hinton 和 Bengio 也是。他们是对的，他们都得了图灵奖，上帝保佑他们。还需要一系列其他解锁，尤其是 GPU，才能让那项技术蓬勃发展。所以，在足够长的时间线上，我不会赌 Yann LeCun 输，但我还不知道那一块的解锁是什么。

And everyone had to build a handwriting recognition neural network and it worked. And they're like, yeah, that's all you can do with it. That's it. Congratulations, here's your degree. And now neural networks run the world and Yann LeCun was right, and Geoffrey Hinton and Bengio. They were right and they all won the Turing Award, God bless them. And it took a series of other unlocks, GPUs in particular, to get to the point where that technology could flourish. So, I wouldn't bet against Yann LeCun over a long enough timeline, but I don't know what the unlock is yet on that piece.

我认为具身部分可以两者兼顾。具身部分将受益匪浅，也许还能大大帮助世界建模。一旦你把那些传感器放出去，你就有更好更丰富的数据。当你有机器人数据，会给你本体感觉，给你摩擦力。我觉得那将是一个巨大的解锁。所以，我认为具身那个实际上——我把它们按顺序放错了。具身那个可能是中间环节。获得那些数据并处于那种情境中，对当前模型有巨大好处。而且它可能也是你开始理解构建世界模型需要什么的一部分数据。而我们似乎天生就有世界模型。

I think the embodied part can do both. The embodied part will benefit a lot and maybe help a lot with world modeling. Once you have these sensors out there and you have better richer data. When you have robotics data, which will give you proprioception, which will give you friction. Like, I think that is going to be a big unlock. So, I think the embodied one is actually—I put them in a sequence the wrong way. The embodied one is probably the in between. It benefits the current models a huge amount to get that data and to be in that context. And it also probably is some of the data that you need to start to understand what it takes to build a world model. Which we appear to be born with.

你知道，其中一个可穿戴项目，我想你称之为 Orion 项目。没错。我其实不知道你为什么叫它 Orion。我不是怀疑，我只是不知道。

And you know, one of the wearables project, I think you call it the Orion project. That's right. I actually don't know why you named it Orion. I'm not saying that skeptically, I just don't know.

是的，没错，非常酷。我带来了。是的，没错。

And yes, exactly, very cool. I brought them with me. Yes, exactly.

我有幸来到 Meta 总部，体验了它，并得到了一些详细的了解。那么，你——跟我说说 Orion 项目吧？告诉我你认为重要的用例是什么，以及元宇宙在 Orion 的推动下正走向何方？

I had the pleasure and honor of coming by Meta HQ and playing with it and getting some of the detailed exposure. So, what do you—tell me about the Orion project? Tell me about what you see as the significant use cases and where the metaverse is moving towards with Orion?

是的，这是——所以，我们现在戴着 Orion 眼镜。我还戴着这个腕带，这个神经接口腕带，实际上——我看看能不能稍微展示一下。它背面沿着带子有这些小金属凸起，那些是测量 EMG 的传感器，肌电图。它们测量我手部传下来的电脉冲。我能做的是——如果我做得对，希望你们看不到任何东西——所以，如果你直视我，我面前有一个屏幕，我有一个界面，所以我可以处理邮件。我可以看 Instagram。我现在没在做那个。我发现——我们有可以玩的小游戏。我以前在开会时被抓到玩游戏。公平地说，那不是我的会议。我只是在旁听，但我被抓到过。而且我可以——我可以用手。它有摄像头，所以我可以用手朝前。那有点尴尬。所以，我可以手腕静止，用少量手势来做。这是主页手势。我把自己带回了主页。这是选择手势，我用眼动追踪来尝试指向东西。

Yeah, this is—so, we've got the Orion glasses are on right now. I've also got this wristband, this neural interface wristband that is actually—I'll see if I can maybe I can going to a little bit. It's got these little metal bumps on the back of it all down the band and those are measuring the EMG sensors, electromyography. They're measuring electrical impulses going down my hand. And what I'm able to do—and you won't be able to see anything, hopefully, if I'm doing this right—is so, if you're looking straight at me, I've got a screen in front of me and I have an intro present so I can do my email. I can do Instagram. I'm not doing that right now. I found—we have little games that you can play. I have been caught playing the games in meetings before. They weren't my meetings, to be fair. I was just listening in, but I have been caught. And I can do it—I could use my hands. It's got cameras, so I can use my hands forward facing. That's a little awkward. So, I can do it with my wrist at rest by using a small number of gestures. This is the home gesture. I brought myself home. And this is the select gesture and I'm using eye tracking to try to direct things.

所以，我们必须做的是在光子和光学方面解决很多难题。有些东西我们理解。我们理解如何构建应用。我们理解如何做这些。这里有一些新颖的交互设计，涉及腕带和眼动追踪。但我认为我们已经把它做得相当直接。做神经接口超级难。你在基于从表面观察到的电脉冲构建一个关于手在做什么的 AI 模型。所以，你只需要很多人来构建一个通用的模型，让任何人都能戴上它。

And so, what we have to do is we have to do a lot of tough problem solving on photonics and optics. Some of the stuff we understand. We understand how to build apps. We understand how to do these things. And there's some novel interaction design here with the wristband and eye tracking. But I think we've made that pretty straightforward. Doing the neural interfaces was super hard. You're building an AI model of what the hand is doing based on these electrical impulses that you're able to observe from the surface. And so, you just need a lot of people to build a generalized model that works so that anybody can put this on.

我认为我们在你身上取得了成功，Reed。我想，戴上它的人中，有超过 95% 的人能立刻知道手部的形状。这让我们能够实现基于手势的控制，即使你的手放在口袋里、背后，非常细微的动作也能识别。

And I think we had success with you, Reed. I think we're well into the 95th percentile of people who put this on and can right away know what shape the hand is in. And that allows us to do these gesture-based controls even with your hands in your pocket, behind your back, very subtle.

没错，就是滑动手势。

That's right, the swipe gesture.

是的。上下左右滑动。点击是这样。这很有效。你可以双击来唤醒你的 AI。所以，这需要大量的工作。

Yep. Swiping up and down, left and right. Tapping is this. This works. You can do a double tap to get your AIs. And so, that was a lot of work.

最难的是光学和光子学。所以，这里面有一个微型投影仪。如果我把这个拿得离摄像头太近，就会失焦，所以请原谅我这套极其复杂的视频通话设置。但这里面是一个微型投影仪。它使用的是微型 LED。我们自己制造这些微型 LED。我们自己生产。三个这样的 LED 可以放进一个红细胞里，而且每一个都比太阳还亮。它们必须极其高效，因为头显的电池容量不大。实际上，比电池更糟糕的是，散热空间也不大。你不能辐射太多热量，因为我们不能烧伤你的脸。我听说烧伤脸是非常糟糕的。我听说这不推荐用于消费电子产品。所以，我们的热窗口很窄。它们产生光的方式必须超级超级高效。有些东西真的很难。红色波长非常难。红色是一种非常长的光波长。在非常非常小的空间里产生它很难。你必须使用大量的镜子和反射以及复杂的几何结构。你制造的是绝对微小尺度的消费电子产品。所以，你产生这些光子。然后你必须把光子弯曲成光。我们使用一种叫做波导的东西。波导利用光的全内反射原理，就像光纤一样，试图以一定的速度弯曲光线。但是，当然，当你的眼镜不是光纤，而是包裹在盖子里的，就会有几个其他问题，那就是其他光子从外界进入并被困在同样的光管中。这会产生彩虹和雾霾。以及这些分散你注意力的伪影。然后，还有一些光子会逃逸，所以外面的人有时能看到它们。在这种情况下，我们有一种出射光栅，所以如果我把脸放在正确的角度，你可以看到那里有一点光。哦，那是杂散反射。是的。如果我把脸放在正确的角度，你可以看到一点蓝光从光栅中逃逸出来。所以，你想尽量减少伪影，因为这些首先需要是普通的眼镜。当它们关闭时，我必须能够看到你的眼睛，看到我的眼睛。这种人与人之间的联系很重要。否则，我不会使用这些眼镜。你必须在舒适的全天可穿戴形态中完成所有这些。有很多挑战。我认为，这是 10 年努力的成果，当我们开始这个项目时，我们认为成功的机会不到 10%。所以，它的存在证明了马克·扎克伯格的愿景，以及我们的首席科学家迈克尔·阿布拉什和他的团队长期以来的愿景。我们正在做从材料科学到开发新化学物质、新玻璃的一切。光子学工作是前沿的。而且超级难。

The hardest one was the optics and photonics. And so, inside of this, there's a tiny projector. And if I hold this any closer to the camera, it'll be out of focus, so you'll forgive me for my absurdly elaborate VC setup. But inside here is a tiny projector. And it's using micro LEDs. We build these micro LEDs ourselves. We manufacture them ourselves. Three of them would fit inside of a red blood cell, and they're individually brighter than the sun. And they have to be incredibly efficient because you do not have a lot of battery on the headset. And actually, worse than battery, you don't have a lot of thermal space. You just can't radiate a lot of heat because we can't burn your face. Burning faces is terribly bad, I'm told. I'm told it's not recommended for consumer electronics. So, we have a tight thermal window. It has to be super super efficient in how they generate things. Some of these things are really hard. Red wavelengths are really hard. Red is a really long wavelength of light. It's hard to generate it in a very very small space. You have to use a lot of mirrors and reflections and complex geometry. You're making these absolutely tiny scale consumer electronics. So, you're generating these photons. Then you have to bend the photons into the light. We use a thing called a waveguide. And waveguides use the concept of total internal reflection of light, the same way fiber optics work, to try to bend the light in a certain pace. But, of course, there's a couple other problems when you have glasses that aren't fiber optics, which are encased in a cover, which is that other photons enter from the world and get caught in those same light pipes. And that can create rainbows and haze. And these kind of artifacts that are distracting to you. And then also, some of those photons escape, so people outside can sometimes see them. In this case, we have a kind of an exit grating, so if I put my face at just the right angle, you can see a little bit there. Oh, that's a stray reflection. Yep. If I put my face at just the right angle, you can see a little blue light escaping from the gratings. So, you want to minimize the artifacts because these need to be regular glasses first and foremost. When they're powered off, I have to be able to see your eyes to see my eyes. And like that has to be a that human connection that we have is important. Otherwise, I wouldn't use the glasses. You have to do all of this in a comfortable all-day wearable form factor. There's a lot of challenges. This was I thought, you know, this was 10 years in the making and we thought when we started this program that we had less than 10% chance of being able to build it. So, the fact that it exists is a true testament I think to a vision that Mark Zuckerberg had that the research team Mike Michael Abrash, our chief scientist, and his team had for a long, long period of time. We're using we're doing everything from material sciences, you know, developing new chemicals, new forms of glass. You know, the photonics work was cutting edge. And it was super hard.

