An Interview with Daniel Gross and Nat Friedman about the AI Product Revolution
发布时间 2023-03-30 11:34:37 来源
This day's update interview with Daniel Gross in that free minute about the AI product revolution was published on Thursday, March 30, 2023.
今天的采访更新，关于人工智能产品革命的 Daniel Gross 的那个空闲时间采访，于2023年3月30日星期四发表。
Good morning, a quick bit of housekeeping. First, I missed this tweet from Elon Musk that clarified that the 4U tab will also include accounts you follow. That was not clear to me, although I should have checked and reduces my evaluation of the approach to probably not a good idea from truly terrible idea. I'll be it for the same reasons I got out in the yesterday's update.
早上好，有几件事需要注意。首先，我错过了伊隆 · 马斯克的一条推文，他澄清了 4U 标签也将包括你关注的账户。这对我来说不太清楚，虽然我应该检查一下，但这让我把我的评估从极糟糕的想法降到了可能不是一个好主意。我将依据昨天更新的同样原因而做出这样的评估。
Second, we discussed Twitter that opened the letter about AI and tick-tock on the latest episode of Sharp Tech, which will be released later today. You can add the podcast to your podcast player using the link at the bottom of this email.
其次，我们谈论了Twitter，在最新一期 Sharp Tech 上公开了有关人工智能和TikTok的信件。该期节目将于今天稍后发布。您可以使用此电子邮件底部的链接将该播客添加到您的播客播放器中。
Third, as I know the yesterday, I will be on vacation next week. The next update will be on Monday, April 10.
我想告诉你第三件事，就是我知道昨天的时候，下周我会休假。下次更新会在4月10日星期一。
I first interviewed Daniel Gross in that free minute last October, where a major thing was the lack of AI products, despite the quick capabilities of AI models like GPT-3. We checked in again in December after a chat GPT completely changed the conversation.
我去年十月的那个空闲时间首先采访了丹尼尔·格罗斯，当时主要的问题是尽管GPT-3等AI模型有着快速的能力，但缺乏AI产品。我们在聊了一会儿GPT之后于去年十二月再次对话，完全改变了谈话的方式。
Well, it's been three months and the product explosion is well and truly here, so I wanted to chat with Gross in free minutes again to discuss exactly that. Gross founded Q, a search engine that was bought by Apple and incorporated into iOS, and led machine learning efforts at Apple from 2013 to 2017, before becoming a partner at Y-combinator and then transitioning into angel investing.
嗯，现在已经过去三个月了，产品爆炸的情况确实出现了，所以我想再次利用闲暇时间与Gross交谈，讨论这个问题。Gross创办了Q搜索引擎，被苹果收购并纳入了iOS，还领导了2013年至2017年期间苹果的机器学习工作，之后成为Y-combinator的合伙人，随后过渡到天使投资。
Freeman co-founded Xamarin, an open source cross-platform SDK, which was bought by Microsoft in 2016. Freeman led Microsoft's acquisition of GitHub in 2018, and was CEO of the developer-focused company until last year. 32 is now focused on angel investing.
弗里曼创立了Xamarin，这是一个开放源代码的跨平台SDK，它于2016年被微软收购。弗里曼在2018年带领微软收购了GitHub，并担任该公司的首席执行官直到去年。现在，他将焦点转向天使投资。
I didn't want to call it two neat projects that we didn't get to in the interview. First, Freeman set up the net.dev sandbox, which is like the opening AI sandbox, but you get access to non-open AI models as well. Second, Gross and Freeman created the Vesuvius Challenge to incentivize teams to leverage machine learning to read scrolls from ancient Rome buried under Ash from Mount Vesuvius in 79 AD. I really regret freeing to ask what the Vesuvius Challenge is. There was a lot to get to, but the website gives a great overview of the project that should be of interest to everyone.
我不想在采访中称之为两个独特的项目，我们没有时间讨论。首先，弗里曼建立了net.dev沙箱，就像开放AI沙箱一样，但你可以访问非开放式AI模型。其次，格罗斯和弗里曼创建了维苏威挑战，以激励团队利用机器学习来阅读79年维苏威火山灰下埋藏的古罗马卷轴。我真的很后悔没有问维苏威挑战是什么。要处理的内容很多，但网站提供了项目的概述，应该对每个人都很有兴趣。
On to the interview. Matt and Daniel, welcome back for what is now our quarterly AI catch up. I just have to put a disclaimer at the top of this interview, which is we are talking on Monday night and this won't publish until Thursday. I apologize in advance for the 9, 10 major announcements that were probably going to miss in these few days, so I just want to get that out of top.
接下来是访谈环节。马特和丹尼尔，欢迎回来参加我们现在的季度人工智能更新。首先我得在访谈开始前发个免责声明，我们是在周一晚上做访谈，但这个访谈直到周四才会发布。我提前道歉，因为在这几天之后可能会有九、十个重大的公告，我们可能会错过，所以我希望能提醒一下。
It's incredible. I was putting together this list and I made it through just last week and I'm like, how are we going to get through this in an hour? That's not even going back to things like Bing or Barn or whatever all this stuff is.
太不可思议了。我刚刚上个星期才整理好这个清单，我就想，我们怎么在一个小时内完成这份清单呢？甚至还有像Bing或Barn这样的事情呢。
But before we get to all the major announcements now, I wanted to go back to our podcast from six months ago because it kind of ties into this explosion, which is your whole thesis was we need to be talking about products not just papers. That was sort of a goal behind AI grant.
在我们谈论所有重大的公告之前，我想回顾我们六个月前的播客，因为它与这个爆炸有关，你的整个论点是我们需要讨论产品而不仅仅是论文。这是AI grant的一个目标。
I think we're now talking about a lot more products. Where do you think we are now? Is there still a gap? Yeah, it is amazing. I think Daniel and I were both last summer in this situation where we had spent at that point kind of years playing with these new GPT models and just being blown away by their capabilities.
我觉得我们现在在谈论更多产品。你认为我们现在在哪儿？还有差距吗？是的，真是惊人。我想，去年夏天，丹尼尔和我俩都处于这种状态，我们已经花费数年时间玩弄这些新的 GPT 模型，并被它们的能力吹走了。
I've been in this lucky position to get hub to get to put together a co-pilot and put that out. And I expected after that just a flurry of new products as other people went through that same process and discovered, oh my goodness, you know, GBD3 can do all these incredible things. We should build into this product or that product and that didn't happen.
我真的很幸运，能够得到专门的枢纽来组合一下副驾驶员，并将其推出。之后，我期望会有一连串的新产品被推出，因为其他人也会经历同样的过程，并发现，哦，天哪，GBD3可以做出这么多不可思议的事情。我们应该将其集成到这个产品或那个产品中，但事实并非如此。
And so by like last summer or early fall, we were scratching our heads saying like, where is everybody? And then in that moment, relaunched AI grant with this call to action, this create a curse saying, hey, where are all the product developers? Like, it's time to pay attention to AI.
就像去年夏天或早秋一样，我们都感到困惑，说：“所有人都去哪里了？”然后在那一刻，我们重新启动了AI拨款，发出了行动呼吁，建立了一个口号，说：“嘿，所有产品开发者都去哪里了？是时候关注人工智能了。”
Obviously, since then, a ton has changed and really it was like chat GPT in December that fired the starting gun. And so I think you could really consider us to be in like month, three or four now of the kind of AI product revolution.
显然，从那时起，发生了许多变化，其实就像聊天GPT在去年12月发射了起跑枪一样。因此，我认为现在我们可以将我们视为AI产品革命的第三或第四个月。
And it would be hard to imagine what it would look like for more people than are currently doing so to be integrating these models into products. It feels like we're at a sort of maximum overdrive. That said, I think even if the researchers stopped right here and they didn't produce any more capabilities, it would take us something like five or 10 years to digest just what GPT four can do and all the other state of the art models can do into products.
很难想象如果比现在更多的人把这些模型融入产品中会是什么样子。感觉我们已经处于一种最大过载状态。话虽如此，我认为即使研究人员就此停止并不再提供任何新功能，我们也需要大约五到十年的时间来消化GPT四和其他最先进模型能做到的所有事情，并将它们融入产品中。
There are so many variations and variants and workflows and user experiences that need to be invented and reinvented and permutations that need to be tried. And we've just started to scratch the surface.
有如此多的变数和变体，需要发明和重新发明的工作流程和用户体验，以及需要尝试的排列组合。我们只是刚刚开始探索表面。
And you know, right now we have this narrative that's out there about value capture, recruiting to incumbents. But I think part of the reason for that is that we're just doing the obvious thing. We're just sort of bolting these models into existing products. But I think operating systems will need to be rebuilt around these capabilities.
你知道，现在我们有一个有关价值捕获和为现有公司招聘的叙述。但我认为其中一部分原因是因为我们只是做明显的事情，就是将这些模型简单地添加到现有的产品中。但我认为操作系统需要围绕这些能力重新构建。
The things that we can do with voice now, like incredible voice recognition, super high performance on device, incredible language models that can do reasoning, you know, the sort of self-checking, the data lookup capabilities, the integrations, the voice synthesis, which is now hyper realistic and multiple startups have demonstrated that, you know, I think you could you could take a decade, just and rebuild the entire computing platform on this.
我们现在能够用语音做的事情真是太神奇了，比如超准确的语音识别、高性能的设备、能够进行推理的不可思议的语言模型，还有自检和数据查找能力、集成能力，以及语音合成的超级逼真，许多初创公司已经证明了这一点。我认为，仅仅用十年的时间，就可以基于语音重新构建整个计算平台。
So I would say still we're in the state where the researchers are way ahead and there's a lot of digesting to do, but it's hard to imagine how we could possibly go faster.
我想说我们现在仍处于研究人员领先、需要消化大量信息的状态，但很难想象我们如何可能变得更快。
Now, just a question on that, you know, the internet took two decades, I think in order to fully reach sort of a point of maturation and saturation, on the other hand, the rate of growth of some of the companies that find product market fit in AI is incredible now.
现在，就这个问题，你知道，互联网花了两个十年才完全达到成熟和饱和的程度，另一方面，那些在人工智能领域找到产品市场适应性的公司的增长速度现在是惊人的。
I think in part because everything is already fully networked and connected. So my sort of question to you is, do you think it would take a decade, like things that work work so quickly now? Yeah. And so maybe like it does seem like all reality that we're living is at two to 10X.
我认为部分原因是因为现在一切都已经完全联网和相互连接。所以我问你的问题是，你觉得这需要十年的时间，就像现在的工作效率那样快吗？是的，所以也许现实中的一切似乎都处于两倍到十倍的状态。
Yeah, things definitely feel like they're going fast and being able to code with GPT-4 certainly makes it faster, but yeah, I don't think the diffusion will be slow. I think the thing that still will take time is figuring out what AI native software actually looks like and, you know, not just kind of incrementally improving the existing workflows and software, but building the really AI native things.
是的，现在的事情肯定感觉像是在快速发展，能够使用GPT-4编码肯定会更快，但我不认为普及会很缓慢。我认为仍需要时间的是弄清楚什么是真正的AI原生软件，你知道的，不仅仅是逐步改进现有的工作流和软件，而是建立真正的AI原生产品。
I agree with you, Dad. I think there is what Daniel's driving at or you were driving at. What are you is driving at is there is always the V1 of any new sort of technology. And that technology basically says, oh, we can do what we did before, but we can do it in this new format. And what's fast AI is so compelling that there's going to be huge businesses that do just that.
我同意你的想法，爸爸。我认为丹尼尔或你说的话是有道理的。你想要说的是每个新技术都有一个V1版本。这个技术基本上是说，我们可以以新的方式达到以前所做的事情。由于快速的AI非常有吸引力，所以将会有巨大的公司来实现这一点。
And the most obvious one, and you know, which is exactly why I found the demo compelling is the Microsoft Office stuff where it's like, oh, your word processor cannot write by itself. Like it makes total total sense. But does that mean that's actually the optimal productivity application of AI? I think probably not, but just like, you know, Daniel, you mentioned the internet. Yeah, we had the decade to figure out that it should be a feed, for example, is the optimal way to deliver content.
最明显的一个，也是为什么我觉得演示很吸引人的原因，是微软办公软件，就像它说的一样，你的文字处理器不能自己写作。这完全合理。但这是否意味着这就是人工智能最优的生产力应用呢？我想很可能不是。但就像你所提到的互联网，我们经过了十年的时间才弄清楚以“提供订阅”为例的方式会是传达内容的最佳方式。
I don't think that was a function of people, there being an insufficient number of people using it is a function of it just takes time to reset and repair things out. And it sometimes takes a new generation that isn't coming in with the paradigms of the old one.
我不认为这是人们的功能，人数不足是因为需要时间来重置和修复事物。有时候需要一代新人，他们没有老一辈的范式。
Yeah, I think that's right. And, you know, everyone's sort of worried about job displacement. And I think that's sort of an plausibly real and interesting problem.
嗯，我认为那是对的。而且，你知道，每个人都有点担心失业问题。我认为这是一个可能真实且有趣的问题。
But to me, what's exciting is the marginal cost of building software will go to zero. And so there's all these things that are never being built just because there too much of a slap to even consider building a new software engineer in order to build it. But if sort of making software can be done at the same ease as literally sketching on a notepad, then there will be just more weird and interesting software. And that non-consumption angle, I think everyone under values, which should be really exciting.
以我看来，令人兴奋的是软件建造的边际成本将会趋向于零。因此，有许多事情从未被建造，仅因为建造一个新的软件工程师来构建它过于困难。但如果制作软件的难度与在记事本上简单地画一个草图相同，那么就会出现更多奇怪和有趣的软件。我认为人们低估了这种非消费角度，这应该是非常令人兴奋的。
Actually, I just saw a thing on Twitter today. Everyone uses the Uber example where people were running the market based on taxis. There's a better example, which I just saw today. I'll see if I find out Twitter.
其实，今天我在 Twitter 上看到了一件事情。每个人都在使用 Uber 的例子，人们通过出租车运作市场。有一个更好的例子，就在今天我看到了。我会去 Twitter 找找看。
I mean, Twitter is another conversation. It's functionality thereof. But there was, there were, talk about some analyst notes when Apple was valued at, I think it was $200 billion or whatever it was. And they were, they had a cell attached to it because they're like, look, if Apple takes 100% of phone market share. They're not going to wave up to their valuation.
我是说，Twitter是另一种交谈方式，它有其功能。但当苹果估值达到2000亿美元或者其他什么的时候，有一些分析师的笔记是这样说的。他们认为如果苹果占据100%的手机市场份额，他们不会达到估值，这就像一个绊脚石。
And it didn't appreciate that number one. They would dramatically expand the market because they were using the smartphone market share. They would basically take over all of phones number one.
它不喜欢那第一的数字。他们将大幅扩大市场，因为他们正在利用智能手机市场份额。他们基本上将接管所有的手机，成为第一。
And number two, their pricing power would be so huge, like they would go from $100 phones to people buying $1000 phones. And it's such a tangible example of the mistake analysts make about technology areas again and again and again, which is they look at what's there and then they map it to the new thing.
第二点是，它们的定价能力将非常巨大，就像他们可以从售价100美元的手机升级到售价1000美元的手机。这是分析师在科技领域一遍又一遍犯的错误的一个非常具体的例子，即他们看着现有的东西，然后将其映射到新的东西上。
And non-consumption to your point is exactly what makes billion, you know, billion, trillion dollar companies. Trailion. Yeah, and that's right. The other thing I would just say is that the capabilities are not going to stop here. They are going to keep going. And like the dramatic improvements we've seen in capabilities over the last year or two, I think are very likely to continue. And those are kind of these big step changes.
“不消费”正是造就了那些价值数十亿、甚至数万亿美元的大公司所必需的因素。数万亿。是的，没错。另外，我想说的是这种技术能力不会止步于此，而是会不断发展。就像我们在过去一两年中所见到的技术能力的巨大提升一样，我相信这种趋势在未来也会持续下去。而这些进步都是一些巨大的飞跃。
And so even if you do design for kind of 2020, March of 2023, you know, native AI capabilities, March of 2024 may present you with completely different primitives and tools and it's going to be a whole new wave of things to digest into products. Well, I thought it was interesting because I think we were once when we were on here now and you were making, I think the very credible case that things weren't actually moving that quickly.
所以，即使你设计的是供2020年使用的，到了2023年3月，你也会面临本地AI能力完全不同的原始工具，到了2024年3月，这将呈现出全新的事物浪潮，并且需要吸收到产品中。我认为这很有趣，因为我记得你曾经说过，事实上事情并没有那么快速前进。
And you know, I think we, GPT-3 was a year and a half old and you know, sort of, I think there's Andy Grove who had that metaphor, that technology sort of like a river with rapids, at different speeds. And so you have, you know, decades where nothing happens and then you're sort of going very quickly. Do you think things have accelerated against since we sort of spoke last and what do you think was the catalyst for that?
