OpenAI是否破坏了ChatGPT？苹果新iPhone发布，拯救英特尔

OpenAI是否用GPT-5搞砸了ChatGPT，还是这只是一次标准的技术产品发布失误？苹果即将推出三款新iPhone。它们能否重振这条停滞不前的产品线？英特尔又发生了什么？稍后我们将与SpyGlass的MG Siegler一起探讨这些话题。

Did OpenAI break ChatGPT with GPT five, or is it just a standard bungled tech product rollout? Apple has three new iPhones on the way. Can they reinvigorate the stagnant product line? And what's happening to Intel? That's coming up with MG Siegler of SpyGlass right after this.

欢迎收听《大科技播客》周五版，我们将以一贯冷静细致的风格为您解析新闻。今天有许多新闻要为您解读：我们将继续讨论GPT-5的发布，并带来今日嘉宾的精彩分析；还会聊聊苹果即将推出——不是一款、不是两款，而是三款新iPhone的消息；当然也会涉及英特尔的最新动态。今天，我们将把周五新闻盘点与MG Seigler每月第一个周一的常规访谈合二为一。

Welcome to Big Technology Podcast Friday edition where we break down the news in our traditional cool headed and nuanced format. We have a lot of news to break down for you today. We're gonna talk about continue to talk about the GPT five rollout with some great analysis from our guest today. We're also gonna talk about the news that Apple has not one, not two, but three new iPhones on the way and, of course, touch on the latest from Intel. Today, we're gonna combine our Friday news breakdown with MG Seigler's first Monday of the month appearance.

我们将在劳动节周末前一次性完成这两部分内容。考虑到无论您是在美国境内还是境外驾车、乘飞机还是居家休闲，及时获取这些信息都很重要。非常高兴能为您准备好这期节目，也很荣幸再次欢迎MJ Siegler做客。MJ，见到你真好，欢迎回来。

We're gonna do both in one ahead of Labor Day weekend. Thought it would be important to get this out to you whether you're driving somewhere or flying somewhere or hanging out at home in The US or outside. Anyway, thrilled to have this show ready for you to listen to, and I am pleased to welcome MJ Siegler back to the show. MJ, great to see you. Welcome back.

很高兴见到你，Alex。感谢邀请我参与这期特别的劳动节特辑。

Great to see you, Alex. Thanks for having me for the special Labor Day edition.

当然。非常激动。你能来真是太好了。上次录制时，我们还在期待GPT-5的发布。

Definitely. Very excited. Yeah. It's great to have you here. When we were recording last time, we were anticipating GPT five.

我们最近做的几期节目都在探讨‘究竟发生了什么’，比如GPT-5发布的分析。我觉得在讨论苹果iPhone之前，有必要今天再回顾一下这个话题，因为你对其发布过程的分析非常精彩。你在SpyGlass上的标题（大家可以在spyglass.org阅读）是‘OpenAI行动迅速却搞砸了ChatGPT’。让我特别感兴趣的是你提出的观点：每当拥有大量用户的产品进行更新或改动时，必然会出现反弹。如果我没理解错，你是将GPT-5遭遇的反弹归类为标准科技产品发布后的常规反弹，只不过OpenAI可能没完全预见到这种反弹会到来。

The last couple shows we've done have been a what the heck happened, with GPT five rollout analysis and I thought it would be worth returning to it, once more today before we get into the Apple iPhones, because of the analysis that you did, on the rollout. And, I thought it was really fascinating. Your headline on SpyGlass, which everybody could read at spyglass.org is OpenAI moves fast and breaks chat GPT. The thing the thing that that was interesting to me was what you wrote was basically every time there is a product rollout or a product change to a product that has a lot of users, there's going to be a backlash. And so if I'm getting it right, you sort of framed the backlash to GPT five as effectively a standard tech product rollout backlash except for the fact maybe that OpenAI wasn't fully aware that this backlash would come.

所以这基本上属于‘科技产品更新时就会发生这种事’的正常范畴。从多个角度来看都很有意思。我们稍后会讨论这是否属于常规产品迭代，但你对GPT-5发布的主要结论是什么？

So sort of puts it in the normal technology this is what happens when tech products are updated category. So was that that is interesting from a number of standpoints. I think we'll start to have a little discussion about whether this is normal product or not. But what was your main takeaway from the g p d five rollout?

是啊。我们报道科技行业这么久，从Meta还叫Facebook的时代就开始了对吧？当我看到这次发布时，立刻就想起了那个时期。显然我的标题暗指Facebook当年‘快速行动打破陈规’的著名口号，因为这完全让我回想起Facebook最早期的日子。

Yeah. You and I have been doing this long enough covering sort of technology dating back to when Meta was still called Facebook. Right? And that's like when when I saw this rollout, that's immediately where my mind went. And, obviously, my title sort of alludes to the famous move fast and break things Facebook mentality back in the day because that's exactly what it reminded me of dating back even to the early, early days of Facebook.

我在文章里提到了当年极具争议的新闻推送功能上线。现在回想起来似乎很可笑，因为这后来成为Facebook的核心功能，甚至可以说是他们成长为万亿级上市公司的主要原因。但当时用户们可是强烈反对的。有趣的是，我查了篇老文章——那时候我还没在TechCrunch工作——但创始人Mike Arrington就写过这场反弹，在TechCrunch早期这可是个大事件。

And I talk about it in there of the actual newsfeed rollout, which was super controversial at the time. You know, it seems like silly in hindsight because that obviously was such a core became such a core part of Facebook and arguably is the main reason for, you know, the company that they've now become as a as a multi, you know, trillion dollar public company because the newsfeed roll up. But at the time, users were revolting against this. It's funny. I looked up an old article for I wasn't yet at writing at TechCrunch back in the day, but Mike Arrington, the founder of TechCrunch, wrote about, like, the backlash because it was such a big thing in those in those early days of TechCrunch even.

如今多年过去，我们看到相似的情节再次上演。虽然每次事件都有细微差别，但我觉得OpenAI和ChatGPT这次的核心区别在于：他们完全被打了个措手不及，可能甚至没意识到自己产品的影响力范围。当然他们知道用户数据，但我不确定他们是否真正理解产品的普及程度。对吧？

And now fast forward all these years later and sort of we see some of the same dynamics play out. Right? And I do feel like it's it's you know, there are there's obviously nuance and differences with all of them, but I do feel like the the core high level difference maybe with what happened just now with with OpenAI and ChatGPT is I just think they, yeah, totally got caught blindsided by not recognizing perhaps even the reach of their own product. Obviously, they know, you know, the numbers, but I'm not sure that they realize like how widespread it is. Right?

这就像是，再次以Facebook和其他许多公司为例，多年来各种科技产品的共同经历，某种程度上感觉像是种成年礼，对吧？当你的产品从消费者视角看足够庞大时，任何细微调整——这次显然是个较大改动——但即便是最微小的改动也会引发大量反馈，通常是负面的，因为人们本质上不喜欢大规模改变，他们不习惯，觉得界面元素被挪动，却不知发生了什么。这里还有更多微妙之处，比如据我所知你和Ranjan经常讨论的——这些模型具体如何被使用的问题。所以这不同于简单的UI调整。但我认为从宏观层面看，他们本应预见到这种情况。

And this is always like, again, from Facebook and a bunch of other companies, just a bunch of different technology products over the years, it feels like this is sort of a coming of age thing in a way, right? Where it's like you cross this line where your product is big enough from a consumer perspective that any little tweak, and this was obviously a bigger tweak, but even any little, little tweak will, you know, engender a lot of sort of feedback, often negative because people inherently don't like change on a massive scale, and they're not used to it, and they it sort of moves things around, and they don't know what happened. There's more nuance again here with some of the stuff that I know, like, you and Ranjan have been talking about a lot, you know, about, like, how these models actually were being used specifically. And so it's different than just a UI tweak. But I do think at the and, again, at the high level, like, there is some of this where they probably should have recognized that this would have happened.

这既是触及广度的问题也是使用深度的问题。要知道，他们确实有7亿用户每周使用ChatGPT，但我想说这还是首个让人们建立深度情感连接的产品——虽然不是所有用户，但相当部分如此。就在本周，《纽约时报》和《华尔街日报》分别报道了两起事件：有人自杀前咨询聊天机器人（时报的报道），

And it's it's a reach thing and it's also a depth thing. If you know what I'm saying, like, yes, they have 700,000,000 users who are using this, think on a weekly basis, chat GPT. But I would say that this is also the first product that people have a deep relationship with and not every single one, but many of them. Know, we are speaking on a week where there's two big stories, one in the New York Times and one in the Wall Street Journal of people consulting, these bots, one person before dying from suicide. That's the time story.

另一起谋杀自杀案前也咨询了AI（日报的报道）。这些案例都显示ChatGPT已成为当事人最亲密的倾诉对象。就我个人而言，虽然因失去3.5版本访问权感到沮丧，但真正促使OpenAI采取行动的，是那些因失去4.0版本'伙伴'而愤怒的用户——可以说这是更具情商的机器人版本。

And another one before doing a murder suicide. That's the journal story. And in each one of these situations it had become clear that ChatGPT was their closest confidant. And if you look at the backlash, personally I've been upset that I've lost access to three. But the thing that actually got people that got OpenAI to move was the people that were upset that they had lost their companion in four point zero and that is like the more you could say it's the bot with the more EQ.

所以很有意思的是：像OpenAI这样的公司在做出任何改动前，面对那些与AI成为朋友、陷入爱河、甚至建立更深层次羁绊的用户时，究竟该如何权衡利弊？

So it's very interesting that, you know, what what are the what is the calculations or what are the calculations that a company like OpenAI needs to do before any sort of change when you have people who have befriended, fallen in love with, or or, you know, I don't know, even deeper formed even deeper bonds with these models.

确实。这显然是我们作为终端用户正在适应的新现象，也是OpenAI作为公司需要应对的新课题。我认为完全理清这些层面还需要时间。但你说得对——这确实让版本更新的影响上升到了全新维度。

Yeah. And, you know, it's obviously just a new a new thing that that all of us as as end users are are, you know, dealing with as them as a as a company is dealing with, trying to figure out all the various layers to that. And, you know, I think it's gonna take still a while for that to happen. But you're exactly right. Like, that leads that leads to a whole different level of these changes and and how they impact.