现在，我们一直认为，就你提到的元宇宙而言，首先会出现的是世界中的全息图。起初它们只是与你相关，所以那是你的个人界面，然后随着时间的推移，它们会附着在世界上，最终你会有一个 AI 来做这件事。令我们惊讶的是，AI 先来了。就像 AI，你知道，提前了。我们一直有这个愿景。这是马克和我从第一天起就有的愿景。这些是——AI 和 AR 是重点。这就是为什么我们在 Facebook AI 基础研究 FAIR 上投入巨大，并在 Reality Labs 上投入巨大。我们把顺序搞错了。AI 比预期更早出现。所以，令人兴奋的是，就像我之前提到的，Ray-Ban Meta 有实时 AI 会话，现在有很多产品是完全可行的，易于使用，介于全 AR 眼镜（很壮观，但会很昂贵）和 Ray-Ban Meta（非常实惠，但功能稍有限）之间。整个光谱现在向我们开放，这真的非常令人兴奋。

Now, we always thought, to your point about the metaverse, that the first thing would happen would be these holograms in the world. And at first they'd be just referenced to you, so that's your personal interface, and then over time they would be attached in the world, and then eventually you would have an AI that was doing it. What's been surprising to us is the AI came first. Like the AI, you know, moved up in the We always had this vision. It was a vision that we've had from Mark and I had from day one. These were the It was AI and AR were the things. That's why we had this big investment in Facebook AI fundamental AI research in FAIR and this big investment in Reality Labs. We had the sequencing wrong. The AI showed up earlier than expected. So, what's been so exciting is, just like I mentioned earlier on the Ray-Ban Metas have live AI sessions, there's a lot of products now that are totally valid products that are easy to use between full AR glasses, which are spectacular, but will be expensive, and the Ray-Ban Metas, which are super affordable, but like on a little bit more limited in their functionality. That entire spectrum is now open to us, and it's really, really exciting.

好的，我有一个自私的问题，也许我们会剪掉，然后是一个真正的问题。所以，我是一个 90% 的视力都来自一只眼睛的人。我能使用这些吗？

Okay, I have a selfish question, which maybe we'll cut, and then a real question. So, I am someone who sees 90% of my vision out of one eye. Will I be able to use these?

是的。所以，这些是双目的。所以，显然，你会经历与正常情况相同的深度感知限制。

Yes. So, these are binocular. So, obviously, you'll experience the same limited limits to depth perception that you would experience normally.

没错。

That's right.

尽管当然深度感知主要不是基于双目视觉，事实证明它是基于物体在空间中的运动和大小。但不管怎样，我认为所以你仍然能够使用这些。在未来，我确实预计这里的一个真正选项空间是，如果你变成单目呢？如果你只在一只眼睛里有显示呢？

Although of course depth perception isn't primarily based on binocularity it turns out it's based on movement of things through space and and size. But anyways, but I think so so you'll still be able to use these. Over the future of time I do expect a one of the real option spaces here is what if you go to monocular? What if you just had the display in one eye?

是的。

Yeah.

那里有一些挑战。它会产生一些双目竞争。所以对于双眼视力完整的人来说，他们有时可能会难以知道，嘿，我的眼睛在看同一个空间时看到了不同的东西。你实际上会更好。

And there's some challenges there. It creates some binocular rivalry. So for people who have full vision in both eyes they might struggle sometimes to know, hey my eyes are seeing different things looking at the same space. You would actually be better off

哦，我喜欢。我等不及了。

Oh, I love it. I can't wait.

它会是一个更好的——事实证明对你更好。你会期望——你会有一个小超能力。

It would be a better It turns out it's better for you. You would expect You would have a little superpower.

没错。你会有一点优势，单目显示可能对你来说很棒，更便宜、更轻，而且可能一样好。所以我认为你可能比我们其他人有一点内部机会。我喜欢。我喜欢。

That's right. You'd have a little like advantage the monocular displays would probably be great for you, cheaper, lighter, and and probably just as good. So I think you've got maybe a little inside opportunity on the rest of us. I love it. I love it.

好的，所以对于外面的一些怀疑者，我觉得这对你来说尤其是一个好问题，因为你基本上是创建新闻推送的人。对于我们中的年轻人，他们可能不记得新闻推送刚出来时，每个人都像，“不，这太可怕了。我们在做什么？”而现在我们就像，“哦，新闻推送。这太合理了。这很完美。”他们无法想象没有它的世界。

Okay, so for for some of the skeptics out there, I feel like this is especially a good question for you because you were essentially the person who created the newsfeed. And for for those of the youngs among us, they might not remember this that the newsfeed came out and everyone was like, "No, this is horrible. What are we doing?" And now we're like, "Oh, a newsfeed. This makes so much sense. This is perfect." They can't imagine a world without it.

是的。

Yeah.

所以有些人可能也在说同样的话。

And so some people might be saying the same thing.

他们可能会说：“你为什么要创造这些我们不需要的东西？”比如，你会对怀疑者说什么？然后，当你看到人们使用它时，你认为一些主流应用会是什么？

They might be saying, "Why are you creating these things that we don't need?" Like, what would you say to the skeptics? And then also as you see people using it, what do you think some of the mainstream adoptions are going to be?

是的。嗯，这个答案有两个部分。第一部分是关于 newsfeed 和类似 newsfeed 的产品，尽管人们对此感到惊讶和愤怒，但使用量是立即的。是立即的。人们经常问我们，鉴于 backlash，你们怎么决定坚持使用 newsfeed？其实并不难，因为内部我们看到所有用户数量一夜之间翻倍，而且从未下降。同样，newsfeed 是一个构建的产品——设计界有一个著名的理念，我举最常给出的例子是在公园里。你看一个公园，城市中的自然场所，人们会告诉你路径应该在哪里，通过在草地上踩出路径。所以无论你看到草被踩平的地方，你就在那里铺一条路。这就是 newsfeed 那种创新。我们观察人们在网站上的行为，然后只是构建了一条让他们正在做的事情更容易的路径。像 AI 这样的东西是不同的。这些是不同类型的创新。这些是真正的新工具。这是一个新空间。这是一座建筑，就像它已经被建造好了，你不能只是以后再弄清楚路径应该怎么走。所以这些是不同类型的创新。它们确实来自同一个地方，就像我每天的首要原则，我的团队会告诉你，人类说，这个世界上哪个人因为拥有这个而过得更好？这涉及到思考他们 instead of it 在做什么？他们当前对问题的解决方案是什么？问题是什么？是大问题吗？顺便说一句，有两种大问题。有像罕见但超级大的问题，也有超级常见但小的问题。这些都是我们可以去解决的 collectively 大问题。但我说的是，找到那个人，他们的生活就像，哦我的天，好多了。AI 在这方面一直不均衡。到目前为止，AI 的答案就像作业。像孩子们做作业和得到作业帮助，我认为这是 AI 的头号用例。第二是程序员，他们以前去 Stack Overflow。这些 pretty close 我会说 to the park path story，因为这些人以前使用 Google 或在线资源，现在使用一种 consolidated 形式。但就像智能手机是一种颠覆性的东西。它就像，“嘿，你有一个电话，你有一个 iPod，你喜欢网络。我们要把所有这些放在一个地方，对吧？那就是 pitch。这将类似于那个。这将就像，“嘿，看，你已经有一个电话，你喜欢做这个。你已经喜欢——你已经喜欢 Instagram。这是做你已经在做的事情的更好方式。然后一旦你通过舒适的路径建立了那个滩头阵地，那么巨大的机会就出现了。你知道，你谈到了拥有超能力。这就是我们对这些可穿戴设备的愿景。每个人，超人的视觉，超人的听觉，超人的记忆，超人的认知。这就是我在说的。这就是我真正相信会在这里发生的事情。一种 tremendously 平等化的技术。今天我们在社会中经常谈论出生彩票的变幻莫测。我们正确地这样做。我们过去在种族、性别或物理地理的背景下谈论出生彩票。这些是我们生活如何展开的非常重要因素。但我们不谈论出生彩票的其他部分。谁有不可思议的记忆，不可思议的视觉，不可思议的听觉。有些人天生就有这些才能。不可思议的创造性思维、模式匹配能力。如果我们有可穿戴设备，我们没有理由不能都成为国际象棋中的加里·卡斯帕罗夫。你知道我在说什么吗？现在，这可能让国际象棋失去乐趣，所以我不推荐反竞争性能。但我的观点是，钟形曲线的变幻莫测和人类能力的范围超越了仅仅人口统计学的。它们存在于基本能力方面，如果每个人都完全 access 这些设施，我们作为一个社会会有多有趣。那就是我看到的未来。我认为这是一个非常有说服力的 pitch 让人们尝试新事物。但它仍然必须做 Instagram。

