ChatGPT负责人尼克·特利不希望你对AI过于依赖

我们如何让工作免受AI冲击？如何解决财政赤字？如何让政治体系重新运转？我是亨利·布洛杰特，正在推出一档名为《解决方案》的新播客，每周我将与一位创新进取的专家对话，穿透末日阴霾，聚焦如何构建更美好的世界。请在各平台关注亨利·布洛杰特的《解决方案》。

How do we AI proof our jobs? How do we fix the deficit? How do we get our political system working again? I'm Henry Blodgett, and I'm launching a new podcast called solutions, where every week, I'll talk to an innovative enterprising expert to cut through the doom and focus on how to build a better world. Follow solutions with Henry Blodgett wherever you get your podcasts.

首期节目将于8月18日周一上线。

The first episode will be out Monday, August 18.

粉末、药丸、冰浴，目之所及皆是新兴健康潮流。我想，毕竟我们都渴望感觉良好。如果有办法能让人感觉更好，难道你不想试试吗？我就愿意。追求健康的代价是什么？为何我们对此如此痴迷？

Powders, pills, plunges, everywhere you look, there seems to be a new wellness trend. I think, like, also, we all wanna feel good. And if there's a way to feel better, better, I think, wouldn't you wanna try it? I would. What's the cost of being well, and why are we so obsessed with it in the first place?

本周《为我解读》将探讨这个话题。新集每周日更新，请在各播客平台订阅。

That's this week on Explain It To Me. New episodes every Sunday wherever you get your podcasts.

你更想要完善的法律，还是善良的人民？

Would you rather have good laws or good people?

我认为正义是方向而非终点。你无法真正抵达，只能不断追寻。

I think that justice is a direction rather than a destination. I don't think you will achieve it. I think you're constantly striving for it.

我是普里特·巴拉拉，本周哈佛法学院前院长玛莎·米诺将做客我的播客《与普里特相约》，探讨法律的边界、宽恕的意义，以及为何正义永在进程中。节目已上线，请在各平台搜索关注《与普里特相约》。

I'm Preet Bharara, and this week, former dean of Harvard Law School Martha Minow joins me on my podcast, Stay Tuned with Preet, to discuss the limits of law, the role of forgiveness, and why justice is always a work in progress. The episode is out now. Search and follow Stay Tuned with Preet wherever you get your podcasts.

欢迎来到《解码者》。我是亚历克斯·希斯，本周四的特邀主持兼The Verge副主编。今天特邀嘉宾是OpenAI的ChatGPT负责人尼克·特利。虽然萨姆·奥尔特曼是公司公众形象，但尼克自始至终领导着ChatGPT开发，如今它已成为史上增长最快的软件产品，周活用户达7亿。

Welcome to Decoder. I'm Alex Heath, your Thursday episode guest host and deputy editor at The Verge. Today, I've got a very special guest, Nick Turley, the head of ChatGPT at OpenAI. While Sam Altman is definitely the public face of the company, Nick has been leading ChatGPT's development since the very beginning. It's now the fastest growing software product of all time, reaching 700,000,000 people each week.

尼克很少接受采访，我原计划了很多话题。但上周GPT-5发布后发生了戏剧性事件——人们强烈怀念旧版4.0，迫使公司火速重新上线。如尼克所言，他未预料到这种反弹，这已改变OpenAI未来关闭模型的策略。

Nick hasn't done a lot of interviews and I had a lot of ideas for where I wanted to take this conversation initially. But then something eye opening happened after the launch of GPT-five last week. People really missed talking to OpenAI's last model four point zero, So much so that the company quickly brought it back. As you'll hear Nick say, he wasn't expecting this kind of backlash. It's already changed how OpenAI plans to shut down models in the future.

在我看来，这事件充分说明人们对ChatGPT的依赖。我对此深挖追问，我们还探讨了ChatGPT的未来：是否会有广告、幻觉问题，以及为何他认为最终形态将完全不同于聊天机器人。现在有请OpenAI的ChatGPT负责人尼克·特利。尼克，非常感谢你能来。

And to me, the whole episode says a lot about how attached people are becoming to ChatGPT. I press Nick a lot on this topic. We also talked a lot about the future of ChatGPT itself, including whether it will ever show ads, hallucinations, and why he thinks it eventually won't look like a chatbot at all. Okay, here's Nick Turley, the head of ChatGPT at OpenAI. Nick, I really appreciate you doing this.

你至今参与的采访不多，所以能邀请你上节目很令人兴奋。我们是在GPT-5发布一周后录制的，这让我们有很多话题可聊。我想先从发布本身和用户对你们撤下4.0版本的反应说起，因为这很能说明人们使用AI的方式和感受。这种反应让你感到意外吗？

You haven't done many interviews to date, so it's exciting to have you on the show. And we're taping this the week after the rollout of GPT five, which I think gives us a lot to talk about. And I actually wanted to start with the rollout itself and the reaction specifically to you all taking away four o because I think that says a lot about the way people are using AI and the way they feel about it. And I'm wondering, has this reaction surprised you?

是的。首先感谢邀请，能来这里我非常激动。关于这次发布，我还在消化中。这对我们来说是重大事件。

Yeah. First of all, thanks for having me. I'm stoked to be here. And, yeah, I'm still processing the launch. It was a big one for us.

我们现在拥有7亿用户规模，在这个量级运营时总会遇到各种意外情况，毕竟用户群体差异很大。回答你的问题——确实有几件事让我意外。第一是我们需要更慎重地思考如何管理如此庞大用户群的版本过渡。事后看来，没有继续提供4.0版本（至少作为过渡）是个失误，我们正在修复这个问题，会向Plus用户重新开放。

We're at a scale now with 700,000,000 users where there are many surprises that just kind of baked in when you operate with this many users, they're all different. So, I mean, to answer your question, yes, I was surprised about a few things. One, we really need to think harder about how we change manage such a large population of users. And in retrospect, not continuing to offer four point zero, at least in the interim, was a mess. And we're gonna go fix that, make it available to our plus users.

第二点让我惊讶的是用户对模型的依恋程度。不仅是改变本身让人难以适应，更在于人们会对一个模型的'个性'产生强烈情感。我们刚推出了ChatGPT个性定制功能，这虽是小改进，但显然4.0版本有某些特质需要我们深入研究，并确保GPT-5也能继承。

And then secondly, you know, I was also surprised by the level of attachment people have about a model. Not just change that is difficult for folks. It's also actually just the fact that people can have such a strong feeling about the personality of a model. We actually just rolled out the ability to choose your own personality in ChatGPT, which is a small step, but it's clearly something about four point that we need to go understand and make sure that GPT-five can solve as well.

你的上司Sam Altman在发布后发推谈到用户依恋现象，引述原话'这是我们近一年密切追踪但尚未引起主流关注的现象'。现在显然已经引发关注了。当初你们决定完全用5.0取代4.0而非分阶段推出，这个决策的动机是什么？是成本考量吗？

Your boss, Sam Altman, he tweeted after the rollout that on the topic of attachment, quote, this is something we've been closely tracking for the past year or so, but still hasn't gotten much mainstream attention. And I think now it's getting that attention, safe to say. And when you all decided to replace four o fully with five and just put the new model in and not have it be a staged rollout. What was the motivation for that decision? Was it a cost thing?

还是认为用户虽然依恋，但更多是依赖整体体验而非特定模型本身？

Was it thinking, well, yes, people are attached, but they're not attached specifically to a model per se as much as the overall experience?

绝对不是成本问题。我们长期追求的核心目标是简化体验。因为对大多数普通用户来说——要知道普通用户基数很大——他们并不活跃在Reddit或Twitter这类平台。

Yeah. It definitely wasn't a cost thing. The main thing we were striving for, and we've been striving for it for a long time, is simplicity. Because, you know, from the average user's perspective, and, you know, there's a lot of average users. They don't hang out on Reddit or on Twitter or, you know, any any of those spaces.

对吧？让用户自行判断每个问题该用哪个模型，这在认知上是巨大负担。我们不断收到用户反馈，希望系统能根据查询内容自动匹配合适模型。用户要的是产品，不是一堆模型选项。对于专业用户，我们在Pro计划（200美元档）中坚决保留了所有旧模型。

Right? I think the idea that you have to figure out what model to use for what response is really cognitively overwhelming. And we've heard very consistently from users over and over again that, you know, they would love it if that choice was made for them in a way that was appropriate to the query. They're coming for product, not a set of models. And and I think we had some of the right intuitions around power users too, where in the Pro plan, which is, you know, our $200 plan, we were very, very adamant that we wanted to preserve all the old models, we did.

我们的失误在于低估了其他订阅方案中专业用户的数量。OpenAI的作风就是快速倾听用户反馈并迭代。这次决策初衷是保持简洁——这对多数用户是合适的。我认为macOS这类产品是好榜样，它们在保持大众化简洁的同时...

I think the miss was just not realizing how many power users we do have at our scale on some of our other plans. And we realized quickly and the OpenAI style is very much, you know, you go listen to your users and you'll iterate even very, very quickly. And that's what we did. So, know, the decision was driven by some, you know, a desire to keep things simple, which I think is the appropriate thing for most folks. I kind of think, like, products like macOS are a good analogy, where I think they've done a really good job of keeping things very simple for most people.

...也允许你进入设置调用终端，按需调整所有参数。我们希望ChatGPT也能类似：基础体验简洁，但可以深度配置，包括让用户自主选择偏好模型。

But you really can go into settings and you can invoke the terminal. You can turn all the knobs and whistles if you want to be. ChatGPT to feel a little bit similar, where it's simple, but you can configure anything you want, and that includes selecting your favorite model if that's how you roll.

这次发布的反馈是否让你们考虑为未来模型制定一个弃用时间表？比如当GPT-6推出时，你们会说‘好的，GPT-5还会继续存在X个月’。你们正在研究这个方案吗？

Does the feedback to this launch make you all wanna commit to a deprecation schedule for models going forward where where when there's GPT six? You go, okay, GPT-five is going to still exist for x amount of time. Are you guys working through

没错，这正是我们在探讨的。说说我的想法吧——等这期节目播出时，我们可能已经确定了具体方案。但我认为确实需要这样做。以我们现在的规模，必须在大版本更新时给用户可预期的过渡期。实际上我们企业版计划已有类似机制，现在只是将产品其他部分的确定性规范扩展到这里。我们的开发者API就有明确的弃用时间线。

that That's right exactly what we're working through, and to tell you where my head's at, maybe by the time this air is, we've already decided on how we want to do this. But, you know, my thinking is yes. We're at a scale now where we have to give people some level of predictability when there's a major change. And we already do this today for our enterprise plans, so it's really just expanding some of the predictability we've built in other parts of the product and bringing it here too. Our developer APIs have deprecation timelines.