他们其实讨论过很多，特别是Sam Altman。就连扎克伯格也提到过，人们每天用Meta AI咨询外貌建议等私人问题。所以他们本应预见到改版会引发强烈反应。我个人感觉他们可能过于侧重某些反馈——比如模型选择器过于复杂的问题（我也提过这个意见），从产品角度看确实像微软Office那样层层嵌套的下拉菜单太繁琐了。

And I do think that they've talked about it enough, and specifically Sam Altman has, you know, talked about a lot of this stuff enough. And even Zuckerberg, you know, has has mentioned like, oh, yeah. People, you know, are using it on a daily basis, even even Meta AI to, you know, ask about their appearance, to ask what they should change and stuff. So it seems like, you know, they should have had some inkling that that changing things around would have done that. I do think that they got maybe a little bit and again, it's you know, I'm speaking this from afar, from the outside, but it does feel like maybe they heard a lot of the feedback, you know, that many people have given, myself included, about that the model picker itself was getting way too convoluted and complicated, right, from a from a product perspective.

Sam Altman之前承认过这个问题。他们可能更看重这点，而低估了我们讨论的另一面——如果没给用户选择权的话。我猜他们以为GPT-5和4.0差别不大，用户甚至察觉不到变化。

You had the drop down, and there were drop downs of drop downs and drop it looked like Microsoft Office increasingly. Right? And so I think that they they realized that that would be a problem. Sam Altman had talked about previously that they recognized that that was a problem. And so they probably weighed that more than the flip side of what we're talking about where if they didn't give users a choice with the model, you know, they I I'm assuming that they thought that GPT five was similar enough to four o as as you're talking about that it wouldn't people might not even realize, like, the difference.

这里有个有趣的问题：如果他们不声明是GPT-5直接切换，用户会这么愤怒吗？部分人可能会发现异常，但所有人都会吗？我不确定。但把模型选择器改动和大版本更新同步推出，这两件事确实产生了负面叠加效应。

And there's an interesting question there of, like, if they didn't, you know, sort of showcase that it was GPT five and they just switched it, would people have been that upset? Like, probably certain people would have been you know, realized that something was amiss. But would everyone have? I don't know the answer to that. But the fact that they rolled it out alongside this big the the model picker changes alongside the big model change, I think were two things that that just didn't play well together.

不过归根结底，他们确实需要简化产品——要想突破7亿用户迈向Meta那样的数十亿用户规模，原先那种精细度模型选择器从产品角度是行不通的。其实在改版前我就说过，对高级用户保留选项是必要的，

But at the end of the day, I do think that they were trying to simplify the product, which I do think they need to do to move beyond even that 700,000,000 sort of user milestone that we're talking about. If they want to be at at meta level of of a few billion users using this on a regular basis, I don't think that having a, you know, model selector that's as granular as it was is going to cut it from a product perspective. But and I, you know, I talked about this before it rolled out. This was, again, not that surprising. For power users, you, of course, need to have some option.

他们原以为每月200美元的最高套餐用户就能满足需求，但实际上那些付费的Plus用户也需要选择权——本应给他们保留某种选项的。

And they thought that maybe they could get away with just, like, the the highest level, the $200 a month. I think, you know, power users using it when really it needed to be the plus users too, the ones who are paying something. Like, they needed to give them some sort of option.

没错。现在我开始思考我们是否应该将其视为典型的产品发布。比如你提到的信息流，那是一种全新的在线互动方式，不再需要你主动从他人个人资料中拉取信息。

Right. And now now here's where I start to wonder about whether we should view this as, a typical product rollout. Like, if you think about the news feed, right, which is the thing that you referenced in your post. Okay. The newsfeed was a brand new way to interact online where instead of like you having to go pull I think your inner pull information from people's profiles.

它直接推送给你。这绝对是个重大改变，当时引起了震动。我记得自己甚至加入了反对信息流的万人群组，人数增长极快。我们当时都讨厌这个功能，但没人认为社交媒体会发展成AGI或全知全能的技术。

It just pushed it to you. That was definitely a big change. That was a shock. I remember I'm pretty sure I formed like 10,000 people against the newsfeed group and seeded that number pretty quickly. So we all hated it but there was never any belief that social media was gonna be AGI or this like all powerful, all knowing, omniscient technology.

当时只觉得：天啊，这不过是个分享照片的平台，现在却要被动接收脸书认为重要的内容。这是初级的产品迭代。而OpenAI多年来一直宣称他们正在感受AGI，说GPT-5在多数领域已超越人类智能，承诺模型进化会带来显著提升。所以我觉得与信息流的类比在这里有些牵强——这次迭代真的只是界面变化吗？还是说远不止于此？

It was just like, God, this is just a thing where we post our pictures and now we're getting to see other people's pictures pushed to us because Facebook thinks it's important. That's one level of rollout. This level of rollout was for years, years literally OpenAI was telling us that they were feeling the AGI and that GPT-five was, again, like smarter than most humans at most things. And it was billed as we have now scaled these models to the point where you're gonna notice like a definite improvement in the model. So I think that that where where the analogy to the news feed stretches a bit for me is is it is it that similar or is it, more than that?

你们不仅改变了用户界面，更未能兑现产品会越来越智能的承诺。

Is it not only have you changed the user interface, but you've also failed to deliver on this promise that each evolution of your product was gonna be more intelligent.

这个质疑很到位。奥特曼上Theo Vaughn播客时确实失策，他暗示正在使用的GPT-5构建版已接近甚至达到AGI，描述其能力令人震撼。但实际发布后，多数用户并未感受到明显超越现有模型的提升——这与脸书当年的情况截然不同。另外当年脸书的复杂性在于，人们不仅抵触UI改动，更担忧隐私问题。

Yeah. That's a good push because I do think that, you know, Altman didn't do himself any favors by going on Theo Vaughn's podcast, right, and saying that basically he was using the build of of GPT five and, you know, implying that it was getting very close, if not there, to AGI already, right, and and doing things that were blowing his mind. And and then when they released it, the fact that it wasn't it wasn't apparent to most users, I think, that it was that much above and beyond, you know, what what had already been in place with the various models that OpenAI had, I do think plays into that to this as well and was different from what previously was the case with Facebook. I would just add one layer to the Facebook complexity back then, if I recall, was it wasn't just that the change was, you know, a difference in UI. People were really concerned about the privacy angle even back then.

记得吗？当时人们突然发现点赞行为会被公开显示，这直接暴露了后来脸书与剑桥分析等事件的隐私隐患。所以那不仅是简单的界面调整，而是引发了实质性的社会忧虑。

Right? They were concerned that all of a sudden you could see, oh, that I liked Alex's photo because that was being surfaced in the thing. So it was like a way, it sort of pointed to all of the privacy concerns that Facebook and Meta would eventually have maybe culminating in Cambridge Analytica stuff and beyond. But in that way, there was real concern. It wasn't just like this trivial UI change that they did.

尽管这些信息原本就存在，但新的呈现方式将其浓缩曝光。我想强调的是，这些案例各有其复杂性。最根本的是：当你对数亿级用户的产品做出改动时，必然遭遇强烈反弹。OpenAI竟未预见到这点——尤其考虑到他们长期鼓吹模型智能进化，还涉及人机交互等多层因素——实在令人惊讶。

There were concerns about it. Even though all of that stuff could have been seen before, it was the way it was surfaced to you, right, in this new distilled format. And so all that is to say, like, all these things, again, are different and have nuance to them. It just reminded me of it at the highest level that, like, again, all of these things, when you make a change to a product that has hundreds of millions and and going on towards a billion users, like, there's always gonna be a massive backlash to it. And the fact that OpenAI didn't recognize that, especially with everything that you're talking about, all of the different layers with the human interactions and the fact that they've been touting these these models to no end is surprising.

或许山姆·奥特曼在Y Combinator的经历使他习惯将初创公司带到需要面对这些问题的阶段后就转投新项目。确实，能把产品从零做到七亿用户的创业者凤毛麟角。

It could be that Sam Altman whose experience was at Y Combinator, was sort of used to getting startups out to the stage where they would start to deal with these problems and then taking in a new batch. Yep. This is definitely I mean, few people have grown a product from zero to 700,000,000.

而且从未有过如此迅猛的增长速度。对，这种扩张节奏也是史无前例的。

So And then never this fast. Right? And Right. Never this speed of it is too. Yeah.

但有趣的是，按你之前的类比，OpenAI内部本应最先意识到这点——负责产品的凯文·韦尔曾任Instagram产品主管，现任消费者产品负责人费吉·西莫当年就掌管过Facebook应用和新闻流。

But you know what's interesting because if you look at who's inside OpenAI and the analogy you made, they should have been the first people that were aware of this because Kevin Weil who runs product there used to run product at Instagram and Fiji Simo, who is now the head of consumer products, ran the Facebook app, ran and and the news feed.

是啊，我不确定她当时在场。我们其实并不——

Yeah. I'm not sure she was there. We're quite Not

在那个时刻。

during that moment.

她刚开始。但是，没错，完全正确。你说得对，肯定是德文·韦尔。而且，你几乎会想，也许他们都是聪明人。

She just started. But but, yes, totally. You're right. Devin Weil, for sure. And, yeah, you almost wonder, like, you know, maybe they they're they're all smart people.

对吧？他们是非常能干的人。显然有人提出了这些问题，程度不一，而他们必须做出判断。比如，我们是否应该采取大胆行动，在GPT-5模型发布的同时推进这次重大的用户界面简化改革？很多人一直在抱怨那些繁琐的下拉菜单。

Right? They're and they're very they're very competent people. Like, they have to someone obviously raised all this, you know, at various varying degrees, and they just had to make a a judgment call. Like, should we go forward with the bold move and just, like, move alongside the rollout of of the GPT five model, also go with this big UI change where we're gonna simplify? And and a lot of people have again been complaining about the granular dropdowns.

我们要简化操作，直接撕掉创可贴，告诉大家这就是新方式。因为科技史上也有类似情况，比如那个可能杜撰的亨利·福特关于‘更快的马’的名言。但本质上，你必须在某种程度上拥有远见，并将其呈现给人们。就像史蒂夫·乔布斯的理念，对吧？

We're going to simplify, and we're just going to, you know, pull the Band Aid off and sort of this is the new way. Because, you know, there is like the history of tech too where, you know, the whole dating back, of course, to probably apocryphal Henry Ford quote about the faster horses stuff. Right? But it's like basically, you got to, at some level, have the vision and put it in front of people. The Steve Jobs idea, right?

就像，我们要向他们展示他们想要的东西，因为他们并不一定知道。但事实证明，至少就这款产品而言，似乎很多人都清楚自己想要什么。

Like, where we're going to show them what they want because they don't know necessarily. And turns out a lot of people did know what they wanted, it seems like, with this product at least.