Yeah. Um, so there's two parts to this answer. The first one is with newsfeed and products like newsfeed, as much as people were kind of surprised by it and outraged by it, the usage was immediate. It was immediate. People often ask us, how did you guys decide to stick with the news feed given the backlash? It wasn't actually hard because internally we saw all of our users' numbers doubled overnight and they never went back down. And likewise newsfeed was a product built—there's a kind of a famous idea in design which is—I'll use the example that is most often given is in a park. You look at a park, a natural place in a city, and people will tell you where the path should be by cutting paths in the grass. So wherever you see the grass is trampled down, you put a path there. That was the kind of innovation that newsfeed was. We watched the behavior of people on the site and just built a path that made what they were doing easier. Things like AI are different kinds of things. These are different types of innovations. These are truly new tools. This is a new space. This is a building, like it's been constructed and you can't just figure out where the path should go later on. And so these are different types of innovations. They do come from the same place which is like at my alpha my omega every day and my team will tell you that the human says like what human on this earth is better off because they have this? And that involves thinking about what are they doing instead of it? What is their current solution to the problem they have? What is the problem? Is it a big problem? By the way, there's two types of big problems. There's big problems like rare but super big and there's also super common but small. Those are both collectively big problems that we can go solve. But I'm like find me the human whose life is just like, oh my god, way better. AI has been uneven at this. The answer for AI so far has been like homework. Like kids doing homework and getting help with their homework has been like by far I think the number one use case for AI. Number two has been coders who used to go to Stack Overflow. Those are like pretty close I would say to the park path story cuz these are people who were previously using Google or online resources and are now using a consolidated form of that. But like the smartphone was a kind of a disruptive thing. It was like, "Hey, you have a phone and you have an iPod and you have like you like the web. We're going to put all those in one place, right? That was like that was the pitch. This will be similar to that. This will be like, "Hey, look, you already have a phone that you like doing this. You already like—you already like Instagram. This is a better way of doing a thing that you already do. And then once you've established that beachhead through comfortable paths, then the huge opportunity presents itself. You know, you talked about having superpowers. That is the vision that we have for these wearables. Every person, superhuman vision, superhuman hearing, superhuman memory, superhuman cognition. That's what I'm talking about. That's what I really believe is going to happen here. A tremendously equalizing technology. Today we talk often in our society today about the vagaries of the birth lottery. And we're rightly doing so. We used to talk about the birth lottery in the context of whether it be race and gender or physical geography. And those are hugely important factors in how our lives play out. But we don't talk about the other parts of the birth lottery. Who has incredible memory, incredible vision, incredible hearing. Some people are born they just have those talents. Incredible ability to think creatively, to pattern match. And there's no reason we couldn't all be Garry Kasparov in chess if we have the wearables. You know what I'm saying? Now, it probably takes the fun out of chess, so I'm not recommending anti-competitive performance. But my point is like the vagaries of the bell curve and the ranges of human capabilities are beyond just the demographic ones. They exist in terms of fundamental capabilities and how interesting will we be as a society if everyone has the full access to those facilities. That is the future I see. And I think that's a pretty compelling pitch to people to get them to try a new thing. But it'll still has to do Instagram.

没错。我要借此机会问你一个我讨厌被问到的问题，所以能成为提问者还挺有趣的，那就是，如果你预测未来，比如 3 年，你知道，随着可穿戴设备和 AI 以及其他一切，你认为未来会出现什么样的事情，是人们会做的？你知道，是那种，它给我一个持续的扫描我的生活？它在做主动搜索，或者哦，你在看像胶片冲洗或你在看这个东西。哦，让我告诉你一些关于这个的事情。你对未来的预测是什么，可能会让半技术或非技术人员说，“哦我的天，那会在 3 年内到来？”

Exactly. And I'm going to take this moment to ask you a question that I hate getting, so it's kind of entertaining to be able to be the asker of it, which is you know, if you're predicting out call it 3 years, you know, and with kind of wearables and AI and everything else, what kinds of things do you see in the future that will be the kind of thing that people will be doing? You know, is it kind of like a you know, it's giving me a constant scan on my life? It's doing the proactive search or oh, you're looking at like film developing or you're looking at this thing. Oh, let me tell you some stuff about this. What is your future prediction that might kind of you know, cause call it semi- or non-technologists to go, "Oh my god, that's coming in 3 years?"

是的。是的，我认为有趣的是这个问题的难点在于时间线。我认为你和我可能在 1 年和 10 年上会做得更好。3 年是最尴尬的时间，因为，你知道，你可以有点——我知道 1 年，我对 10 年有很好的感觉。3 年很难。我认为我不认为我们会在 3 年内达到主动的地方。我认为早期采用者可能会有始终开启的系统能够做到这一点。我认为它对于普通消费者来说足够可靠的程度，我认为我们从世界建模和认知的角度可能还有点远。但对于相当一部分人，你知道，数千万，我们将会有人们定期与 AI 进行随意、舒适的对话。从我一直做饭，我是家里的厨师。嘿，就像你知道，一加仑有多少夸脱？我不知道——你知道，汤匙到盎司水的转换是多少，你知道，什么东西。到嘿，我妻子说我需要在杂货店买什么？提醒我，她昨天告诉我了。

Yeah. Yeah, I think what's interesting the hard part of this question is the timeline of it. I think you and I probably would do better in 1 year and 10 years. 3 years is the most awkward time because, you know, you can kind of I know 1 year, I have a good sense of 10 years. 3 years is tough. I think I don't think we'll be at the proactive place in 3 years quite yet. I think I think we'll have early adopters will probably have always-on systems that are capable of it. The degree to which I think it's going to be reliable enough for the average consumer, I think we're probably a little further out from that from a world modeling and cognition standpoint. But for a decent portion of people, you know, tens of millions, we're going to have people who are in regular conversation with AIs on in casual, comfortable conversation. Everything from I cook all the time, I'm the cook in my family. Hey, like you know, how many quarts is it in a gallon? I don't How many you know, what's the what's the conversion tablespoons to ounces of water, you know, what's the thing. To hey, what did my wife say I needed to get at the grocery store? Remind me, she told me yesterday.

嘿，我打开冰箱的时候，我们有我喜欢的那种奶酪吗？那是最好的。

Hey, when I open the refrigerator, do we have that cheese that I like? That's the best one.

听着，有些东西我真的很想推动，但我们必须在监管上倡导。我的意思是，我认为一个经典的例子就是鸡尾酒会问题。你一定经常遇到这种情况。你看到有人朝你走来，你认出了他们，你知道你认识他们。但你不记得为什么认识他们，或者怎么认识的。而现在我们只能这样：“嘿，哥们，很高兴见到你，朋友。”然后你就在寻找线索试图回忆。我希望能让你的 AI 在你耳边低语：“嘿，你知道，这是谁，这个人。上次你见到他们是在这里。他们，你知道，”然后你说：“哦，对，对，对，对。”诸如此类。但那个我们需要帮助，对吧？现在伊利诺伊州和德克萨斯州有法规，BIPA 和 CUBI，使得这种事情充其量是脆弱的，如果可行的话。所以，有很多非常人性化的问题，我们可以更主动一些，而且我认为在完全隐私的方式下安全地做到这一点是令人安心的。我们得做工作。所以，3 年后，我确实认为你会在这种处于前沿的人群中，但不是最早的采用者，不是最前沿，只是前沿，只是早期采用者会从他们的助手那里获得巨大的实用性，比如认知和记忆方面的帮助。

And listen, there's stuff that I really want to push for, but we have to advocate for regulatorily. I mean, I think one of the classic ones is the cocktail party problem. And you must run into this all the time. You see somebody coming up to you, you recognize them, you know you know them. You don't remember why you know them or how you know them. And right now we're just like, "Hey, guy, good to see you, you know, friend." And you're just looking for clues to try to remember. I'd love to be able to have your AI whisper to you like, "Hey, you know, this is who this is, this person. Last time you saw them was here. They, you know," and you're like, "Oh, right, right, right, right." And that kind of thing. But that one we need help with, right? Right now there's regulations in Illinois and Texas, BIPA and CUBI, that make that kind of thing tenuous at best if it's doable. So, there's a bunch of very human problems where we could be a little bit more proactive, and I think comfortably so in a totally privacy way of securing it. We got to do work. So, 3 years from now, I do think you'll be in this kind of people at the leading edge, not but not the earliest adopters, not the bleeding edge, just the leading edge, just the early adopters will be having tremendous usefulness kind of cognition and memory help from their assistants.

那么，你如何看待各种可穿戴设备的全景？比如，眼镜会与手表、手机协同工作吗？还是会出现这样的情况：“一旦你有了眼镜，就不那么需要手表了。”

And where do you see kind of like the panoply of wearables? Like is the glasses going to work with a watch, work with a phone, work or will there be some like, "Well, once you have the glasses, you don't need the watch as much anymore."