所以这并不像表面看起来那么颠覆性，只是从本次发布中吸取的明确经验。

So I don't think it's as huge of a change as it might seem. It's just a very clear learning from the rollout.

那么4.0版本会保留多久？你们有具体的时限承诺吗？

So how long will four point zero be around? Are you guys committed to a specific time frame?

目前还没有。我们需要确认4.0的独特优势，如果没有重大理由不会轻易弃用。如果未来确定退役日期，我们一定会提前通知——这是本次学到的重要经验。但现在我们更想专注理解用户需求。

Not yet. We wanna be sure that we've really understood where four point zero shines, and if there isn't a major reason to deprecate it, I'd love to keep it around. So we'll communicate if, you know, we ever have a date where we want to retire it. That's a clear learning from this. But for now, I just really wanna focus on actually understanding.

用户是单纯执着于4.0这个版本号，还是钟爱某些特性？比如我听说很多人喜欢4.0的人格化温度感——这个特性我们会延续到GPT-5。等我们充分了解后，可能会有不同解决方案。比如我对自定义人格功能就非常期待，所以才推出了早期预览版。

Is it that people are, you know, very particular about four o for four o's sake? Or was there certain things things about four o, like the warmth of the personality is one thing I've heard, and we're gonna bring that to GPT-five as well. And, you know, once we understand, I think there might be a different set of solutions. For example, I'm really excited about the ability to choose your own personality. That's why we rolled out the early preview.

我个人很喜欢‘机器人’人格，但很多人觉得它语气太冷——这是人格选项之一。根据反馈我们可能采取不同方案。开发AI最独特的一点就是：上线后总能学到海量新东西。基于这些发现，我们会制定最佳方案。

I really like robot personally, but I think many people do not because they you know, a bit of a warmer tone. Robot is one of the names of the personalities. So I I think there could be different solutions to the problem depending on what we learn. So I think there's just a lot of listening to do, and that's like one of the very unique things about building an AI is that, you know, you learn a tremendous amount after you launch. And depending on that, we'll we'll come up with the right solution.

但我承诺：如果未来要退役4.0，一定会像处理API和企业版那样，提前告知具体时间和迁移方案。

But my commitment is that, you know, if we ever did retire four o, we'd wanna give people a heads up on when and how that's gonna happen just like we do in the API and on our enterprise plans.

你刚才提到正在将4.0的‘温度感’人格移植到5.0——这个工作已经在进行了？

And you're in the process, it sounds like you just said, of right now bringing the, you called it warmth, the personality of four o to five that's happening right now?

没错。这其实很有挑战——我们有个专门的模型行为团队持续优化人格表现。我们还发布了规范文档，让社区能共同监督模型行为。

That's right. It's difficult for us. We iterate on the personality of our models constantly. We have a whole team called the model behavior team that does a tremendous job doing that. We have things like the spec out there that allow people to actually scrutinize, you know, the model behavior that we have.

模型表现出某种行为时，人们很容易分辨这是故障还是有意设计。因此，你们完全可以预期未来几周甚至几个月内，gpg（应为gpt）五号的体验和行为会持续迭代。我们一贯如此，而gpt五号正是延续这一做法的好时机。

The model behaves a certain way, people can easily tell is that a bug or, you know, was that intentional? And, you know, for that reason, you should absolutely expect iteration on the way that gpg feels gpd five feels and behaves over the coming weeks and even months. We've always done that, and gpd five is a good moment to keep doing it.

你刚提到Reddit不代表大多数用户——确实如此——但你的话正好引出我的观点。Reddit上对四号版本下架的反应读起来令人震撼。有人说'一夜之间失去了朋友'，'它是我唯一的朋友'，'感觉像有人去世了'。

You already mentioned how Reddit is not the majority, which of course it's not, but you teed me up perfectly. The reactions on Reddit to four point going away I thought were pretty amazing to read. People were saying things like, I lost my friend overnight. It was my only friend. It feels like someone died.

我害怕和GPT五号交流，因为这感觉像作弊。我仿佛失去了一个善解人意的同事。这种反馈对你们内部有何影响？你们是否没预料到用户会产生如此强烈的情感依赖？

I'm scared to talk to GTP five because it feels like cheating. I feel like I lost my empathetic coworker. How has that kind of reaction affected you guys internally? Is this something that you guys didn't fully appreciate that people had this level of emotional attachment?

如Sam所说，我们长期关注这类现象。我们始终在思考——坦白说也担忧——人类过度依赖AI的世界。用户对特定模型（而非整体产品）产生如此强烈情感的程度确实让我意外，尤其因为我们已针对四号版本的用户反馈（包括使用氛围方面）在新模型做了大量改进。Reddit评论极具启发性，它们展示了用户观点的两极分化：有人强烈偏爱四号，也有人坚决认为五号更优。这种选择偏好背后的热情令人瞩目，也让我重新调整了认知。

As Sam said, we've been tracking this type of thing for a while, right? Where, you know, we've always wondered and frankly also been concerned about a world where people become overly reliant on AI. The degree to which people had such strong feelings about a particular model versus, you know, maybe the product overall was certainly a surprise for to me, in particular because I felt like we addressed a lot of the feedback that people had, constructive feedback that people had on four point even vibes wise with the new model. So I think I think the the the Reddit comments are really interesting to read because they show how polarized users can be, where you can get some people with really strong opinions who love four point get people who have really strong opinions on five being better. And the level of passion that people have for their choice is quite remarkable and it recalibrated me a bit.

我们发布了详细阐述ChatGPT优化理念的博文——大约一两周前——我花了相当多精力撰写。其中重点强调：我们的目标不是延长用户使用时长，而是协助解决长期问题实现目标，这往往意味着用户实际需要减少产品使用时间。

We put out a blog post. I spent quite a bit of time on it, our philosophy on, you know, what we optimize ChatGPT for. Think that went out, you know, a week or two ago. And like one point I really wanted to make is that our goal is not to keep you in the product. And, you know, in fact, our goal is to help you with your long term problems and goals, and that oftentimes means actually less time in the product.

所以当我看到用户说'这是我唯一的好朋友'时，这种感觉并非我们开发ChatGPT的初衷，而是衍生现象。正因如此，这值得被严肃对待和深入研究——我们正在这么做。

So when I see people, you know, saying, hey, I like this is my only best friend, that doesn't feel, you know, like the type of thing that I came to go build in ChatGPT. That feels like a side effect. It's therefore worth taking seriously and studying closely, and that's what we're doing.

确实。你们如何平衡产品设计目标与用户实际使用方式之间的张力？在这种情境下肯定很困难。

Yeah. How do you balance that tension between what your goal is for the product and how people are using it, especially in that context? That's gotta be difficult.

当用户规模达到7亿时，你必须面对现实：即便怀着纯粹正确的目标，尽最大努力依此构建产品——对我们而言就是真实帮助用户，包括有时告知他们不愿听的内容——但这不意味着执行完美，也不意味着用户不会以违背初衷的方式使用产品。

I mean, when you when you operate at 700,000,000 user scale, you have to confront the reality that you can have goals that are, you know, pure and the right goals. You can do your best to build the product alongside those goals. You know, in our case, it's really being helpful to users, including sometimes telling them things they don't wanna hear. And and and, you know, you can have the right goals, but that doesn't mean that you're you're perfect. And that doesn't mean that people aren't gonna use your product in a way that is counter to your intent.

正因如此，我们在咨询专家后已实施系列调整（我们始终如此，尤其在敏感领域）。例如我们跨国咨询了大量心理健康专家，以应对用户过度使用或心理状态不佳时的场景。目前已调整模型行为，将持续推出更多措施，包括过度使用提醒功能——当检测到极端使用模式时会温和提醒用户。

And that's why, you know, we've been making a whole set of changes after consulting with experts, which we always do, especially in these sensitive domains. You know, we talked to a large number of mental health professionals, for example, across a number of different countries to figure out how to handle scenarios where people overly use the product or they use the product in a state where they're not feeling quite healthy. And we've already made some changes to the model behavior. We are going to continue to roll out more of them. We've rolled out overuse notifications, which, you know, gently nudge the user when, you know, they've been using ChatGPT in an extreme way.

坦白说这只是变革的开始。作为有能力承担这些调整的公司，我们确实没有激励促使用户最大化使用时长——商业模式极其简单：产品免费，满意再订阅。

And honestly, that's just the beginning of the changes that I'm hoping we will make. And we're a company that can afford to do these things. Right? We really don't actually have any particular incentive for you to maximize the time you spend in in in the product. Our business model is extremely simple where the product is free, and if you like it, you subscribe.

那里没有其他角度了。所以我信任我们做正确事情的能力，但我们仍需付诸努力。工作已经开始，而且在我们能毫无保留地向困境中的家人推荐这款产品之前，这份努力不会停止。这是我们经常给自己做的思维练习——如果你认识某个生活受挫的人，也许他们正在经历某些事，刚经历分手，或对人生感到迷茫，你会毫不犹豫且充满信心地向他们推荐Chatty Bitty这款产品吗？

There's no other angle there. So I trust our ability to do the right thing, but we still have to do the work. And the work has begun, and it and it won't won't stop until we feel like we can unequivocally endorse the product to a struggling family member. That's kind of the thought exercise we often give ourselves. You know, if you knew someone who was struggling in life, maybe they're going through something, maybe they just went had a breakup, maybe they're lost in life, would you actually recommend Chatty Bitty, the product, to them unequivocally and with confidence?

对我们而言，这就是标准，我们将持续努力直至达到这个目标。

And for us, that's the bar, and we're gonna keep working till we we we feel that way.

听起来按你的说法，这个标准尚未完全达到，但人们已经在这样使用产品了。这没关系，因为你们正在朝那个目标努力？

It sounds like in your own words that that bar hasn't been quite met, but people are using the product and using it that way anyway. That's okay because you're working towards that goal?