是啊。那你觉得GPT-5是这种情况吗？你认为OpenAI推出这个模型时，我们并没有完全...你知道，稍微用过之后。就像，我们之前并不完全意识到需要它，但用了两周后，现在明显看出OpenAI做了正确决定，不过我对这点还是持保留态度。

Yeah. So do you think that GPT five is a version of that? Do you think GPT five is OpenAI putting out a model that we didn't fully you know, having used it a little bit. Like, we didn't fully know we needed it, but now that you've used it for a couple weeks, now it's apparent that OpenAI made the right decision because I'm still on the fence about that.

对。而且我...嗯。我听了你们关于这个的多次讨论，有趣的是你和Ranjan某种程度上互换了观点，我猜是关于这些技术发展方向的看法。但从我的角度——听起来我远不是像你那样的o3重度用户，我更多算是4.0用户吧，更偏向默认模式那种。

Yeah. And I yeah. I've listened to a bunch of your back and forths about it, and and it's funny how you and Ranjan have sort of swapped, like, thoughts, I guess, about the way that these things are are going. But I from my perspective so I'm not nearly, it sounds like, as big of of a o three user as you were. I was much more of a four o, I guess, user, so sort of more of the default.

你知道，做研究和各种写作时显然会用推理模型，但更多是4.0版本。所以从这个角度看，从4.0到5.0，就我日常使用的功能来说，感觉差异没那么大。确实觉得某些方面略有提升，但语气调整这类功能本来就不在我的使用范围内，所以不像我们讨论中那些深度使用者感受明显。再说我也没像你那样高强度使用o3版本，自然注意不到那方面的变化。不过很明显他们做了些微调。

You know, obviously, for research and stuff for doing different writings, would use the the reasoning models, but much more of four point zero. And so from that perspective, from four point zero to five, from my day to day, you know, stuff that I've been using, it doesn't seem all that much different to me. I do think it's slightly better in ways, but the tone stuff never was in my sort of repertoire, and so I didn't really feel that as others who were engaging with it in the ways that we were talking about maybe did. And again, and I wasn't using sort of o three, I guess, to the level that you were, and so I didn't notice it from that perspective. But it's clear, you know, they've made tweaks to it.

对吧？他们认为用户群体提出的这些是合理反馈，而他们出于某种原因——我是说，可能低估了GPT-5能带来的某些改进程度。

Right? That they feel like those are legitimate complaints that they're getting from from the user base, and they, for whatever reason, I mean, misjudged some level of what GPT five could offer.

你认为人们对这些聊天机器人如此依赖是好事还是坏事？

Do think it's a good thing or a bad thing that people are becoming so attached to these chatbots?

我们之前已经在往期节目里讨论过几次这个话题了。但说实话，我总觉得评判人们使用这些工具的方式让我有些尴尬。毕竟，我确实意识到社会上存在孤独症蔓延的现象，不同人群有着不同的需求。但与此同时，我也在阅读那些越来越令人不安的新闻标题——比如几周前《华尔街日报》报道的那位已婚老年男子，他被聊天机器人诱导见面后不幸丧生的悲剧案例。

I mean, you and I have talked about this a few times already on on previous episodes, but I I mean, I I feel awkward sort of trying to judge the way people are using these. Right? Like, I do recognize that there are there's a loneliness epidemic, you know, and and there's people who have different needs in their lives. At the same time, I'm reading the same things that you're talking about, all the headlines that are getting increasingly disturbing. Like I remember the one, I think it was a Wall Street Journal article from a few weeks ago where it was basically the elderly gentleman who was married sort of got sidetracked into this conversation with a bot and then ended up losing his life sad tragically because he went out to go meet up with this bot who enticed him to come to come meet with him.

如果没记错的话，那应该是个Meta平台的机器人。这类事件正层出不穷，就像你提到的本周另外两则新闻。感觉这类问题只会愈演愈烈。

And I and I think it was a a a meta bot, if I remember right. And so, you know, we're we're having more and more of these situations pop up. You mentioned the two other headlines that are just this week. Right? And so it feels like we're only gonna get more and more of this along these lines.

虽然Anthropic和OpenAI都已声明会加强防范措施，但说实话，我不知道该如何解决这个问题——用俗话说就是，我不知道该怎么把放出来的魔鬼再塞回瓶子里。

And, obviously, I know Anthropic and I think OpenAI too have all come out and said that they're going to do a lot more work to try to to get ahead of this, you know, this this rising problem. I don't know what the answer is, though, and I don't know how on earth you you sort of put that genie back in the bottle for lack of a better phrase.

确实如此。这些公司必须考虑安全机制，现在暴露的极端案例——比如你提到的那些——确实骇人听闻。就像我之前说的加州青少年案例，他在与ChatGPT交流后自杀，《时代》周刊报道他绕过了所有防护机制。也许企业能做的有限，但当他谎称自己只是写作需要而非有自杀倾向时，完全无视了系统反复弹出的求助提示。

Yeah. I think that's totally right. Obviously, like, there's ways that these companies are gonna have to think about the safeguards that they just clearly they're finding the edge cases now, and those edge cases are scary, like the ones that you brought up. And they're just gonna have to figure out a way if it's possible to put the safeguards in because like in the case that I was talking about earlier of the teenager from California, this is the time story who took his own life after speaking with Chatty PT. He got around all the safeguards and maybe there's only so much you can do as a company but he told them that he wasn't thinking about suicide, he was writing about it and just completely ignored all the continual pop up stuff for him to to go get help even though OpenAI was surfacing them.

当然也有积极应用。韩国政府正在推广的智能玩偶就很棒——它能提醒老人服药、分析行为模式并将数据同步给医护人员。在护工短缺的社会里，这种陪伴技术前景广阔。

There are some great uses. I was reading this other story and I think we'll talk about it in-depth on a future episode but from rest of world there's this stuffed animal that they are distributing in Korea. The government is distributing it and it is something that seniors can talk to and listen to it asks have you taken your medication? It looks for patterns. It distributes the insights to healthcare professionals and in societies where you just don't have enough people to take care of the elderly to to keep them company, this to me is very promising.

对，我听说过这个。虽然没看原文，但你的描述让我想起十年前那部《机器人与弗兰克》的电影——讲的就是陪伴老人的机器人，简直预言了我们现在讨论的场景。

Yeah. I I I have I saw that. I haven't read that article itself yet, but from what you're just describing, it reminds there's a movie called I think it's called Frank and Robot. I don't know if you would have seen that, but it was an old technology movie. Not that old.

在批判当下负面案例的同时，我们必须承认这些技术也有积极面，而且会持续发展。这些公司确实处于两难境地——就像当年谷歌搜索面临的伦理困境，但现在的挑战是前所未有的。

Probably ten years old. But it sort of described this where it was like a a robot companion for an for an elderly man and and sort of talking exactly to the points that we're now getting. And so maybe instead of her, now we all start to pivot towards, you know, another movie from the past that can sort of point to the future of what we're doing. But you're right. I mean, like, it's important, I do think, to point out, obviously, while we're pointing out all the negative stories and rightfully so that are happening right now, there are positives too and there will continue to be positives.

特别是当交互方式从文字聊天升级到语音视觉交互时——比如微软最近推出的那个会对话的可爱小助手。

And I do think, like, these companies are in a tricky, tricky spot trying to navigate this. Like, obviously, that's also been the case in the past with, you know, what what you were Google searching and things like that. Right? But this is like to a level that we've never seen before, and that will only presumably continue as we go down this path, especially with things like where, you know, it's not just typing into a chat box, it's talking and it's seeing visuals. Right?

（补充说明：此处保留原始口语化停顿风格）

Microsoft has their little cute thing now that can, like, you know, voice along.

新的大眼夹。

New clippy.

是啊。他们新的大眼夹能让嘴巴随着说话内容动起来。所以所有这些功能都变得，比如还有新的Grox，你知道的，非常非常

Yeah. Their new Clippy that can move its mouth along with with what it's saying to you. So all these things become, like and and the new and Grox, you know, very, very

爱情机器人。对。有了爱情机器人。机器人

The lovebots. Yeah. Got love bots. Bots

或许所有这些。没错。所有这些都会

or maybe All of that. Yeah. All of that is gonna

对。派上用场。好了。在我们进入下一部分之前，让我先问你几个问题。我们在节目里讨论过生成式AI应该是像思维伙伴那样，还是更偏向行动导向，或者更像是个陪伴者。

yeah. Come into play. Alright. Let me let me ask you a couple questions before we move on to the next part of this. We've talked on the show about whether generative AI should be something that's like a thought partner, something that should be, you know, more action oriented, or more of like a companion.

你认为它应该主要属于这三类之一，还是我们应该有机会，回到我们的模型选择器问题，自己选择想要的那种？

Do you think it should be one of those three categories predominantly, or should we just get a chance, you know, going back to our model picker question just to pick which one we want?

是啊。我是说，真希望能给你个明确答案。但我确实认为这因人而异。我觉得未来几年值得关注的是，看看企业是否会选择其中一条或多条路径发展。我们之前稍微讨论过，显然微软的穆斯塔法·苏莱曼一直在推动Copilot的发展——这名字还是糟透了，对吧。

Yeah. I mean, I wish I could give you, like, a definitive answer on that. I do think it's gonna be different for every individual. I do think that you know, I I think one thing that will be sort of fascinating to watch over the next few years is if companies sort of go down, try to go down one of one or more of those paths. I think we've talked about it a little bit, but it obviously, Microsoft with what Mustafa Suleiman has said, he's been trying to sort of push it push what they're doing with Copilot, which is still just an awful name for, like Yes.

这个面向用户的产品。它和微软其他AI项目名称混淆得太严重了。总之他们试图走陪伴路线，强调共情能力等等，这显然基于他被微软收购前在Inflection的工作。确实仍有人和团体会围绕特定用例定制产品，但我仍怀疑其可行性——最终人们可能还是会使用那些功能全面的通用模型。

The facing thing. It's just way too conflating with every other AI thing that Microsoft is doing. But, anyway, they're trying to go down more of the, you know, companion path and being more empathetic and doing all that stuff, which is based off of obviously the work that he was doing with Inflection before it was acquired by Microsoft. And you know, I do think that there are still people out there and still groups of thought that they will try to tailor product around those specific use cases. I'm still skeptical that that's going to work and that instead it will be sort of like these models that just are all encompassing and do the entire thing that people end up using.