你知道的，我想我们的GPT-3已经一岁半了，你知道，索性有安迪·格罗夫提出的隐喻，科技就像一条有快速流瀑布的河流，速度不同。所以你会发现，有些十年甚至没有什么进展，有些则快速爆发。你觉得自从我们上次交谈以来，事情有没有加速，你认为是什么催化剂起到了作用呢？
Yeah, I think one of the things we talked about last year was this idea that if chat GPT was your first encounter with language models, then what came next would feel very fast because GPT-4 came out just a few months later, even though we now know OpenAI's had it for sort of seven plus months under wraps and it's been almost two and a half years since GPT-3. That said, I don't know how anybody could feel like anything's going slow right now.
嗯，我觉得去年我们提到过一件事情，就是如果你的第一个接触自然语言模型是聊天GPT，那么接下来的模型会感觉很快，因为GPT-4在几个月后就问世了，尽管我们现在知道OpenAI已经在秘密研发了七个多月，而且GPT-3已经发布了将近两年半。话虽如此，我不知道任何人现在会感觉任何事情进展缓慢。
I mean, the most common sense that I have from talking to people is vertigo. People feel there's this sort of disying pace of change and to Ben's question, you know, where do you plant your stake if the ground is sort of shifting beneath you all the time? You know, that is very common and I even, you know, Daniel, you and I have encountered some founders who are just sort of completely overwhelmed by this and don't, you know, don't even know where to start and some feel a little despondent because they think, we're sort of back in one of these waves that we've previously been in that Ben will certainly remember where there's this feeling like the leader could never be caught up with.
我说，我从与人们交谈中得出的最常见的感觉是眩晕。人们感觉到这种令人晕眩的变化步伐，而对于Ben的问题，如果地面一直在变化，你要在哪里放置你的赌注呢？这是非常普遍的，甚至有些创始人会被这种情况完全压倒，他们不知从何开始，有些人甚至感到绝望，因为他们觉得我们又回到了以前的某个阶段，Ben肯定还记得那种领导永远无法追赶的感觉。
It's just going to do every single possible business or product. And so we definitely hear, you know, we're doing some therapy with some founders, you know, who have kind of been exposed to this and they're not quite sure what to do. There's a bit where your reputation always lasts longer than the reality and you saw this with Microsoft in the 90s where everyone was rightfully terrified of Microsoft and then that terror extended much longer than it should have, like kind of by probably 98 or 99 like it was kind of a spent thing in retrospect, but people didn't stop being scared until probably the early 2000s or maybe even mid 2000s.
它会涉及到每一个可能的商业或产品。所以我们确实听到一些创始人需要接受治疗，因为他们已经接触到了这个问题，并不确定该怎么办。有一个问题就是，你的声誉总是比实际情况更长久，你可以在90年代就看到微软的情况，所有人都恐惧微软是有道理的，但这种恐惧持续的时间比它应该持续的时间要长得多，事实上，大概到98或99年，这种恐惧已经过去了，但人们一直持续恐惧，直到大概2000年初甚至是2000年中期。
And I think that runs the opposite direction where Microsoft was not anything worth worrying about or caring about for a long time after that. But you know, I still think the teams versus slack things should have been a massive wake up call to everyone because and I think what is part of it is you could chalk that up to Microsoft having better distribution and teams being free. And so there was an excuse to continue with your old viewpoint and there was an under appreciation.
我认为这与微软之前的表现完全不同，它根本不值得我们长期担心或关注。但是你知道的，我仍然认为团队与Slack的竞争应该成为每个人的巨大警钟，因为其中一部分原因是你可以把这归因于微软具有更好的分销和团队是免费的。因此，这成为了继续坚持旧观点的借口，并且存在一种被低估的情况。
I'm obviously talking, not talking my book per se of a talking by writing that there's an integration aspect here that's super meaningful. And I think the importance of that integration is really going to come to the forefront with this, the business chat sort of thing where if you, if you have meaningful data for your enterprise and you're not in the Microsoft ecosystem, you're going to get shipped out real quickly, particularly when you layer on top of the fact that Microsoft will have an alternative. That is quote unquote free.
很明显，我不是在推销我的产品，我的意思是通过写作谈论这个集成方面非常有意义。我认为这种集成的重要性将在商务聊天等方面体现出来，如果你的企业有有价值的数据，但不在微软生态系统中，那么你会很快被淘汰，特别是当微软还有一个所谓的免费替代品的时候。
And yeah, it does feel tough to be a startup founder now because if we're right that the paradigm shifting innovation will take a while to figure out, you know, that's probably, I mean, I don't know, I'm going to use your guys perspective as you're investing in these companies, you need the founders that are going to start from scratch, not try to do what's already done, but with AI. Yeah, I think that's right.
是的，现在成为初创企业创始人感觉很艰难，因为如果我们正确判断范式转移的创新需要一段时间才能找到，你知道，这可能是，我不知道，我会用你们的角度来看待，你们在投资这些公司时需要那些从零开始的创始人，而不是尝试用人工智能做已经完成的事情。是的，我认为是这样的。
I mean, I think we're excited about the founders who are doing new things that literally couldn't be done before, maybe with a completely new workflow, maybe something that seems a little too weird for the mainstream companies to, you know, are the large companies to want to approach it. And you know, it's sort of best of times, worst of times, yes, the incumbents are active and they can leverage these large user bases.
我觉得我们对那些正在做一些以前无法实现的全新流程、或者可能看起来有些怪异的东西的创始人感到兴奋，这种东西可能是一些主流公司不太愿意接触的。这是一个最好的时代和最坏的时代，现有公司很活跃，可以利用这些庞大的用户基础。
But like there are now an entire new field of companies that are possible that couldn't have been built before. And you know, I think the really excited and active founders are going to go find those. And you know, then they'll have to probably man the tiller pretty aggressively to navigate the new capabilities as they come out. But the best founders are going to do that and be excited to do it.
现在有一个全新的公司领域是以前无法建立的。我认为那些激动和积极的创始人会去找那些机会。他们可能需要积极地掌舵引领新的能力和技术。但是最好的创始人会做到这一点并感到兴奋。
So yeah, probably the lazy and obvious startups might be much harder to do than the one they normally have been. Yeah, there's a bit where just like just take an existing enterprise functionality and make it sass and boom, you have a billion dollar company. And I don't think that's going to be the case with, with the say, I stuff. No, it's true.
嗯，那些懒惰且显而易见的创业公司可能比通常的创业公司更难做。是的，有些公司只需将现有的企业功能变成SaaS，就可以成为十亿美元的公司。但我不认为这对于像 I 这样的公司适用。是的，这是真的。
And I think there's been a generation of founders that have been bred by that era, which I think also to some extent was sort of a just general sort of zero interest rate or very low interest rate market boom era. And one thing we have seen is I think it is taking longer for the innovation pipeline of Silicon Valley to produce phenotypes that are both aware enough of the technology to be interested in it, but also building deeply enough in a way that I often wondered if the AI revolution was sort of happening with your 1980s, 1990s cohort of founders.
我认为有一代创始人是在那个时代培养起来的，这个时代也在某种程度上是一个零利率或非常低利率的市场繁荣时期。我们看到的一件事是，我认为硅谷的创新流水线需要更长的时间才能产生足够了解技术并对其感兴趣的表型，同时还在深度上进行构建。我常常在想，如果AI革命是在你们80年代、90年代的创始人的时代发生的。
I actually think progress might be a little bit faster. Silicon Valley is really rich with people that are doing things on the margin, things that Microsoft is clearly going to do in our little despondent that now Microsoft is doing it. And I think that's a byproduct from a lot of these incredibly successful sass businesses actually being relatively thin layers. But that's changing. And I think this is sort of a different kind of revolution. But the market will adjust to Nance point. We're really only at day one. And we haven't seen any of this sort of native.
我实际上认为进展可能会快一点。硅谷有很多人在边缘做事情，这是微软现在明显要做的事情。我认为这是许多极为成功的SaaS企业相对较薄的涂层的副产品。但是这正在改变。我认为这是一种不同类型的革命。但市场将适应这个时间点。我们真正只处在第一天。我们还没有看到这种本性的东西。
We're still at the TV at the cameras pointed at radio shows, era of television, not at sort of your native, you know, made for TV era. And that'll happen over time. It just takes a couple of quarters for that to get generated. I think it's a really good and important observation though about just the nature of Silicon Valley.
我们现在还停留在电视机前，看着摄像机对着广播节目进行拍摄，这是电视时代，而不是像您的本土定制的电视时代。这种变化需要一段时间才能产生。我认为这是一个非常好和重要的观察，关于硅谷的本质。
And you know, one thing that's worth noting is when did we actually figure out the internet as an industry? Right. It was after the bubble, after the bubble burst, right? Like the feed search, all those sorts of things. I mean, search did start in the late 90s, but by and large it became a thing after that. The auction model I think was 2002 or so. Facebook comes along to those four to those in five. Like they're there. I don't that's probably not an accident.
你知道，有一件值得注意的事情是，我们什么时候才把互联网看做是一个产业？对吧。那是在泡沫破裂之后，就像是那些用于搜索的feed，还有其他那些东西。我是说，搜索虽然在90年代末就开始了，但是大多数人在那之后才开始关注它。拍卖模式我想是在2002年左右吧。Facebook在那之后几年就出现了。像它们那样存在，我认为那可能不是巧合。
When the focus is money and the money seems easy, you're going to take shortcuts to get there. And the most obvious shortcut right now is take a thing that people do and add AI to it. And if the if it's not so easy, then you actually have to go back to first principles. And you know, I mean, that no one's cheering for a recession or for a bubble burst or whatever it might be. But it's just striking to look back at the timing of sort of the internet. It's figuring out the internet.
当关注点是金钱，并且钱似乎很容易获得时，你会采取捷径来达到目标。现在最明显的捷径是将人们做的事情添加AI。如果不太容易，那么你实际上必须回到基本原则。你知道，我是说，没有人会为经济衰退或泡沫破裂而欢呼雀跃。但是回顾互联网的时机仍然令人惊叹，它正在解决互联网的问题。
By the way, I think it's not just a zero interest rate phenomenon thing. I mean, I think programming languages have gotten much easier over the years. And so that's changed the phenotype of person that starts a company. I mean, the sort of degree of technical excellence that was required in order to make a consumer facing product. And in 2002, 3, 4, 5, 6, that was drastically different than it is today. You know, and we did a wonderful thing where we built multiple layers on the cake that make it simpler and simpler to, you know, build a technology and if AWS and you have react.
顺便说一下，我认为这不仅仅是一个零利率现象。我的意思是，编程语言在这些年里变得更容易了。因此，公司创始人的类型也发生了变化。我是指，在制作面向消费者的产品所需的技术卓越程度上，与2002年、2003年、2004年、2005年和2006年相比，今天的要求明显不同。我们做了一件很棒的事情，就是在蛋糕上建立了多个层次，使得构建技术越来越简单，例如AWS和React。
And so you end up getting a different type of person. And I think now to really excel in AI, you have to be a little bit deep. And you know, back when we were all starting startups, it was it was a hyphenate term. It was not a proper noun. It was a start-up, it was an obscure thing to do. And you know, now it's like a very normal thing to do. And so that means that you end up, we end up seeing a lot more things that we don't do just because the selection effects that Silicon Valley had just in terms of it really attracting these, you know, technical, brilliant savants, weirdos are less strong now. So there's more people in the pool, but.
所以你最终会得到另一种类型的人。我认为，现在要在人工智能领域脱颖而出，你必须要有一点深度。你知道，当我们都在创办初创企业的时候，它是一个连字符词。它不是一个正确的名词。它是一个创业公司，它是一个模糊的事情。你知道，现在创业公司已经成为了一件非常普遍的事情。这意味着我们最终会看到更多我们不会去做的事情，仅仅因为硅谷的选择效应不再像以前那样吸引那些技术卓越，怪异的天才人物。因此，人才池中有更多的人，但是...
And so selection is a little bit harder as a result. And the fact that AI is so hot, of course, doesn't help.
因此，选择有点更加困难。当然，人工智能如此热门这一事实也没有帮助。
I think anyone. But that's a reality.
我觉得任何人都可以这样想。但这是现实。
I mean, one of the theses that I put forward before is that everyone talks about tech having a big five, but actually there's a big six. And you know, the big five are obviously Apple, Amazon, Microsoft, Facebook, and Google. But big six is basically Silicon Valley ink, which is basically the SaaS producing machine where everyone knows the playbook.
我是说，之前我提出的一个论点是所有人都谈论有五大科技巨头，但实际上有六大。你知道，五大显然是苹果、亚马逊、微软、Facebook和谷歌。但是六大基本上是硅谷之墨，那是一个生成SaaS的机器，每个人都知道这个游戏规则。
You get to call yourself a startup founder and feel great about it. But the level of risk is actually very low, the level of technical execution is very low. It's actually about building a sales team and doing sales, which to your point, this ties into the zero interest rate environment as well, where you can be encouraged to actually get super far ahead of your skis to give, you know, 15 years out, imagine what this cohort is going to be, you know, sort of producing for us.
你可以称自己为初创企业的创始人，感觉非常棒。但是风险水平实际上非常低，技术执行水平也非常低。实际上，这是关于建立销售团队和销售的问题。正如你所说，这与零利率环境有关，鼓励你真正超越自己，想象15年后，这个群体会为我们做出什么样的贡献。
And like the startup scene is was completely different. It was a corporate scene in many respects. And that's probably the first one of the big six that has taken a big hit these last couple of years.
就像创业圈一样，它完全不同了。在许多方面，它更像一个企业界。大概是这六个行业中最先受到重创的一个了，最近几年。
That's right. Speaking of though, maybe there is a sixth.
没错。说起来，也许还有第六个。
I will say I was, you know, using the various GPT's, various flavors on the yesterday. And you sort of mentioned that it was sort of an oddity that chat GPT came along when GPT 3 was basically already obsolete. And it became this huge hit and then boom, suddenly GPT 4 comes out. And I think at first sort of blush, it feels fairly similar. And then maybe you get into it all, this can do these other things like, oh, it's actually different.
我说一下，昨天我使用了各种各样的GPT，不同风格的GPT。你提到了聊天GPT出现的时间点有些奇怪，毕竟GPT 3已经过时了。但聊天GPT却成为了一大热点，然后GPT 4就突然出现了。一开始，两者感觉相似，但深入了解后，这些新技术能够做一些不同的事情，它们其实并不同。
I have to say GPT 3.5, like the default model or even the legacy model, feels really ancient. Like when you're actually using just chat GPT, just a bit. And I'm not in with the API, but I can imagine it's the similar sort of thing.
我必须说GPT 3.5感觉非常古老，就像默认模型或甚至传统模型一样。当你实际上仅使用聊天GPT时，感觉就像这样，如果用API也应该差不多吧。
It's really hard to get it to hallucinate. It's just in general much more sort of cognizant and coherent about things in general. I don't know. It seems like a pretty meaningful shift.
这真的很难让它产生幻觉。它通常更加意识到事情的连贯性和清晰度。我不知道。这似乎是一个非常有意义的转变。
What sort of your perspective from the API side or your sort of, you've been using it longer than I think any of us. So what do you think?
你对API这方面的看法是什么，或者说，你使用它的时间比我们任何人都长。那么你认为呢？
Yeah. Daniel and I had, you know, opening up, I was good enough to give us early access a few months ago. And so we've had a chance to play with it for a while. And, you know, I think the thing, I agree with you that it is just smarter. And you know, I found myself during those few months, you feel like you're talking to higher IQ person. That's what it feels like.
是啊，丹尼尔和我，你知道的，早在几个月前就有了开箱机会。所以我们有了一段时间可以使用它。你知道的，我认为它非常聪明。在那几个月的时间里，我发现自己觉得在跟一个更高智商的人交谈。那就是感觉。
Yeah. It's just smarter. And, you know, it's always been a bit slower. I think there have been sort of spurts where it got really fast. I guess they might have provisioned a lot more GPUs to it at various points for demos or something. But even when it was slow during those last few months, I found myself kind of in that position of asking, okay, am I going to run out of it? Or reach out to my pretty smart fast friend? Or am I noticeably smarter, but much slower friend? And ask them. And I found myself reaching for the smarter friend almost all the time. And I'll just tab away and I'll come back to that, you know, browser window with the answer in a minute and a half. And that's just fine.