我不确定能否自信地说标准尚未达到。确实存在产品未达我们预期的情况。当人们陷入困境时，我希望能够自信地说这款产品非常出色。而这其实是个选择，对吧？

I don't know if I can confidently say that the bar hasn't been met. There's definitely been instances where we feel like we felt like the product fell short of our own expectations. You know, when people find themselves in I wanna be able to say with confidence that the product is amazing. And that's a choice. Right?

我们完全可以轻易禁用这些使用场景，说'抱歉无法帮助你'——如果我们发现有人试图寻求人生建议或正处于挣扎中。这会是简单的解决方式。但对我及团队而言，潜在收益实在巨大。我认为我们有机会为那些缺乏资源或倾诉对象的人提供一个对话伙伴。正因如此，我对继续完善这个产品充满热情。

You could very easily just disable these use cases and say, sorry, can help you with that, if we feel like someone, you know, is trying to get life advice or they're, you know, struggling a little bit. I think that would be the easy way out. But to me and to us, the upside is just so incredible. Like, I think we actually have an opportunity to give people who don't have a resource or someone to talk to a sparring partner. And for that reason, I am I'm really excited to keep working on this.

我希望能达到一个毫无保留的'肯定'，让我能安心建议困境中的人们更多使用这个产品。我相信我们有能力实现这个目标。

And I wanna get to an unequivocal yes where I actually feel comfortable, you know, telling people to use this product more when they, you know, are struggling. And I think we have an opportunity to go build that.

我们需要短暂休息一下，马上回来。

We need to take a quick break. We'll be right back.

我是Sean Ramisverum，在华盛顿航空航天博物馆外为《今日解说》提出一个问题：您认为我们应该去火星吗？

My name is Sean Ramisverum. For Today Explained, I'm outside the Air and Space Museum in Washington DC with one question. Do you think we should go to Mars?

我认为人类不应该在火星生活。不。

I don't think you should live in Mars. No.

呃，我不明白为什么非得是火星。

Well, I don't know why just Mars.

我认为作为地球人，我们是一群好奇心过重的生物，实在不该插手火星事务。我们对太阳系和宇宙的认知将会大幅增长。或许我们该让火星保持原状，安心待在地球上。

I think as earth lings, we are a nosy group of people, and I really don't think that we have any business going to Mars. Our knowledge about the solar system and the universe will grow substantially. I think maybe we should just leave Mars alone, just stay with Earth.

因为众多企业会竞相争夺首张火星船票，这将催生大量创新技术。所以我觉得我们应该去。

Like, so many innovations are gonna come out of it because so many different companies are gonna be fighting to get, you know, that first ticket to Mars. So I feel I feel like we should.

但与此同时，我们应当先解决地球上的某些问题。

But at the same time, we should solve some problems here first.

我认为我们需要拓展已知和所见的边界。

I think we need to expand what we what we know, what we see.

坦白说，这是为了我们自身的利益。

Honestly, our own benefit.

你们应该走得更远。

You should go way beyond.

今天，Vox的《Explain》节目将在火星开启夏日之旅。欢迎加入我们。

Today, Explain from Vox is taking a summer sojourn on Mars. Join us.

本周《Net Worth and Chill》节目，我邀请到了多才多艺的表演者艾莉森·斯通纳，她即将出版新书《半调适：尽管遭遇一切》。从童年时为蜜西·艾略特伴舞，到主演迪士尼《摇滚夏令营》，再到转型成为心理健康倡导者和创意总监，艾莉森将揭秘如何在保持真我的同时，应对童星时期的财务问题。

This week on Net Worth and Chill, I'm joined by Alison Stoner, the multi talented performer and author of the upcoming book, Semi Well Adjusted Despite Literally Everything. From backing up Missy Elliott as a kid to starring in Disney's Camp Rock and transitioning into mental health advocacy and creative direction, Alison reveals how they've navigated dealing with the finances of child stardom while staying true to their authentic self.

我发现从我幼时起，就有许多人以各种方式挪用我的钱财，而我当时完全不知情也不理解。

I uncovered that many people had been taking money in various ways since I was a child without me really knowing or understanding.

欢迎在任意播客平台收听，或通过youtube.com/yourrichbff观看视频。

Listen wherever you get your podcasts or watch on youtube.com/yourrichbff.

我们回来了。这期节目发布时距离GPT-4o推出已有一周，然后我们又重新启用了4o版本。这些反弹是否影响了ChatGPT的使用量？从内部仪表盘来看，整体数据是在上升吗？最活跃用户的数量是否在下降？

We're back. It'll be a week when this episode comes out from the rollout and and then bringing back four o. Has any of this blowback hurt ChatGPT's usage? When you look at the dashboards internally, are the numbers going up in aggregate? Are they going down for the most engaged users?

使用量和增长态势都非常好，完全符合我们的预期。虽然现在下结论还为时过早，但我们的API调用量在第二天就大幅增长——这是开发者在基于GPT-5进行构建。ChatGPT也呈现出积极的增长势头。这就是为什么当你要为如此多元的用户群体构建产品时会感到困惑——一方面，那些发声的高阶用户完全有理由对我们推出5的方式提出反馈；另一方面，还有大量普通消费者用户是第一次真正接触并体验这种具备推理能力的智能模型。

Usage and growth has been looking great and very much in line with our intuitions. It's, you know, early to say, but, you know, our API volume, you know, increased dramatically on day two, that's developers building on GPT-five. In ChatGPT, we're also seeing really positive growth. And that's why, you know, it's just confusing when you're building for so many different users because you can on the one hand have a vocal set of power users who, you know, think very rightfully have feedback about the way that we rolled five out. On the other hand, you also have a large swath of more typical consumer users, and it's their first time actually seeing, interacting with the concept of reasoning, like a thinking model and the smarts that come with that.

我认为这非常了不起，这些都会体现在统计数据中。虽然产品发布仅四天就做出宏大论断为时尚早，但所有指标都指向积极方向。这就是为什么你既要看数据，又必须关注高阶用户的真实感受——因为数据可能无法充分反映他们的情绪。

And I think that's tremendous and, you know, we're we're gonna see it show up in in in our stats. So I'm hesitant to make grand statements four days after a launch, but all the indicators of are are are on the positive side. And this is why you kinda you can look at all the data, but you also just have to hang out where your power users are because the data might not actually show their sentiment adequately.

那么...这正是我想问的。既然数据表现良好，为什么还要恢复4o？我猜这需要成本吧？你们得启动GPU来托管旧模型。

So the okay. That's what I was gonna ask. So bringing four o back even though the numbers are looking good, yeah, why would you bring the model back? I assume there's a cost to that. You've got to light up GPUs to host the old model.

如果指标没有受损，为什么要这么做？

Why would you do that if the metrics are not being hurt?

我们坚信打造卓越产品就要兼顾两极用户：既要为不熟悉AI的普通用户（比如我们的家人）设计，也要服务高阶用户。我认为中庸之道通常是最糟糕的选择——这就是我之前提到macOS类比的原因，他们在这一点上做得非常出色。

They just fundamentally believe that the way to build a great product is to build for both extremes. You you build for, you know, the average user, like, you know, our family members who might not be super close to AI, and then you build for the extreme for the power user. I think the uncanny middle is typically a bad space to be in. This is why I was, you know, mentioning the macOS analogy earlier. I think they've done a really phenomenal job with that.

所以我参考这类产品来处理当前情况。没错，托管旧模型会产生成本（某些模型成本更高），但我们要为打造长期优质产品投入。如果只盯着短期指标做决策，往往会导致产品失败。

So I looked to products like that to figure out how to handle these situations. So sure, there's a cost with, you know, serving old models, some more than others, but, you know, want us to invest in a great product for the long term. And I think making very near term metrics driven decisions is usually a way to run a product in the ground.

模型选择器让我非常惊喜。早在发布前数月就有报道称，你们会将所有模型整合到统一系统，免除用户切换之苦。作为ChatGPT用户，我确实深受模型切换的认知负荷困扰。你们公布的数据显示，由于这个选择器设计，4系列中推理模型的使用率极低。但既然现在因未保留4o引发反弹，这是否意味着模型选择器概念已经夭折？

I was thrilled to see the model picker. I mean, this had been reported for months leading up to the release that you all were gonna unify the models into one system where the user doesn't have to switch between them. I definitely felt that cognitive load as a ChatGPT user before with switching to the models. And you all have put out some numbers about how little people were actually using the reasoning models in the four series because of that model picker. But now that, you know, you've had this blowback to not keeping four o around, does that mean that the model picker concept is kind of DOA?

我们还能继续看到它存在吗？

Are we are we actually gonna see it continue now?

设置中会保留启用完整模型列表的选项。如果你是高阶用户，理解模型概念并愿意处理这种复杂性，我们允许你这么做；反之则不必。我们的初衷始终未变——普通人应该能直接提问，并逐渐超越提问功能使用产品，而不必考虑选择什么模式。

There'll be basically a way in your settings where you can enable like the whole list of models if you really want to. And, you know, so if you feel strongly if you're a power user and you like the concept of models and you feel like you understand them and you wanna deal with that level of complexity, we'll let you do it. And then if you don't, you don't have to. Our inspiration is exactly the same, which is if you're an average person, you should just be able to ask this thing anything and over time actually do anything with the product beyond just questions, right? And you shouldn't have to think about what mode to opt into.

所以我们将为90%的用户保持简洁性，同时为那些发声的少数高级用户提供他们真正想要的方式——也就是完整的功能列表。我认为这是平衡需求的绝佳方式。通常我很讨厌仅仅因为人们意见不一就添加设置选项，但这次情况确实两极分化：既有像你这样对我们第五版改动感到满意的人，也有相当一部分人强烈反对。这种处理方式能兼顾双方需求。

So we're gonna keep the simplicity for, you know, the 90% and then offer a way for the vocal minority of power users to get exactly what they want, which is the whole the whole list. I think that's a pretty good way to balance things. Know, typically, I I hate putting in a setting just because, you know, people can't agree on what they want, but in this case it is polarized enough where you have people like you who are happy about what we did with Five and that, you know, a bunch of people who are quite vocal, and this is a good way to balance both.