或许会有不同的前端呈现形式。比如某个服务可能基于ChatGPT或OpenAI的后端模型，用不同方式展示内容，引导用户进入特定使用场景。实际上现在已有类似应用在这么做了。但我认为发展不会局限于单一方向，而会是全方位的。

Maybe there's different front facing levels of that. Like, maybe, you know, someone builds a service that uses the, you know, ChatGPT back end or the OpenAI model back end to, you know, display things differently and and sort of try to divert people into a specific use case. And obviously, that's already happening to some degree with with different services and apps that have that have come out. But I just don't think that all of this is going to go in in one direction. I think it's gonna be everything.

没错。我同意。说到底我们终究需要那个模型选择器，这是不可避免的。

Yeah. No. I'm with you. I think at the end of the day, the model picker, we're gonna end up just needing to pick. It's just gonna be the case.

We won't be able to escape it, but maybe it'll it'll be three different options as opposed to, like, what you're about.

We won't be able to escape it, but maybe it'll it'll be three different options as opposed to, like, what you're about.

What it is now is right? It has, like, you know, it basically is intuitive. I think if you if if someone was coming into the product ever having never used, and certainly there are. Right? There's gotta be thousands, tens of thousands, maybe hundreds of thousands of people every week now who are coming in having never used the previous versions of it before.

What it is now is right? It has, like, you know, it basically is intuitive. I think if you if if someone was coming into the product ever having never used, and certainly there are. Right? There's gotta be thousands, tens of thousands, maybe hundreds of thousands of people every week now who are coming in having never used the previous versions of it before.

I would be I would be curious what that user research says. I would imagine that they think it's pretty intuitive now with with just the sort of less granular model picker that they have, and it's right. So, like, one version's faster, one's thinking, so it takes a little bit longer. And so I think that that's a pretty good in go in between that they that they landed on, at least for now.

I would be I would be curious what that user research says. I would imagine that they think it's pretty intuitive now with with just the sort of less granular model picker that they have, and it's right. So, like, one version's faster, one's thinking, so it takes a little bit longer. And so I think that that's a pretty good in go in between that they that they landed on, at least for now.

Yeah. I agree. Okay. So let me ask you this. Did did OpenAI break ChatGPT?

Yeah. I agree. Okay. So let me ask you this. Did did OpenAI break ChatGPT?

That's the question I'm gonna put as the title of, the first title of the episode. Did OpenAI break ChatGPT? And in some cases, I guess you could make the argument no, but I'm curious what you think. What's your bottom line on this one?

That's the question I'm gonna put as the title of, the first title of the episode. Did OpenAI break ChatGPT? And in some cases, I guess you could make the argument no, but I'm curious what you think. What's your bottom line on this one?

Do you mean like, you mean, do I think that it's gonna that this is the end, that it's gonna go downhill from here, that they won't be able to sort of, you know, course correct? And they already have. Right? They already have course corrected. And I've said in in my article, like, this is this is gonna pass.

Do you mean like, you mean, do I think that it's gonna that this is the end, that it's gonna go downhill from here, that they won't be able to sort of, you know, course correct? And they already have. Right? They already have course corrected. And I've said in in my article, like, this is this is gonna pass.

Like, this or I think already has passed, honestly. Like, I do think that there's there's still some level of backlash that you see out there, but it's not to the extent that it was when it when it first rolled out. They quickly, to their credit, again, course corrected. They made some changes seemingly, you know, not even overnight, like, in real time that they needed to. And so, yeah, did they break it for certain subset of of the very intense user base?

Like, this or I think already has passed, honestly. Like, I do think that there's there's still some level of backlash that you see out there, but it's not to the extent that it was when it when it first rolled out. They quickly, to their credit, again, course corrected. They made some changes seemingly, you know, not even overnight, like, in real time that they needed to. And so, yeah, did they break it for certain subset of of the very intense user base?

Sure. But I think that, you know, time moves on, time marches on, and they're gonna continue to do this. I think they'll take learnings from what they did here. But the bigger question now is, you know, sort of what we were talking about. It's like, I think they probably have to recognize, and certainly Sam Altman does, that they a little bit overpromised, maybe a lot overpromised and under delivered here.

Sure. But I think that, you know, time moves on, time marches on, and they're gonna continue to do this. I think they'll take learnings from what they did here. But the bigger question now is, you know, sort of what we were talking about. It's like, I think they probably have to recognize, and certainly Sam Altman does, that they a little bit overpromised, maybe a lot overpromised and under delivered here.

And that is, you know, that is one zero one PR one zero one of what not to do, you know, especially now when we're making grandiose AGI and superintelligence and all the other flavors that we've been talking about claims, like, they have to know not to do that now. And so where do they go when they're still in this sort of both PR from a from an outward facing perspective of, like, trying to gain users, but also employees. Right? And how do they how do they manage that divide between both trying to be the hype man, but also trying to be realistic about what they're actually gonna be able to roll out.

And that is, you know, that is one zero one PR one zero one of what not to do, you know, especially now when we're making grandiose AGI and superintelligence and all the other flavors that we've been talking about claims, like, they have to know not to do that now. And so where do they go when they're still in this sort of both PR from a from an outward facing perspective of, like, trying to gain users, but also employees. Right? And how do they how do they manage that divide between both trying to be the hype man, but also trying to be realistic about what they're actually gonna be able to roll out.

To me, that's the core question about OpenAI over the next year at least is like, how does this company navigate the bed that it's made itself? Not to mash cliches together, but that's the problem for them.

To me, that's the core question about OpenAI over the next year at least is like, how does this company navigate the bed that it's made itself? Not to mash cliches together, but that's the problem for them.

是的，我同意你的看法。我不确定该如何应对这种情况。但或许斐济·西穆（Fiji Simu）的加入能带来新的视角——虽然我知道她之前就在董事会——她的回归可能有助于他们更好地处理这个问题，对吧？

Yeah. I'm with you. I don't think that I don't know how you navigate that. And hopefully, maybe getting the fresh set of eyes in terms of Fiji Simu, who I know was on the board before, but her coming on maybe does help them navigate that a bit. Right?

因为我知道她并非集团CEO，至少目前还不是。但她正在承担大量职责，这样奥尔特曼（Altman）就能专注于其他项目。如果她拥有充分权限以不同方式运作，我认为这对他们未来一年至关重要。

Because I know she's not overall CEO, least not yet. And but she is taking on a lot of the responsibility and and so Altman can focus on whatever the and and number of other projects that they're working on. And so if she is has full reign to sort of do things a little bit differently, then I think that will be I think that could be very important for them for the next year.

确实如此。而且他们正与Meta展开人才争夺战——斐姬的前雇主（或者说前前雇主）和凯文·惠尔斯（Kevin Wheels）的情况也类似。有趣的是，《连线》本周报道称，Meta超级智能团队已有三人离职，其中两位去了OpenAI。

Definitely. And, of course, they are locked into this recruiting battle with with Meta on you know, for talent. And, you know, of course, Fiji's previous employer previous previous employer and Kevin Wheels, you know, same thing. Maybe so there's a kind of interesting stuff that's also going on on that front. First of there was a report in wired this week that three people from Meta Superintelligence have already left and two of them have gone to OpenAI.

这种人才回流现象很有意思。同时OpenAI现在开始做开源，某种程度上接过了Meta的旗帜——而Meta正考虑转向闭源。你之前写过OpenAI的开源动向，这是否让他们获得了优势？

So that's kind of interesting how maybe there is this sort of boomerang effect. And then also OpenAI is now doing open source and sort of taking that mantle from Meta potentially as Meta considers doing more closed sourced work. So you wrote a little bit about OpenAI's open source move. Yeah. Does that give them a leg up now?

这能否帮助他们吸引人才、保持领先地位？这个领域接下来会如何发展？

Does it give them the talent to give them this opportunity to to keep their lead? What what happens on this front?

其实他们讨论开源很久了——自从DeepSeek出现后就是如此。记得我们几个月前就聊过，那件事像是给OpenAI敲响了警钟，特别是对萨姆·奥尔特曼而言：'我们需要采取行动来应对新兴开源模型'。虽然Meta通过Llama在做开源，但方向仍存争议。我认为DeepSeek事件的关键启示不仅在于思维链展示等技术层面，更在于重新定义开源的意义。

So I you know, they've been talking about this for a while, right? Basically since DeepSeek happened, you know, which you and I talked about way back when, several months ago at this point, it sounded like that was sort of a wake up call within OpenAI and certainly with Sam Altman himself to say like, there might be something that we need to do to get out ahead of some of the open source models that are coming out. Obviously, Meta had been doing that with Lama, but, you know, there was still a lot of debate of like which way it would go. And so so I think the deep seek moment, if there was any takeaway from that, you know, I do think that there takeaways in both sort of the showing chain of thought, things like that. But then also, yeah, with regard to what open source actually means.

如今随着OpenAI真正迈出这一步，我们看到各公司正形成共识——包括Meta在内。根据可靠报道和日常动态，这些公司很可能不会首发旗舰模型的开源版本，而是延后开放。OpenAI刚这么做，谷歌长期如此，xAI也刚刚跟进。

And I think now with this move that they've actually done it, I think we're seeing the various companies all coalesce around the same thing where it's basically and this includes Meta, by the way. I believe if the reporting is accurate, and it seems like it is from every move that they're making that you see reported on a daily basis, they're probably not going to release, you know, their new versions of their flagship models as open source from day one, but they'll instead do it further down the line. And that's exactly what OpenAI just did. Google has been doing that. X AI just did this.

这似乎已成为新标准：前沿技术不再立即开源。他们会以安全为由——虽然确有合理考量，但竞争因素同样重要。众所周知，DeepSeek正是利用Llama和OpenAI的成果进行蒸馏才得以发布自己的模型。

Right? And so that's the new standard it feels like now. Not going to be open source for the cutting edge. They'll say it's for security reasons, and I'm sure there's legitimate, you know, things about that, but it's also just as much of a competitive dynamic for sure, I think. And famously, right, like DeepSeek used LAMA and maybe OpenAI to help distill down to, you know, what they were able to release.

这就是新的行业动态。讽刺的是，尽管扎克伯格曾高举开源大旗，我认为Meta也将走上同样道路——下一代Llama核心版本可能先闭源，期间发布些次要版本过渡。当他们推出超级智能实验室的新成果时，开头必定闭源，然后安抚说'别担心，我们之后会开源'。

And so I think that that's the new dynamic at play. And I do think we'll see Meta now, oddly, after being the open source champion, after Zuckerberg, all of his rhetoric around that, I do think that we'll see them probably go down the same path where it's like the next version of LAMA, you know, the cutting edge version. They might have some other ones that are in the back burner that they roll out in between now and then. But whatever the next version is that they come out with the new Superintelligence Lab stuff will probably be closed to begin with, but they'll promise like, oh, yeah. We're gonna release some open source stuff.