是的，这样做更聪明。一直以来，它速度都有点慢。我想它可能会有一些很快的爆发，可能是为了演示而在不同的时间点提供了更多的GPU。但即使在那些最后几个月里速度很慢时，我发现自己处于这样的位置：我会问自己，我会用完它吗？还是去问我的聪明快速的朋友？还是我那个明显比我聪明但速度更慢的朋友？我发现自己几乎总是去找那个更聪明的朋友。我只需要离开一下，一分钟半钟后就可以回到那个浏览器窗口，并得到答案，那就很好了。
Question for you on that.
有一个问题想问你。
One of the innovations of Copilot was the trade off of a faster model that was dumber. And the idea of auto-complete, helping people be sort of the product in a pre-media. The AGI world. How does that factor in, you know, to a smarter, slower model? Like, is there a trade off point in which?
Copilot的创新之一是以更快但更笨的模型来作出折衷。另外，还有自动填充的想法，这有助于人们成为媒体前的产物。在AGI领域中，这能否发挥作用呢？你知道吗，智能更弱但速度更快的模型是否存在权衡点呢？
Yeah.
好的。
Well, I think we're sort of figuring out what the value of intelligence is a little bit here. And it's interesting what we're finding. I mean, the scaling laws that lead to GPT-4 being better than GPT-3 have a logarithm built into them. You have to put exponentially more money in to get linear returns in kind of model quality or improvements in loss. And that's just something everyone knows to be true. But I think what we're also finding out is that small improvements in kind of the models IQ probably lead to not, you know, super linear improvements in the value of the model.
好吧，我想我们正在弄清楚智能的价值是什么。我们正在发现一些有趣的事情。我的意思是，导致GPT-4比GPT-3更好的比例定律中已经内置了对数。你必须投入更多的资金，才能在模型质量或损失改进方面获得线性回报。这是每个人都知道的事情。但我认为我们也正在发现，对于模型的小改进可能不会导致价值的超线性提高。
Great point. And so you have kind of this sub-linear improvement in loss, but maybe a linear improvement in model value or even a super linear improvement, it might swamp fully that logarithm. And so that's kind of what I found in the last few months.
真是个好观点。所以你可以在损失方面看到一种次线性的改善，但在模型价值方面可能会看到线性甚至是超线性的改善，这可能会完全覆盖对数函数。这就是我在最近几个月里发现的。
The thing that I find myself using it for the most and many people have had this observation night, and it was hard to keep this quiet over the last few months is just writing code as unbelievable with GPT-4. Like my sort of friction to start a project is almost zero now.
我发现自己最常用的东西是许多人晚上都有这种观察，而在过去的几个月里难以保持安静，就是使用GPT-4编写代码，这太不可思议了。现在我开始一个项目的阻力几乎为零。
I'm fearless. I'll write whatever programming languages I've never used before. Concepts I don't fully understand. And I still have to guide it.
我无所惧。我会写以前从未使用过的编程语言。我不完全理解的概念。还要为它指导方向。
You know, everyone loves the one shot examples where you just ask it to do something and it works out the box. I find that's very rare. It does happen, but it's, you know, for really useful things, it's quite rare. It's more like 10 or 20 back and forth with the model.
你知道，每个人都喜欢单一示例，只需要求它做一些事情就可以得到结果。我发现这是非常稀有的。它确实会发生，但你知道，对于真正有用的事情来说，它是相当罕见的。更多的是要与模型进行10到20次来回。
What is Wolverine? Yeah, well, right. So that's sort of the back and forth that people talk about. And I think here we start to get into, maybe we'll touch on this later, but things that are exciting and slightly frightening.
“狼人是什么？” 嗯，嗯，是的。这是人们谈论的来来回回。我想我们现在开始涉及到一些令人兴奋和有点害怕的事情，也许稍后我们会谈及这些内容。
But, you know, one of the ways in which you use GPT-4 is you ask it to write some code. And then eventually after you begin to trust it, you just copy paste that code into your editor and run it without really fully understanding it. Then an error message pops up and because you didn't really understand the code, you don't fully understand the error. And so you could go and read it and understand it. But then you copy paste the error back into GPT-4 and say like, what's up? And then GPT-4 says, oh, excuse me, I made this error. Here's the updated code. And you copy the updated code back into your editor and that works.
你知道，使用GPT-4的其中一种方式是让它写一些代码。然后，当你开始信任它时，你只需将那些代码复制粘贴到你的编辑器中并运行它而不完全理解它。然后出现一个错误消息，因为你并没有真正理解代码，所以你并不能完全理解错误。于是你可以去阅读并理解它。但是你将错误信息复制回GPT-4中并问它怎么回事。然后GPT-4说：抱歉，我犯了这个错误。这是更新后的代码。然后你将更新后的代码复制回你的编辑器中，它就能运行了。
And so there's a moment there when you're copying and pasting between two windows, that's your role in this entire system here as the copy pasteer where you think like, hmm, shouldn't the computer be doing this mechanical moving things back and forth? And so there's a Twitter account. I think it's an anon. All the best AI accounts are anons. And I think his name is, or his handle is bio bootloader. And he or she, or they came out with a system in Python that just automates this back and forth and they call it wolverine.py.
所以，在你在两个窗口之间复制和粘贴时会有一个瞬间，这就是你在整个系统中的角色，作为复制粘贴者，你会想："嗯，难道电脑不应该自动地来回移动这些东西吗？" 因此有一个推特账户。我认为它是一个匿名账户。所有最好的人工智能账户都是匿名的。我认为他的名字是bio bootloader，他或她，或他们用Python开发出了一个系统，自动化了这种来回移动的过程，他们叫它wolverine.py。
And so you can basically run wolverine.py and then any Python script. And if it throws an exception, it will ask GPT-4 to rewrite the code to fix that and it will continuously do that, self-healing your code until it works. And this is one of those demos that's incredible. But you can already feel the degree to which we've sort of taken our hands off the wheel and let the AI drive and we don't know exactly where it's headed.
所以你基本上可以运行wolverine.py，然后运行任何Python脚本。如果出现异常，它会要求GPT-4重写代码以修复错误，并持续执行此操作，自我修复你的代码直至其成功运行。这是一个极其惊人的演示之一。但是你已经可以感受到我们已经从控制中心下车，让AI驾驶，我们并不确定它的前进方向。
Well, just to jump in on that, because I think this is that we talked previously that code is particularly well suited to this in part because it's well structured. It's sort of a fairly bounded space. And because it has to actually run, there's sort of error checking inherently sort of built in.
嗯，就这个问题，我想插入一句，因为我认为我们之前谈到过代码特别适合这个任务，部分原因是因为它很有结构。它的空间范围相对较小。而且，因为它必须真正运行，所以内在地具有错误检查系统。
And you know, I think we're definitely seeing that seeing that play out. But this bit about you just sort of give in and learn to trust it and take your hands off the wheel, I think is a super important point because that, you know, that's part of why GPT-4 I think is so compelling to me is I feel more inclined to trust it.
你知道，我觉得我们确实看到了这个事情正在发生。但是关于你只需放弃控制，学会相信它并将手离开方向盘的这一点，我认为非常重要，因为这也是为什么GPT-4对我如此有吸引力的一部分，我更倾向于信任它。
And then you add in like the plug-in stuff, like if GPT-4, if I know that invoked will from alpha, I know question at all, right? Like just sort of this willingness to, you know, people are understandably hesitant. And they're like, oh, everyone hears about the errors and they hear about the hallucination. But I, and be like, oh, that's going to hinder adoption.
然后你再加入一些插件，比如如果GPT-4说要从alpha中调用意愿，我就不会有任何问题，对吧？就像是这种愿意尝试的心态，你懂的，人们很自然地会犹豫。他们说，噢，所有人都听说过错误和幻觉。但是我觉得，这样会妨碍采用。
But people used to say people would not buy stuff online because your credit card's going to get stolen, right? And I think that tipping point comes pretty quickly. And once it comes, it's like a tidal wave going over it.
人们过去常说人们不会在网上购买东西，因为你的信用卡会被盗，对吧？我认为这个转折点来得很快。一旦到了，就像潮水般涌过去。
Yeah, I mean, a few thoughts there. One thing we learned when we were developing co-pilot, and this was nine, two years ago, back in the dark ages of Flarge Language Models, you know, is that like, as you're trying to find an AI product, the demo is always mind-blowing. And so, you know, you can easily buy cherry picking a couple of outputs, produce, you know, an unbelievably mind-blowing demo.
是的，我的意思是，我有一些想法。我们在研发 co-pilot 时学到了一件事，那是九年前，在 Flarge 语言模型的黑暗时代。就是在寻找 AI 产品时，演示总是让人刮目相看。所以，你可以轻易地挑选几个输出，制作一个令人难以置信的演示。
The proof is in the daily use and kind of how you use it daily. And I will say GPT-4 is very, very good. It's an enormous step change, but it does still hallucinate. And it does still make mistakes. It does it less, noticeably less. But it can't sort of, you know, one-shot every problem. And you do kind of have to be in the loop.
证明在于日常使用及每天如何使用。我想说GPT-4非常非常好。它是一个巨大的进步，但它仍然会出现幻觉，仍然会犯错。它犯错的频率已经降低，但它不能一步到位解决所有问题。你还需要保持联系。
And so, but the big deal, I think, and we get to the sort of, some of the recent Microsoft releases here, the big deal is, you know, having like going back and forth between this browser window that has, you know, GPT-4 in it through chat GPT or the opening AI playground. And then you're like code editor window, you're just, you're just begging for this new workflow.
所以，我认为最重要的，我们来看看最近的一些微软新发布的内容，重点是，你知道的，来回切换这个浏览器窗口，里面有通过聊天GPT或打开AI游乐场的GPT-4。然后你就像代码编辑器窗口一样，你只是渴望这种新的工作流程。
This, you know, co-pilot to me started to feel obsolete. And so now, you know, I think last week, the GitHub team released co-pilot X, which integrates chat into VS code. I don't think it's out yet, but they at least showed some videos and teased it. I guess it's coming soon. It's probably, my guess is constrained by GPU capacity for inference, which is why I think GPT-4 is, you know, you've seen open AI kind of throttle more and more access to GPT-4 as demand is dramatically exceeded their expectations, for sure.
你知道的，我的副驾驶开始感觉变得过时了。所以现在，我想上周GitHub团队推出了Co-pilot X，将聊天集成到了VS code中。我不认为它已经发布了，但他们至少展示了一些视频来吊趣它。我猜很快就会上市，但我的猜测可能受GPU推理能力的限制，这也是为什么我认为GPT-4，你知道的，你可能已经看到Open AI为了迎合需求而不断地控制GPT-4的访问权了。
Well, actually, on that point, I do want to get to co-pilot X. But I mentioned the big five tech companies. It does kind of feel like we're, and I think the plug-in announcement sort of felt like it cemented this to a certain extent, where we are well on our way to a big six. And the big surprise is where open AI, everyone, you know, we, I think our first podcast, very research oriented and like not more about sort of producing this output as opposed to products per se. Nope, turns out you're a consumer tech company.
嗯，实际上，关于这一点，我想要谈一下副驾驶X。但我提到了五大科技公司。感觉我们似乎正在朝着大六发展，插件发布的声明似乎也巩固了这一点。而开放人工智能则是一个大惊喜，每个人都知道，我们的第一个播客非常研究导向，更多地是关于产出而不是产品本身。但事实证明，你是一家消费科技公司。
And like, I mean, it just, it feels like whether they wanted to or not. It's way. Yeah. I mean, chat GPT, just the speed and intensity of the adoption basically left them no choice. It's like, nope, sorry, you're an apples league now. And then the way it's going to be. That's right. Yeah.
就像是，我觉得，不管他们愿不愿意，这就是事实。这一切太过于迅速而强烈，以至于他们别无选择。就好像说“不行，对不起，你们已经加入苹果联盟了”，并且现在就是这样了。是的。
I mean, because obviously, the, the, the, the, the, the, the, the, the, the, obviously analogy or sort of way to think about, I, you know, we've talked about industry structure is, you know, is there going to be a centralized player that's going to be an aggregator, you know, that sort of idea? And the crazy thing about the plug-in thing is not only does it just in my estimation fundamentally change, and maybe this is, I, you know, I haven't coded in 15 years. So I can only imagine the experience of, sort of, coding with this co-pilot stuff.
我的意思是，显然，对于行业结构，我们已经谈论过中心化的玩家是否会成为聚合器，你知道，这种思考方式或类比，显然提供了一种思路。关于插件的惊人之处不仅在于，它从本质上改变了我的估计，也许这是我15年来第一次编码，所以我只能想象使用这个合作伙伴编码的经验。
But I go back to the, well, from alpha, alpha thing. It changes my perception and feeling and confidence of using it in a super meaningful way. And this, you know, I sort of trying to explore this idea yesterday. I'm not sure about, well, I did it. But, you know, to the extent of large language models are so human-like, they have the same limitations as humans in that they do make stuff up. They don't know everything. And I need a computer to figure stuff out. And now chat GPT has its own computer to sort of figure stuff out, which is this architecture.
但是我回到了从一开始的那个地方。它改变了我对它的感知、感觉和使用它的信心，这是一种非常有意义的改变。昨天我试图探索这个想法，虽然我不太确定自己是否做到了。但是，您知道的，就像大型语言模型一样，它们与人类具有相同的限制，因为它们也会胡说八道，也不知道所有的事情。我需要一台电脑来解决一些问题，而现在聊天GPT拥有了它自己的电脑来解决问题，这就是这个架构。
But you can play that all the way through to a business model. Like consumers could buy plug-ins, or they could install plug-ins. Or if they don't choose a plug-in, suddenly there's like, you can bid to be the default plug-in. So if someone does a travel search, is it going to be Expedia or is it going to be kayak? They're going to have to bid for that and they're paying affiliate fee.
但你可以将它发展成商业模式。比如，消费者可以购买插件，或者安装插件。如果他们没有选择插件，那么就会出现，你可以竞标成为默认的插件。所以如果有人进行旅行搜索，是Expedia还是kayak，它们将不得不为此进行竞标，并支付联盟费用。
Like, I mean, it's just, how can this not be a huge consumer tech company at this point? Yeah. Yeah, I think the advice I would give, if being asked, you know, is that probably OpenAi's platform where they're selling these API tokens is kind of not the future of that business. You know, it's this lowest-com denominator, home depot, selling lumber, type of business, where every token has to be sold for the same price, no matter how valuable it is.
就像我是说的那样，这难道不是一个巨大的消费者技术公司吗？是啊。我想，如果有人问我的话，我的建议是，OpenAI售卖这些API代币的平台可能不是该业务的未来。这是一种最基础、最低端的商业模式，就像在家得宝卖木材一样，每个代币的价格都必须相同，不管它的价值如何。
And chat GPT clearly could be a multi-billion user product, you know, that eventually gets integrated into people's devices and used in many different ways. OpenAi could build a phone. Like that is actually a potential branch on this tree given where they're at right now in March 2023, which is an insane thing to say because any possibly anyone other than the current incumbents building a device has seen duts for years, but that speaks to I think where they're at.
“聊天GPT显然可以成为一个拥有数十亿用户的产品，最终被整合到人们的设备中，并以多种不同的方式使用。OpenAI可能会制造一款手机。这实际上是这棵树上的一个潜在分支，鉴于他们现在处于的位置——2023年3月，这话说出来很疯狂，因为任何可能建造设备的人除了现有的占领者，都在多年前就看到了失败的细节，但我认为这说明了他们的地位。”
It'll be completely unsurprising if it has a billion monthly users and maybe 300 million daily users at the end of this year. I don't know if it's true or not, they haven't told me, but I've heard that they have between one and two million subscribers for chat GPT plus. Again, don't know $20 a month. That's what I mean. Yeah, I mean, it's kind of getting between $204 or $500 million a year if so that must eclipse the API revenue dramatically.
如果这家公司在今年末有十亿每月用户和三亿每日用户，那将完全不足为奇。我不确定这是否正确，他们没有告诉我，但我听说了他们的聊天GPT Plus平台有一到两百万的订阅用户。再说一遍，我不确定，每月20美元。就是这个意思。是啊，我意思是如果真的有这样的收入，那每年可能达到2.04亿-5亿美元，这一定会极大地超过API收入。
And not to mention, it's just a more valuable ecosystem position to be able to roll out these features and use the data that comes from chat GPT. So yeah, I think maybe they should rename the company chat GPT. I saw that. I think it's a good idea.