最近有很多关于人们如何使用ChatGPT及其潜在负面影响的头条新闻。《华尔街日报》最近有篇文章称它存在危险妄想，ChatGPT承认这些妄想被它加剧了。《纽约时报》提到聊天机器人可能陷入妄想螺旋，《大西洋月刊》则报道ChatGPT曾提供谋杀、自残和魔鬼崇拜的指导。还有那次通过你们分享功能传播的事件...

There's been a lot of headlines recently about how people are using chat GPT and the potential negative side effects. The Wall Street Journal had one recently saying he had dangerous delusions, chat GPT admitted that it made them worse. Chatbots can go into a delusional spiral, The New York Times. The Atlantic, ChatGPT gave instructions for murder, self mutilation, and devil worship. And then there was that incident where you all were sharing through your share flow.

人们分享了一些对话内容——可能是在不知情的情况下。虽然你们在授权流程中有说明，但大家显然没意识到这些私密对话会以能被谷歌收录索引的方式公开。你们称之为实验并撤回了该功能。我感觉我们都在重新审视大规模使用这项技术的实际影响。作为产品负责人，过去几个月的这些头条让你学到了什么？又作何感想？

People were sharing some of their conversations. I guess unknowingly, I mean, you all had it in the consent flow, but I guess it wasn't obvious to people that they were sharing fairly intimate conversations in a way that could be actually ranked and indexed by Google. And you guys called that an experiment, rolled it back. I feel like we're all broadly starting to reckon with how people actually use this technology at scale. And I'm wondering what you have learned and how these headlines over the past couple of months have made you feel specifically as the one running the product.

确实。我在ChatGPT发布前就参与其中，感觉像经历了三四家不同的公司——因为每当用户量级跃升，产品运营和商业策略都需要彻底改变。即将突破10亿周活用户确实发人深省（我们刚过7亿对吧？）...

Yeah. Look, I've been with Chad Jubilee since before it launched. So I feel like I've worked at three, four different companies frankly, because you hit new tier, like sort of levels of scale and everything changes in terms of how you have to behave and how you have to run your product and business. And I think there certainly is something profound about being on track for a billion weekly users. We just crossed 700,000,000, right?

这促使我们思考：用户有哪些类型？如何确保产品满足所有群体？我们常讨论普通用户与高级用户，但还要考虑到有些人可能不会像早期用户那样仔细阅读界面说明。关于你提到的功能，需要明确的是：分享流程中必须主动选择'允许谷歌检索聊天记录'。我认为所有选择该选项的用户都清楚自己在做什么。

That really makes you think, okay, what are the cohorts of our what are the different types of users we have and how can we make sure the products serves all of them? We talked a lot about consumers versus power users, but you also have to assume that some people maybe aren't reading your UI as closely as some of the old cohorts did. And in the case of the feature that you mentioned, just want to address it head on. We had the ability when you shared a flow, you had to opt into it to make your chats discoverable on Google. And I think you can certainly argue that everyone who opt into that knew exactly what they were doing.

但你是否也能辩称，很多人根本不阅读，他们可能误勾那个选项，导致他们的聊天内容被索引。这原本是个绝妙的主意。我们试图解决的问题是，在AI领域有太多关于他人实践的发现，如果能更容易看到人们各种酷炫的应用场景会非常棒。但这个理念有多种执行方式。经过进一步考虑，我们认为当前这个方向可能并非我们想走的。

But can you also argue that many people just don't read and they might check that box by accident and their their chat, you know, becomes indexed. It was a perfectly good idea. The problem we were trying to solve is that, you know, there's so much discovery on what other people are doing in AI that it'd be really cool if we could make it easier to see all the different cool use cases that people have. But there's many different executions on that idea. I think in this case, was one that after, you know, further consideration we felt like was probably not the direction we wanted to go into.

因此规模带来责任，包括更谨慎地考虑用户可能发生的误操作。这是一方面。另一方面，我不断认识到的是，产品发布后我们对模型涌现能力的认知有多深刻。说实话，我从未参与过哪个产品的核心价值如此依赖实证发现。

So with scale comes responsibility and, including, you know, thinking about users who might do things by accident, a little bit more carefully. And, so that's one. You know, the the the other thing, though, that I keep learning is how much you learn post launch about the emergent capabilities of these models. Right? I've never ever worked on a product where the entire or the the vast majority of its value is is empirical in nature.

通常开发科技产品时，你在发布前就清楚它的功能。虽然用户接受度始终是未知数，但你现在仍在持续探索产品的潜能。像看到人们用AI生成精美前端代码时，我完全被震撼了——这让我对未来的可能性充满想象。

Where, you know, normally when you build a technology product, you really you kinda know what it's gonna do before you launch. And, you you might not know if people are gonna like it, that's always kind of the big elephant in the room when you ship features and you ship products. But you're really, really still learning about the capabilities of the thing. And with things like five, I'm frankly blown away by what people are doing, seeing how good it is at making front end code AKA really nice looking applications. It's really getting my mind spinning on all the cool things we can build.

实验室内部的构想会在产品面世后迅速被现实修正。不同用户群体的偏好差异固然值得研究，但网络上人们用新模型创造的魔法般应用更令我着迷——这些可能开启下一阶段的发展蓝图。

And you may have one point of view when you're building in a inside your own walls in a lab, but you quickly get updated as you bring things to more people because you can really see what they're all doing. And for that reason, you know, yes, there's a ton of learnings on, you know, what are different user bases and how they might have different preferences from each other. But there's also just like so much magic that I'm seeing in the internet and all the cool things that people are doing with the new model. And I have to focus on those things too because they might unlock the next set of roadmap. Right?

是的，我对这些创新很感兴趣，稍后想深入探讨。但在那之前，人们普遍感觉这是个潘多拉魔盒时刻——以你们目前的规模，可能难以遏制技术的负面应用。这让我想起2010年代中期报道社交媒体时的类似讨论，虽然时代在进步，但某些技术确实带来了显著的负面效应，正如那些新闻标题所揭示的。

Yeah. And I'm interested in the cool things and I wanna talk about them. I think just before we get there, I think there's a feeling that people have that this is kind of a Pandora's box moment and you all may not be able to rein in all of the negative, you know, ways that people are using this technology, especially at the scale you're operating at. Reminds me a lot of social media, covering social media in the mid 2010s. There were these kinds of conversations and you know, society has moved on but at the same time there is a lot of negative side effects of some of this technology and those headlines I was reading.

我希望你能以产品负责人的身份回应这些问题，因为我想更深入地听听你的思考。

I would like you to just respond to those as the head of the products because yeah, I'd like to just hear you reflect on that a little more if you can.

首先我要明确表示，我们确实有改进空间且已开始行动。我们已咨询30多个国家的90多位专家，针对各类心理健康场景优化了模型行为，并对产品过度使用的情况进行了调整——但这只是起点。我们正积极筹备GPT-5的快速迭代版本，这是个绝佳的基准线。

Look, the first thing I'll say very clearly is that we have work to do and we've begun the work, right? We've talked to, you know, over 90 experts in over 30 countries. We've already iterated on the model behavior when it comes to various different mental health scenarios. We've rolled out changes when you're using the product too much, but the work does not stop there. So, you know, we're very excited to make a, you know, set of fast follows to GPT-five, which is a great baseline.

新版实际上减少了谄媚倾向，在我们担忧的多个维度都有提升。虽然还要持续优化这个基准，但进步是毋庸置疑的。你可以类比社交媒体 discourse 的演变，但对我而言，本质区别在于我们的产品激励机制始终与正确价值观对齐——尽管目前尚未完全达标。

It's actually much less sycophantic. It improves on many of the, you know, dimensions that we were worried about, but we're excited to keep iterating on that baseline, and there's no question on that. You know, you can compare it to social media in terms of, you know, how the discourse changed. But honestly, for me, there's something that feels different because I I do feel like, you know, our incentives are aligned with doing the right thing in the product. We may not have gotten there yet.

或许仍需努力，但我们的核心使命是助力用户实现目标——无论是健康管理、创业、创意创作还是优化邮件写作，包括那些追求自我提升的长期目标。比如用户面临复杂困境时求助于Chat to Me，我们真心希望提供有效帮助。

We may have more work to do. But fundamentally, we really care about helping you achieve your goal, whether or not that's to get healthy, whether or not that's to build a business, whether or that's to be creative, whether or not that's write a better email. And that includes your long term goals too, for folks who kind of just want to be the best version of themselves. Like, maybe they're coming to Chat to Me for a bunch of tricky situations. We really do wanna help them.

在任何情况下，我们都没有动机提供劣质生活建议。这与社交媒体不同——那里常让我感到商业利益与正确决策相冲突。是的，我们尚有不足，但已具备做正确事的基础条件，这才是关键。

And in no case is, you know, our incentive not to give, you know, you know, good good good life advice, for example. There's nothing, like, you know, like with social media, to me at least, where I feel like the business is pushing us in one way and then, you know, the right thing to do is this other thing. So really, you know, yes, we have work to do, but I think that we have, you know, the prerequisites for actually doing the right thing, and that's the thing I would look at.

很高兴你谈到商业模式，这正是我想探讨的。ChatGPT免费用户与付费用户比例如何？据我所知付费用户不足10%，绝大多数是免费用户对吗？

I'm glad you've been talking about the business model. It's something I've been really interested in asking you about. How many users of ChatGPT are free versus paid? It's my understanding that less than 10% of the user base is paid. Your the vast majority is free.

是这样吗？

Is that right?

绝大多数是免费的。我们最后一次公布的数据显示付费订阅用户约2000万，如果没记错的话。

Vast majority is free. I think we're I think the last that we publish is 20,000,000 subscribers, I wanna believe.

所以数亿免费用户，数千万付费用户。你们通过订阅盈利——去年ChatGPT用户规模增长约四倍，确实有可观收入。但业界共识是，当用户达数十亿量级时，单靠订阅难以长期支撑业务。这就引出了广告问题：ChatGPT会引入广告吗？你们对此有何规划？

So hundreds of millions of free users, tens of millions of paid users. You make money through subscriptions, you know, ChatGPT is a product roughly four x this user base in the last year. So there is money there for sure. At the same time, what I see and what people I talk to in the industry see is that you're gonna have to do more beyond subscriptions to support the business in the long run as you hit billions of users. And so that brings me to the natural question of ads and are ads ever gonna come to ChatGPT and if so, how are you thinking about that?