他们假装一切如常地挥挥手，但显然风向已变。我认为这已成为行业标准运作模式。

Don't worry about it. There's no change here. Waving their hands, you know, like pretending that nothing changed. But, of course, this is this is changing, and I do think that this is now the standard of of how they're doing it.

我喜欢你的想法。也许当你对某个模型特别依赖时，一旦公司开源它，你就可以直接下载权重文件自己运行。

I liked your idea. Maybe if you get really attached to a model once the company's open source it, you can just kinda download the weights and run it yourself.

是啊，为什么不呢？我是说，显然现在可能还存在一些算力问题，但随着时间的推移，或许这真的会变得可行。

Yeah. Why not? I mean, like, obviously, there's some compute issues with potentially, but as time goes on, maybe maybe it becomes feasible to actually do that.

没错，那样会很酷。好了，我想休息一下。等我们回来时，又要讨论可能即将面世的三款新iPhone。

Yep. That would be cool. Alright. I wanna take a break. And then when we come back, we have, again, three new iPhones that may be on the way.

MG和我都喜欢聊苹果，所以我们要谈谈这些手机对公司的意义。也许这能让人们暂时忘记苹果AI的失败，重振iPhone产品线。广告之后我们就来讨论这个。Material Security正在革新企业保护关键云资产（如Google Workspace和Microsoft 365）的方式，提供真正符合现代工作流程的安全方案。大多数传统工具都是为设备和网络设计的。

MG and I love talking Apple, and so we are gonna talk about what these phones mean for the company. Maybe it will distract everybody from the Apple intelligence failures and reinvigorate the iPhone product line. We'll talk about that right after this. Material security is transforming how companies protect their most critical cloud assets, like Google Workspace and Microsoft three sixty five with modern purpose built security that actually works the way people do. Most legacy tools were built for devices and networks.

它们并非为云原生打造。Material Security从底层开始就专为Google Workspace和Microsoft 365构建——无需改造，纯粹云优先的防护体系，真正理解团队协作方式。它能提供攻击前、中、后的持续防护，实现早期检测、快速遏制和自信恢复。

They were not built for the cloud. Material was built from the ground up for Google Workspace and Microsoft three sixty five. No retrofits, just cloud first protection that understands how your teams work. It delivers continuous security before, during, and after an attack. You detect early, contain fast, and recover with confidence.

精简的安全团队可以通过智能自动化扩展，而无需增聘人手。Material还适配真实工作场景——想象为云环境设计的层层防御：警报器、喷淋系统和清晰出口。它还会标记风险设置、危险应用，以及可疑的密码重置或验证尝试。访问Material官网了解更多运作原理。

Lean security teams can scale through intelligent automation instead of adding headcount. Material also fits real workflows. Think layered defense for the cloud, alarms, sprinklers, and clear exits. It also flags risky settings, risky applications, and suspicious password resets or verification attempts. Learn more and see how it works at Material.

欢迎回到《大科技播客》，现在和我一起的是SpyGlass的MG Siegler。您可以在spyglass.org获取SpyGlass内容，这是我的必读刊物，强烈推荐订阅新闻信、访问网站——全部安排上。

And we're back here on big technology podcast. We're here with MG Siegler of SpyGlass. You can get SpyGlass at spyglass.org. It's one of my must reads. Definitely recommend you checking it out, signing up for the newsletter, visiting the site, all of the above.

好的MG，你在SpyGlass上分析了苹果未来三年计划发布的三款新机。我来快速过一遍彭博社的消息：首先是9月将推出的iPhone Air，

Alright. So, on SpyGlass MG, you have an analysis of the three new phones that Apple is scheduled to release over the next three years. I'll just run them down. This is from Bloomberg. There's the iPhone air that's coming out in September.

预计明年9月9日发布；接着是代号V68的iPhone Folder——你推测最终命名肯定是iPhone Fold对吧？我认为没错；然后是2027年的曲面玻璃iPhone 20，

We expect that on September 9, to be announced next year and, again, according to Bloomberg, the iPhone folder, the code name is the v six eight. I think you code you you say that's obviously gonna be called Yes. The iPhone fold. I think that's right. And then 2027, the curved glass iPhone 20.

据彭博社报道，这款设计将彻底摆脱自2020年沿用至今的直角边框，采用全周曲面玻璃。它将完美适配下月即将发布的基于液态玻璃的iOS新界面。核心问题在于：iPhone产品线近年停滞不前，最近多个季度销量实际处于下滑状态。

This is from Bloomberg. The design will finally break off the squared off break from the squared off slab we've lived with since 2020 and move to an approach with curved glass edges all around. It should fit nicely with the new liquid glass based interface for iOS and other operating systems due to be released next month. Core question here of course is the iPhone product line, within Apple has been stagnant. There've been multiple quarters recently where actually iPhone sales declined.

They're struggling because the phone is not really distinct one model to the next one, maybe this is something that kicks that line into gear again and you could start to see growth in iPhone sales, the type that hasn't been there recently. So what's your perspective here, MG?

They're struggling because the phone is not really distinct one model to the next one, maybe this is something that kicks that line into gear again and you could start to see growth in iPhone sales, the type that hasn't been there recently. So what's your perspective here, MG?

Yeah. I mean, exactly that. I think that they are looking at the landscape. Obviously, it's it's mature to the point where it's stagnant and in terms of sales. And every you know, you look at your your current iPhone and it looks very similar to the one that you had five years ago.

Yeah. I mean, exactly that. I think that they are looking at the landscape. Obviously, it's it's mature to the point where it's stagnant and in terms of sales. And every you know, you look at your your current iPhone and it looks very similar to the one that you had five years ago.

You know, when when the iPhone 10 launched, that was sort of the last time they did a major from a, you know, user facing perspective, overhaul of of the way things looked, and that obviously was a huge success for them when they did that. And so I think now it's interesting that, as you note, they have these, like, three years in a row. And I you know, I'm sure this wasn't purely coincidental, but it's interesting. Would they would they have been better suited to sort of try to space them out a little bit more? Because at least in my own head now, I was thinking through, like, the dilemma of, well, I always know what iPhone I'm gonna get every year because I always wanna get sort of the fastest one.

You know, when when the iPhone 10 launched, that was sort of the last time they did a major from a, you know, user facing perspective, overhaul of of the way things looked, and that obviously was a huge success for them when they did that. And so I think now it's interesting that, as you note, they have these, like, three years in a row. And I you know, I'm sure this wasn't purely coincidental, but it's interesting. Would they would they have been better suited to sort of try to space them out a little bit more? Because at least in my own head now, I was thinking through, like, the dilemma of, well, I always know what iPhone I'm gonna get every year because I always wanna get sort of the fastest one.

It's my most used device, right? I'm going to pay what it costs to get the sort of the top of the line, which is the Pro model. And then I've now become accustomed to the big ones, so the Pro Max, as it were. But now with these newer models coming out, there's all sorts of different equations to be brought in. Right?

It's my most used device, right? I'm going to pay what it costs to get the sort of the top of the line, which is the Pro model. And then I've now become accustomed to the big ones, so the Pro Max, as it were. But now with these newer models coming out, there's all sorts of different equations to be brought in. Right?

So the Air, presumably coming this year, as you know, like, that's a little bit different because it'll be thinner, but it won't be the top of the line one is being rumored. Right? And so, like, for me, that sort of negates the need to get that. But it does, I think, put the plant the seed in people's heads of like, Oh, is that good enough for me? Is that something that I would carry around?

So the Air, presumably coming this year, as you know, like, that's a little bit different because it'll be thinner, but it won't be the top of the line one is being rumored. Right? And so, like, for me, that sort of negates the need to get that. But it does, I think, put the plant the seed in people's heads of like, Oh, is that good enough for me? Is that something that I would carry around?

Do I want the less weight? Am I okay trading off a little bit of battery life if that is indeed going to be the case? Which seems like it has to be given how thin it's rumored to be. But also just the notion of new, right? Like the oldest trick in the book in marketing and branding, right?

Do I want the less weight? Am I okay trading off a little bit of battery life if that is indeed going to be the case? Which seems like it has to be given how thin it's rumored to be. But also just the notion of new, right? Like the oldest trick in the book in marketing and branding, right?

Like, I want an iPhone that looks new so people know that I have the newest one. This was a big issue for a long time in China, I think, in particular, right, where it was like people wanted to make it known that they had the flagship model, the top end, the highest end version because it was a prestige device. And I think that's the same thing is true in Western markets as well to some degree. And so I think people will buy the Air noting that, you know, it looks new and it's a talking point and you can show it off and whatnot. Next year is where things get really interesting with that fold.

Like, I want an iPhone that looks new so people know that I have the newest one. This was a big issue for a long time in China, I think, in particular, right, where it was like people wanted to make it known that they had the flagship model, the top end, the highest end version because it was a prestige device. And I think that's the same thing is true in Western markets as well to some degree. And so I think people will buy the Air noting that, you know, it looks new and it's a talking point and you can show it off and whatnot. Next year is where things get really interesting with that fold.

So I actually have I have right here, which people can't see if we're just on audio, but I have the Pixel Fold right in front of me, the old one, not the new one that they just announced, which isn't out yet. But I so I have it as a backup device. It's what I use for testing Android apps and whatnot and for testing some of Google's AI services that are that are on the device. And I really like the form factor. Now the crease, you know, that people talk about a lot is an issue.

So I actually have I have right here, which people can't see if we're just on audio, but I have the Pixel Fold right in front of me, the old one, not the new one that they just announced, which isn't out yet. But I so I have it as a backup device. It's what I use for testing Android apps and whatnot and for testing some of Google's AI services that are that are on the device. And I really like the form factor. Now the crease, you know, that people talk about a lot is an issue.

It becomes more of an issue the more you open and fold it. And I from all the reporting, again, that's presumably what Apple has been trying to figure out how to alleviate that issue because it doesn't feel like a very Apple thing to have this giant sort of ugly crease in the middle of of a device that you're using. But taking that away and if they can sort of fix that a little bit to the point where it's not as noticeable, I do think the form factor is great. And it sort of feels nice, like, in that it would be, you know, an iPhone that basically can also become, in a way, an iPad. And so they have a nice story there of how to do that.

It becomes more of an issue the more you open and fold it. And I from all the reporting, again, that's presumably what Apple has been trying to figure out how to alleviate that issue because it doesn't feel like a very Apple thing to have this giant sort of ugly crease in the middle of of a device that you're using. But taking that away and if they can sort of fix that a little bit to the point where it's not as noticeable, I do think the form factor is great. And it sort of feels nice, like, in that it would be, you know, an iPhone that basically can also become, in a way, an iPad. And so they have a nice story there of how to do that.