不，我认为它们希望协同工作。我的意思是，从热管理的角度来看，我们可能还需要超过 10 年的时间才能让眼镜独立运行。如果它们要带另一个东西，那么今天 Orion 就带了一个我们称之为 stage 或 puck 的东西。它是一个小电池，随眼镜一起。你把它扔进包里，扔进口袋里，不用拿出来。这完全没问题。实际上，它在这方面做得相当好。而且它足够好，可以拿出来。顺便说一句，这是其中一个问题，一个普遍存在的问题，听起来不像问题，但实际上我认为奇怪的是，从口袋里拿出手机。看起来这就像，它真的毫不费力。但当我戴着 Orion 时，我发现我会做一些我本来不会从口袋里拿出手机去做的事情，但因为它就在那里，我就能做。这很有趣。一旦你深入其中，你会惊讶于这种摩擦实际上创造了多少。听起来完全疯狂。我们是最特权的物种。我接受这一点，但这是现实。有没有你会使用的东西。所以，至少对我来说，我认为在可预见的未来，你确实希望这是一个设备星座。而且戴在手腕上有很多价值，你知道，我们在这里开发的神经接口表明，我们可以给消费者提供大量的增量信号和控制，而不必劫持他们的眼睛或让他们伸手去敲击太阳穴臂上的东西。我认为眼镜系列也有巨大的机会。所以，如果你的眼镜没有显示屏，你想把它们与有显示屏的手表或手机配对。如果你的眼镜有显示屏，那么也许你只需要在手腕上戴一个更简单的表带，另一只手腕上戴一个传统手表。我们处于可穿戴设备领域。所以，我们实际上有一段时间没有在行业中处理过的事情是奢侈品呈现，人们想要向世界展示的自我身份。他们想看起来某种方式。所以，如果我是想要看起来某种方式的人，我必须要有选择来维持对自己的那种愿景，同时也要成为现代的一部分。

No, I think they want to work together. I mean, we probably are more than 10 years away from having the efficiency of compute from a thermal perspective to have the glasses stand alone. And if they're going to have another thing with them, so today Orion comes with this kind of we call it a stage or a puck. It's a little battery that comes with it. You toss in your bag, you toss in your pocket, you don't take it out. And that's totally fine. Actually, it works pretty well for that. And it's nice enough to take it out. It's one of the things, by the way, one of the problems that is this universal problem that doesn't sound like a problem, but actually I think weirdly is taking your phone out of your pocket. Seems like it's like it's such a having it's literally no effort at all. And yet what I found when I'm wearing the Orions is I will do things that I would not have taken my phone out of my pocket to do, but because it's right there, I can do it. It's interesting. It's surprising how much that friction that actually creates once you get into it. It sounds totally insane. We are the most entitled species. I accept it, but it is the reality. Is there are things that you'd use. So, I think for me at least, I think for the foreseeable future you do want this to be a constellation of devices. And there's a lot of value in being on the wrist, you know, the neural interfaces that we've developed here show that there's a tremendous amount of incremental signal and control we can give consumers without having to hijack their eyes or make them reach up and tap something on their temple arm. I think there's also a huge opportunity for the range of glasses. So, if you have glasses that have no display, you want to pair them with a watch that has a display or with a phone that has a display. If your glasses do have a display, okay, now maybe you just have a simpler band on your wrist and you're wearing a maybe a conventional wristwatch on your other wrist. We're in this space of wearables. So, a thing that we actually haven't had to grapple with in the industry for a while is the luxury presentation, the identity that people want to bring to the world about themselves. They want to look a certain way. And so, if I'm somebody who wants to look a certain way, I have to have the options to maintain that vision of myself while also being a part of the modern era.

有趣的是，关于 Google 和 Apple 设备上的应用商店有很多讨论。我真的不觉得那些有问题。那不是我看到的问题。更让我担心的是，这些设备，作为可穿戴设备星座的自然算力中心，在多大程度上锁定了对第三方可穿戴设备制造商的访问，比如关键蓝牙通道，你知道，我用的著名例子是 AirPods。听着，你可以从我的耳机看出，我是个音频爱好者。我喜欢，你知道，我关心音频质量。有些耳机比 Apple 制造的许多耳机更好。AirPods 很棒，是个好设备。它们有 70% 的市场份额，它们拥有这个份额不是因为它们有最好的产品。当然不是最有价值的产品。它们拥有它是因为它们有一个专有蓝牙通道，使配对超级容易。而且因为我认为它们容易丢失，你得买更多。所以，这真的让我烦恼。不应该这样。听着，我是 90 年代的人，所以我是老派的，你知道，我是老派计算机结构的人。我可能不完全像 Tim Sweeney 和 Epic 那样，但我介于两者之间。所以，这就是让我担心的事情，你的手机应该比现在更有用，他们没有理由做不到，除非他们想把你留在生态系统中。

It's funny, there's been a lot of discussion about the app stores on Google and Apple devices. I really don't have a problem with those. That's not the issue I see. What worries me more is the degree to which these devices, which are the natural center of compute for a constellation of wearables, are locking down the access to third-party wearables manufacturers, to things like critical Bluetooth channels, you know, the famous example that I'll use is the AirPods. Which, listen, you can tell by my headphones, I'm a bit of an audio guy. I like, you know, I care about the quality of my audio. There are better headphones that you can get than the ones many have to manufacture by Apple. AirPods are great it's a great device. They have 70% of the market share, and they don't have that because they have the best product. Certainly not the best value product. They have it because they have a proprietary Bluetooth channel that makes it super easy to pair. And also cuz I think they're easy to lose you have to buy a lot more. So like the but it really bothers me. Like that shouldn't be that way. And I'm listen, I'm a '90s guy, so I'm an old school You know, like I'm an old school computer construct guy. I'm not maybe full all the way to Tim Sweeney and Epic, but I'm somewhere in between the two. And so that's the stuff that worries me is that your phone should be more useful to you than it is, and there's no reason they couldn't do that except that they want to keep you in the ecosystem.

对，绝对。我的意思是，我一直在想这个。我买了非 AirPods，但它们太难用了。所以我想，算了。这对我来说不行。

Right, absolutely. I mean, I think about that all the time. I bought non-AirPods, but it just they were too hard to use. And so I was like, forget it. It's like this is not working for me.

完全同意。你说到手机也很有趣。我丈夫终于给我买了一个智能手表，因为他不想让我掏出手机，但我的肌肉记忆，如果我想看时间，我 literally 从后口袋里掏出手机。我 literally 需要教自己看手腕，所以有道理。

Totally. And it's also funny what you say about your phone. My husband just got me a finally a smart watch because he doesn't want me pulling out my phone, but my muscle memory, if I want to check the time, I literally pull my phone out of my back pocket. I literally need to teach myself to look at my wrist, so fair enough.

是的，有时我发现自己戴着 Orion 时还在看手机，而 literally 有一个永久的时钟在我脸前。我还是会这样。所以，不只是你。

Yeah, every now and then I've caught myself looking at my phone when I have the Orions on, and there's literally a permanent clock in view of my face. And I'll still So like it's not just the it's not just you.

好的，很好。谢谢。我很感激。你之前说过一件事，你一遍又一遍地告诉你的团队，什么样的人类生活会更好。地球上什么样的人会使用这个？所以你能谈谈有一次，可能是 Orion 或不同的产品，你真的看到用户测试让你啊，这就是顿悟。你看到了一些有趣的东西，它让你有了新的飞跃。

Okay, good. Thank you. I appreciate that. One thing you said earlier is that you tell your team over and over again like what human life will be better. Like what human on this earth will be using this? So can you talk about a time, it could be for the Orions or a different product, that you really saw like user testing got you ah, this is the aha. You saw something interesting, it made you have a new leap.

有很多，但我要告诉你一个最近让我惊讶的例子，事后看来完全合理。购买我们 Ray-Ban Meta 眼镜的最热门人群之一是盲人。盲人买眼镜，这很意外。但如果你在用户研究中观察他们，就完全说得通了。实际上，我们团队里就有盲人成员。我们与 Be My Eyes 合作，那是一项很棒的服务。但在我这个能看见的人看来，我会想，他们去餐厅或过马路会不会有困难？他们没有。他们有解决方案。他们有 Google Maps 把方向读进耳朵里，他们有手杖或导盲犬，或者一堆系统。他们能到餐厅。你知道他们不能做什么吗？他们找不到门。就像，门在哪儿？所以他们做的是问眼镜：“嘿 Meta，看看并告诉我门在哪里。”然后它说，哦好的，在你左边。如果他们做不到这种事，他们可以呼叫 Be My Eyes，现在他们有一个实时视频流，先连接到 AI 智能体，如果失败就转接给人工帮助。所以有些时刻你知道会发生，因为你自己是人，你会想，是的，我作为人和其他人一样，我会想要这个，所以其他人也会想要。但也有一些非常迷人的时刻，你构建了能力，然后这些东西从木工里冒出来，你从未预料到。

There are quite a few, but I'll tell you one that surprised me recently, which makes total sense in retrospect. One of the most popular demographics purchasing our Ray-Ban Meta glasses are blind people. Now, blind people buying glasses is a surprise. But if you watch them in the user research sessions, it makes total sense. Actually, we have members of the team who are blind. We have a partnership with Be My Eyes, which is a great service. But in my head, as a seeing person, I have no problem—I'm like, oh, I wonder if they have problems navigating to the restaurant or getting across the street. They don't. They have solutions for that. They've got Google Maps reading directions into their ears, they've got a stick or a dog, or there's a bunch of systems they have. They can get to the restaurant. You know what they can't do? They can't find the door. It's like, where's the door? And so what they do is they ask the glasses, "Hey Meta, look and tell me where the door is." And it's like, oh okay, it's to your left. And if they can't get that kind of thing done, they can call into Be My Eyes and now they've got a live video stream going to an AI agent at first, and if it fails over to a human who helps them out. And so there are these moments where you know it's going to happen because you yourself are a human and you're like, yes, I as a human am like other humans and I would want this, so other humans will want this. But there are also these really fascinating times where you build the capability and this stuff comes out of the woodwork that you never saw coming.