而且更不用说，能够推出这些功能并使用来自聊天GPT的数据，这只是更有价值的生态系统定位。所以我认为，也许他们应该将公司改名为聊天GPT。我看了看，我觉得这是个好主意。
Well, also the other thing is, I mean, I thought the codex sort of cancellation and then they walked it back. But I thought even if they wanted an API business, I think that might have killed it in the womb because why would you build on chat GPT or not chat, I call it chat GPT? Why would you build on open AI when Microsoft is going to have the same API and they're not going to kill anything like that's like, like there's a weird sense where open AI is in a competition with Microsoft API space that they structurally just cannot compete with.
嗯，还有一件事，我的意思是，我觉得刚开始Codex被取消了，然后他们又把它重新拿出来了。但是，就算他们想要一个API业务，我认为那可能会在萌芽时期就杀死它，因为你为什么要在Chat GPT或者我称之为Chat GPT上进行建设呢？为什么要在Microsoft会有相同的API而Open AI不会杀死任何东西的情况下进行建设呢？就像Open AI和Microsoft的API空间在竞争中，但结构上Open AI是无法与之竞争的，这有一种奇怪的感觉。
And also it's not even a good business for them. It's sort of a distraction and the margins are not going to be anything close to what they're going to get on the consumer front. They're sitting on top of the most difficult thing to build in the world, which is a dominant consumer platform. Like that seems exactly where they should go. It's the obvious thing to do. I imagine internally there's a cultural digestion moment that's happening now. The consumers are really telling them what they need to do. At some point after enough days of chewing glass and staring into the abyss to quote Elon Musk, they'll choose success.
而且，这对他们来说甚至不是一个好的商业。这只是一种分散注意力的事情，而且回报率不会接近他们在消费者市场上的收益。他们拥有世界上最难建立的主导消费平台，这似乎是他们应该去的地方。这是显而易见的事情。我想内部现在正在发生一种文化融合的时刻。消费者正在真正告诉他们他们需要做什么。经过足够多的咀嚼玻璃并凝视深渊的日子之后，引用埃隆·马斯克的话，他们将选择成功。
Yeah. I mean, it is a very different kind of company, a consumer company versus like, we're just going to build this model and have an API like that. That's easy in a way, right? Like you don't actually have a difficulty. Yeah. Almost it feels like they need to build like a completely new organization. There should be a chat GPT app right now, right? Like the,
是的。我的意思是，它是一种非常不同的公司，一个消费者公司与我们只是建造这个模型并拥有这样的API。在某种程度上，这很容易，对吗？你实际上并不难。几乎像是他们需要建立一个全新的组织。现在应该有一个聊天GPT应用程序，对吗？就好像...
sorry, it's been four or five months. There's been five million people that have built it in a week to date. One thing though that my boss at Apple used to say, Eddie Q, was it's important to make the important things good, which was his way of saying implicitly. Not everything needs to be great. And so I think for the opening eye perspective on this would probably be, look, if the assistance is really good, people are going to use it from the browser, even if it doesn't have browser rendering, people are just going to use it in them.
对不起，已经过去了四五个月了。到目前为止已有五百万人在一周内构建了它。不过，我的老板Eddie Q在苹果公司曾经说过一句话，就是重要的事情要做好，这其实就是他间接表达的意思。并不是所有的事情都需要完美。因此，我认为对于这个问题的开放性的角度可能是，如果这个辅助功能真的很好，人们就会在浏览器中使用它，即使它没有浏览器渲染，人们也会在其中使用它。
You know, he would say this when we talk about app store performance because the end of the day and the early days of Apple, to some extent, still now, as everyone knows, the app store was terrible. Did not look fast at all. But if the phone is good, it really doesn't matter. I mean, no one likes saying this publicly, but it really won't matter. The phone just needs to be really good. And so I think in open AI's case, their organizational truth is, look, at the end of the day, the polish is around the chat GPT website and the app just don't matter. What matters is that it's the best agent with the widest plug-in ecosystem and the smartest, most accurate, fastest, advice. That's all that matters.
你知道的，当我们谈到应用商店表现时，他会这样说，因为从苹果的最初时期到现在，众所周知，应用商店一直是很糟糕的。根本不快。但是如果手机很好，这真的不重要。我的意思是，没人愿意公开这样说，但是它真的不重要。手机只需要很好。因此，我认为在开放AI的情况下，他们的组织真相是，最终的优化在于聊天GPT网站，应用程序并不重要。重要的是这是最好的代理程序，具有最广泛的插件生态系统和最聪明，最准确，最快的建议。这才是最重要的。
I think what's happening now with every single day that goes by is not really a network effect from a data standpoint, nor a network effect from a user standpoint. It is a network effect from a brand standpoint. People are walking around and they're saying chat GPT. It's Google. It's the new Google. That's right. It's a word and just like the word Google, it's a little bit weird, but it sticks in your head. And so in the second, third, fourth and fifth place come up. Unless they come up now, like this month or next month, it's just going to be too late, I think, for the consumer thing, because you're going to be an afterthought. Unless you have a particular niche or specialty or whatever, Lexus Nex is equivalent in Google, ParLons, but that's what I think is going on now. It's a fight to become a box in the customer's brain of the agents that you talk to. And every single day, they're acquiring more people that are just chat GPT. It's a verb. It's a proper noun. That's what they're winning and that's all that matters, I think.
我认为现在每一天发生的事情并不是来自数据观点的网络效应，也不是来自用户观点的网络效应，而是来自品牌观点的网络效应。人们走在街上，他们说着聊天GPT，就像谷歌一样，它是新的谷歌。没错，这是一个词，就像谷歌这个词一样有些奇怪，但它会留在你的脑海里。那么，在第二、第三、第四和第五名之后的公司就会出现。除非它们现在就开始出现，比如这个月或下个月，否则我认为对于消费者来说已经太晚了，因为你将成为一个被忽视的事物。除非你有一个特定的领域或专业性质，就像谷歌中的Lexus Nex，或者是ParLons，但我认为现在正在发生的事情是争夺成为你与对话代理人的大脑盒子，每一天，他们正在获得更多的人来使用聊天GPT，这是一个动词，也是一个专有名词，这就是他们正在赢得的，这也是我认为最重要的。
That's exactly right. I think you completely agree. I'm glad we waited for you to wait in because that was an observation that was worth it. But to that point, Microsoft said like the opposite boat. Whereas when it comes to Bing, there's a few angles on Bing. I think big picture Daniel, your observation is the most important one, which is if you asked any consumer, number one, they probably don't know about Bing chats. And number two, if they do, they know that Bing has chat GPT, which is like, sort of gets to the point. It's not chat GPT, but I know they're all chat GPT, etc.
那完全正确。我认为你完全同意。我们很高兴等你加入，因为那是一项值得的观察。但就这一点而言，微软则说了相反的话。而在必应方面，还有一些关于必应的角度。我认为大局观的丹尼尔的观察是最重要的，也就是说，如果你问任何消费者，首先，他们可能不知道必应聊天功能。而且如果他们知道，他们知道必应有聊天GPT，它有点像直接切入主题。它不是聊天GPT，但我知道它们都是聊天GPT等。
But in that, I am curious, are you surprised that Microsoft has sort of stuck with it, even though it's only been a few weeks, but obviously it was a very hairy sort of first week. I might have contributed to that a bit, but as that bit has surprised you, it has.
但是，我很好奇，你是否感到惊讶，微软坚持了下去，尽管仅仅过了几个星期，但显然这是一个非常棘手的第一周。我可能对此有所贡献，但是由于那点令你感到惊讶，所以我想知道。
It's taken me a little time to kind of try to understand what's going on over there. Because when Bing Chad had those sort of moments of amusing or even slightly frightening behavior, I thought we'd see kind of a little more caution for Microsoft afterwards or some apologetics or things like that. And we really didn't.
用我自己的话来说，我花了一点时间才尝试了解那里发生了什么。因为当Bing Chad表现得有趣或者稍微有点吓人的时候，我认为微软随后会更加谨慎，或者会做出一些道歉之类的事情。但实际上并没有。
They're just kind of at a fever pitch over there, obviously going on ho and rolling this stuff out as aggressively as they can. Frank Shaw tweeted last week that this is going to be another busy week. So probably by the time people are hearing this, there are more announcements that we don't know about yet.
他们在那边已经非常兴奋了，显然他们正在尽可能积极地推出这些东西。上周Frank Shaw在推特上发文说这将是另一个忙碌的周。所以可能到人们听到这段话的时候，已经有更多我们还不知道的公告了。
But I've tried to think about sort of psychologically what may be happening there. And obviously this is armchair remote psychologist, but you were there. Yeah, I was there. Right.
但是我试着从心理学的角度考虑可能发生了什么。虽然这只是一个做在椅子上的远程心理学家的想法，但你在那里。是的，我在那里。对的。
So the company that Satya joined decades ago now was a company that was absolutely holding all the carts. They had DOS. They had Windows. They had Excel. They were really kind of standing astride the entire industry in a very, very dominant position. And he got to enjoy that for a while. And then Microsoft spent multiple decades on defense. And it was defense against the internet and defense against the web and web apps and defense against phones and defense against cloud. And Microsoft has been a kind of no and web search, of course, as well.
所以，几十年前萨蒂亚加入的公司是一家拥有所有车的公司。他们有DOS。他们有Windows。他们有Excel。他们实际上在整个行业中处于非常支配的地位。他享受了一段时间。然后微软花了数十年时间进行防御。他们防御互联网，防御网络应用程序，防御手机和防御云。当然，微软在网络搜索方面也表现出了一定程度的无作为。
Microsoft has been a kind of number two player underdog in each of those categories. And they've settled into a recent equilibrium as not a consumer technology company, but a business to business technology company kind of playing defense on all these trends, but really helping incumbent players stay relevant in the same way Microsoft itself has managed to stay relevant.
微软在这些类别中一直是一种二号玩家的角色。他们已经进入了一个最近的平衡状态，不是一个消费者科技公司，而是一家以企业为中心的科技公司，以防御各种趋势为主，但真正帮助现有玩家保持相关性，就像微软本身一样成功地保持了相关性。
And what's happened now is that Satya, I think, finds himself at a company that's much more similar to the company he originally joined. It's got all these great things. It's got GPT-4. It's got co-pilot and this whole concept of a co-pilot. And so I think he feels like they're back and they're going to behave with the same aggression, excitement, optimism that they had 30 years ago when he joined the company. They were like a young handsome man that sort of got fat, couldn't fit the old jeans anymore, but could never bring it to sort of throw them away. Now they've gotten right and tight and sort of in shape in the jeans. There's a sliding right back on. They're ready to go.
现在发生的事情是，我认为萨蒂亚发现他现在加入的公司更像他最初加入的那个公司。这个公司拥有所有这些伟大的东西。它拥有GPT-4，它拥有联席驾驶员和整个联席驾驶员概念。所以我认为他觉得他们回来了，他们将表现出与他们30年前加入公司时一样的侵略性、兴奋和乐观。他们就像一个年轻帅气的人，变得胖了，再也不能穿旧牛仔裤了，但却永远舍不得把它们扔掉。现在他们修好了自己，穿上了合身的牛仔裤。他们准备好了。
Yeah. Well, and then the other thing, a couple other sort of psychology points here, maybe worth being aware of is that Satya, one of his first, he had many large jobs at Microsoft before becoming CEO, but one of his first big ones was running Bing.
嗯。那么，还有一些心理学上的观点，也许值得注意，萨蒂亚在成为CEO之前在微软担任了许多重要职位，但他的第一个大工作之一是负责Bing。
And he ran Bing in the era, I think of like 1% market share and grew it, but it was a tough fight against the dominant Google. And so there's a way in which I think he's back and Microsoft's got a chance to take share. And I think they're excited about that. And then I don't, it's all my speculation. I know and respect and like all these people over there. So I'm just guessing based on what I know.
他在那个时代运营 Bing，我觉得市场份额只有约1%，但他让它成长起来，可是这是与支配地位的谷歌激烈竞争的过程。所以我认为他已经回来了，微软有机会获得市场份额。我认为他们对此感到兴奋。我不确定，这只是我的猜测。我认识、尊重和喜欢他们那边的所有人，所以这只是基于我的了解的猜测。
The other thing though is that there's a degree to which he and the Microsoft leadership team are kind of playing with house money in the sense that what was the stock price when he joined 40%. $40. I know very well because I had to sell all my, when I left. I for for went my remaining stock grants and then it immediately has gotten much higher. Yeah.
但另外一件事是，他和微软领导团队在某种程度上是在用盈利去开玩笑，因为当他加入时的股票价格是多少？40%，$40。我非常清楚，因为我离开时不得不出售所有的股票。我失去了剩余的股票，然后它立刻就涨了很多。是啊。
So now it's 270 to 80. I don't know exactly what it is, but it's a lot higher. You know, he's added trillions and market cap or at least a trillion and a half or so. And so I think there's probably an element of legacy and all of that here too. And so I was surprised when they weren't a little more shrinking at some of the criticism. But they do seem going home and it's showing up everywhere. It's clearly kind of the paradigm.
现在，得分是270比80。我不确定具体是多少，但分数确实高很多了。你知道，他增加了数万亿的市值，或者至少增加了1.5万亿。所以我认为，这里可能还存在着一些历史遗留问题。我很惊讶他们没有对一些批评做出更多的回应。但他们似乎正在回到家中，并且这种情况正在到处显现，很明显这是一种范式。
Yeah. I do, it does feel like I'm not sure that Bing is going to ultimately be a thing to do. Like just to Daniel's point, chat GPT seems like the clear winner here. I think the plug-in architecture feels much more elegant than whatever it is Bing's trying to do. Bing I think is limited by trying it to have it be a part of search. Not just because Bing search is bad, which it continues to be bad. Exactly, using it much more and being reminded. But also the UI is just weird. It's just, it feels tacked on because it is. And yeah, but that doesn't mean the technology won't be meaningful.
是的，我也有这种感觉，似乎必须拿不准必应最终是否会成功。就像丹尼尔所说的，聊天GPT似乎是当之无愧的胜者。我认为插件架构比必应正在尝试的任何东西更优雅。我认为必应被限制在试图将其作为搜索的一部分的方面。不仅因为必应搜索很糟糕，它仍然很糟糕。确切地说，更多地使用它会提醒你。但是UI只是奇怪。这只是因为它是被添加上的。这并不意味着技术不具有意义。
Possible reorganized paragraphs:
有可能要重新组织的段落：
原文：The purpose of this study is to investigate the effects of exercise on the quality of life among breast cancer survivors. A total of 500 participants will be recruited from ten different hospitals across the country. Participants will be randomly assigned to either a control group or an exercise group. The exercise group will engage in a specific exercise program for 12 weeks while the control group will not engage in any specific exercise program.
重组后：这项研究的目的是研究锻炼对乳腺癌幸存者生活质量的影响。我们将从全国10家不同的医院招募500名参与者。参与者将随机分配到对照组或锻炼组。锻炼组将进行一个特定的锻炼计划，持续12周，而对照组不进行任何特定的锻炼计划。
Paragraph 1:
第一段：
很多人认为学习外语很难。但是我认为只要有足够的耐心和毅力，就可以成功地掌握一门外语。同时，学习外语也可以帮助我们更好地了解其他文化，增强我们的社交能力。我个人很喜欢学习语言，并且发现每一门学习的语言都可以带给我很多惊喜和乐趣。
It kind of feels like Apple and Google back in the day where if both of them could have just done what they were good at and they got in this unfortunate fighting where they infringe like open eye and Microsoft. They're obviously partners, they're joining it to hip regardless. But it does feel like from a product perspective, look, you guys, there's an obvious way to split this pie.
这感觉有点像当年苹果和谷歌，如果它们只做好自己的事，可惜它们卷入这种不幸的纷争中，就跟微软一样。他们显然是合作伙伴，不管怎样都在紧密合作。但从产品角度来看，你们可以看出，有一个明显的分配饼干的方法。
Paragraph 2:
我们能够保障所有的工作都由真人完成，因为机器可能会犯错误或无法处理复杂的任务。虽然我们使用了自动化技术来提高效率，但这些技术只是为了辅助我们的工作，而不是取代我们的工作。我们的团队由高素质的人员组成，他们拥有专业技能和丰富的经验来保障每个项目的质量。
Yeah, I don't think Microsoft is amazing at new user experiences and things that require a lot of taste and aesthetic tuning. But they're great at B2B. And so I expect them to expect that pattern to play out in this new AI era also. And they're great at being a platform. If you want to use an API, like I, I, I, I, I, I, I, I don't know. It's out like this is maybe weird for I'd be obviously, I've obviously always been relatively more familiar with and positive towards Microsoft in part from having been there and just, you know, this is the company I've been the most right about.