And if that is indeed like at flagship specs, like, that's probably the device I get, and I probably pay 2,000 plus, right, or whatever it's going to cost to get that. And then finally, the third year one, that becomes interesting too because they're releasing a new overall design edge to edge as you talk about glass. And then it becomes like, can if you if you bought the Fold the previous year, and this is again something that I was weighing in my head, can you really go back to sort of having just one the one screen again? And is the newness of it and the new design of it enough to tempt people over? The high level point, though, is that people finally have more than like one choice, more than just size choice basically and price point choice.

And if that is indeed like at flagship specs, like, that's probably the device I get, and I probably pay 2,000 plus, right, or whatever it's going to cost to get that. And then finally, the third year one, that becomes interesting too because they're releasing a new overall design edge to edge as you talk about glass. And then it becomes like, can if you if you bought the Fold the previous year, and this is again something that I was weighing in my head, can you really go back to sort of having just one the one screen again? And is the newness of it and the new design of it enough to tempt people over? The high level point, though, is that people finally have more than like one choice, more than just size choice basically and price point choice.

Now there are a ton of price points and now there's all different sorts of form factor choices, which I think is interesting for Apple.

Now there are a ton of price points and now there's all different sorts of form factor choices, which I think is interesting for Apple.

Yeah. I would say listening to you, the bottom line that I get out of this is that, again, going back to the my setup here, Apple has struggled, really struggled to grow, these, to grow these, these, the iPhone line. And getting these options, is really gonna help grow it. I I think there's no way we go through the next three years and we see iPhone sales flat.

Yeah. I would say listening to you, the bottom line that I get out of this is that, again, going back to the my setup here, Apple has struggled, really struggled to grow, these, to grow these, these, the iPhone line. And getting these options, is really gonna help grow it. I I think there's no way we go through the next three years and we see iPhone sales flat.

Yeah. Yeah. And again, part of it is just new. Right? People will decide that, oh, the new thing is enticing them to to upgrade their devices for the the first time in a few years, right?

Yeah. Yeah. And again, part of it is just new. Right? People will decide that, oh, the new thing is enticing them to to upgrade their devices for the the first time in a few years, right?

That's the other thing. The other major thing at play here that sort of has only been alluded to in this discussion is the fact that a lot of these devices now are so fast that they're fast enough for most people to have for a few years. Like my wife's iPhone, I think, is three years old she and doesn't complain about how slow it is. I mean, you know, maybe if she used to compare it to mine, she would notice a little bit, but they're really fast enough for the most part. And so you need other things to entice people to upgrade.

That's the other thing. The other major thing at play here that sort of has only been alluded to in this discussion is the fact that a lot of these devices now are so fast that they're fast enough for most people to have for a few years. Like my wife's iPhone, I think, is three years old she and doesn't complain about how slow it is. I mean, you know, maybe if she used to compare it to mine, she would notice a little bit, but they're really fast enough for the most part. And so you need other things to entice people to upgrade.

And as we all know, famously last year, Apple thought that perhaps AI would be that thing. Right? They only put Apple intelligence on the latest models that you could do it, and they were hoping, I think, that that's more people would would be enticed to upgrade. But there was nothing, know, enticing enough about Apple intelligence, I think, to get people to do that on a mass scale.

And as we all know, famously last year, Apple thought that perhaps AI would be that thing. Right? They only put Apple intelligence on the latest models that you could do it, and they were hoping, I think, that that's more people would would be enticed to upgrade. But there was nothing, know, enticing enough about Apple intelligence, I think, to get people to do that on a mass scale.

Definitely. And, you know, going, you know, model by model, I do think that the air is gonna be kind of I don't know. I I still think it will lead to an uptick in sales because, as you mentioned, new sales. But it's gonna be a tough sell. I think especially if you know people who are early adopters know that the new stuff is coming the next year and it's gonna be tough to settle for something with low battery life and just one camera.

Definitely. And, you know, going, you know, model by model, I do think that the air is gonna be kind of I don't know. I I still think it will lead to an uptick in sales because, as you mentioned, new sales. But it's gonna be a tough sell. I think especially if you know people who are early adopters know that the new stuff is coming the next year and it's gonna be tough to settle for something with low battery life and just one camera.

Yeah. And it's interesting. It sounds like it's replacing the the Plus in the lineup. Right? And so, like, the Plus was the bigger of the regular, quote unquote, regular iPhone line.

Yeah. And it's interesting. It sounds like it's replacing the the Plus in the lineup. Right? And so, like, the Plus was the bigger of the regular, quote unquote, regular iPhone line.

And so clearly, Apple had the numbers to showcase that, you know, people just weren't buying that model relative to the other ones that were on the market. And so they're trying to slot this in there and seeing if it that'll work. Now they've done stuff like this in the past and it hasn't worked. Right? I I recall, like, when they released, if you remember the iPhone, I think it was five c, which was the color version Yeah.

And so clearly, Apple had the numbers to showcase that, you know, people just weren't buying that model relative to the other ones that were on the market. And so they're trying to slot this in there and seeing if it that'll work. Now they've done stuff like this in the past and it hasn't worked. Right? I I recall, like, when they released, if you remember the iPhone, I think it was five c, which was the color version Yeah.

Where they did all these fun colors. And it was actually, like, a beautiful device in a lot of ways, but it just didn't resonate in the way that they thought it would. They marketed it all over. It was on all the billboards because it was so fun and colorful, you know, in some ways harkening back to the old iMac designs and things like that. But yeah, it just didn't sell in the ways that they had hoped.

Where they did all these fun colors. And it was actually, like, a beautiful device in a lot of ways, but it just didn't resonate in the way that they thought it would. They marketed it all over. It was on all the billboards because it was so fun and colorful, you know, in some ways harkening back to the old iMac designs and things like that. But yeah, it just didn't sell in the ways that they had hoped.

And so you wonder about that here. Obviously, the Air, if they do go with that is the name of it, that ties directly into the MacBook Air, which has long been their sort of, you know, very popular Mac device lineup to the point where they've changed the form factor where it's like doesn't really even seem like the old Air anymore, but they still call it that because and they had to bring it back that name back. Right? Because it was it was so popular. And so does that level lead to some level of, like, brand affinity that people have with their MacBook Airs and they they want the iPhone Air and all those things come into play?

And so you wonder about that here. Obviously, the Air, if they do go with that is the name of it, that ties directly into the MacBook Air, which has long been their sort of, you know, very popular Mac device lineup to the point where they've changed the form factor where it's like doesn't really even seem like the old Air anymore, but they still call it that because and they had to bring it back that name back. Right? Because it was it was so popular. And so does that level lead to some level of, like, brand affinity that people have with their MacBook Airs and they they want the iPhone Air and all those things come into play?

好的。你能再详细说说为什么喜欢Fold吗？我显然见过它的实际使用。有趣的是，当我看到人们使用折叠手机时，他们似乎总是对能展开屏幕观看内容感到特别开心。它主要是个媒体消费设备，还是Fold有其他吸引你的地方？

Yeah. Can you talk a little bit more now going to the Fold about why you like the fold? I've obviously seen it in use. It's funny because when I see people with folding phones, there's they always seem to be pretty happy about the fact they can fold the screen out and watch something on it. Is it mostly a media consumption device, or is there something else about the Fold that that you like?

我想说两点。其一，媒体消费绝对是它的核心用途之一。不过另一个我觉得很有意思的是多任务处理——展开时它就像并排放置的两块手机屏幕。你可以想象成两块拉长的iPhone屏幕并列。就像现在iPad上那样，你可以同时运行两个独立应用，还能更轻松地在它们之间复制粘贴。

I would say two things. One, media for sure is is one of the key use cases of it. The other thing, though, that I think is pretty interesting is multitasking because you can it's basically like having two phone screens side by side, you know, when it's unfolded. And so it's like two elongated iPhone screens you would imagine sort of being there. And so just like you can on an iPad now, you know, you could run two separate applications and and sort of, yeah, copy and paste between them more easily.

所以从生产力角度看，这确实是个有趣的应用场景。另一个方面确实与媒体相关。我认为苹果会巧妙处理像FaceTime这类功能，他们肯定会推出能充分利用这种屏幕转换特性的新版应用。谷歌也做过类似尝试，对吧？

And so I do think that that's like an interesting use case from a productivity perspective. Yeah. And the other stuff is is media related. And I do think that Apple will be smart and clever about how they do things like FaceTime, right, and things like that where I'm sure they'll have new versions of of certain applications that can fully utilize and take advantage of that screen translation, things like that. Google's done stuff like that, right?

比如你可以把设备平放，一边显示内容给这侧看，另一边显示其他内容。我当初是心血来潮买的，想看看是否只是噱头。三星做折叠屏有段时间了，早期确实像噱头，初代机型也有各种问题。但现在它已经发展到相当成熟的地步了。

Where it's like you can put something on the table and see one side and the other person can see the other type stuff. And so it's one of those things where I bought it on a whim just seeing like if it was a pure novelty. You know, Samsung has obviously been doing them for a while, and it seemed like it was a novelty at first. And they had all sorts of issues with those first models. But now it's gotten to the point where, again, I think it's pretty good.

考虑到苹果的iPad是目前市场上唯一成功的专用平板（至少就拥有独立应用生态而言），我认为这会帮助他们。毕竟用户已经习惯iPad的使用方式，当然也熟悉iPhone。如果苹果能巧妙融合这两者的特性，会很有吸引力。不过这会蚕食iPad mini的销量吗？因为两者尺寸相近——尺寸差异到底有多大呢？

And I think the fact that Apple, again, has the iPad as really the only successful, at least in terms of, like, purpose built tablet with its own applications ecosystem on the market right now, I think that that will help them because, again, people are trained sort of on how to use an iPad, and obviously, they're trained on how to use an iPhone. And so if Apple is able to do this delicate dance well between those two and sort of meld those devices together, I think it could be pretty compelling. And there's a question of, like, does that cannibalize certainly, it seems like it would cannibalize some level of iPad mini sales because they would be similar ish Yeah. What's the difference? In size.

我想区别就在于有人不会选择折叠屏吧。我倒不担心这个。确实如此。是的。但除此之外，我觉得人们最终会适应它的。

One doesn't fold, I guess, is the difference. I wasn't worried about that. That's true. Yeah. But otherwise, yeah, I think I think people will eventually get the swing of it.