显然，Meta、Jan LeCun，你们是 AI 领域的领导者。那么，你认为还有什么额外的事情让你们与众不同？未来几年你们将专注于什么？

And so obviously Meta, Jan LeCun, you guys are leaders in AI space. And so what do you think are the things additionally that set you all apart? And what are you going to be focusing on for the years to come?

嗯，我们非常自豪于我们在 Llama 上的开源立场，我认为这真的很——我认为我们是最早的之一。在我看来，有太多战略利益。就像如果有人构建了伟大的 AI，我们的产品会变得更好。但人们构建伟大的 AI 并不能让他们复制我们的产品。所以我们从 AI 中获得了这种不对称的利益。所以这里有一个战略性地商品化你的互补品的结构。但对我们来说，这真的更深层。如果你听过 Jan 的任何讲话，你就听过这一点。我们真的认为这是加速进步的最佳方式。你开源这些东西，人们从中学习。然后你得到十倍的回报。当我们推出 Llama 1 并开源它时，我在内部是大力倡导者，我想没几天就有人让一个版本在笔记本电脑上运行。我想几周内就有人让一个版本在手机上运行。太神奇了。从资源角度来看，这本来会花费我们——我相信我们有这个才能。但我们不会去做。我们本来要花几年才能做到，因为我们只是有其他事情要做。所以哇，几乎是不经意间 stumbled into 这样一个壮观的闭环，凭借这个强大的政策。

Well, we're pretty proud of our open source stance with Llama, which I think is really—I think we were one of the earlier ones. In my opinion, there's so much strategic benefit. Like if anyone builds great AI, our products get better. But people building great AI doesn't let them replicate our products. So we have this asymmetric benefit from AI. So there's a strategically commoditize your complements construct here. But it really is deeper than that for us. And if you've spent any time listening to Jan, you've heard that. We really think that this is the best way to accelerate progress. You open source these things and people learn from them. And you get back tenfold. When we launched Llama 1 and open sourced it, which I was a huge advocate for internally, I don't think it was a matter of days before somebody had a version of it running on a laptop. And I think it was a matter of weeks before someone had a version of it running on a phone. It's amazing. Which would have taken us—just from a resource standpoint, I'm sure we had the talent to do it. We weren't going to. It would have taken us years to get to go do that because we would just have other things that we were doing. And so wow, what a spectacular closed loop to have stumbled into almost with this powerful policy.

嗯，我的意思是，对上一个问题的一个明显挑战是，你知道，当开源的东西流向初创公司、企业家、学者和构建东西的人时，这很棒。但如果它流向流氓国家、恐怖分子、罪犯，就更具挑战性。你知道，如何 navigate 这一点，确保更多那种创新利益以那种方式在循环中发生，而更少地出现像朝鲜黑客勒索医院或类似的事情。

Well, I mean this is an obvious kind of challenge question to that last one is that you know, look it's great when the open source stuff goes to startups and goes to you know, kind of entrepreneurs and academics and you know, people building stuff. More challenging if it goes to rogue nations, terrorists, criminals. You know, what's what's the way to kind of navigate that making sure more of that kind of innovation benefit in the loop happens that way and less of it with, you know, like North Korean hackers holding hospitals ransom or, you know, other kinds of things.

是的，嗯，这有两个部分。我的意思是，再次回到 AI 的这种不对称结构。我想有两个部分。第一部分是我确实再次非常认同我 90 年代的传统。信息想要自由，我绝对不相信最闭源的东西实际上不会在那些倾向于攻击它们的民族国家中广泛可用。无论是通过间谍活动，直接或间接，我相信这很可能是事实。但即使抛开这一点，我仍然坚持这一点，因为我认为你试图 handicap 自己以减缓敌人进步的机会，更大的风险是你被敌人超越。我们讨论的人，尤其是中国，非常有能力。他们有一支极其有才华的工程师队伍。他们正在看和我们一样的东西。你很可能最终相对于他们 handicap 了自己，而不是相反，就像，是的，你必须参与这场竞赛。这就是竞赛。这是我们的太空竞赛。这就是我们时代的样子。而且很少有秘密，只有进步。你要确保你永远不会落后。所以我会尽可能多地浇油。Deep Seek 开源他们的技术对整个行业来说是一件极其积极的事情。从他们在内存方面做出的伟大创新中学习，他们在管理方面做出的伟大创新，你知道，在缩小这些模型方面，在哪里和何时截断。就像他们做出的非常非常聪明的决定。他们做出这些决定是因为我们给他们施加了压力。他们可以说击败了一堆资金更充裕的美国公司，因为他们资金不那么充裕。因为他们必须在芯片紧张的环境中创新。所以我们为他们创造了超越我们的条件。我们很幸运他们开源了。我认为你必须把这件事竞赛到顶端。我认为这是我们作为一个国家必须认真参与的事情。我认为《芯片法案》是一个好的开始，但我们需要更多这样的东西来将国内制造业带到前沿，以减少我们在南中国海都感受到的紧张。所以至少对我来说，我认为除了通过，别无出路。

Yeah, well, there's two parts to this. I mean, I think again, getting back to this kind of asymmetric construct of AI. I guess there's two parts. First one is I do kind of again, very much do get my 90s heritage here. Information wants to be free and I have absolutely zero faith that the most closed source thing that exists isn't actually widely available in the nation states that are inclined to attack those. Whether it be through espionage, directly or indirectly, like I believe that's likely the case. But setting that aside even, I still stand by this because I think the opportunity that you have to try to handicap yourself to slow the progress of your enemies, the far bigger risk is that you just get lapped by your enemies. The people we're discussing, especially China, are highly capable. They have a tremendously talented pool of engineers. They're looking at the same thing we're looking at. And you're very likely to end up having handicapped yourself relative to them rather than the opposite, which is like, yep, you got to engage in the race. It is the race. It is, you know, this is our space race. This is what it looks like in our era. And there's very few secrets and there's just progress. And you want to make sure that you're never behind. And so I would pour as much fuel on it as we can. And the fact that Deep Seek is open sourcing their technology is a tremendously positive thing for the entire industry. Learning from the great innovations they made in memory, the great innovations they made in how they manage, you know, to shrink these models in terms of where and when they truncate. Like really, really smart decisions that they made. They made those decisions because we put pressure on them. They arguably beat a bunch of American companies who were more lavishly funded because they were less lavishly funded. Because they had to innovate inside of a tight circumstance around chips. And so we created the conditions for them to outpace us. And we're lucky they open sourced. I think you got to race this one to the top. I think it's a thing that we've got to engage in seriously as a nation. I think the Chips Act is a good start, but we need a lot more of that to bring domestic manufacturing to the forefront to reduce the tension that we're all feeling in the South China Sea. So for me at least, I think there's no way out but through.

嗯，我只是——我只是因为我想在这件事上多停留一会儿，因为顺便说一句，我同意你的两个反驳点是重要的点，Boz。但我确实继续停留在——你知道，我明白，我们只是加速，我们试图加速以超越坏行为者可能在做的事情。

Well, I would just—I just because the one thing I want to linger on a little bit in this, because I agree with, by the way, your two counterpoints as important points, Boz. But I do continue to linger on the—you know, and I get the look, we just accelerate and we try to accelerate to get past what might the bad actors might be doing.

但同样重要的是，在某种程度上要放慢、遏制、限制坏人。我倾向于认为，我们一定还是能做一些事情的。我在 Mozilla 董事会待了 11 年，是开源的大力支持者。在 LinkedIn，我们实际上有一些开源项目后来变成了独立的数十亿美元公司。所以它在很多方面都非常积极。但也要看到，好吧，如果这东西要搞网络攻击、钓鱼，如果它要搞生物恐怖主义，那我们就需要做一些事情。我知道你们很聪明，也很在意这些。所以我很好奇，你们对这个问题是怎么想的？

But it is important to also, to some degree, slow, contain, limit bad actors. And I tend to think that there must still be some things we can do. I was on the board of Mozilla for 11 years, a huge proponent of open source. At LinkedIn, we've actually had open source projects that have gone and become their own multi-billion dollar companies. So it's very positive in a lot of different ways. But it's also like, okay, if this is going to do cyberattacks, phishing, if it's going to be doing bioterrorism, there's some stuff that we need to do. And I know you guys are smart and care about this stuff. So I'm kind of curious about what's the thought about the navigation?