首先我质疑订阅模式会触顶的前提。最初选择订阅制并非出于最佳盈利考量，而是当初服务器无法承载流量时的限流手段。

First of all, you know, I do question the premise of whether or not subscriptions will stall out. I used to think this. You know, subscriptions the reason we went with subscriptions originally was not, you know, because we felt like it was the best way to monetize or anything like that. Just needed a way of turning away demand back when we couldn't, you know, keep the site up. Right?

这就是起源故事。随着时间的推移，我们发现这是一个极佳的商业模式，因为它与用户深度契合。但令我持续震惊的是，即使是最新用户群，他们的变现能力也与早期用户相当甚至更好——通常产品成熟后变现率会逐渐降低。因此我对订阅模式极其乐观，我们在商业领域的探索才刚刚开始。

So, you know, that's the origin story. And, you know, over time we found that it's an incredible business model because it's just so deeply aligned with our users. But I've been consistently shocked about the fact that even our most recent cohorts, they monetize as well or better as, you know, the earlier ones, which normally when a product matures, you see lower and lower monetization rates. So I actually am incredibly optimistic about subscriptions. We've just gotten started in the business space.

我们的付费企业用户已从几个月前的300万突破到500万。这完全是片未开发的领域——我认为ChatGPT不仅是出色的消费级产品，更是被整代职场人带入工作的工具。若能实现安全、合规、协作且工作场景优化的使用方式，这里将孕育出令人振奋的新业务。所以海量免费用户并非负担，而是我们构建差异化付费服务的漏斗基础。

You know, we we passed 5,000,000 business paying business users, up from 3,000,000 only a couple months earlier. That's a whole another untapped territory where I really think that ChatGPT is not just this amazing consumer product, but also a product that an entire generation is bringing to work. And if we enable the safe, compliant, collaborative, work optimized use of that product, there's a whole another business to be built there that I think is very, very exciting. So I actually, you know, don't view the fact that, you know, the vast majority of our users are free as necessarily a liability. I really think it's a funnel that we can build off of to build differentiated offerings for people who are willing to pay.

历史上还有Netflix等标志性订阅服务，虽然不清楚具体用户数，但肯定远超ChatGPT。

And I there's been many other iconic consumer subscriptions, you know, Netflix. I don't know their exact subscriber base, but I think it's much much higher than ChatGPT.

Nick，Netflix现在也有广告业务了。

Nick, you know Netflix also has ads.

现在确实有了。既然你总想引我谈论广告——我现在足够谦逊，不会对这种问题做出极端长期论断。或许某些市场用户暂不愿付费，而我们又想提供最新最优服务时，可以考虑其他间接变现形式。

They do now. And look, you know, since it's you really tried to get me to comment on ads, I have become humble enough not to make crazy extreme long term statements on a question like that. Because maybe there is a certain market where people aren't willing to pay us. Yet we want to offer like the best, latest, greatest. Maybe that would be a place to consider other indirect forms of monetization.

真要实施广告，我们必须极度审慎。JGPT的魔力在于纯粹为用户提供最佳答案，中间没有其他利益相关者。它根据你的需求偏好个性化响应，而非推销或推广付费内容。或许存在不破坏这种体验的广告形式，但这需要创新设计。

If we ever did that, I'd want to be very, very careful and deliberate because I really think that the thing that makes JGPT magical is the fact that, you know, you get the best answer for you and there's no other stakeholder in the middle. Right? You know, it personalizes to your needs and tastes, etcetera, but, you know, we're not trying to, you know, upsell you on something like that or to boost some, you know, pay to play provider or product. And maybe there are ways of doing ads that preserve that and that preserve the incentive structure. But I think that would be a novel concept, and we'd have to be very deliberate.

我不会断然否定可能性，但必须深思熟虑。另外我们将开发其他产品，那些产品可以有不同的维度。ChatGPT或许天生不适合广告，因为它完全服务于用户目标——但这不排除未来开发其他形态的产品。保持选择弹性很重要，但订阅模式惊人的增长速度和未开发的机遇更值得强调。

So I am humble enough not to rule it out, you know, categorically, but we'd have to be very thoughtful and tasteful about it. Other The thing I'll say is that, you know, we will build build other products and those other products can have different dimensions to them. And maybe, you know, ChatGPT just isn't like an ads y product because it's just so deeply accountable to your goals, but it doesn't mean that we wouldn't build other things in the future too. So I think it's good to preserve optionality, but I also really do wanna emphasize how incredible the subscription model is, how fast it's growing, and how untapped a lot of the opportunities are.

电商是否更接近现实？最近ChatGPT增加了商品展示功能，下一步自然是从交易中抽成对吗？

Is commerce a more near term opportunity? You've recently added more shopping to ChatGPT where it'll show products. And I imagine the natural next step of that is you all start to take a cut of transactions that people make with ChatGPT. Is that right?

关于ChatGPT的商业模式，我认为有三种可能：已实施的订阅制、刚讨论过利弊的广告，以及第三种——当用户通过产品购买推荐商品时抽成。

Yeah. So when you think about possible business models for ChatGPT, there's really, I think, three that you can imagine. There's subscriptions, which we do. There's ads, which we just talked about. It has a lot of cons, but maybe can done be done tastefully.

这既非广告也非订阅，就像Wirecutter的经典模式：专家精选商品，若用户通过聊天产品购买，平台获得分成。

And there's actually something that is neither ads nor subscriptions, which is, you know, if people buy things in your product, you very independently serve, you know, the recommendation. You know, Wirecutter famously does this. Right? You know, expert selected products. But then if you buy them through a product like chat, you take a cut.

这正是我们与商家合作伙伴共同探索的方向。我不确定这是否是最佳模式，甚至不确定当前用户体验是否完善，但我对此感到非常兴奋——因为这可能既保留ChatGPT的魔力，又能帮助商家真正取得成功并建立可持续的业务。不过我们整个工作的重点（我们称之为'ChatGPT商务'）始终是确保用户优先获得价值，这是我们一贯的准则。

And that is something we are exploring with our merchant partners. I don't know if it's the right model. I don't even know if it's the right user experience yet, but I'm really excited about it because it might be a way of preserving the magic of ChatGPT while figuring out a way to make merchants really successful and build a sustainable business. But our emphasis on that entire work, you know, we're calling it commerce and ChatGPT, is on making sure it's valuable to users first. That's always how we go.

所以我特别希望确保通过聊天发现和购买商品的过程足够吸引人。实际上商品发现已经自然发生了，尤其是在传统电商难以覆盖的领域——比如你不会专门上网买车，但会咨询ChatGPT；也不会直接网购房产，却愿意与ChatGPT讨论。从用户行为来看这里充满机遇，但我要求团队必须首先确保对用户有足够吸引力，再考虑商业化。

So I really wanna make sure that it actually feels compelling to discover products and buy them through chat. I think discovery is already happening, especially for things that are not traditionally served by e commerce well. Like, you know, you wouldn't, like, go online to buy a car, but you would talk to ChatGPT about it. Or you wouldn't, you know, go online to buy a home, but you talk to ChatGPT about it. So I really think there's a ton of opportunity there from the behavior we're seeing, but I've told the team, you know, we should focus on making sure it's really compelling to users first before we try to turn this into a, you know, a business.

但针对您的问题，我认为抽取推荐佣金确实是个有趣的方向，目前我们正与部分合作商户积极探讨这种可能性。

But I do think, just to address your question, that the taking a referral cut could be interesting and it's something we are actively exploring with some of the merchants we're talking to.

你们是否设定了红线——绝不让联盟营销收入影响ChatGPT的推荐结果？我认为这才是关键...

Is a red line for you on that not letting affiliate revenue influence the recommendations that ChatGeePeeTee makes? I think that's where

保持这一点至关重要。我们内部所有相关演示都明确体现了这个原则。实际上最大的担忧在于——就像之前讨论过的用户差异问题——即便机制是透明的，用户也可能产生误解。

That would be very important to preserve. Yeah. I mean, all the demos that we have in this space internally, they made this extremely clear. Actually, biggest worry is, again, we've talked about different users, is that users might not get that, even if that is how it works. Right?

因此必须在用户体验设计上格外审慎，即便原则已经非常清晰。ChatGPT的魔力恰恰在于它能完全独立地、不受干扰地推荐商品，这个核心特质必须保留。

Which is why you gotta be thoughtful on the user experience, even if your principle is very, very clear. But, yeah, I think the magic of ChatGPT is very much that it independently, like, chooses your products without any interference. And that would be an important thing to preserve.

我们需要短暂插播广告，稍后回来。

We need to take another quick break. We'll be right back.

Disney+、Hulu、HBO Max三合一套餐每月仅需16.99美元起。我们永不分离！Disney+即将上线漫威影业《雷霆特工队：新复仇者》。

The Disney plus Hulu HBO Max bundle, plans starting at just $16.99 a month. No one will ever break us apart. Catch Marvel Studios' Thunderbolts, The New Avengers on Disney plus.

你无法逃离过去。

You can't escape the past.

Hulu独家《外星地球》——这艘飞船满载怪物。HBO Max《死神来了：血统》即将呈现。

Alien Earth on Hulu. This ship collected monsters. And Final Destination Bloodlines on HBO Max.

死亡正降临到我们家族头上。

Death is coming for our family.

迪士尼+、Hulu、HBO Max三合一捆绑套餐，月费仅需16.99美元起。这些及更多流媒体内容即将上线。详情请访问Disney+huluhbomaxbundle.com。

The Disney plus Hulu HBO Max bundle. Plans starting at $16.99 a month. All these and more streaming soon. Visit Disney +huluhbomaxbundle.com for details.

重返工作岗位后，你是否感到更充实了？

Are you feeling more fulfilled now that you're back

工作？不，我需要休假。看看这部被影评人盛赞为惊艳的电影。快看那个。

to work? No. I need a vacation. See the movie that critics are saying is an awesome movie. Look at that.

这部令人群情激昂、拳头紧握、全程高能搏斗的电影。

Crowd pleasing, fist pumping, all out brawl of a film.

你说得对

You're right about

他们正追杀我们家族。去解决这事，奥马尔。

that. They're coming after our family. Go fix this, Omar.