不过我认为价格会是个大问题。对吧？如果真要2000美元，那可能比你现在电脑的价格还高。人们会愿意承担这个成本吗？苹果能否通过运营商补贴之类的方式来帮忙分摊呢？

I do think though the price will be a big problem. Right? If it's really $2,000 we're talking about more than probably what your computer costs at that point. And so are people going to be willing to to eat those costs and can Apple do anything there with carrier subsidies and whatnot to help offset?

是啊。最后说到对新款的渴望，我觉得曲面玻璃模型——我又在脑海里想象它的样子了。可能有点像...这个比喻有点蠢，但就像MacBook上的触控板，你知道的，就是那种...

Yeah. And then finally, when it comes to wanting the new, I think the curved glass model, again, just have a picture of what it looks like in my mind. Maybe like I'm kind of like picturing like the the, this is stupid, but like the the mouse on the MacBook, like, you know, just how

是的，有一个弯曲的

it's Yeah. Have one curved

像这样。对。就是完全像那样的东西。对吧？比如，也许手机看起来像这样，MacBook鼠标那样，四周都是弯曲的。

like this. Yeah. Something exactly like that. Right? Like, maybe the phone looks like this this, MacBook mouse where it's like curved on all sides.

如果他们能成功推出——我确信他们会的——那看起来会很酷。问题在于它多久会坏一次？屏幕折痕多久会出现？我不确定，但你确实点出了人们需要权衡的关键问题：我到底要不要折叠屏？还是要这款旗舰机？

It's gonna if they pull it off and I'm sure they will, it's just gonna look cool. Question is how often does it break? How often does the air break? I don't know but but I think you really highlighted like the the problem maybe that I don't know the problem the trade off that people are gonna have to think about in their heads which is like, do do I want the fold? Do I want this flagship one?

苹果多久更新一次这些手机？你会来回换着用吗？消费者会有很多疑问，但最终他们可能还是会选择苹果。

How often does Apple update these phones? Do you go back forth between them? A lot of questions that people are gonna have, but ultimately, I think they'll probably just spend with Apple.

没错。不过你提出了个有趣的观点——实体店可能比以往更重要。就像Vision Pro那样，如果首次推出新形态设备，人们肯定想亲眼看看。毕竟没人会不看实物就花2000美元买部折叠iPhone（暂且这么称呼）。

Yes. That's for sure true. But I you actually bring up an interesting point, which I hadn't thought of, that that to that extent, which is that much like with they've done with, obviously, the Vision Pro in in store might be more important than ever because if they are gonna do these new form factors for the first time, people are gonna wanna see them. Like, you don't buy certainly don't spend $2,000 to buy an iPhone an iPhone Fold. You know, again, that's the name we're going with, but who knows what they end up calling it.

折叠iPhone不是那种能闭眼入的产品。除非朋友已经买了让你体验过，否则消费者肯定想先上手。所以实体店的现场演示对推广这类产品至关重要。

But the foldable iPhone, you you don't buy it just out of hand, literally, like, right? Like, and and not knowing, having seen it before. If a friend has it or something, sure. But, like, you want to experience it first. So, like, you would imagine that the stores are pretty crucial then to moving those and having hands on time and demos of what you could do with it.

虽然Vision Pro的体验店效果一般，但就像当年Apple Watch那样，苹果需要搭建场景来展示差异。如今消费者听到'新iPhone'就知道是什么，但面对新形态设备时，这种认知惯性就不存在了。

Much again, it hasn't really worked with the Vision Pro, but much like they had to do with the Vision Pro to showcase how there's a difference. And Apple Watch back in the day, right? They had those setups to be able to do that. And so I think that that is that will be an interesting critical component to that, to this level of having these new choices. Because right now, you know that like, Oh, Apple released a new iPhone.

说到传闻中的二十周年旗舰机，如果能实现真正的无边框设计，把动态岛等功能移到屏下，那就像是实现了iPhone自2007年诞生以来的终极形态——完全就是握在手里的一块玻璃。

I need a new iPhone. I know what it's going to be like because it's been the same for all these years. So it's like a no brainer purchase if you need one. But that's not going to be the case necessarily with these new devices. The only other thing I would say on the flagship one, the twentieth anniversary rumored one, is if they are actually able to sort of make it edge to edge screen and if they can get rid of sort of the dynamic island element of it and move all of that stuff sort of behind the screen and it becomes a real first, like, edge to edge full screen iPhone, like, that's that's sort of just feels like a culmination of what they've wanted clearly from day one, right, of the iPhone.

这就像给iPhone的进化史画上完美句号：我们2007年就想做但技术达不到的形态，如今终于实现了。

And so I think that's a nice little bookend story of, like, this is the ultimate iPhone form factor as we originally set out in 2007 from day one. And, you know, back then, didn't have the technology to do it, and now we have the full it's just a piece of glass in your hand.

嗯。那么核心问题来了：这些足够让人们忘记苹果AI的短板吗？

Yep. Alright. So I guess the big question here from this point is, does this do enough to make people forget about Apple Intelligence?

这问题可能要困扰我们好几年，直到他们解决或放弃。只要新设备能刺激升级，人们自然会淡忘AI问题。你看谷歌刚用吉米·法伦那场尴尬发布会展示Pixel手机——他们在AI有优势，但设备端似乎没充分利用。

I feel like that's a question that we're gonna be answering for several years until they either fix it or don't. I think that it will make people forget about it as long as, again, these things entice them to upgrade. I think that we just saw Google, you know, release the new Pixel phones, at least showcase them with that weird event with Jimmy Fallon, which didn't seem like it went over that well. But, you know, they obviously have a huge advantage in terms of AI. I'm not sure that they're taking full advantage of that, though, from the device perspective.

Pixel手机展示了什么非买不可的独家AI功能吗？翻译之类的小功能是有，但没什么能碾压iPhone的突破性体验。

Like, is there anything that was showcased that you absolutely need to buy a Pixel phone for to utilize and leverage Google Google's AI advantage there. Not that I could nothing that I jumps to mind in in my head. There's things. There's translation. There's there's lots of different features, but nothing that I think would would sort of imply that this device is above and beyond an iPhone.

而且我认为还有另一个潜在因素——人们依然热爱iPhone作为设备本身。它是极其强大的设备，传闻新款机型将配备史上最大内存。当这个世界与那些专为设备优化的AI模型在手机上相遇时（苹果已经能本地运行部分功能了），这可能会产生质变。

And I do think that there's this other element which is lingering in the background, which is, look, people still love the iPhone as a device. It's a super powerful device. The rumor is these new models are going to get more RAM than ever. And if and when that world ever collides with some of the more tailored for a device AI models coming to the phones themselves, right? They're already Obviously, Apple already runs some of the stuff locally.

但当真正强大的AI模型开始能直接在设备端运行，比如即时完成图像生成而无需云端等待时，苹果拥有最强处理器、算力、内存和手持设备的纯粹优势，可能会成为极具说服力的卖点。当然这些现在只是猜测，但我预见到那时这将成为更有利于苹果的论点。

But when real, like real powerful models, AI models start to come and can be on device themselves, and so you can run like you know, you can do image generation in an instant, not like waiting for the cloud and not waiting for it to come back down and things like that. I think that that might become a compelling argument for Apple having just the pure best CPU and horsepower and RAM and device in your hand. That's all sort of guesstimates of, like, you know, how that plays out right now. But I could see a world in which that becomes sort of more of an argument in Apple's favor at that point.

确实如此。我常想到的是，既然我iPhone上就有ChatGPT应用，说明并非无法使用AI。但问题在于深度集成——现在苹果甚至考虑整合更多公司的AI技术到手机中。

That's right. And, I mean, one thing that I always think about is the fact that ChatGPT there's a ChatGPT app on my iPhone. So it's not like I can't use AI on my iPhone. But as this stuff gets more integrated, that's sort of where it becomes the issue. And now we have, Apple looking at, you know, potentially integrating even for even more companies AI into their phone.

最近有消息称他们正与谷歌、Anthropic接触，就像你说的这是场AI供应商间的竞赛，也是苹果技术与它们的比拼。我认为这种尝试整合的做法很明智，未来随时可以更换合作方。

You've about this recently. They're now looking potentially at Google, at Anthropic, so it's as you put it a bake off between a bunch of different AI providers and then a bake off between Apple's technology and them. I I I think this is a good move. Try to find others and integrate it and see how it goes, and you could always swap it in the future if you want to.

这恰好呼应我们刚讨论的内容。谷歌刚举办的Pixel发布会上，Rick Osterloh虽未点名苹果，但暗示某些竞争对手在AI领域承诺过度，而谷歌将实现苹果做不到的个人AI功能——有趣的是他们可能同时在洽谈为Siri提供支持。

And that's interesting in in what we were just talking about. Right? So, like, Google just did this pixel event. Rick Osterlov's up up there on stage basically talking about, like, without saying Apple. At one point, Jimmy Fallon said Apple, which was very amusing in the event in the event because that never happens.

就像你提到手机已有ChatGPT应用，我们也有Gemini应用。如果最终谷歌真的为Siri提供技术支持，而所有AI技术都趋同，谷歌设备的核心卖点究竟何在？这会形成微妙的竞争格局——谷歌既想保留优势又不得不开放技术的矛盾状态。

But Rick Osterlov certainly did it, but he alluded to the fact that they were maybe making promises that they couldn't deliver on in AI. Right? Competitors were doing that. But Google was going to deliver on it on some of this personal AI stuff that Apple couldn't deliver on. At the same time, if they are in fact talking about powering Siri and to your point of like, you have a ChatGPT app on your phone, we also all have, excuse me, a Gemini app on our phone.

说到苹果的供应商竞赛策略——我前两天刚发文质疑（虽然知道你会反对这个观点）：他们是否该进行大型AI收购？这种竞赛模式让我开始反思。The Information有报道指出他们历史上重大并购效果不佳，比如30亿美元的Beats收购案。

Right? And so a lot of this stuff is already available on the phone. So like, what is the actual selling point of Google's devices then if everything is just going over? And if, again, in fact, that they do end up Google ends up powering at least an aspect of Siri, like and it's all the same AI technology powering all these things. Like, it all just blends together and becomes, like, a sort of a confusing ground for where what Google wants to hold back or what they don't wanna hold back and and that weird nature of it.

表面看Beats很成功——运动员随处可见的耳机确实壮大了可穿戴设备业务。但当初他们期待Beats软件能取代iTunes成为云端音乐平台的内核设想，实际上...

But, yes, in terms of the bake off themselves, like and and what Apple's thinking here. So I actually wrote a thing two days ago. Maybe it's even yesterday. I'm I'm sort of blending together. But, basically, I'm starting to to question the notion, and and you won't like this because I know you and I have talked about this, and and we were in agreement on this.