当然。有意思的是，我们这个领域很多人都在讨论安全这个话题，我认为这很重要。最常被提到的就是生物、网络和核武器。核武器有意思的地方在于，核武器最难的部分是钚。钚和铀，这些才是难点，而我们作为一个社会把这些锁得很死。我觉得生物领域也有一个类似的模式。我认为知识是存在的，而且我听到的很多被归咎于 AI 的威胁，其实都过不了谷歌测试。我能用谷歌搜到这个东西吗？很多时候我能搜到，那这种情况下 AI 就不是真正的威胁。信息不是威胁。真正的问题是，你可以邮购炭疽。我不确定这是不是真的——别替我背书。别去核实我这句话。但没错，你可以邮购这些东西。我有一些在生物领域工作的亲友，他们对自己能在实验室里获取到的东西——完全没有任何管控——一直感到相当惊恐。所以我认为生物领域有监管解决方案。在网络领域，我其实对 AI 检测网络攻击的能力比生成攻击的能力要乐观得多。当然，它会生成更多攻击，但我确实认为我们在检测方面一直很吃力，有很多证据表明这一点。如果你看看民族国家行为体对美国做过什么——OPM 黑客事件、对某些加密货币交易所的黑客攻击、史上最大的盗窃案，朝鲜。我认为 AI 在防御方面的价值远远不对称地高于攻击。我觉得我们在这件事上已经站在错误的一边有一阵子了。所以那个我比较看好。如果要我说我真正最担心的是什么，不是上面任何一个，而是欺诈。就是老式的欺诈。我已经跟我父母谈过了。嘿，如果有人打电话给你，看起来像我，听起来像我，但他们在问你要钱，你就问一个只有我知道的事实。对吧？这需要真正的教育。我经常跟人聊这个话题，这很难让人理解，但我必须提醒人们，我们成长的那个时期在历史上其实非常不寻常。在照片和视频出现之前，所有媒体都被认为可能是假的。信件、报纸都被认为——你不知道它的真实性。曾经有一个非常独特的时期，可能再也不会出现了，那就是你可以制作一份媒体——一张照片或一段视频——人们根本无法想象它是假的。造假的成本比真实存在的成本高出好几个数量级。所以这些东西被推定是真的。这种情况不会再有了。所以我们要回到 1900 年代以前那种人与媒体的关系。顺便说一句，孩子们已经到那儿了。孩子们已经到那儿了。他们已经知道了。我们有一代人需要照顾。是我们。是我们这一代人。我们有一代人需要照顾，他们没有这方面的抗体。他们没有这方面的抗体。所以至少对我来说，我认为这是一项教育，我会投入大量精力把它作为国家政策来做。要理解，嘿，我们得重新教育人们，你看到的所有媒体，不管它看起来、听起来多么逼真，都可能是假的。

Yeah, for sure. Yeah, it's funny. I think a lot of people in our space have had this conversation around safety. And I think it's an important one. The ones that come up most often are bio, cyber, and nuclear. Now, what's funny about nuclear is the main thing that's hard about nuclear is plutonium. Plutonium and uranium, those are the hard pieces, and we lock those down as a society. I think there is a model there with bio. I think the knowledge exists, and a lot of times the threats that I hear ascribed to AI fail the Google test. Can I Google for this thing? And very often I can, in which case the AI isn't really the threat. The information isn't the threat. It's the fact that you can mail order anthrax. I don't know if that's something — don't back me up on that. Don't fact check me on that. But yeah, you can mail order these things. And I have some friends and family who work in bio who are pretty consistently alarmed at what they're able to acquire for their labs without any kind of control. So I think there are regulatory solutions in the bio space. On the cyber space, I actually feel way more optimistic about AI's ability to detect cyberattacks than to generate them. Of course, it will generate more, but I actually think we have been struggling on the detection side, and there's a lot of evidence of that. If you look at what nation-state actors have been able to do with the US — the OPM hack, the hack on some of the crypto exchanges, the biggest heist of all time, North Korea. I think AI is a much more asymmetrically valuable tool in defense than it is in attack. I think we have been on the wrong side of that for a little while. So that one I'm more bullish on. If I were to tell you the one that I'm actually the most worried about, it's not any of those, it's fraud. It's good old-fashioned fraud. I've already had the conversation with my parents. Hey, if somebody calls you and it looks like me and it sounds like me, but they're asking you for money, ask about a fact that only I would know. Right? There's a real education that has to happen. I talk often with people about this one, and it's a hard one to wrap their heads around, but I have to remind people that actually the period that we grew up in was very unusual historically. Before the photograph and before video, all media was presumed to be possibly fake. Letters, newspapers were presumed — you didn't know the veracity of it. There was a very unique period, never probably happened again, where you could produce a piece of media, a photograph or a video, that it was impossible to imagine faking it. It was orders of magnitude more expensive to fake than to have it be real. And so these were presumptively true. That's not going to be the case anymore. So we're going to return to a pre-1900s media relationship that we have with media. The kids are already there, by the way. The kids are already there. They already know. We have a generation to look after. It's us. It's our generation. We have a generation to look after who didn't have the antibodies for that. They don't have the antibodies for it. So, for me at least, I think that's a piece of education that I would put a lot of energy into as a national policy endeavor. Understanding, hey, we've got to re-educate people that all media that you see, no matter how realistic it looks and sounds, may be fake.

所以我觉得这就引出了很多潜在听众关于新 AI 世界的问题，以及它和信息准确性的关系。有时候是 AI 生成的，有时候是用户生成的。你怎么看这件事？显然，很多人看到了 Meta 的新消息，以及你们对事实核查和内容审核的改动。你们这么做的正面理由是什么？你怎么看？

And so I think that begs the question that a lot of possible listeners have about the new AI world and how it relates to information accuracy. And sometimes that's AI created, sometimes that's user generated. Like, how do you think about this? Obviously, a lot of people have seen the new news out of Meta and your changes to fact-checking and content review. Like, what's the positive case for what you guys did? And how do you think about that?

嗯，我觉得社区笔记就是比事实核查更好的功能。它能在更大的规模上运作。你现在用百科全书还是用维基百科？就是这样——绝对。这不是什么难做的——社区笔记已经证明了自己是更好的功能。我们在做一个更好的功能。它会做得更好。我对此很兴奋。这甚至不是什么难做的决定。所以至少对我来说，这是我更宏观的想法：我认为作为一家美国公司，这并不令人意外。我觉得在世界其他地方，我说“是的，人们被允许说一些不真实的话、相信一些不真实的事”，可能会更令人意外。我认为我们都学到了一个艰难的教训：真相的本质、什么是真的，也没有我们希望的那样坚固。我们必须适应一个事实：我们成长在一个相对黄金的时代，顺便说一句，那时两个政党实际上几乎是同一个政党，而我们正在离开那个时期，进入一个——比我们愿意承认的更接近——正常的美国或全球民主动荡期，充满紧张和对未来走向的不同想法。而技术在其中扮演了巨大的角色。

Yeah, well, I mean, I think community notes is just a better feature than fact-checking was. It works at a larger scale. Do you use an encyclopedia these days or do you use Wikipedia? It's just — absolutely. It's not that hard of a — community notes has just proven itself to be a better feature. We're building a better feature. It's going to do a better job. I'm excited about that. That's not even a tough one. So, for me at least, this is kind of my broader thought, which is, I think it's not surprising for us as an American company. I think it may be more surprising in other parts of the world for me to say, yeah, people are allowed to say things, believe things that aren't true. And I think we've all learned a tough lesson that the nature of truth and what is true is also not as firm as we'd like it to be. We have to adjust to the fact that we grew up in this relatively golden era where, by the way, the two political parties were effectively almost the same political party, and we're exiting that period into what is probably — more than we'd like to admit — a normal period of American or global democratic upheaval, of tension and different ideas about the future of where things are going. And the technology is playing a huge role in that.

回到创业生态这个话题，我经常被问到的一个问题是：由于算力、数据规模等的重要性，AI 是不是一个只有超大规模厂商才能赢的游戏？我认为 Meta 在开源方面做的事情非常有益，对吧？这属于正面的一类，显然是让这个领域变得——让很多不同的人都能参与进来。

Going back to kind of the startup landscape, one of the questions I often get is, because of the importance of compute, size of data, etc., is AI a game that is only going to be won by the hyperscalers? And one of the things I think is very helpful about what Meta is doing with the open source stuff, right? This is in the positive category, is obviously making that a much more — a field where a lot of different people can play.

你觉得超大规模厂商，比如 Meta 和其他公司，会在哪些领域大规模部署？另外，你认为初创公司有哪些有趣的方向，以及这会如何发展？

What do you think are going to be the things that the hyperscalers, you know, Meta and the others, are going to be kind of like, this is the area where we're going to be deploying a bunch of stuff? And what are the things that you think are, you know, kind of some of the range of interesting things with startups and, you know, how will that play out?

是的，是的。我们非常兴奋 Llama 在构建初创公司生态系统方面所发挥的作用，让它们有更好的创新机会。我们看到这一点正在切实发生，因为超大规模厂商被迫采纳来自这些小初创公司的创新，反之亦然。听着，每一代人的智慧，Reed，你比任何人都更了解这一点，每一代人的智慧都是：上一代的大公司显然会赢得下一代。但这几乎从未发生。几乎从未发生。我们从来不知道原因和方式，直到它发生，然后我们才恍然大悟：哦，显然这就是原因和方式。所以我不知道原因和方式，但我怀疑真正颠覆性技术还有很大空间。顺便说一句，有趣的是，ChatGPT 就是一个颠覆者。OpenAI 在这个领域是一个颠覆者。他们不是传统意义上的超大规模厂商。现在，我要说，从结构上看，我们实际上更了解超大规模厂商面临的挑战。谷歌有商业模式挑战，对吧？他们是否愿意削弱并蚕食有史以来最成功的商业模式之一，如果不是最成功的？他们拥有技术、能力，但这种矛盾很棘手。对我们来说更容易，这一切对我们都是锦上添花。我们所有的产品都变得更好。难道不是变得更好吗？对我们来说全是好消息。我认为微软实际上处于类似的强势地位。他们的产品变得更好。使用 Office 产品的消费者变得更好。拥有所有 AI 并不能让你构建 Office，但拥有 Office 加上 AI 会更好。所以，我觉得我们和微软无论如何都会赢。你知道，对于黑子们，我很抱歉地告诉你们，我们就在那里。我认为谷歌有这种矛盾。我认为亚马逊介于两者之间，AWS 当然可以得到巨大帮助，但这是否是一场逐底竞争，他们只是增加一个增量服务？所以，也许对他们来说是无操作。他们宣布与 Anthropic 合作。他们对 Anthropic 有巨额投资。Alexa 有巨大的覆盖。他们能否通过这个新项目重振 Alexa？你知道，Panos 在那里，我认为他显然是个天才。所以我支持他们。我认为把 AI 带入家庭和更多有趣的地方会很棒。所以，讽刺的是，我们对超大规模厂商及其面临的格局有了更多的了解。初创公司完全是未知数。这就是我喜欢它们的地方。它们不知从何而来。DeepSeek 有点奇怪，它突然席卷全球。实际上，模型发布 4 周后，我们在圣诞节期间研究了 R1。我们认为它很酷很有趣，从中学到了很多。然后一个月后全世界才醒来。我不确定为什么会有这种延迟，或者是什么导致的。但是，初创公司……所以，我们一直看到这种情况。说实话，如果 Llama 没有出现，我不认为这种情况会发生。如果 Llama 没有出现，我不认为超大规模厂商的定价会下降这么多。我认为定价也促成了 AI 领域之外的许多创新，坦率地说，这没有得到足够的关注。但是，人们将这些 AI 付诸实践并解锁新用例，这很重要。我认为，当 Gemini 推出超大上下文窗口时，有很多用例以前是不可能的，现在 Gemini 通过在该领域的创新赢得了客户群。所以，我认为这很令人兴奋。是的。