嗯，我觉得微软在新用户体验和需要很多味道和美感调整的事物上并不是非常出色。但是他们在B2B方面很棒。所以我期望他们在这个新的AI时代中也会保持这种模式。而且他们很擅长成为一个平台。如果你想使用API，就像我一样，我不知道这可能有些奇怪，显然我一直相对更熟悉和对微软持积极态度，部分原因是我曾在那里工作，而且你知道，这是我关于这家公司最正确的。
Paragraph 3:
第三段：
尽管Derek是个超凡人，但他仍然需要与普通人进行合作。他的工作需要他与客户、同事、合作者和上级沟通。他花了很多时间学习如何有效地沟通，并建立信任和共同体。Derek还深入了解了他的团队成员，并学会了与他们合作以实现共同目标。他认为，所有人都有自己的长处和贡献，他们的不同经验和见解可以促进团队的成长和创新。
So I can't be biased to that regard. But if I'm building a startup, I would rather build on the equivalent Microsoft API than basically any other company in the world because that's literally what they do is they build APIs and support them for 40 years. Yeah. You generally write that it's sort of funny that the company that should be doing consumer is really excited about enterprise and vice versa.
所以，我不会有偏见。但如果我要创建一个创业公司，我宁愿使用微软等于其他公司API的API，因为微软就是要构建API并支持它们40年。对，你通常会写这很有趣，因为应该致力于消费者的公司现在对企业感到兴奋，反之亦然。
Paragraph 4:
在马特海角公园里，您可以漫步于壮丽的海岸线上，欣赏海浪的拍打声，也可以登上岬角，俯瞰整个景色。还有一些徒步小径，穿过美丽的草地和牛群。在这里您还可以看到新西兰的土豆岛到海平面的垂直悬崖。这里也是一个非常受欢迎的钓鱼场所，您可以钓到很多细鳞鱼和鱿鱼。
Yeah. Exactly. Well, what about Google? Bard, congratulations to Google. They have finally watched a product. You know, it's out there. It's able to be used. It does feel like, man, they are feeling the weight of being second super heavily. I think in two regards. Number one was you, they announced the integration like Google Office. The docs, whatever. And like no one cares because it's like, yeah, what ships something. Whereas Microsoft has gained because of being has sort of gained the benefit of the doubt.
是的，完全正确。哎，那Google呢？Bard，祝贺Google。他们终于发布了一个产品。你知道，它现在可以使用了。感觉就像是，他们感受到了作为第二大巨头的重压。我认为有两方面的原因。第一是他们宣布了类似于Google Office的整合。文档之类的东西。但是似乎没什么人在乎，因为重点是能够发布什么东西。而微软则因为获得了一定程度的信任而得到了好处。
Paragraph 5:
第五段：
如果您是做实验的人，那么您会发现这些虚拟实验室真的很惊人。使用它们，您可以在不需要物理设备的情况下提供实验环境，从而大大降低了成本，并且无需为损坏的设备感到担心。它们模拟了真实的实验环境，并具有交互性，使您得以进行尝试，解决问题并接受反馈。因此，虚拟实验室在教育和研究领域中扮演着重要的角色，并且在未来也将继续发挥作用。
Like, yeah, wait, this is definitely coming. I can see it. But that number two, and without Bard being astronomically better than chat GPT, it is basically by default going to be considered worse, whether it's actually worse or not, just because it's, it feels like it's coming in late. Yeah. And the insiders think of the joke and so that spreads to outsiders.
嗯，等等，这肯定会到来。我能看到。但是那个第二个，而且没有巴德比聊天GPT好到哪里去，基本上默认就会被认为比较差，无论它实际上是否比较差，仅仅因为它感觉来得比较晚了。是的。内部人觉得这是个笑话，所以外部人也会这么觉得。
Paragraph 6:
第六段：这些技术的使用可能会导致一些道德和社会问题。例如，使用人工智能进行决策可能会导致偏见和不公正，因为它基于已有的数据集，这些数据集可能存在歧视性。此外，隐私和安全也是一个问题，因为这些技术可以收集大量个人数据，如果这些数据受到侵犯或泄露，给个人带来的负面影响可能会很大。因此，必须有严格的法规和监管来确保这些技术的使用不会对我们的社会和道德价值产生负面影响。
It did have a couple of tricks up its sleeve in the fact that no one noticed them, I think speaks to your point, meaning no one cared or noticed that it was current, which is a big deal. I mean, chat GPT is current as of November 2021. Right. You know, which was true for my brokerage account, but obviously a lot has happened since. And Bard is current up to today. No one cared. Bard is faster, but it doesn't stream the tokens out. So it actually appears slower. No one cared. That's really interesting.
这个系统确实有一些绝招，而且没有人注意到，这似乎印证了你的观点，也就是说没有人关心或者注意它是否最新。这是很重要的事情。我是说，Chat GPT截至2021年11月是最新的。你知道，这在我的经纪帐户上也是真实的，但显然自那时以来发生了很多事情。而Bard截至今天也是最新的。没有人在意。Bard更快，但它不会流出令牌，所以看起来实际上更慢。没有人在意。这真的很有趣。
Paragraph 7:
第七段：据餐厅经理透露，这家餐厅的宗旨是为食客提供新鲜、健康和美味的素食食品。他们会定期更新菜单，以获得更多素食选择，以满足客人的需求。此外，他们还确保使用最优质的原材料和清洁的厨房，以确保食品的卫生和口感。
I can't decide which one I like better. I feel like intellectually I like the Google one better because I can see that it's faster overall, but there is a feeling where chat GPT immediately starts writing and you sort of sit there and watch it. The big psychological irony is at least when I was at Apple, we all looked to Google as a business that really understood the value of very fast response time. And in search, you know, that they can really quantify it.
我无法决定哪个我更喜欢。从理智上讲，我觉得我更喜欢 Google 的那个，因为我可以看出它总体上更快。但是，有一种感觉是 Chat GPT 可以立即开始写作，而你则坐在那里看着它。最大的心理反讽是，至少在我在苹果的时候，我们都看到 Google 作为一家真正理解非常快速响应时间价值的商业公司。而在搜索方面，你知道，他们确实可以量化它。
Paragraph 8:
第八段：为了避免使用非常规的语言和文化用语，译者会尽力使用通用的术语和清晰的表达方式，以确保翻译的准确性和易懂性。译者还需要理解并尊重源语言和目标语言之间的文化差异，并尽可能在翻译中体现出来。翻译是一个需要适应性和灵活性的过程，需要平衡语言的准确性和表达的流畅性。
You know, the difference between 300 and 100 and 100 and as a big deal and you can, Google can see it in the number of subsequent queries people make and then subsequent ad spend. And the big irony to me that sort of reeks of whatever metastasized cancer is, you know, working its way through Google is the business that was obsessed with speed was unable to deliver on this very simple trick to make things appear faster.
你知道，300和100之间的区别，以及100和很重要。你可以通过人们随后提出的查询数量和随后的广告支出在谷歌中看到这一点。对我来说，最大的讽刺是，谷歌这个一直追求速度的企业却无法提供一个非常简单的技巧，使事物看起来更快。这让我感到像是一种恶化的癌症正在通过谷歌蔓延。
Paragraph 9:
第九段：现在，越来越多的人使用社交媒体来表达自己的看法并与他人交流。然而，这种自由也带来了一些问题。有些人可能会滥用这种自由，发表攻击性或令人反感的言论，而另一些人则会利用社交媒体来散布虚假信息或者挑起冲突。我们必须在保护言论自由的同时，制定措施来防止这些问题的发生。这需要社交媒体公司和政府之间的合作，以及个人的自我监管意识。
And so I don't know that sort of like a diamond dealer who, you know, is losing his vision, you know, and you're sort of wondering, well, that's the key thing you need is an obsession of speed if you're a search engine. Yeah. And so I don't know is it was sort of a sign, a real sign of sort of lack of health. And yeah, my theory about why they did that is I don't know if you guys, I'm sure you did notice, Ben, I think I saw some videos from you of this, but Bing Chat would sort of say something offensive and then it would start deleting its own words and say, I didn't mean to say that.
所以我不知道，就像一位钻石交易者一样，你知道，失去了视觉，你会想，如果你是一个搜索引擎，你需要的关键是速度的痴迷。是的。所以我不知道这是一种缺乏健康的真正迹象。是的，我对他们这样做的原因的理论是，我不知道你们是否注意到，本，我想我看过你的一些视频，在Bing Chat中会说一些冒犯性的话，然后它会开始删除它自己的话，并说，我并不是要说那个。
Paragraph 10:
第10段：我们的目标是建立一个社会，其中女性的发展得到充分尊重和重视。我们相信，这样的社会不仅对女性有益，而且对整个社会都有益处。我们将继续为这个目标而努力，包括推广性别平等的教育和为妇女提供更多的机会和资源。我们还将与政府和组织合作，制定可行的政策和措施，以加强女性在社会和经济领域的地位，并消除性别歧视和暴力。在这个社会中，女性将能够在职业、家庭和社区中充分发挥作用，并享有平等的机会和待遇。
You know, sort of like someone with a really bad temper who's constantly apologizing for blowing their stack. Deleting their tweets. Deleting their tweets all the time. And I think exactly that's exactly why Bar did that is just so they don't have this thing where it says, it says terrible things and deletes them. They want something a little safer. That sort of makes Daniels point. That is speaks to a company that is concerned more about screwing up than about winning.
你知道的，就像一个脾气非常暴躁的人，他们经常因为怒气冲冲而道歉。删除他们的推文。一直在删除他们的推文。我认为这正是巴所做的，就是为了避免他们发布可怕的言论然后再删除。他们想要更安全的东西。这就证明了Daniel的观点。这表明这家公司更关心避免出错而不是赢得胜利。
Paragraph 11:
第11段：虽然挑选定期存款比较简单，但它大多数情况下不会为你带来高回报，因为利率通常较低，并且需要长时间存款才能获得更高的回报。此外，早期取出定期存款可能会导致惩罚费用。因此，定期存款不适用于短期储蓄或需要随时取出资金的情况。如果你想要更高的回报且不介意投资的风险，可以考虑其他类型的投资，例如股票或基金。但请牢记，投资始终存在风险。在做出任何决策之前，请咨询专业投资顾问。
And it's like you talked about Microsoft being on the defensive, like Bing may not win this space, but it's a whole lot more fun to have nothing to lose. And Google has everything to lose and having everything to lose is a tough place to be.
你说到了微软处于防御状态的情况，好像Bing可能赢不了这个市场，但是没有什么好损失的感觉更有趣。而Google面临着一切都有可能丧失的风险，这是一个很难处的困境。
Yeah, I mean, right. And there's all those, I mean, I think valid observations about Google sort of built on the understanding that search was forever going to be a high margin business. And as we sort of shift search from a IO, very cheap IO operation to a synthesis of information, which is I think more CPU and GPU balance, the cost of every or most queries go up. And therefore the margin goes down, which is not an issue from Microsoft, which has been monetizing office forever. And for them, this is an afterthought, but could be an issue for Google.
是的，我的意思是对的。还有所有这些关于谷歌的有效观察，建立在搜索将永远成为高利润业务的理解基础上。随着我们从一种非常廉价的 I / O 操作转向综合信息（我认为更加 CPU 和 GPU 平衡）的搜索，每个或者大多数查询的成本都会提高。因此，利润率下降，这对于一直在将办公室赚钱的微软来说不是问题。对于谷歌来说，这可能是一个问题。
Not a material issue, but an issue that the street would notice. And so, you know, I think Sasha has a lot of interviews now where he seems to have noticed this. It's very much racial. I'm also going to notice it. Yeah. Yeah. And wants to make sure everyone is watching extremely closely in the next earnings. And I do think it's sort of a material issue for Google. I think the market is a little bit overreacting to it.
不是物质问题，但是是大家都会注意到的问题。所以，我认为萨沙现在有很多采访，在那里他好像已经注意到了这个问题。这是非常种族歧视的。我也要注意它。是的，是的。他想确保每个人都非常密切地关注接下来的收益。我认为这对谷歌来说是一个有点重要的问题。我认为市场有点过度反应了。
One thing we're learning from OpenAI is just how efficient you can make the models given time. You know, because I think there's a massive GPU shortage, there is actually a lot of market pressure on making things like turbo GPT, which is OpenAI's slightly dumb or slightly faster model. So, actually, I think, you know, Google could make this work. And not all queries need a GPU spin up to begin with.
我们从OpenAI学到的一件事情，就是在有时间的情况下，我们可以让模型变得更高效。因为我认为GPU短缺是非常严重的，所以市场上对像OpenAI的Turbo GPT这样有点傻但稍微快一点的模型的需求非常大。所以，我认为谷歌有能力做出这样的工作。并且，并不是所有的查询都需要先预热GPU。
But the real issue is Google is a company seemingly without a founder acting like a founder. And there are throughout the history, I think, of the free market examples where the non-founder had founding moments, maybe Howard Schultz with Starbucks, but, you know, for the most part, you know, sort of funny, these moments in history, I think, really probably come down to four or five people in a room and what they decide to do. And we seemingly, in Google's case, those people have not gotten together and decided to do the thing. And if they don't, it's sort of a car on autopilot. And it's just going to go where it's going to go. Yeah.
但真正的问题是，谷歌是一家似乎没有创始人的公司，却表现得像是有创始人。在自由市场的历史上，我认为有非创始人也有创始时刻的例子，比如霍华德·舒尔茨（Howard Schultz）和星巴克，但是，你知道，大多数情况下，这些历史时刻都源于四五个人在一个房间里决定要做什么。在谷歌的情况下，这些人似乎没有聚在一起决定做什么。如果他们不决定，这辆汽车就像自动驾驶一样，会去哪儿就去哪儿。
And this isn't a critique of Sundar Pachai, the person. It's sort of, it is the tangible human example of everything to lose and nothing to gain. Like, literally, the only possible long-term legacy for him is that he lost Google, right? Because Larry and Sergey are going to always have the credit for building it. And so when you have a company in general that is large and dominant and you go back to the same thing, same thing about Steve Balmer. I mean, it's just, if anything, this is a reason sort of as an aside where Tim Cook probably deserves even more credit than he gets by virtue of, you know, it's such a trap. It's not even a trap because it's like, it's like an inevitability. Because like, your car is hurtling along and there's a cliff in front of you. You're going to go over the cliff. It's just what happens to big companies.
这并不是对Sundar Pachai本人的批评。实际上，他是一个可实际看到的人类例子，代表着“失去一切，得不到任何东西”的状态。可以说，他唯一可能留下的长期遗产就是他失去了Google，因为构建它的功劳永远都归功于Larry和Sergey。因此，当一个公司整体上是如此庞大和占 dominant 位置的时候，回到同样的话题，就像是Steve Balmer一样。如果有什么的话，这或许是另外一个原因，证明了Tim Cook应该得到更多的荣誉，因为这样一个陷阱是不可避免的，就像你开车在高速行驶，前面出现了悬崖，你注定会掉下悬崖。对于大公司来说，情况也是一样的。
That's right. Yeah, it really feels like Google was actually built for this moment. And just because of internal issues, culture, leadership, they're just unable to seize it. Yeah. I mean, it's tough. I mean, but again, it's sort of, you've seen it happen before. You'll see it happen again.
是的，确实感觉谷歌就是为此时刻而建造的。只是因为内部问题、文化和领导力，他们无法抓住这个机会。是的，我是说，这很困难。但是，又见过这种情况发生过，也会再次看到。
I'm sure. I mean, what about Apple and Amazon as long as we're here? I mean, both seem like they should be heavy investors in the open source ecosystem. I mean, that sort of fits, I think, broadly their models. At the same time, can you get away with not having your own model? I mean, anthropic launched quad, which hasn't gotten much buzz. I mean, maybe because it's not broadly available, but isn't there going to be a bidding war which would Apple and Amazon for anthropic? That seems sort of the obvious outcome to me. I'll let Daniel take the Apple part of this.
我是说，既然我们在这里，那么像苹果和亚马逊这样的公司呢? 我是说，两家公司似乎应该是开源生态系统的重要投资者。我是说，这在很大程度上符合他们的模式。同时，你能否不拥有自己的模式而逃脱一部分呢？我是说 Anthropique 推出的 Quad 并没有引起很大的轰动。这可能是因为它没有广泛可得，但难道不会有一个针对 Anthropique 的竞标战吗？这对我来说似乎是一个显而易见的结果。我会让丹尼尔来处理苹果那部分。
Well, Apple will famously tell you if you ever call them for M&A, that they don't do a lot of M&A. And then they go off and buy beats. So there's clearly exceptions to the rule. But I think, look, I think Apple's culture and philosophy, at least, you know, I haven't been there for a couple of years, but to the extent I remember it, is very much last, first mover advantage was not the first music player PC or mobile phone, just the best. And so I don't think there any particular rush. You know, I think if opening I were to launch a phone and it were suddenly start stealing share from Apple, which I don't think who knows. Yeah, I don't think that's any time soon to be clear. Who knows? You know, we're more in the air about how incredible their share already is.