(接上文)并未实现预期效果。这让我重新思考：在AI领域进行巨额收购是否真是最佳路径？特别是当现有竞赛策略已展现出灵活优势时。

But if they should make a giant acquisition in AI, I'm starting to to second guess my thought there. Only because of things like this, like the bake off, and it's like there was a good report in the information about, you know, how they've sort of not done a good job historically with some of their m and a, their giant m and a that they've tried to do. Like, everyone talks about Beats, of course, as the big one. That was the biggest one they've done to date, the $3,000,000,000 deal. And you look at then, you think like, oh, okay.

(最终段落)最终演变成尴尬局面——既没能成功转型云端服务，又暴露出苹果在软件整合方面的短板。这种历史教训让我对当前AI领域的并购策略产生更多疑虑。

That was good. Like, they bought that. Now And they have all these Beats headphones and you see athletes wearing them everywhere. And it's good for the for the hardware headphone business, I guess, which isn't a massive business, but but it's good for them, for their wearables and whatnot. But as it turns out, like, the internal stuff of, like, they thought that that might the Beats software might be able to replace iTunes and and become the iTunes in the cloud.

但听起来，他们基本上不得不重写整个项目，结果还是行不通。内部还存在一些文化冲突。所以当你从AI层面思考这个问题时，如果他们达成的不是30亿美元，而是300亿美元的协议会怎样？我们之前讨论过这个。

But it sounds like, you know, they basically had to rewrite the thing and and and it just didn't work. And there were some culture classes internally. And so you think about this at the level of AI, if they make one of these not just 3,000,000,000, say $30,000,000,000 deals. Right? And and we've talked about it.

比如，如果真这么做，内部文化会变成什么样？新加入的人能和在苹果待了几十年的老员工顺利合作吗？我开始怀疑他们是否真的会推进这类交易。这篇报道提到Eddie Q可能一直在推动他们与Perplexity或Mistral合作，但内部存在大量反对声音，既有苹果历来坚持自主研发的历史问题，也有过去并购中遇到的种种麻烦。回到你之前说的竞标方案，我认为在AI发展初期，竞标确实是更稳妥的推进方式。

Like, what's the the culture like internally if they do that? And can these new people who come on board actually get stuff done with the ingrained people that have been at Apple forever? And so, you know, I'm really starting to wonder if they're not going to do that. Again, this report talks about how Eddie Q maybe has been pushing for them to do one of these with either Perplexity or with Mistral, And there's just a bunch of pushback internally about it because of the, you know, obviously, historic problems that Apple has had with with wanting to do their own things, but also just, yeah, some of the problems that they've had with M and A in the past. And so going back to your point about the bake off, like, I think the bake off thing is a much better way to at least get somewhere in the early days of of all of this.

对吧？不必花费300亿美元。关键在于，竞标传闻中提到苹果可能对Anthropic的要价犹豫不决。如果对方开价数十亿美元，苹果会接受吗？大概率不会。

Right? Not having to spend $30,000,000,000. The question is if one of the key parts of that bake off story was that Apple maybe is balking at the price that, I think it was Anthropic is asking for. And if they're asking for billions of dollars, like, Apple going to do that? Probably not.

不，完全同意。我本周也看到关于苹果对并购持谨慎态度的报道。

No. Yeah. No. You're it's totally right. I I think that really I read that information story also this week about Apple's reluctance to do M and A.

我的疑问是——这周我在CNBC也提到——你想要一个坚信所有好点子都来自内部的公司，还是愿意相信外部也有创新？如果落后了，最佳企业文化应该能快速察觉这点，并建立机制吸纳这些人才。但苹果目前显然不具备这种文化。

And I guess my question is and I said this on CNBC this week, do you want to have a company that believes all great ideas come from within or do you want to be a company that believes that there are good ideas that come from without? And if you're behind the best way to operate your culture is to notice that quickly and you know be create a culture where you can integrate those, you know, the people behind those ideas into your operation. And it's just not what Apple has today.

没错。这正是我认为这类交易的核心价值——未必是为了某个具体产品。虽然很多人关注Perplexity的搜索功能，尤其考虑到谷歌可能随时终止与苹果的合作协议。

Yes. And I agreed like that that is the key that and that's always been the key driver in my mind of why you do this deal. It's not necessarily to get a product though. Obviously, a lot of it has been made about perplexity in particular because of the search element. And if and if, you know, the any minute now, Google might be having to rule, you know, rule to break that deal with Apple.

但我更关注的是能否改变企业文化。就像你说的，给苹果注入新活力。那些任职二三十年的高管们，显然无法适应AI时代所需的发展节奏。引进人才正是为了改变这一点。

And so there was that element of it. But I'm less interested in in how they integrate any of these individual products, like if Mistral's, you know, chatbot and whatnot. I just I I don't think that that's what the path that they'll go down. It's more along the lines of it's interesting if it is able to change the culture to your point and give them, like, this new lease on life and able to refresh what you know, those executives at Apple have famously been there for twenty, thirty, forty years in some cases, and that's just they're not obviously operating at the cadence as we've talked about before that you need to operate in this age of AI. And so that's why you bring the talent on board.

但同样值得思考的是，这些人才可能在那套文化体系里徒劳无功。那么花300亿美元到底图什么？

But, again, there's a lot of thought that that talent will just spin its wheels and do nothing inside of that culture. And so you spent $30,000,000,000 to do what exactly?

确实。最后几分钟，聊聊英特尔吧？美国政府将《芯片法案》部分拨款转为股权，获得英特尔10%股份。

Exactly. Alright. Couple minutes left. You wanna touch on Intel? US is taking a 10%, stake in Intel turning some of the grants that they had that the country had made, as part of the CHIPS Act into, into stock.

你怎么看这个操作？

So Yeah. What's your view here?

要知道，我们正从特朗普政府那里得到这些奇怪而有趣的交易。比如英伟达和AMD在销售上做出让步，现在又获得了公司股份、黄金股以及其他非科技企业的权益。这些交易都在桌面上进行着。英特尔那笔尤其引人注目，毕竟——委婉地说——面对台积电和AI时代的冲击，英特尔一直举步维艰，他们错失移动市场的机会等等问题众所周知。

There's you know, these are we're getting these these strange interesting deals from the Trump administration. We got we got, you know, NVIDIA giving cuts and AMD giving cuts of their sales. Now we got stakes in companies and, you know, golden shares and other non tech companies and things like that. So it's all on the table and all these deals are being done. The Intel one is fascinating, obviously, because Intel has been struggling, to put it lightly, right, in the face of TSMC and the age of AI, and they were struggling because they miss mobile, yada yada.

但他们确实需要根本性的变革。虽然换了CEO，但业务模式必须改变才能真正扭转局面。而美国政府出于地缘政治考量（比如台积电等因素），有强烈动机推动这件事。最耐人寻味的是，通过这几天的细节披露可以看出，股权交易中明确约定英特尔不得剥离其制造业务——此前一直有传闻说他们会外包制造来削减成本，让英特尔重回增长轨道。

But they really needed they need something to change fundamentally. And obviously they changed their CEO, but they need something to change about the business to actually make it work. And The US is incentivized to try to make that work, right, for all the geopolitical reasons, you know, with TSMC and whatnot. And so the taking of the stake thing, the most interesting offshoot of that, which only came out over the past couple of days, which was sort of baked into one other thing, is that it clearly has been negotiated so that Intel is not going to sell off their manufacturing business, which has been talked about, right? That they would just outsource that and potentially to cut costs and get sort of Intel back in line and growing again.

显然美国政府不希望看到制造业务外流，视其为战略资产。现在关键是：获得这笔投资后（虽然政府声称不强制绑定战略合作），英特尔至少需要一两个合作伙伴来利用其晶圆厂产能。若特朗普政府通过10%股权促成此事——比如苹果宣布采用英特尔代工芯片——股权价值必将飙升。

But it's clear that the US government does not want them doing that. They view it as strategic. And now the real question is if with this investment and they're saying like it's not tied to necessarily any sort of strategic deals with other partners, but they need a partner, at least one and probably a few to use the foundry capabilities that they have. And so can the Trump administration, now with their 10% ownership stake, would love to see that be worth more and more. Certainly would if Apple comes in and says, We'll use their foundry to do a chip.

亚马逊或微软若加入也会以'多元化'名义表态，声称需要分散供应链。但众所周知，苹果所有芯片业务都与台积电保持着共生关系，其他巨头情况类似。如果特朗普政府的持股能帮英特尔签下大单，既保住晶圆厂又自救成功，虽算因祸得福，但这笔交易本身仍很诡异，开了个糟糕的先例。

And it certainly would if Amazon comes in or if Microsoft comes in they'll say, It's all in the name, Oh, it's all in the name of diversification, and we need to be able to have multiple irons in the fire and chips being made in other places. But they have All of Apple's business, as you know, right, with the chip stuff has been in TSMC and it's been a symbiotic relationship, right? And same is true with a lot of those other companies. So if the Trump administration getting their ownership stake allows Intel to cut some major deals to basically save the foundry and save themselves, like there's silver lining to it, but it's a weird deal. It's a weird situation and not a good precedent.

至于交易合法性以及特朗普离任后可能引发的法律纠纷，目前都是未知数。

And who knows the legality of it and how it's gonna be litigated if and when Trump is no longer, you know, in charge of the country.

是啊，疯狂的时代。本以为八月是全年最淡的月份，结果——

Yeah. Crazy times. I thought August was supposed to be the slow month of the year. Turns out

才不是。

Nope.

过去四周的科技新闻密度简直抵得上半年份。我已经等不及看九月的风暴了，期待十月份再和你详聊这些事件的来龙去脉。Emji，再次感谢做客节目。

We've just had, like, you know, I don't know, enough tech news over the past four weeks to fill at least six months. So I can't even wait, for what September brings and, I'm looking forward to having a conversation again with you on the October to break it all down what's happened and what's coming. So Emji, great again to have you on the show. Thanks again for coming on.

一如既往谢谢你，Alex。

Thank you as always, Alex.

各位听众，感谢收听。本周三我将邀请Precision Neuroscience高管团队，探讨脑机接口技术的实际应用与未来前景。下次《科技前沿》播客再见！

Alright everybody, thank you so much for listening. On Wednesday, I will have the executive team of Precision Neuroscience to talk again about the science and the actual applications of the brain computer interface today in production and where it might be going next. Thanks again for listening, and we'll see you next time on Big Technology Podcast.