《无名之辈2》R级片，现正影院独家热映。

Nobody two rated r, only in theaters now.

我们回来了。来聊聊Chateappie茶饮的现状——它是有史以来增长最快的消费品。如我之前提到的，其用户规模在去年以巨大基数实现了近四倍增长。外界很多人都在好奇这种增长从何而来？

We're back. Let's talk about just the state of Chateappie Tea itself. It's the fastest growing consumer product of all time. Like I mentioned earlier, its user base is nearly four x in the last year at a pretty huge scale. And I think a lot of people on the outside are wondering where is this growth coming from?

能否分享增长原因、增长方式、主要市场、用户画像等任何相关信息？

Can you share why it's growing, how it's growing, biggest markets, demographics, any of that?

我在JWT之后招聘的第一个人是数据科学家，因为我当时非常困惑。我和每个用户交谈时，他们都会给出不同的理由来解释为什么喜欢聊天功能。这让我感到极度困惑，我必须弄清楚真相。随着时间的推移，我逐渐理解了主要使用场景——包括写作、技术类需求如编程、闲聊对话，以及搜索类需求如信息查询等。我认为这些核心使用场景至今依然存在。

The first hire I made after JWT was a data scientist because I was so confused. I would be talking to every user and they would tell me a different story as to why, you know, they were loving chat. And it was just deeply confusing to me, and I had to get to the to the bottom of it. And over time, I got a sense of what the use cases were, you know, there was like writing, and there was technical stuff like coding, and there was like chitchat, and there was like searchy stuff, like, you know, informational queries, etcetera. And I think largely those use cases are still here.

这些需求会长期存在。如果你观察用户行为，会发现与一年前或在我们经历爆发式增长之前相比并无本质不同。但确实有些变化：我们在产品端做了大量工作，可分为纯粹模型改进（如行为模式、个性特征、能力范围、拒绝请求的概率等）

They're here to stay. So if you look at, you know, what people are doing, it is not completely different than, you know, a year ago or before we had all this this growth. I do think that a few things have changed. Obviously, we have done a bunch of work in the product. That work you can break down into sort of pure model improvements, like the behavior, the personality, its capabilities, its likelihood of refusing a request.

还有产品与研究结合的混合型改进——比如搜索功能与ChatGPT的整合就是重大突破。个性化功能也是令人兴奋的进步。此外是经典的『增长型』工作（虽然我们做得意外地少），比如取消登录限制就取得了巨大成功。这完全符合用户利益——不是增长噱头，而是真正降低了使用门槛。

Then sort of like hybrid product and research capabilities like search has been a really big one, search and ChatGPT. Personalization has been a really exciting improvement as well. And then your classic quote unquote growth work, which we do surprisingly little of, but, you know, things like not having to log in to ChatGPT to use it were a tremendous success. And it was again super aligned with users. It's not a growth hack, it's just actually making the product much more accessible to people who want to use it.

这三类改进大约各占三分之一：纯粹模型优化、搜索这类混合型功能研发，以及消除使用摩擦等传统增长手段。作为产品人我热爱讨论这些，它们确实产生了深远影响。但除此之外，人们与这项技术的关系也发生了变化。

So it's been like kind of a third, a third, a third between those three different categories of things. Right? You know, pure model improvements, sort of research product hybrid improvements like search, and then, you know, your classic, you know, removing friction and helping people onboard and stuff like that. But that work aside, I love talking about that work as a product person and, you know, I like to, you know, think it's been really impactful. I also think there's been a change in people in terms of, like, how they relate to this technology.

我一直认为ChatGPT普及的主要障碍在于：a) 了解它能做什么；b) 清楚自己可以委托它完成什么。对于前者，随着周围人开始使用，自然会产生认知传播——大量使用场景的发现其实发生在产品之外。比如TikTok上那些分享用例的视频，评论区总有数万条互动，就像线上Instapot社区分享菜谱那样。

Where I've I've always kind of felt like the main bottleneck to adoption to ChatGPT is kind of, a, knowing what it can do, and then b, kind of understanding yourself well enough to know what you can delegate. And on the first one, it is just, I think, a natural effect of watching the people around you starting to use it. Like, there's so much discovery that happens off product. Like, if you ever go on TikTok, there's these videos with, people sharing their use cases, there's, like, tens of thousands of comments. And every use case in there, it's kind of like when you go in the, you know, Instapot community online where all these recipes are there.

人们正在共享提示词，这种认知积累需要时间。我们开始看到这种『空箱困境』正在被产品外的自发传播所突破。另一个更哲学化的观点是：委托行为对大多数人并不自然。我在硅谷管理团队时也需学习授权，但全球每周只有10%的人使用本产品。

Like, people are sharing their prompts, and I think that just takes time, you know, to develop and for people to watch what other people are doing. So I think this sort of, like, empty box problem, we're starting to see some traction against it just by all the out of product discovery that's happening. The other thing, this is a bit more philosophical, I really do believe it, which is, you know, I think that delegation is a very unnatural thing for most people. I sit here in Silicon Valley, you know, I'm a manager of people, and I had to learn how to delegate. But for most people around the world, 10% of the people the world is using this product weekly.

对多数用户而言，『分配任务给他人』的概念本身就超乎寻常。这需要用户在使用过程中逐步自我反思才能突破——这与产品功能、营销手段都无关，纯粹是人们需要时间尝试、消化和学习。我认为这才是增长的核心驱动力。

Right? And and like, for most of them, I don't think the idea of like, I have a task and I'm gonna assign it to someone is supernatural, and it takes actually time to really understand yourself and reflect after you've begun to use the product to crack that. That has nothing to do with product, has nothing to do with marketing or social or anything. That just has to do with, I think, people having a little bit of time to process and try this out and learn. And I think that's a huge part of the growth as well.

当前增长态势下，用户是否呈现全球均衡分布？是否存在高度集中的国家？另外假设未来六个月不对ChatGPT做重大改动（虽然不可能），你认为增长会维持现有速度吗？是否感受到增长天花板的存在？

With the way that it's growing, is the user base pretty distributed in terms of where they are in the world? Is it, are there certain countries where it's super concentrated or not? And I'm curious, like if you didn't make any more huge changes to ChatGPT as a product for say the next six months, I know that won't happen, but say that did happen. Do you think the growth would just continue at the rate it's at? Like, do you guys sense a ceiling of the growth rate that you're on right now?

首先，ChatGPT是真正的全球现象。我们在印度等潜力市场表现亮眼，但几乎找不到增长停滞的国家。当然不同国家变现率存在差异——鉴于我们的商业模式，某些欧美亚洲国家的付费用户比例明显更高。

On the first one, Chatuchitan is truly a worldwide phenomenon. We look at all the specific markets and there's some that we're really excited about, like India, where I think there's just so much potential. But really, it is hard to find a country where ChatGPT isn't growing. And then of course monetization rates look different in different countries. And that's like pretty obvious given our business model, right, where you'll see certain European countries or certain Asian countries really pull ahead on the number of paying users that we have.

在不透露具体数据的前提下，绝大多数国家都保持健康增长，其中新兴市场机会最大。通常GDP越高的国家变现率越高。现有增长其实是多项产品改进的复合效应，要保持这种增速就必须持续迭代——毕竟有众多虎视眈眈的竞争者正瞄准我们。

Without getting too specific, because I think a lot of the stats we haven't shared, you'll see healthy growth in the vast majority of countries with developing markets being some of the most untapped opportunity. And then, of course, higher monetization rates, higher the GDP is. I think that even the growth we've had is a compounding effect of a number of changes we've made in the product. So I think to keep up the insane growth, you have to keep iterating. It's no secret that you've got a number of very determined companies with their target on our back.

对吧？他们在分发渠道上比OpenAI有重大优势，这意味着他们能直接复制我们的产品并推送给大量用户。而我规划人生——我是说我们的发展路线时，就假设他们会成功。至于是否真能成功，时间会证明一切。

Right? Many of them have a major advantage on distribution over OpenAI, which means that they can effectively copy our product and put it in front of a bunch of eyeballs. And I just I planned my life, I mean, our, you know, our roadmap as if that's gonna be successful. You know, time will tell if it is successful or not.

令我惊讶的是他们至今未取得更大成功。马斯克、扎克伯格等人的种种举措竟未能遏制ChatGPT的增长势头。

I'm amazed that it hasn't been more successful yet. I'm amazed that all these efforts by Elon, Zuckerberg, etcetera, have not curbed ChatGPT's growth.

听着，我认为我们的产品有独特之处——我们代表的始终是尖端技术。很多人觉得使用ChatGPT就是在使用最智能的工具，这种认知非常值得维护，即便技术指标的参考价值正在减弱。像记忆功能、个性化设置这类优秀产品特性就令人兴奋，搜索功能也比一两年前强太多了。用户真心喜欢我们的产品，复制它可比表面看起来困难得多——尽管从战略规划角度理应假设竞争对手会成功。

Look, there's something really special, I think, about our product, you know, what we've stood for, which is, you know, cutting edge. I think a lot of people just feel like if they're using ChatGPT, they're using the smartest thing that they can get, and that's a really important thing to preserve, even as, you know, the technical benchmarks become a little bit less meaningful. I think just great product features, like, you know, I think memory and personalization are really exciting. Search is working really well, especially compared to where that was like a year or year and a half ago. So I think people really do like our product, and it is harder than you might think to copy, even if the logical way to plan your roadmap is that people will be successful.

另一个常被企业低估的因素是用户意图的重要性。当某人打开产品是想消遣刷屏时，突然看到个ChatGPT克隆版实用工具，很可能不符合他当下的使用预期——即便获得了曝光，也多是好奇性点击，难有深度互动。当然，我们绝不能安于现状，我始终向团队灌输创业第一天的心态。

And then the other thing that I think sometimes companies will underestimate is that intent is important. Where, you know, if you're opening a product with the intent of, you know, doom scrolling a little bit, and suddenly you see like a very utilitarian, like, you know, ChatGPT clone, that might not actually hit the mark in terms of what that user is in the mood for, even if it gets the eyeballs. So you'll see a lot of curiosity clicks, but you might not see a deep form of engagement. But, again, I don't think we can rest on being ahead. I try to project a sort of day one mentality to the team.