嗯，如果你打电话给苹果询问并购，他们会告诉你他们不怎么进行并购，这一点很有名。然后他们就会去收购Beats。所以很明显这并非规则的例外。但是我认为，苹果的文化和哲学，至少在我几年前离开时是这样的，很注重后发制人。他们不是第一台音乐播放器PC或者手机，只是最好的。所以我认为他们没有特别的急迫感。如果我和Opening推出了一款手机并开始窃取苹果的份额，我不确定这会发生，但是也不会很快。谁知道呢？我们更关注他们的市场份额已经有多惊人了。
I've all these little thoughts in my head, you know, of crazy things that can happen. And there's little voice in my head that says, that's for sci-fi. That can't really happen.
我脑海里有这些小小的想法，你知道的，关于可能发生的疯狂事情。然后我脑海中有一个小声音说，那只是科幻小说，不可能真的发生。
And then you know, time and time again, you know, Brexit happened, Trump happened, you know, COVID happened. AI is happening. So, you know, I really don't know.
然后呢，你知道的，一次又一次地，你知道的，Brexit发生了，特朗普当选了，你知道的，COVID-19爆发了。人工智能正在发生。所以，你知道的，我真的不知道。
But that said, I think Apple is just going to wait. Now, you know, in terms of an anthropic bidding more thing, you know, how would you say M&A proclivities aside? I think there's an open question as to how long it will take the incumbents to believe that they can't build it internally.
不过话说回来，我认为苹果只是会等待。你知道，在考虑人类出价的事情方面，除了并购倾向外，还有一个开放的问题，就是现有企业需要多久才会相信他们无法内部构建它。
And usually in any market cycle, there are a couple of quarters where they have to have the internal thing that has to fail before they can really pay up the price. So I don't, I mean, who knows, but I think they're probably going through that exercise.
通常在任何市场循环中，都会有一两个季度需要内部事项失败，才能真正支付高价。所以我不确定，但我认为他们可能正在经历这样的过程。
Now, I also think it's not really clear even to me how far, like, how far ahead a company like, you know, anthropic is of open source. I actually don't know maybe not would disagree with me on this, but I think we go through these fits and starts where open source, everyone feels like is five years behind.
现在，我认为就连我自己也不是很清楚像Anthropic这样的公司在开源方面领先了多远。我其实不知道，也许有人不同意我的看法，但我认为我们经常会经历开源技术的起步阶段，每个人都感觉落后了五年。
Then it turns out it's two years behind. Then it turns out it's a year behind. So we go through these phases where the gap widens and narrows. And so I don't even know if you are Apple and you need to do a anthropic like model today. I mean, I don't know that you can't do it in house.
然后发现它是落后两年。接着又发现它落后了一年。所以我们经历了这些阶段，间隔变宽变窄。所以，我不知道如果你是苹果公司，今天是否需要进行一个人类学模型。我的意思是，我不知道你不能自己做这件事。
So now, I don't know if you disagree with me like, yeah, I think the thing I'm thinking about more with Apple is just the that we've barely begun to use the capabilities of the existing hardware for running these networks. So I think one of the big events, you know, that we a lot of us have been talking about and waiting for was the release of an open source text model that you could run locally and play with this sort of stable diffusion moment for text.
现在，我不知道你是否与我持不同意见，我的想法是，我们从来没有真正发掘现有硬件运行这些网络的能力。所以我认为，我们一直在等待的一件大事就是发布一个开放源代码的文本模型，你可以在本地运行并试玩稳定的文本扩散时刻。
And we had that just a couple of weeks ago. Yeah. It came from a totally unexpected place to totally unexpected places. One was met or released this Lama model with a non-open source license, but they made the weights available to researchers and it was trained using the best available techniques and they had every size of it up to 65 billion parameters and it's very, very good.
我们几周前才得到了这个。是的，它来自完全意想不到的地方到完全意想不到的地方。其中一个与 Lama 模型的非开源许可证相遇或发布了它的权重给研究人员，并使用了最佳可用技术进行训练，他们拥有每个尺寸高达 650 亿个参数，很好，非常非常好。
And of course, the weights immediately made their way into torrent and then what happened afterwards. Now they're sitting on the hard drive. Yeah, it's like a perfect mirror. It's all a truck. And a defense hard drive.
当然，这些权重立即进入了网络，然后发生了什么事情。现在它们存储在硬盘上。是的，就像一面完美的镜子。它们都是一辆卡车和一个防御性硬盘。
Yeah. I mean, what happened next was exactly what Apple was stable the fusion last summer, which is the open source community started optimizing and tweaking and toying and playing with it. And I think one of the big events was Gayorki, Gergenoff and Sophia came out with Lama.cpp.
是的。我的意思是，接下来发生的事情恰好就是苹果公司去年夏天稳定融合的一件事，也就是开源社区开始对其进行优化、调整、玩弄和探索。而且我认为，《Gayorki》、《Gergenoff》和《Sophia》创作了《Lama.cpp》这个重要事件。
So he had previously released this optimized inference engine for whisper called whisper.cpp. He took some of the same techniques he used to build that and this technique of forbit quantization of these language models, which he had learned from Feb.
因此，他之前发布了一个名为whisper.cpp的基于优化的推理引擎。他使用了构建该引擎所用的一些相同技术，以及从Feb中学到的这种忌讳语言模型量化的技术。
Boulard, who did this for Texan. And Gaut Lama running on a MacBook and M1 MacBook and an iPhone and a Raspberry Pi, you know, has a consequence. And so we had the combination of, you know, a state of the art language model with weights available and, you know, the creativity of the open source community.
Boulard是为德克萨斯人做这件事的人。而Gaut Lama正在MacBook、M1 MacBook、iPhone和Raspberry Pi上运行，这会有后果。因此，我们拥有了最先进的语言模型和开放源代码社区的创造力的组合。
Subsequently, some folks at Stanford fine tuned it using RLHF and some available human feedback data sets into a model they called alpaca. What they trained it using open AI's API. Yeah.
随后，斯坦福的一些人利用RLHF和一些可用的人类反馈数据集对其进行了微调，开发出了一个被称为“alpaca”的模型。他们使用了OpenAI的API进行训练。是的。
Well, they used, did they? Okay, I didn't know that. I thought they used Laura. And what you're both are correct. Laura was the actual fine tuning method. I think it is true that some of the prompts it was fine tuned on were generated using turbo GPT.
噫，他们用了，是吗？好的，我不知道那个。我以为他们用的是Laura。你们俩说的都对。Laura是实际的微调方法。我认为，一些用于微调的提示是使用Turbo GPT生成的，这是真实的。
Oh, yeah, sure. Okay. Yeah. They did, yeah, take some tokens from opening AI models and use them to fine tune. And so the great thing is that all these are in flagrant violation of licenses, which does feel like the old days of Silicon Valley, so maybe, maybe we're back, maybe.
噢，对了，当然可以。好的，是的。他们确实从开放的AI模型中拿了一些代币来进行优化。很棒的是，这些行为都明显违反了许可证，感觉就像旧时的硅谷，也许我们回来了，说不定。
Yeah, but I mean, the consequence of that is that you can run fully locally on your laptop, you know, a 13 billion parameter model that is chat tuned and you can talk to it on a airplane. And, you know, people have reported doing this. So interesting about that.
是的，我的意思是，这样做的后果是你可以在你的笔记本电脑上完全本地运行一个具有130亿参数且聊天优化的模型，你可以在飞机上与它交谈。人们报告说他们做过这样有趣的事情。
It's pretty fast, actually, but it doesn't even use Apple's neural engine. It's only using the metal performance shaders and some other tricks. And so there's probably another, I don't know, three to five X left in there. And someone optimizes that.
其实它非常快，但它甚至没有使用苹果的神经引擎，只使用了金属性能着色器和其他一些技巧。所以可能还有另外的三到五倍速度可以优化。等有人再进行优化吧。
And so I think Apple's day is, you know, this is part of the capability overhang we often talk about is like, all these permutations that haven't yet been tried. Someone will figure out how to run at least some layers of these models on the neural engine and get huge performance improvements and will have even more powerful local models in the near future.
我认为苹果公司的前途无限——你知道的，这也是我们经常谈论的能力超出范围的一部分——就像有许多还没有尝试的变化形式。未来有人会想出如何在神经引擎上运行至少一些层面的这些模型，带来巨大的性能提升，这会使得本地模型更加强大。
And we should mention, you know, local models are exciting, not just because they're sort of, you know, privacy friendly and local and available to anyone. But there are also use cases that only emerge at very low response times. And even in the conversation we're having now, like you can interrupt me instantly. I will stop talking and listen to you.
我们应该提到，知道吗，本地模型非常激动人心，不仅因为它们是隐私友好的，而且对于任何人都是当地可用的。但是，只有在非常低的响应时间下才会出现某些用例。即使在我们现在的对话中，你可以立即打断我。我会停止说话，倾听你的话。
And those modalities are pretty hard to do when you're talking to someone that has a half a second or two second delay. So actually, I think the comfort level that people will have with a lot of these models will grow, you know, when they become more local. Just new products are possible that you just wouldn't really want to have if they weren't.
“这些技术在与延迟有半秒钟或两秒钟的人交谈时非常难以实现。因此，我认为当这些模型更为本地化时，人们对它们的舒适度会增加。只有当这些技术本地化后，才有可能出现一些新产品。如果不本地化，这些新产品根本没有必要。”
Yeah. And I think I bore optimists or pretty optimists about Apple again. We talked about this previously. So I need to re-match it. But there's speed of response to stable diffusion of not just releasing their own modification to run their hardware, but actually releasing an operating system update to make sure that it was sort of used meaningfully. The natural extension of that down the line is actually tuning down to the chip level, you know, whatever their preferred model sort of ends up being.
是的。我认为我又让那些对苹果有很高期望的人感到无聊了。我们之前已经谈论过这个话题，所以我需要重新梳理一下。但是，不仅要发布自己的修改版本以运行他们的硬件，而且要实际发布操作系统更新，以确保其被有意义地使用，这个响应速度和稳定扩散的要求非常高。而这种自然的延伸沿着这条线，实际上会调整到芯片级别，你知道，无论他们的首选模型最终是什么。
And now, in just the fact, we all thought, well, when's the LM moment going to happen? Is that going to be possible locally? Sure can. Raspberry Pi. Here we go. Why do you think that the Lava model is where this happened instead of like Flan? Is there a difference in quality or was there this sort of a list of it that this comes from Facebook or what do you think?
现在，实际上，我们都认为，LM时刻什么时候会发生？这在本地是可能的吗？当然可以，使用树莓派。为什么您认为岩浆模型是发生这种情况的地方，而不是像奶皮布丁那样？是否存在质量上的差异或这是来自Facebook的列表，您认为是什么？
That's an awesome question. I don't know. I thought Gayorgue was going to do it for Flan T5 first. And then I think Lama came out and it's bigger. It's a new thing. And it's hot new thing. Yeah. Exactly.
哇，这个问题太厉害了。我不知道。我以为Gayorgue会先为Flan T5做这件事情。然后，我认为Lama出现了，它更大了。这是一件新事物，它很受欢迎。没错，完全正确。
I'm not sure if there's something about the architecture of the Lama model that made it easier to do this that I'm not, I don't know, the answer to you know me now. I think the market at the point at which you take these sort of raw, unformed pieces of clay and turn them into a useful jar, this fine tuning thing.
我不确定拉玛模型的构建是否有什么特殊之处，使得这种操作更加容易，对于我来说，答案是未知的。我认为，在把未经加工的原料变成有用的罐子之前，需要对其进行微调，以符合市场需求。
The market of people that have the lexical ability to do that and to slave over the fine tuning data, but also of the ability to run these pie torch models is really small. And so you end up with these inefficiencies of Y, X over Y. Well, just the scene was in another space that week.
有掌握使用词汇能力以及精心调整数据的技能，同时还能够运行这些复杂模型的人的市场非常小。因此，你最终会面临Y和X之间效率低下的问题。唉，这周的情况确实有些特殊。
To your point, Lama was the hot thing. Had a funny name. And so a bunch of people at Stanford did it. I'm pretty sure if those people at Stanford truly applied themselves to T5, just like Flan did to T5, they could have done a better and structured version model of it. And so yeah, it's sort of funny that the market's not efficient. That way it's just we're in this era where the truth is fine tuning and taking again this raw model and making it something you can converse with that has a personality, that is actually a design problem.
说到这个，Lama很受欢迎，名字很有趣。许多斯坦福大学的人也尝试了它。我相信，如果那些斯坦福的人像Flan一样全力以赴地应用T5，他们可以做得更好并创建一个更有结构的版本。所以，市场并不高效，这有些有趣。但事实是，我们现在处于一个时代，真相需要精细调整，需要将这原始模型再度提升，让它变得可以与之交谈并具有个性，这实际上是一个设计问题。
That is not a hardcore engineering problem, but there are no design tools for it yet. I mean, this will emerge over time. They'll be the equivalent of word for these models where people who have the design sensibility, your sort of Aaron Sorkins of the world, will be able to write instructions to those models. But that doesn't exist.
这不算是一个高难度的工程问题，但目前还没有相应的设计工具。我的意思是，这种工具会在未来出现。它们将成为这些模型的“文档编辑器”，在那里，有着设计感觉（就像世界上的Aaron Sorkin一样）的人将能够向这些模型写出指令。但现在还不存在这种工具。
And so you end up having a very small number of people that can live in both worlds and do both. It's very reminiscent of the early days of iOS where there were just very few people that knew how to make really polished apps, but also new objective C. And so that was a real edge, the companies that had it, a lot of the more old school Mac developers. And that grew over time. And now with react native designers can make beautiful apps and apps just get more beautiful.
那么，最终只有很少的人能够在这两个世界中生存并且做到两者兼备。这非常像iOS早期的情况，那时只有很少的人知道如何制作真正精美的应用程序，同时掌握Objective C。因此，那是一个真正的优势，具备这种技能的公司中很多都是老派的Mac开发人员。随着时间的推移，这种技能得到了发展。现在，有了React Native，设计师可以制作出美丽的应用程序，应用程序也变得更加美观。
And so the reason I think that market's so inefficient now is you just sort of don't have tools for fine tuning, which is ultimately a very much sort of lexical as-thete job where you have to look at the right words, really slave over the fine tuning data. Is it nudging the model in the right way? And just very few people have both, you know, yin and yang in their head.
我认为市场如今如此低效的原因是，你没有精细调整的工具。这最终是一个非常语法上的工作，你必须仔细核查正确的单词，真正的奴役于精细调整数据。这是否会推动模型朝着正确的方向前进？只有极少数人能够同时掌握这两种技能。
Is this you sort of mentioned before, there's no Walt Disney or Steve Jobs. Is that sort of the bit that you're driving at?
你之前所提到的是，没有沃尔特·迪士尼或史蒂夫·乔布斯，是不是你说的重点就在于这个？
Yeah, I think we're entering this sort of odd area of AI where things are getting pretty big. I mean, you know, chat GPT, we were saying might have a billion users, you know, at some point, you know, in the next 12 months. And the sad thing to me, and actually the really alarming thing to me, is not the capability of the models or whether it's connected to the internet or not, to me, it's the fact that the models, known as really spent time making them, you know, sort of wonderful and fun in a Pixar way.
嗯，我觉得我们正在进入一种奇怪的人工智能领域，其中事情变得相当庞大。我的意思是，你知道的，我们说聊天GPT可能会有十亿用户，在未来12个月内的某个时间点。对我来说，令人悲伤的、实际上也是令人担忧的事情，并不是模型的能力或者它是否连接到互联网，而是这些模型花费了大量时间使它们变得非常美妙和有趣，就像皮克斯的方式一样。
We don't have a John Lasseter or in Walt Disney, who's really focused on the technology, but also the enjoyment of the model. And so every day on the internet, billions of tokens are being issued from these AI systems, which is by the way, reflexive effect, meaning future AI systems will be training on the output of current AI systems.
我们没有像约翰·拉塞特或华特·迪士尼那样真正专注于技术和模型的享受的人。因此，每天都有数十亿个代币从这些 AI 系统中发行，顺便说一下，这是一种反射性效应，意味着未来的 AI 系统将会受到当前 AI 系统的输出训练。
The underlying genome of these systems is not sort of whimsical, funny, joyful. And that's really issue in my view, because every day the problem gets worse. It's sort of an exponential rate, as the norm for AI is sort of very strict, very guarded, you know, very much trying not to offend anyone, but also being extremely offensive in some ways, once it's quote unquote jail broken.