Yeah. Yeah. Uh we're super thrilled about the role that Llama has played in building up the ecosystem of startups and giving them a better shot to innovate. And we're seeing that really play out materially as hyperscalers are forced to take on innovations that came out of these little startups and obviously vice versa is happening. Um listen, the wisdom of every generation, and you know this better than anyone, Reed, the wisdom of any generation is these big companies from the last generation are obviously going to win the next generation. And it almost never happens. It almost never happens. And we never know why or how until it happens and then we're like, oh, obviously that's why and how. Um and so I don't know why or how, but I suspect there is a lot of room for truly disruptive technologies. And by the way, it's funny, you know, ChatGPT is a disruptor. Like that, you know, OpenAI is a disruptor in the space. They're not a hyperscaler in the conventional sense. I think Now, I will say, structurally, we actually know a lot more about the challenges the hyperscalers face. Google has a business model challenge, right? Like, are they willing to undermine and cannibalize one of the most successful business models, if not the most successful business model of all time? Um that's a boy, they've got the technology, the capability, they've got this tension, that's tough. Easier for us, this is all gravy for us. All of our products just get better. Like, doesn't they just get better. Like, it's all good news for us. Microsoft, I think, is actually in a similar strong position. Their products get better. The consumers who use Office products get better. Having all the AI doesn't make you able to build Office, but having Office and with AI is better. So, I feel like us and Microsoft win kind of no matter what. Like, you know, with respect to the haters, I'm sorry to tell you, like, we're there. I think um Google's got the tension. I think Amazon's somewhere between AWS certainly could be helped tremendously, but is it a race to the bottom and they're just adding one more incremental service? So, maybe it's a it's a no-op for them. They're announcing their partnership with Anthropic. They have a huge investment in Anthropic. Alexa's got a huge footprint. Can they rejuvenate Alexa um with this new program? You know, Panos is there, I think he's obviously a a talent. So, I'm I'm rooting for them. I think that'd be great to have these AIs in the homes in more interesting places. Um so, ironically, what we have is a lot more visibility into the hyperscalers and the landscapes they face. The startups are total wildcard. And that's what I love about them. You know, they they come out of nowhere. Deep Seek It It is a little bit weird that Deep Seek kind of took the world by storm. Actually, 4 weeks after the model actually dropped, we were studying R1 in the you know, Christmas. Um and we under we we thought it was cool and interesting and learned a lot from it. And then kind of the whole world woke up to it a month later. I'm not sure why that delay happened or what was causing that. But, the startup So, you know, we've we're seeing that consistently. And I don't think that would be happening if Llama hadn't been out there, if I'm being totally honest with you. And I don't think the pricing would have come down as much on the hyperscalers if Llama wasn't out there. And I think that pricing is also enabling a lot of innovation above the AI space, which is frankly not getting enough attention. But, people putting these AIs into practice and use case and unlocking new use cases um has been big. I think, you know, when when Gemini went out with the really large context window. Um there was a bunch of use cases that were impossible for that that now Gemini's got a customer base that they've earned by innovating in that space. So, I think I think it's a it's exciting. Yep.

我喜欢这种现状和现实检验。这让我想起那个梗，就像：是的，我知道搜索会很大，所以我投资了雅虎。然后我知道智能手机将巨大，所以我投资了 Research in Motion。然后社交显然巨大，所以 MySpace 是我最大的投资，你知道？就像真的很难预测什么会成功。

I love that state of the state and reality check. It reminds me of that meme that's like, yeah, I knew search was going to be big, so I invested in Yahoo. And then I knew smartphones were going to be huge, so I did Research in Motion. And then social was obviously huge, so MySpace was my biggest investment, you know? It's like it's really hard to predict what's going to hit.

没错。我们低估了最后一英里。我们低估了界面设计。我们低估了用例。这就是我认为 ChatGPT 拥有巨大心智份额的原因，因为他们做了最后一英里的工作。

That's right. And we underestimate the last mile. We underestimate the interface design. We underestimate the use cases. Uh and that's where I think ChatGPT has this huge mind share cuz they did the last mile work.

在这个播客中，Reed 总是有机会问关于科幻的问题，而我总是远远落后。但我要猜一下，既然你有年幼的孩子，你肯定读过或看过《野生机器人》。当然，是的。我忍不住，你知道，字面上他们描绘了那个倒塌的农场，然后因为机器人和自主机器人，它变成了这个美妙美丽的农场。你有一个有同理心的机器人，在森林里和动物说话，就跟我谈谈《野生机器人》吧，你认为它是对 AI 和机器人技术的正面描绘还是负面警告？你怎么看？我相信你的孩子们看过，这是塑造他们对机器人看法的一件事。

So, on this podcast, Reed always gets to ask the questions about science fiction, and I am always woefully behind. But, I'm going to take a gander, since you have youngish kids that you have either read or seen Wild Robot. Of course, yeah. I cannot help, you know, literally they have the depiction of the farm that would was falling down and then because of robots and autonomous robots it you know became this wonderful and beautiful farm. You have this robot who has empathy who is you know talking to the animals in the forest like just talk to me about wild robot and do you think it's a positive depiction of AI and robotics negative warning like how do you see it? I'm sure your kids have seen it and that's one thing that's shaping their vision of robots.

是的，他们……当然，它基于一本儿童书，他们读过，然后看了电影。首先，你知道，我是个超级影迷，我看那部电影时哭了三四次。我又看了一遍，在同样的地方又哭了，这对我来说不寻常。我有点爱哭，我不觉得有什么问题，但……不，这是一部感人的电影，真的不是关于机器人的电影。真的是关于母性的电影。真的是关于为人父母。人性。所以触及了这些。如果我要批评的话，是的。我不太确定 Roz 的“嘿，让我们和睦相处，不要互相吃”如何能持续超过一季。这很公平。当食肉动物需要吃东西时，我不知道，电影中有点未解决的是食肉动物究竟如何在她试图为它们打造的勇敢新世界中生存。所以，我认为那部分的道德说教对我来说有点过于沉重，我更希望他们拥抱更多生命循环的构建，就像：是的，这是自然的方式，我作为你的母亲保护了你，但无关紧要。我不确定那对狐狸会有效。

Yeah, they they there's of course it's based on a children's book and they which they've read and then they they saw the movie. First of all you know I'm a huge film buff film film fan and I cried like three or four times I watched that movie. Um and I watched it like again and cried again at the same points which is unusual for me. I'm I'm a bit of a crier and I got no problem with that but but no it it was a so it was a touching film and it really a film not about robots. Really a film about motherhood. Really a film about Parenthood. Humanity. And and so touching on on that. Um you know if I'm being a critic here Yeah. I'm not exactly sure how Roz's hey let's all get along and not eat each other works out more than one season. That's fair. When the carnivores need to eat things like I don't know kind of unresolved in the film is how exactly the carnivores survive Roz's brave new world that she's trying to craft for them. So it's a I think the the the morality of that part is a little heavy-handed for me and I would have loved them to embrace more the circle of life construct like yep this is like this is the the natural way of things and and I protected you as your mother and but neither here nor there. I'm not sure that would have worked for the fox anyways.

就机器人学论著而言，那是一个无限的魔法机器人，拥有永久能量，手臂可以无限延伸，诸如此类。所以对我来说，我在科幻作品中思考更多的是工程导向的科幻。安迪·威尔是这方面最棒的，对吧？《挽救计划》。如果人们熟悉《火星救援》，那是一部非常有趣的作品，而且我认为书比电影好，恕我对马特·达蒙不敬。我认为《挽救计划》甚至更好。这不仅仅是因为准确性，而是因为它属于近未来科幻，所以触手可及。但它对人类通过工程摆脱严重问题的能力持乐观态度。我喜欢这些，这些是我兴奋地开始读给我孩子们听的。

As far as a treatise on robotics goes, it's an infinite magical robot that has permanent energy and arms that extend to infinity and all of that kind of things. So for me, what I've been thinking a lot more about in science fiction is engineering-oriented science fiction. Andy Weir is the best of this, right? Project Hail Mary. If people are familiar with The Martian, which is a great fun one, and I do think the book is better than the movie, with all respect to Matt Damon. I think Project Hail Mary is even better. And it's not just the accuracy, but it's near-future science fiction, so it's tangible. But it's optimistic about humanity's ability to engineer our way out of grave problems. Those I love, and those are the ones that I'm excited to start reading to my kids.