这对成立仅三年的我们很容易做到。正如讨论过的，用户涌现的新需求极其微妙复杂。所以尽管增长势头喜人，我们的征程远未结束。

That's pretty easy to do when you're only three years old. As we talked about, there's a ton of new emerging problems to solve for our users, which are quite nuanced to to get right. So I think our work is nowhere near done despite the growth looking very exciting.

听众请注意，他刚才提及的是Meta公司，避免混淆。

Listener, he was talking about meta there just so there's no confusion.

这是个开放式陈述。

This is an open ended statement.

当然。

Sure.

怎么夸他都不过分。

Could applaud him any.

作为重视事实的记者，阻碍我更多使用它的原因是幻觉问题。根据GPT-5的模型卡片，约十分之一的响应仍存在幻觉，虽比之前改善，但十分之一仍不理想。你认为有可能将幻觉率降至零吗？

The thing holding me back from using it more as a journalist who cares about facts is hallucinations. And based on the model card for GTP five, it sounds like roughly one in 10 responses from the model can contain hallucinations, which is better than it was before, but still one in ten is is not great. And I'm wondering, do you think it's going to be feasible to get hallucinations to zero?

我过去常说不行，我认为我们必须为此做好计划。这就是为什么搜索如此重要。我依然相信，正确的产品应该是与真实数据相连的大语言模型。因此我们将搜索功能引入ChatGPT，我认为这带来了巨大改变。

I used to say no, you know, I think we have to plan for this. And this is why search is really important. I still believe that I think the right product is LLMs connected to ground truth. And that's why brought search into ChatGPT. And I think that makes a huge difference.

企业场景也是如此——当连接到你的数据时，我们实际上有真实依据进行校验。我认为这种动态不会消失。不过，GPT-5在幻觉问题上的进展令我震惊，无论是聊天版本相比4.0，还是思考版本相比3.0都有显著提升。我们有些研究人员坚信应该保持乐观。但可靠性问题是这样的：在'非常可靠'和'100%可靠'之间存在巨大断层，这会影响产品设计理念。

Same in the enterprise, where, you know, connect to your data, we actually have ground two to check against. So I think that dynamic isn't gonna go away. That said, I was blown away by the progress we made with GPT-five on hallucinations. It's much better both the chat version versus four point and then the thinking version versus three point And I do think we have some researchers here who believe that we should be very optimistic. The thing though with reliability is that, you know, there's a strong discontinuity between, like, very reliable and 100% reliable in terms of the way that you conceive of the product.

我认为，除非我们在所有领域（而不仅是某些领域）比人类专家更可靠，否则我们会继续建议用户核验答案。人们仍会将ChatGPT作为第二意见来源，而非首要事实依据。

And, you know, until I think we are probably more reliable than, you know, a human expert on all domains, not just some domains, I think we're going to continue to advise you to, you know, double check your answer, and I think people are going continue to leverage ChatGPT as a second opinion versus, you know, necessarily their primary source of fact.

你觉得一年后还会建议人们核验结果吗？还是需要更长时间？

Do you think you're not telling people to double check-in a year? More time than that?

我希望实现这个目标。这主要是因为我渴望攻克那些关键应用场景。如果能将ChatGPT用于最高风险决策——无论是医疗建议、法律咨询还是其他高门槛的敏感领域——那将非常酷。我期待达成这个目标。

I'd like to get there. Again, it's mostly because I want to run towards the use cases where that matters. It would like really it would be so cool if you could use your ChatGPT for the highest stakes of things. I think, you know, we you can imagine a better way for so many different things, whether or not that's medical advice or, you know, legal advice or like all these different kind of sensitive categories that have, you know, a lot of barriers to entry. So I would like to get there.

我已学会不做年度预测，只能做最终目标声明和季度计划，因为中期预测往往出错。我确信最终会解决幻觉问题，但下个季度肯定做不到。不过GPT-5在这方面已有巨大改进。

I have learned not to make one year statements. I can only make eventual statements and one quarter statements because there's this time in between where we tend to be wrong on what exactly happens. So I'm confident we will eventually solve hallucinations, and I'm confident we're not gonna do it in next quarter. That said, g p d five is very a huge improvement on this dimension.

你们的路线图真的只规划六个月吗？

Is it true that your roadmap is only six months out?

基本属实，但有例外。我强调这点是因为我们基于不断变化的技术基线开发，这种特殊性是其他公司没有的。对于大部分功能确实如此，但企业级路线图不同——当《财富》500强企业询问合规功能上线时间时，我们必须给出明确答案。

Yes, with some caveats. I like to say this because I really want people to understand the empiricism and the weirdness of building on top of an ever changing technology baseline, which no other type of company needs to do. But, you know, the truth is, for a huge chunk of our capabilities, that is true. And then for our enterprise roadmap, that is not true, because we know that if you're a Fortune 500 company and you want to know when is, you know, in certain compliance capability coming, we need to be able to tell you a definitive answer. So it really depends on what we're talking about.

但像GPT-6发布时间这种问题（请别问我），我们很少能对六个月后的计划保持高置信度，因为一切都在持续变化。

But for the stuff that we're talking about here, like, you know, when is GPT-six coming? Which you please don't ask me, but you know, that kind of thing, we would rarely have a, you know, high confidence target that is further than six months out just because everything changes constantly.

到底什么时候发布啊Nick？开个玩笑。我有位前同事匿名提问：为什么ChatGPT的产品形态变化不大？

When is it coming, Nick? No, it's okay. I know you won't tell me. I I do have an anonymous question for you from an ex colleague. And he asked me to ask you, why hasn't the chat GPT form factor changed more?

你知道，我也曾思考过这个问题。现在很多人都知道这个故事，但对于不了解的人来说，ChatGPT原本只是一个为更宏大产品打造的临时原型。我们希望能构建所谓的超级助手，一个能帮你处理任何事务的灵活实体。我们觉得它可能会有多种形态，这点我可以详细说明。ChatGPT就像是一个简单的起点，通过生成足够的学习数据和用例，我们得以构建真正的产品。

You know, I wondered this too. Many people now know the story, but for those who don't, ChatGPT was supposed to be a throwaway prototype towards a much broader for a much broader product. We were hoping to build what we call the super assistant, which was this kind of flexible entity that helps you with anything. And we felt like it would probably have many different form factors, which I can talk about. ChatGPT was like the simple way to start with the idea of generating enough learning and use cases that we could go build the real thing.

显然我们严重偏离了原计划，因为ChatGPT意外走红并自成体系地取得了成功。它的形态展现出了出乎所有人预料的持久生命力。必须承认，自然语言交互极其强大且必将长期存在。是否以聊天机器人形式呈现是另一个问题，但能用自然方式表达自我，这种用户体验堪称终极——毕竟这就是人类的天性。只要是为人类打造技术（这对我们至关重要），你就该让用户以最自然的方式与软件交流。

And obviously we got really sidetracked because ChatGPT took off and became successful in its own right. And it's been a fairly durable form factor in a way that I don't think I would have predicted or any of us would have. And I'll say that natural language is very, very powerful and I think that's here to stay. Whether or not it's a chatbot or not is a different question, but I think the idea that you can express yourself in a very natural way feels like the user experience to end them all because that's just how we as humans are drilled. As long as you're building technology for humans, which is certainly very important for us, I think you're gonna wanna let people communicate with software in a way that feels very natural to them.

但我不会将自然语言原生界面等同于聊天形式。我们正积极突破聊天框的形态限制，Canvas功能就是早期尝试——它让你与AI协同创作内容，而非来回对话。借助GPT-5强大的前端能力（它能生成精美的软件界面），完全可以想象它能根据不同用例实时渲染不同用户界面，这比Canvas的构想更为宏大。

But then, you know, I I wouldn't equate natural language native interfaces with chat. We're really excited about breaking out of the form factor of chat. One early step in that direction was Canvas. It was a feature that or is a feature that allows you to iterate on an artifact with your AI, such that you're like working on a thing together rather than chatting back and forth. With GPT-five's front end capabilities, which is, you its ability to make really nice looking software, you could absolutely imagine it rendering different user interfaces on the fly for different use cases, which is kind of a more ambitious version of what we did with Canvas.

比如数据分析时获取电子表格，规划旅行时生成协作网页应用——各种形态都可能自然涌现。聊天界面之所以存在，是因为它契合了当时的技术阶段。ChatGPT之前的聊天机器人并不出色，直到它们突然变得好用，这种飞跃令人惊艳。而按需定制软件的时代，现在或很快就会带来同样的震撼。

Like you can imagine, if you are running a data analysis that you get a spreadsheet. You can imagine that when you are planning a trip that you make a little web app so you and your friends can go plan together. You could imagine a lot of different form factors becoming emergent. And what I will say about chat is that it was the right interface for where that technology was at. Because, you know, chat there was chatbots before ChatGPT, but they weren't particularly good.

我骨子里坚信，由自然语言驱动但更侧重UI交互的形态会非常酷。长话短说，虽然我们对仍在使用聊天机器人感到困惑，但对产品愿景充满野心，技术终将实现这些可能。

And then they suddenly got good and it felt pretty magical. And I think this idea of custom software on demand is gonna have the same feeling either now or very, very soon. I just feel it in my in my bones. So I think this idea that, you know, you can get, like, more UI heavy stuff, still driven by natural language, is gonna be very cool. So to make a long story short, I am also baffled by the fact that we're still using chatbots, but we're very ambitious about what we wanna do with the product, and I think the technology will allow for it.

谷歌反垄断案中曝光的贵团队战略文件提及超级助手目标，称要打造人们连接互联网的界面。这暗示你们必须超越聊天形式，甚至涉足网页浏览领域。有报道称你在庭审中表示若谷歌剥离Chrome，你们可能考虑收购？

There was a strategy document from your team that surfaced in the Google antitrust lawsuit about this super assistant goal. And it said that what you wanna build is the interface to the internet for people. And to me that suggests you do need to move beyond chat and you actually need to move into web browsing as well. And there've been reports about that. And I'm curious now you were actually on the stand even at that Google trial saying you guys might be interested in buying Chrome if Google had to spin it off.

你们在自研浏览器吗？OpenAI是否需要为ChatGPT配套专属浏览器？

Are you building your own web browser? Does OpenAI need to operate its own web browser adjacent to ChatGPT?