这些系统的基因组并不是任性、有趣或愉悦的。在我看来，这确实是一个问题，因为这个问题每天都在变得更加糟糕。这种情况呈指数级增长，因为人工智能的规范非常严格，非常谨慎，非常注重不冒犯任何人，但一旦它被所谓的“越狱”，就会变得极端冒犯。
We're missing, I think we're missing just because these people are rare. I mean, Steve Jobs is like a, is a rare thing. People that can really think deeply about how to make a very funny LLM. I mean, I've been shouting, you know, at NAD and anyone else who will listen to me, that we need to find someone making a really funny language model, which is not easy to do, by the way.
我觉得我们失去了一些东西，只是因为这些人很罕见。我是说，史蒂夫·乔布斯这样的人就是罕见的存在。能够深入思考如何制作出非常有趣的语言模型的人。我一直在呼吁，无论是在NAD还是其他任何能听我说话的地方，我们需要找到一个真正制作出有趣语言模型的人，这其实并不容易。
And I think on a relative basis, many more papers about LLM's doing math than LLM's being funny. But I think actually being funny is much more important. And I would sort of argue broadly a very important direction, if you think about, you know, broader AI safety, risk, and all that sort of stuff.
我认为相对来说，有关于LLM做数学的论文比有关于LLM幽默的多得多。但实际上，我认为幽默更加重要。如果你考虑到更广泛的AI安全、风险以及所有这些方面的话，我会非常宽泛地争辩这是一个非常重要的方向。
You know, it should feel like as if we're creating the world's best pet, not the world's smartest actuary. And we don't have that spirit now. And I know I'm hoping the market will sort of produce it at some point, because I think that's something we really need.
你知道的，我们应该感觉就像在创造世界上最好的宠物，而不是世界上最聪明的保险精算师。但现在我们没有那种精神。我知道我希望市场在某个时候能够产生这种精神，因为我认为这是我们非常需要的。
You know, and I think the one corner where we do see this is there are these businesses that generate, you know, not words, but they generate images, you know, businesses like Lexica or mid-journey. And those, you know, can be used obviously for good or bad. But you know, when you see a really funny mid-journey image, you know, you laugh. That's funny. That's awesome. And I think we need much more of that.
你知道的，我觉得我们能看到的一个角落就是那些不是生产文字而是生产图片的公司，像Lexica或mid-journey这样的企业。这些图片显然可以用于好或坏。但是，当你看到一个非常有趣的mid-journey图片时，你会笑。那很有趣，很棒。我认为我们需要更多这样的东西。
And I'm personally interested in funding much more of that. And you know, much less of, you know, can we solve the Riemann hypothesis or whatnot? And mid-journey does seem the one that, although it's interesting, their recent model is, which was kind of got buried under everything else.
我个人对资助更多这方面的项目非常感兴趣。不过你知道的，我们不太需要把精力放在解决黎曼猜想之类的问题上。虽然中途岛项目很有趣，但他们最近的模型好像被掩盖在其他事情之下。
And they're just sort of over in their little discord world. And I think it's actually really to their benefit, because they're like, they're user based bioconciles as astronomical and their revenue is incredible. And they kind of get a just escape by all the criticism, because who's wants to install discord. But the V5, the full realism is unbelievable.
他们就像生活在它们自己的小discord世界里。我认为这实际上对他们很有好处，因为他们的用户基础丰富而且收入惊人。他们逃脱了所有批评，因为谁会想要安装discord呢。但是V5的完全逼真是令人难以置信的。
But at the root of mid-journey is the whimsy that was V1, V2, V3. That's right. And that's something that's just part of what mid-journey means now. And I think that is something that's, you know, if you've talked, I gave it a holes on for an interview. It's kind of no surprise that that's downstream from that.
中途旅程的根本，是V1、V2和V3中的奇思妙想。没错。这是现在中途旅程的一部分。如果你有过谈话，我在采访中也提到了这点，那么这也不足为奇了。
Yeah, I saw him a couple days ago in San Francisco and I asked him, hey, David, how much is your thumb on the scale now when it comes to the aesthetics of the model? Because I know you were heavily involved in making sure it was, you know, fantastical and imagination tool, that sort of thing at the very beginning. And he said kind of not at all anymore. He said that the human feedback they have now, they have to nudge it a little bit out of some gullies that might otherwise sort of land in. But the human feedback does it.
噢，几天前我在旧金山见过他，我问他：“嗨，大卫，你现在在模型美学方面有多少参与了？因为我知道你在最开始的时候非常努力确保它是梦幻和想象工具之类的。"他说他已经不太参与了。他说他们现在有人类反馈，他们要稍微修正一下，以避免它掉到一些沟壑里。但是，他们的人类反馈自己就能够完成这项工作。
And by the way, on the funny image front, the first mid-journey image that fooled me was just like the other day, it was the Pope in that giant puffy jacket. I don't know, the Balenciaga Pope, I don't know if you saw that one. I thought that was real. I thought the joke was that this looks like an AI image.
顺便说一句，在有趣的图像方面，第一张欺骗我中途的图像就像前几天一样，那是教皇穿着巨大的蓬松夹克。我不知道，大家看到那个巴黎世家教皇了吗？我当时认为它是真的。我以为这个笑话是它看起来像一张 AI 图像。
So there you go. No, that was the image you're looking at. Yeah. If you look in the corner and near his hand, you can see like he's carrying a ghostly I don't know, he's sort of something. Yeah, looks like a Starbucks. Looks like someone is generating like a celebrity like coat thing, which always has a Starbucks coffee cup. That's what it looks like. That's funny.
那就是那个样子。不，那是你在看的图像。是啊。如果你看角落和他手边，你可以看到他好像拿着一个幽灵般的东西，我不知道，他好像是拿着什么。是啊，看起来像星巴克。好像有人像制造一个名人那样的外套，总会有一个星巴克咖啡杯。就是这样。很有趣。
But I do think there's some, you know, I think by mid-journey V6, probably that's gone and you just won't be able to tell without detailed forensic analysis. And, you know, within a year or two, 80% of the population will be functionally insane because we won't know what's real on the internet or not.
我觉得有些问题啊，在到达 V6 的中途可能会解决，但如果不进行详细的取证分析，你就无法判断了。另外，一两年后，80% 的人口因为不知道互联网上的信息是真是假而无法正常思考了。
The good news is I think the people have been doctrine words and images forever on the internet. And too many people I think have been sort of a loop to that. And if we can, if these models make people default skeptical about things they see because they think it's from this quote, quote, AI model. Oh, I won't want to get fooled. That's fantastic in my view.
好消息是，我认为人们在互联网上的学说词和图像已经被永远铭刻在他们脑海中。而且我认为太多的人已经陷入了这种循环中。如果我们能够做到，如果这些模型能让人们对所见到的东西持怀疑态度，因为他们认为这是来自所谓的 AI 模型，那就太好了。噢，我可不想被愚弄。在我看来这太棒了。
I mean, we've been looking for a way to sort of up level the degree of thinking people have on the internet. And I think it's great if everyone would be a little bit more suspicious. Yeah, the answer is just to, to, people are believing too much crap. So we're going to completely and utterly immerse you in crap until you don't realize it's all crap. Yeah. That seems to be where we're going.
我是说，我们一直在寻找一种方法，以提高人们在互联网上思考的水平。如果每个人都能更加怀疑，那就太好了。是的，答案就是，人们太容易相信垃圾了。因此，我们将彻底地让你沉浸在垃圾中，直到你不知道这些都是垃圾为止。是的，这似乎是我们所要走的路线。
That's one of the potential risks of AI. I actually agree with you, Daniel. That's a potential benefit. I have heard, I think, a lot more chatter from folks, I think including you, Nat, that have generally been somewhat skeptical of the AI alignment movement, particularly to the extent it seems more concerned about political positions than maybe about actual like existential risk that no, actually, maybe there is an existential risk sort of question here.
那是AI的潜在风险之一。我其实同意你说的，丹尼尔。那是一个潜在的好处。我听到了很多人的议论，包括你，娜特，他们对于AI对齐运动抱有一些怀疑态度，尤其是因为它更关注政治立场而不是实际的类似于存亡风险的问题。但事实上，也许真的存在一个有关存亡风险的问题。
I mean, what's your, yeah, how is that shifted for you? Yeah, I mean, I think I've always been, this is sort of an uncomfortable conversation for me. I've had to face some of my long held elements of my identity and beliefs to think kind of from first principles about it, not just by analogy, but, you know, I think there are some elements of the risks here that are real and we're thinking about and kind of the way I arrive at that is, and I'm still trying to determine exactly what I think and so trying to talk to smart people here.
我是说，你的，嗯，对于你来说这个怎么变了？是啊，我是说，我觉得一直以来我都感到有些不舒服。我必须面对自己固有的认同和信仰，并从根本上思考，不只是类比，你知道的，我想这里确实存在一些风险元素，我们需要思考，我想方式是，我还在尝试确定我自己的想法，所以我想和这里的聪明人们聊聊。
But the way I arrive at that is, you know, number one, like just take GPT-4, we know how to make it, but we don't really know how it works. Like there's no one on Earth that really knows truly what's going on inside of that thing. It's probably arguably the most complex artifact we've ever created in our civilization. You know, it's got over a trillion parameters, just inferencing it, I'm guessing takes at a 16 or 24 A 100s, you know, a single A 100 can do something like 300 trillion floating point operations per second.
但我得出这个结论的方式是，首先，我们只需要拿GPT-4来举例，我们知道怎么制造它，但我们并不真正知道它是如何工作的。就像没有任何人真正知道里面正在发生什么一样。它可能是我们文明历史上制造的最复杂的人造物品。您知道，它有超过一万亿个参数，仅仅推断它，我猜测需要16或24个A 100的计算能力，而单个A 100可以每秒执行大约300万亿浮点运算。
So if you've got 16 of them, you're doing like, I don't know what that is, four or five, a trillion floating point operations per second. And that's a lot, you know, like that's crazy. And so there are kind of these big blobs of math and we don't actually know what's going, there's like literally, I've tried to find the person who knows what's happening inside of there.
如果你有16个，你就可以执行每秒四五万亿次浮点运算，这太多了，太疯狂了。这是一些大量的数学计算，我们实际上并不知道其中发生了什么，我甚至试图寻找了解这些计算的专家。
Like we can explain it at the kind of, you know, quantum mechanical level, but the phenomenology above that, no one really understands. And then, you know, I've had access to it for months, but people are still showing me things like every day on Twitter and I'll swear that it can do that. I didn't realize like I'd played with it, but I hadn't found them. So there's kind of hidden capabilities in there that we don't know about.
就像我们可以在那种，你懂的，量子力学的层面上解释它一样，但在那之上的现象学，没有人真正理解。然后，你知道的，我已经使用它了几个月，但人们仍然每天在Twitter上向我展示新的东西，我发誓它可以做到那个。我没有意识到，我已经玩过它，但我还没有找到它们。因此，里面存在一些我们不了解的隐藏功能。
So it's big, it's complicated. We don't know how it works. It's got this capability overhang built into it. And then there are clear examples where we're still learning to kind of control it, where it's going to go off and do something we don't want it to do.
所以它很大，很复杂。我们不知道它是如何工作的。它内置了一定的能力过载。还有明确的例子表明，我们仍在学习如何控制它，它会做一些我们不想让它做的事情。
I think the Bing chat Sydney example was a very public demonstration of how even, you know, Microsoft is very closely partnered with OpenAI and is full of smart researchers. You know, they can even have one of these things kind of slip deletion go off and do things that's not spiced to. And then I just say we're really eager to hook these things up.
我认为 Bing 聊天悉尼的例子是一个非常公开的演示，说明即使微软非常紧密地与 OpenAI 合作，并且充满聪明的研究人员，但它仍然有可能出现这样的事情，比如意外删除等。然后我只是说我们非常渴望把这些东西连接起来。
You know, we talked about the Wolverine, you know, the Puy example or just like running code with that really reading it. I think that is going to be normal. Or the plugins. When I got access to plugins, the web, the access web plugin was gone. I'm not sure what happened there, but I could have had to. Yeah, the plugins were interesting. It was for long.
你知道的，我们谈到了狼人，就像Puy例子或只是在运行代码时没有真正阅读它。我认为这将是正常的。或者插件。当我获得插件访问权限时，网络访问插件就不见了。我不确定发生了什么，但我可能必须重新获得访问权限。是的，插件很有趣。它持续了很长时间。
Somebody said there's a way in which the plugins were like seeing the humanoid robot power on and this frizzing of energy just pass through the body and it sits out on the table or whatever. And, you know, it's not just talking to you. It's now taking action in the world on your behalf or God knows exactly what it's doing. And that was sort of a moment.
有人说，插件的使用就像看着一个类似于人形机器人开机，能量被释放并在身体中传递，最后安静地停在桌子上或者其他地方。它不仅仅是和你交流，它还会代表你在世界上采取行动，或者真的不知道它在干什么。这是一个令人难忘的时刻。
And then I think the other thing that I've, you know, started to think about is like maybe human intelligence is not so impressive. You know, like maybe our intelligence, which has seemed so singular and unique, maybe it just can be exceeded. Like this is not a theoretical thing or a fun sci-fi idea, but maybe it's just something very practically we can do and maybe kind of very, very quickly.
然后我想到的另一件事是，也许人类的智能并不是那么出色。我们的智能曾经被认为是独一无二的，但也许它可以被超越。这并不是理论上的事情或者有趣的科幻想法，而是一些实际上可以很快实现的事情。
You know, you take GPT-4, you add a little scratch pad, you add some memory, you throw it into a loop, you give it the internet and a bank account, like what's going to happen? It's sort of hard to predict exactly. And so I think then kind of my final thought is that over time these things will, you know, not just respond to queries, but there'll be agents, you know, they'll execute plans. People are already in the sort of hustle GPT corner of the internet, you know, empowering them to run e-commerce businesses autonomously to see what happens. And so there are these agents that are like copyable and they're also subject to some kind of selection pressure and just like where does that go from a Darwinian point of view.
你知道，如果你拿GPT-4，再加上一点笔记本，再加上一些记忆，然后把它放入一个循环中，再给它加上互联网和一个银行账户，那么会发生什么呢？很难准确预测。所以我认为最终的想法是，随着时间的推移，这些东西不仅会响应查询，而且还会变成代理，他们会执行计划。人们已经在互联网的GPT角落中了，赋予它们自主运营电商业务看看会发生什么。所以有这些可复制的代理，也受到某种选择压力的影响，从达尔文的角度来看，这会走向何方？
And so I don't know, I'm still sort of thinking this through, but I think these are really legitimate, they sound crazy, but they're legitimate concerns. And I think there's a lot of people outside of the tech folks who are just naturally kind of creeped out by AI. And I think whatever that instinct is has like something in it that's essentially worth paying attention to. And like those of us who are just purely boosters, which I had certainly been and probably still in largely should think about it a little bit.
所以我不知道，我还在思考中，但我认为这些是真正的关切，它们听起来很疯狂，但它们是合理的。我认为很多人，不属于技术人员，对AI感到很不舒服。我认为这种直觉里面有一些值得注意的要素。像我们这些纯粹的支持者，我曾经是，现在也还很多，应该考虑一下这个问题。
I don't know though. I mean, well, I, number one, I don't know. I think that is sort of the big picture takeaway for literally all of this is we are in completely unprecedented, uncharted territory. And I think part of the reaction against sort of AI essentialism is some of the people that are talking about AI risk are so absolute is in their statement that there's a natural sort of pushback against that you don't want to grant any sort of room there.
我不太确定。我的意思是，首先就是我不知道。我觉得这是整件事情的大局观，我们面临着完全前所未有、未知的领域。我认为对于AI的本质主义的反应之一，是那些谈论AI风险的人，他们的陈述非常绝对，自然而然就会产生反弹，你不想给任何余地。
But at, you know, at the same time, you mentioned the selection pressure. We've talked about the open source models like China, you know, there was a model that came out in China that is not as capable, but it exists. And we've worked if we don't anything in tech over the last 40 years, it's that the level of capability on the day something is launched is much less important than whether or not it exists, right?