好的，快速问答。有没有一部电影，你可以随意长度回答。这只是我们问所有优秀嘉宾的同样问题。有没有一部电影、歌曲或书籍让你对未来充满乐观？

All right, rapid fire. Is there a movie, and you can answer at any length. This is just the same questions we ask all of our excellent guests. Is there a movie, song, or book that fills you with optimism for the future?

哦，太好了。是的，让我想想。有趣的是，我开始涉足电影的一个原因就是我希望有更多乐观的故事。我想我会选择《星际迷航》。顺便说一句，当前的系列非常棒，但你不必选择当前的系列。它确实是乐观科幻的灯塔，在其他相对反乌托邦的作品中，我认为那些作品写起来更容易些。所以，我必须选择《星际迷航》。

Oh, wonderful. Yeah, let me think about this. It's funny, one of the reasons I started getting involved with film a little bit is because I want more optimistic stories out there. I think I'm going to go with Star Trek. And by the way, the current series are fantastic, but you don't have to pick the current series. It really is a beacon of optimistic science fiction in a landscape of otherwise relatively dystopian works, which I think are just a little easier to write, frankly. So, I got to go with Star Trek on that.

太棒了。博兹，你希望人们更经常问你什么问题？

That's awesome. Boz, what is a question that you wish people would ask you more often?

嗯，我们今天已经涉及了一些。有趣的是，人们听说我的工作后，想了解现在正在发生什么。AI 是什么？是大事吗？是坏事吗？但他们很少深入探讨未来的积极愿景。这再次说明，我认为反乌托邦科幻可能更多，因为它的受众更大，因为我们作为一个物种，有时更容易被恐怖吸引。我们有时更多被焦虑驱动，而不是被抱负驱动。但我不常被问到的是：“给我描绘一幅美好未来的图景。”我曾在混合现实和虚拟现实领域花时间做这件事，这个想法是人们不受地理限制。我之前谈到出生彩票，你出生的地方是一个巨大因素。因为它限制了你拥有的机会。如果你没有这些限制，因为元宇宙使你能够充分发挥你的才能，那会怎样？如果人们无论出生在哪里，都能将那种才华带到前沿，人类会受益多少？然后我谈到 AR。如果每个人都拥有那种记忆、认知、听觉、视觉、能力，那会怎样？当我们都平等地拥有这些能力，并且这些能力优于生物学所能提供的，社会如何前进？

Well, we've gotten into some of them today. It's funny, people hear about my job, and they want to understand what's happening right now. What is AI? Is it a big thing? Is it a bad thing? But they so rarely get into what is the positive vision of the future. And again, it speaks to the fact that I think there's maybe more dystopian science fiction because there's a bigger audience for it, because we are as a species a little bit more drawn to the macabre. We're a little bit more motivated by our anxieties than we are by our ambitions at times. But the thing that I don't get asked often is like, "Paint me the picture of the beautiful future." And I spent time doing that in the mixed reality and virtual reality space, this idea of people unbounded by geography. I talked earlier about the birth lottery and where you're born is a huge factor in it. And because it limits what opportunities you have. What if you didn't have those limits because the metaverse enabled you to bring the full strength of your talents to bear? And how much would humanity benefit if people, no matter where they were born, were able to bring that brilliance to the forefront? And then I talk about AR. And what if everyone had that memory, the cognition, the hearing, the vision, the capabilities? How does society move forward when we all have those capabilities in equal measure and at measures that are superior to what biology can provide us?

嗯，这里还有一个积极的问题。那就是在你所在行业之外，你看到哪些进步或势头激励了你？

Well, here's another positive question. Which is where do you see progress or momentum outside of your industry that inspires you?

医学。天哪，感觉我们正在通过细胞模型取得突破，通过 AI 能够承担以前无法比拟的建模任务取得突破。我举个小例子。我们在 Fair 构建了一个定制 AI 来为我们的眼镜光学进行材料探索。解决方案空间是那种比地球上沙粒数量还大的空间，你必须查看每个分子并评估它有什么属性。构建这个极其专用的模型是短暂的工作，它把我们缩小到大约 20 个可能的解决方案，其中看起来有两个会适用于我们的目的。不可思议的结果。所以你做这个，然后你想到医学，你经常尝试做某种蛋白质折叠，你尝试做某种，你知道，向德米斯和他的工作以及那里的团队致敬。祝贺获得诺贝尔奖。我只是认为这两件事的结合将产生爆炸性的健康成果，我真的很期待。

Medicine. Man, it feels like the unlock we're getting to with cell models, the unlock that we're getting to with AI being able to take on modeling tasks that were previously incomparable. I'll take a little example. We built a custom AI at Fair to do material exploration for the optics for our glasses. And the solution space is one of those solution spaces where it's like, you know, bigger than the number of grains of sand on the earth and you have to look at every molecule and assess what properties it has. And it was short work to build this incredibly dedicated model that narrowed us down to like 20 possible solutions of which it looks like two are going to work for our purposes. Incredible result. And so you do that and you think about medicine where you're off so often trying to do a certain kind of protein folding, you're trying to do a certain kind and you know, shout out to Demis and his work and the team there. Congratulations on the Nobel Prize. I just think the combination of these two things is going to produce an explosion in great health outcomes for people and I'm really looking forward to that.

完全同意。巴兹，你完美地为我引出了我们关于可能性的最后一个问题，那就是你能给我们留下一个最终想法吗？你认为如果未来 15 年一切对人类有利，可能实现什么？我们朝那个方向的第一步是什么？

Couldn't agree more. And Buzz, you teed me up so perfectly for our final question on possible, which is can you leave us with a final thought on what you think is possible to achieve if everything breaks humanity's way in the next 15 years and what's our first step in that direction?

我认为 15 年，我们将看到数字与物理的显著共享融合。所以，人们聚在一起，穿着可穿戴设备，戴着眼镜，与在场和不在场的人进行丰富的对话。在模型上合作，并有一种真正的临场感，一种他们都在那里的真实感觉。顺便说一句，有人不在可穿戴设备上，他们在手机上，但通过柯达化身有效地投射到三维空间中。每个人都通过他们拥有的任何工具、他们可以访问的任何模态，获得他们能获得的最大感觉，以在场。并感觉在场。我们与生活背景的联系比我们以为的要紧密得多。我之前谈到意识和潜意识之间的差距，你知道，可能是小脑和杏仁核。我消费信息、交换信息的背景，面部表情和身体姿势的微妙之处是我们大脑的很大部分。你知道，颞下皮层，整个大脑区域专门用于读取面孔。它所做的就是读取面孔。你想要那个。没有人认为这是那个。对。没有人认为视频通话是那个。这使我们因地理而受限于某些事情。你只需要在那里。现在，我不认为有什么能比得上亲临现场。但我们可以在 15 年内比迄今为止更接近。

I think 15 years we're looking at significant shared blending of digital into physical. So, people coming together wearing wearables, wearing glasses and having rich conversations with people who are both present and not present. Collaborating on work on models and with a true feeling of presence, a true feeling that they're all there. And someone, by the way, they're not on a wearable, they're on a phone, but they're being projected into three-dimensional space effectively through a Kodak avatar. Everyone is getting the most sense they can through whatever tools they have, whatever modalities they have access to to be present. And feel present. We're so much more in touch with the context of our lives than we give ourselves credit for. I talked earlier about the gap between the conscious and subconscious, you know, the cerebellum and the amygdala maybe. The context in which I consume information, I exchange information, the subtleties of facial gesture and body posture are huge portions of our brain. You know, the infra-temporal cortex, just entire area of brain dedicated to reading faces. Just all it's doing is reading faces. And you want that. And no one thinks this is that. Right. No one thinks that video calling is that. Which makes us bound for certain things by geography. You just you have to be there. Now, I don't think anything will ever be as good as being there. But we can get a lot closer than we have so far in 15 years.

我喜欢。非常高兴你能来播客。非常感谢。比兹，总是期待我们的对话。

I love it. Such a pleasure to have you on the pod. Thanks so much. Biz, always look forward to our conversations.

是的，谢谢你们两位邀请我。

Yeah, thank you guys both for having me.

《可能》由 Wonder Media Network 制作。由阿拉·芬格和我，里德·霍夫曼主持。我们的节目统筹是肖恩·杨。《可能》由凯蒂·桑德斯、伊迪·阿拉德、莎拉·施莱德、凡妮莎·汉迪、阿莱娅·耶茨、帕洛玛·莫雷诺·希门尼斯和梅利亚·阿古德洛制作。珍妮·卡普兰是我们的执行制片人和编辑。特别感谢苏里亚·亚拉曼奇利、赛达·萨皮耶娃、文卡特·迪利普、伊恩·埃利斯、格雷格·比托、帕斯·帕特尔和本·雷尔斯。

Possible is produced by Wonder Media Network. It's hosted by Ara Finger and me, Reid Hoffman. Our showrunner is Sean Young. Possible is produced by Katie Sanders, Edie Allard, Sarah Schlead, Vanessa Handy, Alea Yates, Paloma Moreno Jimenez, and Melia Agudelo. Jenny Kaplan is our executive producer and editor. Special thanks to Surya Yalamanchili, Saida Sapieva, Venkat Dilip, Ian Ellis, Greg Beato, Parth Patel, and Ben Relles.

还要特别感谢 Meta、John Earla 和 Tatin Yang。

And a big thanks to Meta, John Earla, and Tatin Yang.