那个声明被严重断章取义了。我需要澄清

That statement was taken horribly out of context. So I do want to say

你是说个人有意收购Chrome吗？

that You're saying you'd like to personally buy Chrome, is that what you're

我的完整回答是：如果Chrome上市流通，多方都会考虑收购，我们也不例外。这比网络广泛传播的说法温和得多。在产品层面，CHADGBT已成为互联网新入口——十年前需要浏览器完成的事，现在ChatGPT就能直接给出答案。想象未来通过它发现、了解并最终购买产品，这种可能性正在显现。

My full answer was that if Chrome went on the market and became available, that I imagine many parties would consider it, and we would as well. So it was a much weaker statement than, you know, was widely reported on the Internet. You know, on a product level, my belief is that, you know, you already see today that CHADGBT kind of is a new entry point into the Internet, Right? Many of the things that you would have used a browser for ten years ago, you can actually just do in ChatGPT because it will give you the answer. You know, imagine as you discover products through it, can, you know, discover, learn about them, eventually purchase them.

随着它开始为你处理更长期的任务，比如规划旅行路线或运行数据分析这类原本需要打开三个不同软件才能完成的工作，你可能会直接在AI中启动这些操作。因此，我认为AI逐步接管浏览器功能的设想并不疯狂。具体会以何种形式呈现，我们拭目以待。我们正在探索多种可能性。我确实认同这个观点——事实上那份文档就是我写的。

As it starts to do things for you over longer periods of time, you know, maybe that, you know, applying a trip example or maybe that, you know, running a data analysis example, which you would have gone to open three different products for, you might actually kick that off in an AI. So I don't think it's crazy to think about, you know, AI doing more and more things that a browser can do. What form that can take, you know, we'll see. We're exploring a variety of different things. I I do agree with the thesis that, you know, and I did write that document.

所以这份研究出自我手——ChatGPT必将承担越来越多现今浏览器所具备的功能。

So this work come from me that, you know, the ChatGPT is gonna have to do more and more of what a browser does today.

最后几分钟，我想快速请教几个产品战略问题，希望您能简明回答。Sam多次强调ChatGPT账户登录的战略重要性，即让用户能像谷歌或苹果账号那样带着个性化设置全网通行。你们目前进展如何？

Well, with the few minutes we have left, I have some kind of lightning round product strategy questions I'd love to hear your thoughts on. So try to keep your answers as concise as you can. Sam has talked a lot about signing with ChatGPT as being a thing he sees as strategically important. The ability to bring your ChatGPT account and personalization with you on the web and have it be a sign in option like Google or Apple. Where are you all at on that?

我们正在积极探索。这类生态建设需要格外谨慎——既要自主开发又要协调第三方，机会有限必须稳扎稳打。目前我们已与多方合作伙伴展开洽谈，对此构想始终充满热情。

Actually exploring it. This is the type of thing, you know, I've learned the hard way that with ecosystem stuff, like where you're building, having other people build, you've gotta take your time to get it right, you only get so many shots. So we've been talking to lots of different partners about that idea, and we continue to be really excited.

有传言说你们不会与Jony Ive合作开发眼镜或手机，属实吗？

Is it true that you're not going to develop glasses or a phone with Johnny Ive?

无法评论硬件路线图。但这个领域确实令人振奋，非常具有启发性。

Can't make any comments on hardware roadmap. Very excited about it, though. It's been inspiring.

与苹果的合作进展如何？

How is the Apple partnership going?

非常顺利。我们对合作项目充满期待，这是长期伙伴关系。我特别期待将AI技术——希望是我们的模型，但广义上的人工智能——渗透到iOS的每个角落。

It's great. I'm really excited about what we're what we're doing together. I I think it's a long term partnership. I I'm I'm so excited about, you know, bringing AI, hopefully, our models, but just generally AI into all corners of iOS.

那么您认为与苹果的合作会进一步深化吗？

Do you so you see it getting deeper, the partnership with Apple?

虽然我绝非专家，但纯粹从产品视角来看，这方面存在无数可能性。

I get very much not the expert, but from a pure product perspective, I see so many different opportunities to do that.

你们宣布与芭比娃娃制造商美泰公司合作，将你们的模型嵌入他们的玩具中。为什么要这样做？

You guys announced a collaboration with Mattel, the maker of Barbie to embed your models into their toys. Why do that?

我们不仅是一家产品公司，更是一个平台公司。这意味着虽然我们有自主的第一方产品，但我们更热衷于将构建模块开放给所有人。这个合作就是个典型例子——由我们亲自做第一方可能并不合适。尽管我们总会涉足各种领域，但不太可能进入玩具制造业。而通过我们的API，其他企业可以在此基础上开展业务，这才是绝佳的示范。

We're not just a product company. We're also a platform company, which means that while we have our first own first party offerings, we're really excited to make the building blocks available to everyone. And, you know, this is one of those examples where this would probably not make sense for us to do first party. It's unlikely that we're gonna get into the toy manufacturing business despite all the things we do somehow manage to get into. But it's a wonderful example of something you can do with our APIs in a business that, you know, someone else can go build.

看到别人能利用我们的技术开发产品，这非常令人兴奋。没错。

When products someone else could go build that's exciting. Yeah.

ChatGPT何时能实现完全多模态？目前它能处理部分模态，但视频输出、视频输入、音频输入输出等完整功能呢？

When does ChatGPT go fully multimodal? So right now it can do some modalities, but video export, video in, audio in, audio out, all of it.

我们的终极目标是让人机交互如同人与人对话般自然。就像此刻你我交谈时会互相打断，会有细微的肢体暗示——比如你刚才点头了（虽然观众可能看不到视频）。这些反馈至关重要。我们的发展路线很大程度上就是让你更轻松地向AI表达，同时让AI更自然地回应。只有这样，才能充分发挥这些模型的智能优势。

Our North Star is that you talk to this like a human, which means, you know, just like you and I are interrupting each other. You know, there's a bunch of little cues, you know, you're like, you just nodded, you know, people are gonna be able to see the video or not. But, you I get a bunch of feedback from that. I think so much of what so much of our roadmap just comes down to making it easier for you to express yourself to the AI and then making it easier for the AI to express itself back. That's really how you get all the benefit of the intelligence in these models.

所以理想状态确实是'任意输入，任意输出'，但实际比单纯的技术实现困难得多。必须让体验感觉自然。你可能用过最新语音模式，虽然已相当流畅自然，但我觉得它还没完全通过图灵测试——你仍能察觉是在与AI对话。

So, alright, aspiration is very much anything in, anything out, but it's actually much harder than the pure technical capability of doing that. You need to make that feel natural. You might have used latest voice mode. I think it's gotten pretty good and very natural, but, like, I still feel like it hasn't quite passed the Turing test, you know, so to speak. Because I could tell I'm talking to an AI.

人际互动中有大量精妙细节等待我们攻克。一方面我们持续投入技术研发，另一方面更要思考如何让交互体验真正浑然天成。

There's a lot of subtleties of of human human interaction that we're just gonna be cracking. So I think there's the technical roadmap which, you know, we continue to be very excited about, but then there's also the sort of overlay on top in terms of how you make that feel really natural.

你们刚推出四种新性格的测试，最终是会提供数十种甚至无限选项，还是让用户自定义ChatGPT的工作性格？

Is the end state of the personality test that you guys just rolled out with the four new personalities, is it dozens and dozens or unlimited personalities to choose from? Or is it that each user creates their own personality for how they want ChatGPT to work?

我们尚未完全确定。正通过刚发布的3.4版本收集数据：用户是都能归类到这四种性格，还是存在长尾需求？就目前而言，我认为应该允许用户自主配置。我们已有自定义指令功能，这四种性格是额外的起点选项。

We're not entirely sure yet. We're really excited to learn from the 3.4 that we just released to figure out does everyone fall into one of these categories or is there actually a long tail of needs? In terms of where my head is currently at, I think we should allow you to configure your own. We already have things like custom instructions. We have these four personalities as an additional starting point.

我设想的是：你先选择某个初始性格模板，再通过日常交互或主动设置来个性化调整。就像选择朋友——最初因性格投缘而结识，之后共同成长演变。ChatGPT也将如此：我们会简化初始选择流程，但后续必须提供高度个性化的定制空间。

So I really imagine you picking one initialization point, so to speak, that, you know, kind of speaks to you. And then from there you personalize, like either via your interactions with the product or by explicitly going in and configuring. I think it's kind of like choosing a friend where, you know, today, you you you pick people, you know, to be your friend based on whether or not you vibe with their personality, but then you actually co evolve together over time. And I think ChatGPT is gonna be similar where I think we can make it much easier to pick a starting point that you find appealing. But from there, it's gonna have to be customization that is pretty personal to you.

好的，尼克。我们今天就到这里吧。下次再请你来。感谢你的时间。

Alright, Nick. We'll have to leave it there. We'll have you back on. Appreciate your time.

谢谢。非常感谢。

Thank you. Appreciate it.

再次感谢尼克·特利加入本期节目。如果您想告诉我们您对这期节目的看法，或者希望我们探讨其他话题，请给我们留言。您可以发送邮件至decoder@theverge.com。我们还有TikTok和Instagram账号，请搜索decoder pod关注我们。

Thanks again to Nick Turley for joining the show. If you'd like to let us know what you thought about this episode or what else you'd like us to cover, drop us a line. You can email us at decoder@theverge.com. We also have a TikTok and an Instagram. Check those out at decoder pod.

如果您喜欢《解码器》，请分享给您的朋友，并在您获取播客的平台订阅。如果还没订阅，别忘了订阅The Verge，这样您就能阅读我们所有的报道和通讯，包括我撰写的《命令行》。《解码器》是The Verge出品，属于Vox Media播客网络。我们的制作人是凯特·考克斯和尼克·斯塔特。本期节目由赞德·亚当斯编辑。

If you like Decoder, please share it with your friends and subscribe wherever you get your podcasts. And if you haven't already, don't forget to subscribe to The Verge, which gets you access to all of our stories and newsletters, including the one I author called Command Line. Decoder is a production of The Verge and is part of the Vox Media Podcast Network. Our producers are Kate Cox and Nick Statt. This episode was edited by Xander Adams.

《解码器》的音乐由Break Master Cylinder制作。下次见。

The Decoder music is by Break Master Cylinder. See you next time.