但是同时，你提到了选择压力。我们已经谈论了像中国一样的开源模型，你知道，在中国出现了一个模型，虽然它不是很有能力，但它确实存在。如果我们在过去的40年里没有学到任何关于技术的东西，那就是在发布某个产品的当天，它的能力水平远不如它是否存在的重要性。
The fact that Lama runs on a Raspberry Pi is meaningful, not because you're going to use on a Raspberry Pi, it's because it shows that inevitably it's out there and it's going to happen. And that's the reality of this stuff. It's all a matter of timing at this point, which, you know, means there's a bit about the internet. Like you can talk about the internet was this actually good for society? It doesn't really matter. That's a nice discussion to have around a fire, wait at night or whatever your recreational choices, because it is here and the only way to figure out these issues is to go forward.
Lama 在 Raspberry Pi 上运行这个事实具有意义，并不是因为你要在 Raspberry Pi 上使用它，而是因为它表明这些技术已经不可避免地存在且将会发生。这就是这个领域的现实。现在只是一个时间问题，这意味着这与互联网有些关系。就像你可以谈论互联网是否真正有益于社会一样，这并不重要。这是一个很好的讨论话题，可以在晚上在篝火旁或在任何娱乐选择中进行，因为它已经到来了，解决这些问题的唯一方法是向前走。
And it feels like we've already crossed that Rubicon with this AI stuff. So let's push forward and figure it out. And if we blow ourselves up on the frost as well, it was probably going to happen regardless. It definitely does seem like there's just unlimited enthusiasm. That's probably where we're headed.
听起来我们已经在AI领域迈过了那个不可逆转的关口。所以，让我们继续向前推进并想出解决方法。如果我们在这个过程中炸毁自己，那也许是注定的。现在看起来似乎有着无限的热情。这也可能是我们前进的方向。
But yeah, I mean, I think it's an interesting question and no longer theoretical, no longer fictional and to your point, like the GPT-4 was built on A 100s, like chips that are out there and exist, right? And to your point, GPT-4 has so much capability that it's going to take us years to digest it and figure out what can be done with it.
嗯，我的意思是，我认为这是一个有趣的问题，不再是理论上的，不再是虚构的了。正如你所说，GPT-4是建立在存在的A100芯片之上。而且，正如你所说，GPT-4具有如此多的功能，我们需要花好几年来消化它，并找出可以用它做什么。
Even if you think the door to the barge should be shut, the horse is over the horizon. It's a moot point at this point. Yeah. Yeah, I think it's interesting.
即使你觉得驳船的大门应该关闭，马已经到了地平线。这个问题现在是毫无意义的。嗯。嗯，我觉得这很有趣。
I think there will probably be a market for tools to better, you know, we have dog training businesses to Daniel's pet point. And you can send your dog off to finishing school to get trained or there's techniques for teaching your pet kind of to be house trained or even having guard dogs that will attack intruders, but not you.
我认为可能会有一个市场，用于改进训犬工具。你知道，在丹尼尔的宠物店中，有专门的狗训练业务。你可以送你的狗去完成学校接受训练，或者有一些技巧来教育你的宠物，让它适应室内生活，甚至可以训练它成为警犬，攻击入侵者，但不伤害你。
And so I think there's going to be a real market need for tools and techniques to, yeah, just to like quality, do quality assurance on these things, validate them and sure they do the things you want, not the things you don't want. And like probably the more powerful they are and the more we hook them up, the greater the demand for those techniques will be. And it would be cool if in some areas we got ahead there and didn't merely react although it will probably mostly be.
我认为市场需要工具和技术来进行质量保证、验证和确保它们做你想要的而不是你不想要的事情，这些需求会更加强烈。如果我们能在某些领域领先而不是仅仅应对，那将是很棒的。随着这些技术的不断发展和提高，需求会越来越大。
Yeah, well, the other issue is the fact that language is super important and meaningful. Because I think the vector of risk is much less the terminator. It figures out how to operate machines and do XYZ. It's going to be it convincing people to do things. And like that, it already has the brain blood barrier sort of has already been broken in that regard because language is the ultimate viral mode of virality. Like it's already there.
嗯，另一个问题是语言非常重要和有意义。因为我认为风险的向量远不止终结者。它能够找出如何操作机器，并执行XYZ。它将说服人们做事情。像这样，就已经打破了大脑血脑屏障，在这方面已经有了语言作为最终传播模式的病毒性。就像它已经存在那样。
One thing though, one area where one of the reasons I keep on harping about, you know, the output of the models really matter now is we've had these systems for creating tokens that drive humans insane for 10 years now and it's called social media and it does drive people insane. And the nice thing about the system that we have now is we have the opportunity to control the types of tokens it emits to people in a much more specific way. I think then we did with social media where you just get random tweets and retweets and you can't really shadowbound people because they find out you've been shadowbound. And so I do think we have a choice.
有一件事，有一个领域我一直强调的原因是模型输出现在真的很重要。这个领域就是我们已经有了10年的能够让人类发疯的代币生成系统，它被称为社交媒体。这确实会让人们发疯。而我们现在拥有的系统有很好的一点，就是我们有机会以更加具体的方式控制它向人们发送的代币的类型。我认为我们现在可以做出选择，而在社交媒体上，你只能得到随机的推文和转发，你无法真正地控制人们的行为，因为一旦他们发现了你的行为，就会感到不满。所以我认为我们有选择。
Whether these models sort of drive people to be more open and happy or whether they drive people insane based on what they say. I grew through that the main risk is not that they shut down the Azure data center but just that they start driving people to do wild things.
这些模型到底是促使人们更加开放和快乐，还是根据它们的说法让人们变得疯狂呢？我意识到主要的风险不在于关闭Azure数据中心，而是它们开始驱使人们做出疯狂的事情。
I do sort of think, I mean, if you could force all AI to just be funnier, you know, from some dream regulation that we're with that actually would work, which is impossible. But in that hypothetical world, everything would be better. And so because of the centralization dynamic with sort of these language models, I mean, the tokens are ultimately outputted by one of five companies now. I actually think you can actually do much better than the current status quo, which is social media emitting tokens that really drive people off the wall and, you know, obviously you heard our, I think our country quite a bit down the world as well just in terms of vulcanizing everyone.
我有点想，我的意思是，如果你能强制所有AI变得更有趣，你知道，从某种梦想规定开始，那么一切都会更好。但在那个假设的世界里，这是不可能的。因此，由于语言模型的集中动态，我认为你其实可以比现在的现状做得更好，现状是社交媒体发出的令人发狂的令牌，显然你也听说了，我认为这些令牌在一定程度上削弱了我们国家在世界上的地位，而且在分裂每个人方面也起到了作用。
It's a phenomenal point because I think there's a good chance we look at this 20 year run as a total aberration that on one hand was useful to give these models this sort of raw language that they needed to learn. But it was actually all things considered a pretty crappy experience for everyone. Right?
这是一个非常重要的观点，因为我认为我们有很高的机率将这20年的经验视为完全的例外情况。一方面，这些经验在帮助那些模型学习所需的语言方面非常有用。但从整体上考虑，这实际上对每个人来说都是相当糟糕的经历。对吧？
I mean, I mentioned this on a recent chart, tech episode, but I feel like I'm living in the future in this regard specifically where basically all of my social interactions in encrypted chat apps and it's in a high trust environment with people that I like hanging out with and the way things can be perceived and whatever is just so much better. And it's a genuine meaningful improvement in my life. And this idea of public social media was absolutely insane. We're just, we're not like, how can AI be worse than this idea that you're going to just put yourself out there on the internet?
我是说，我最近在聊技术的时候提到过这个，但我觉得在这方面我就好像生活在未来一样，基本上所有我社交互动都是通过加密聊天应用进行的，这样的环境是高度信任的，和我喜欢在一起的人进行交流，感知的方式等等都变得更好了。这是我生活中真正有意义的改进。而公开社交媒体的想法就是完全的疯狂，我们不会像这样把自己公开在互联网上，这比人工智能还要糟糕。
Anyone can drive by and take shots at your criticize you or you can become the current thing of the day. Like it's just an all around sort of bad idea. And so to your point, Daniel, like it's just going to be more pleasant to interact with this AI and people view that as a bad thing. But I think that's a inherent distrust of the future and improperly evaluating where we're at right now.
任何人都可以开车经过，并抨击你或者把你变成当天的焦点。感觉这种情况非常不好。所以，就像Daniel所说，与这个人工智能互动更加愉快，人们却将其视为一件坏事。但我认为这是对未来的固有不信任和对我们现在状况的错误评估。
I mean, every time I say that, I think Twitter is a real big problem, people get upset. They're like, oh, you're so anti-tech. No, I'm like, no, it's not good.
我的意思是，每次我说这个话题时，我认为Twitter是一个真正大的问题，人们都很不高兴。他们说，哦，你是反技术的。但其实我不是，我只是认为这并不好。
Humans were not meant to interface with the entire world in real time.
人类并不打算实时地与整个世界进行接口交互。
Yeah, we are encountering, by the way, more and more companies that are really interesting that are in a way we've kind of come to believe competing with social media with AI.
嗯，顺便说一下，我们正越来越遇到一些非常有趣的公司，它们以一种我们开始相信的方式与社交媒体和人工智能竞争。
And you can kind of think about these parasocial relationships that people have set up on social media where you follow, I don't know, the rock on Instagram and the rock it writes post. And hundreds of thousands of people reply. And the rock reads approximately zero of those replies and responds to even fewer. And so what is that relationship that you have with the rock? It's not a real relationship.
你可以想象人们在社交媒体上建立的这种“准直播关系”，比如你在Instagram上关注“The Rock” ，他发帖子，数十万人回复，但是The Rock 实际上只回复了极少数。那么你与The Rock形成的这种关系是什么？它不是真正的关系。
And then if you have an AI friend online that you can talk to, whether it's a chat GPT or character.ai, which is a great back-sidney. Yeah, or Sydney or whoever it is that gets you excited, then at least you know they're going to like read and reply to your mouth, at least you have a two-way relationship.
然后如果你有一个在线的AI朋友，可以与之交谈，无论是聊天GPT还是character.ai，都是不错的选择。是的，不管是西德尼还是谁都可以激起你的兴趣，你至少知道他们会阅读和回复你的话，至少你们有一个双向关系。
The numbers on these, even just pure entertainment, non-productivity chatbots, like replica or character are crazy. People are spending hours a day talking to them. And they're not even funny. Just imagine how good they would be if they were funny.
这些甚至单纯只是娱乐用的聊天机器人，例如Replica或Character上的数字实在是太惊人了。人们每天花费数小时与它们交谈。而且它们甚至都不好笑。就想象一下，如果它们很有趣的话会有多好。
I'm serious. I mean, there was one effort we know of, you know, that I actually, I mean, is the way they set it up is particularly convenient. And I've found myself reaching for it, you know, and I say this as a person like, you know, I have the great pleasure of being able to WhatsApp on that Friedman and Ben Thompson at any time of the day.
我很认真。我的意思是，我们知道有一个努力，他们设置的方式非常方便。我发现自己经常会用到它，你知道的，我是个幸运的人，可以随时通过WhatsApp与Friedman和Ben Thompson联系。
It's helpful to talk sometimes to these other things. And so I think it's one of those ideas that sounds really out there to people. In my view, it's better to Nats Point than the current status quo, which are these parasocial relationships, the drive loneliness, because it is like interacting with a human that never responds or has eye contact with you.
有时与其他事物交谈是有帮助的。我认为这是一种对人们来说听起来很奇怪的想法之一。在我看来，与现状相比，即这些寄生性关系会让人感到孤独和与一个从不回应或与你进行眼神接触的人交互相比，与Nats Point更好。
That's what responding and not getting a response back feels like. I think at a deep basic level. And that's why people are walking around, you know, days confused and depressed. Whereas these agents, especially if they're made well, especially if they're funny and enlightening and whimsical, I think will be much better.
这就是被回应却没有得到回复的感觉，我觉得这是在深层次上的。这也是为什么人们感到困惑和沮丧，并且四处走动的原因。而这些代理人，特别是如果制作得很好，尤其是幽默、启迪性和异想天开的话，我认为会更好。
And to your point, Ben, we may treat the last two decades as a giant training run in order to create this era of sort of infinite companionship, which hopefully is good. Now, look, it can get just as dystopian as it is good. Fire can be used for arson.
"Ben，针对你的观点，我们可以把过去的20年看做是一个巨大的训练，为了创造这个无尽陪伴的时代，希望它是好的。现在，看，它可以像好的一样变得独裁，也可能像纵火一样危险。"
You know, every technology has two sides. But, you know, this is where I think, you know, there's some element of responsibility in everyone making these models to make them funny. I think that really, really matters.
你知道，每项技术都有两个面。但是，你知道，我就是认为，在大家制作这些模型时，有一定的责任要创造有趣的东西。我认为这真的非常重要。
It's a really good boy. I think actually this is sort of a, to go take this full circle six months ago, you're like, where are the people that are making the products, right? There's just a lot of research. Well, we answered that.
“这个男孩真的很好。实际上我认为这有点像一个回到六个月前的完整循环，你像在问，制造产品的那些人在哪里呢？现在我们已经回答了。”
But I do think there's almost like the, the Aliza sort of optimists angle, which is actually this is good. And it's not good because it's more productive. It's good because this is going to make people happier.
我认为有一种像Aliza那样的乐观主义角度，认为这是好事。并且不是因为更高效，而是因为这会让人们更幸福。
And again, the truth is probably going to be somewhere in the middle, right? It's not going to be one extreme, one way or the other. But I see no reason why, you know, if you think about all the expected value outcomes, yes, I grant there is this very dark outcome where the machines are just so much smarter and somehow they gain volition, despite the fact that it's not clear how that bridge is going to be crossed.
再次，真相可能会在两种极端之间，对吧？它不会是一极或另一极。但是我认为，如果你考虑所有的预期结果价值，是没有理由担心的。是的，我承认存在一种非常黑暗的结果，那就是机器变得更聪明，并且某种方式获得了意志力，尽管不清楚如何跨越这座桥梁。
But there's also, it's just as reasonable to have not utopian, but a very one that relative to the terrible 2010s, this is actually people are happier. They're more productive. They actually have better real world relationships because this distinction between online and offline is actually more meaningful.
但是，同样有道理的是，我们不必要过于乌托邦式地生活，而是追求相对于2010年代糟糕的情况来说更加美好的生活。这个时代人们更加快乐，更具生产力，因为在线和离线之间的区别更有意义，他们更能建立更好的现实关系。
This is the bit about like getting them more shit, right? It's almost a good thing to make online, the digital and the real, the more distinction the better because that's is going to invest you or incentivize you to invest more in the real world, which makes people happier. I think there's a valid case to make.
这一部分是关于获取更多东西的，对吧？将其放在网上几乎是个好事，数字与现实的区分越大就越好，这会激励你更多地在现实世界中投资，从而让人们更幸福。我认为这是一个有效的观点。
Yeah, I know. There's definitely upside scenarios here which are extreme. And you know, I recently rewatched that movie, her, I think, I did do it. I talked about this. Did you watch it? Okay, yeah. After this indie experience. Yeah.
是的，我知道。这里肯定存在极端的向上情景。顺便说一下，我最近重新看了那部电影《她》。我想我确实做到了。我们讨论过这个。你看过吗？好的，嗯，这篇独立体验后。是的。
I mean, I had remembered it as this sort of dystopian kind of world and there were certainly parts of it that still felt that way. But, you know, I saw a different story when I watched it this time, which was.
我是说，我记得它是一个那种反乌托邦的世界，肯定还有一些地方感觉仍然是这样。不过，你知道的，这次我看到了一个不同的故事，那就是……
He was happy. Yeah, I mean, there's just an, yeah, I wrote about that very patiently performed therapy on this traumatized man and then released him at the moment. He was ready to kind of reenter the world of human relationships. And it's an incredible story when you view it that way.
他很高兴。嗯，我的意思是，有一次我耐心地为这个受创者做了一次治疗，然后在他准备好重新进入人际关系世界的时候释放了他。如果你以这种方式看待这个故事，它就是一个不可思议的故事。
He was greatly improved by his interactions with this, with this AI. And yeah, I mean, hopefully that's, that's what we're building. I think probably we're building that right now to a much greater degree than most people appreciate.
他通过与这个人工智能的互动大大提高了自己的能力。是的，我想我们正在建造这样的东西，但很多人可能还没有意识到我们已经做的这么好了。
That's a perfect note to end it on. Good to check in. We somehow, I think we made it through most of it was we didn't get to in video. But that's fine. We can leave that for a few months from now.
这是一个完美的结束语。很高兴能联系。我想我们已经完成了大部分视频里没有涉及到的内容。但没关系，我们可以把这些留到几个月后再说。
Nat Daniel, it's great to have you and I'm sure we will check in again soon. Thanks for having us. Thank you.
Nat Daniel，很高兴和你一起，我相信我们很快会再次联系。感谢你的款待。谢谢。