第13集 - ChatGPT广告背后的思考

你好，我是安德鲁·梅恩，欢迎收听OpenAI播客。

Hello, I'm Andrew Main, and this is the OpenAI podcast.

今天，我们将与阿萨德·阿赫万讨论ChatGPT中的广告，包括广告的呈现形式、谁会看到它们，以及公司如何维护用户信任。

Today, we're talking to Assad Ahwan about ads in ChatGPT, how they'll look, who will see them, and how the company will preserve user trust.

广告会展示给使用免费版、Goatee版、Pro版、Plus版以及企业版的用户。

Ads are shown to people who are on the free and the goatee for pro and plus and for enterprise.

没有广告。

There are no ads.

如果效果很好，那令人不安也可以接受吗？

Is creepy okay if it is good?

并不如此。

It's not.

我们从事的是信任的生意。

We are in the business of trust.

所以我认为，如果我们必须说明我们的核心业务是什么，那就是建立用户的信任。

So I think if we have to say what is our core business is like doing users trust.

从消费者的角度来看，为什么要做广告？为什么现在要做？

So from a consumer point of view, why ads, why now?

这要回到我们的使命：将通用人工智能带给全人类，并让全人类受益。

It goes back to our mission, which is bringing AGI to all of humanity and to benefit all of humanity.

当你有一个消费者产品，有八亿多人在使用它时，你如何将这个产品的最佳版本带给每个人？

So when you have a consumer product, which like, you know, 800,000,000 plus people were using this, then how do you take the best version of that product to everyone?

广告是实现这一目标的最成熟模式之一。

Ads is one of the most proven models to be able to do that for consumer products.

我认为这一使命的另一部分是如何让全人类受益——你希望将最好的模型、最高的使用限额提供给用户，同时让广告对用户和企业都有实际帮助。

And I think the other part of that mission is how do you benefit all of humanity, which is like you want to take the best model, you want to give the highest limits, usage limits to people, you want to, for the ads to be actually helpful both to the users and the businesses as well.

我认为，对于一家立志将最先进人工智能带给全人类的公司来说，这非常自然。

So I think it's a very natural fit for a company whose ambition is actually to take the best AI to all of humanity.

这是一个非常有趣的决定，因为一方面，你可以说：我们选择走高尚的道路，不投放广告，也不提供大量使用额度，限制功能，或许使用不是最强大的模型，而是采取另一种方式；但另一方面，我们选择拥抱广告。

It's a very interesting decision because on one hand you could say, hey, we're gonna take what we perceive as the high road and say, we're not going to do ads, but also we're not going to give a really good amount of usage for it and limit that and sort of maybe use not the most capable models, but sort of, you know, say take that approach versus embracing it.

是的。

Yeah.

是的

Yeah.

我认为，如果目标是真正实现普及访问，广告是一个不错的模式。

I think if the goal is to truly democratize access, I think ads is a good model.

我认为，那个说法中可能隐藏着一个问题：广告会不会有坏处？

I think maybe what is hidden in that statement is can ads be bad?

现实是，我们该如何思考广告的原则？

And the reality is how do we think about the principles of ads?

我们该如何为这个平台上的广告设定一个极高的标准？

How do we actually set a really high bar for what ads should be on this platform?

我们该如何让它们真正有用？

How do we make them actually useful?

所以当我们刚开始时，我们思考的是：我们会向世界宣布什么样的核心原则，这个原则能让我们感到自豪，并能因此打造出一个出色的产品。

So when we were starting off, we thought like, hey, what would be the core principle that we would announce to the world that we would be proud of that we would stand behind and as a result, create a really great product.

举几个原则的例子，比如第一条：答案必须与广告完全独立，无论是视觉上，还是在模型训练和系统运作方式上，这样用户才能始终信任答案。整个产品，JRGBD产品，都是建立在信任基础上的，因此广告必须服务于这一信任体系。

So so just to give examples of principles like number one, the answers need to be independent from the ads, both visually, but also in how the models are trained and how the system works, so that you can always trust answer Like the whole product, JRGBD product is based on trust, so actually it needs to feed into that.

第二点是您的对话内容是私密的。

The second is your conversations are private.

如果您进行敏感对话，该对话中绝不会出现广告，也不会与广告主共享。

If you have a sensitive conversation that will never have ads in it and the conversation that never shared with advertisers.

因此，虽然我们会匹配最适合该对话的广告——这可能是一种有益的做法——但广告主无法看到这些对话内容。

So while we do the matching between the best ad that can be a useful thing in a conversation, advertisers don't get to see that.

我们会在内部完成这种匹配。

We do that matching internally.

当然，当您引入广告时，一个关键问题是：您是如何了解这些数据的？

And then of course, like, you know, as you introduce ads, a big question is how did you know about this data?

这正是用户信任与仅仅做些与用户相关的事情之间的区别。

That's the difference between the user trust and just like doing something which is relevant to the user.

我们的目标是创造一种用户能够清晰理解的产品。

And our goal was how do we make something which users can transparently understand?

他们如何能够进行控制？

How can they control?

我们可以深入探讨这一点，因为我确实认为这里需要设立一个很高的标准——许多产品可能具备一定的透明度和控制权，但真正优秀的版本是我们一直在思考的。

And we can go into that because I do think there is a high bar to set over there because you could have some transparency, some control that most products have, but what would be a really good version of that is something that we've been thinking about.

最后，一旦引入广告，你就必须为团队和公司设定正确的激励机制，确保始终专注于用户价值。

And finally, once you add ads, I think you have to set the incentives for the teams, for the company in a way that actually continue to focus on user value.

你不想只是为了获得平台上的空置浏览时长而存在。

So you don't want to just like get empty gallery time spent on the platform.

我们要打造一个真正有用的产品，而实际上，一个优质的广告就足够了。

Want to build a very useful product and automatically one good ad is good enough actually.

因此，我们并不以平台上的使用时长为优化目标，而是以用户为中心。

So we don't optimize for time spent on platform and focus on the user.

这些就是我们的原则。

So these are the principles.

所以，联系到你的问题，我想说的是：我们为什么要添加广告？又该如何去做？

So I think like, so connecting this back to your question, is like how do you, why should we add ads and how to do it?

我认为其中一部分是将其推广到全人类，这才是实现这一目标的最佳商业模式，但同时要防止因缺乏深思熟虑而带来的所有负面影响。我们正以清晰公开的原则为起点，接下来将通过测试、改进，逐步成为一个在这一领域持续学习的组织。

I think a part of that is take it to all of humanity and that's the best business model to do that but prevent all the negative things that can happen if you're not doing it thoughtfully and I think being upfront with our principles, very clear with that is is how we're starting and then how we will test, how we will improve and how we become kind of a learning organization with respect to this.

我想这就是它的运作方式。

I think that's how it goes.

所以你基本上是说会有一个分离。

So you said basically there's going to be a separation.

如果我跟ChatGPT说，嘿，我想开始喝奶昔之类的东西。

If I'm talking to chat GBT about like, hey, I want to start drinking smoothies and stuff.

它不会突然模糊地推荐：‘这是你该买的搅拌机。’

It's not going to all of a sudden blurred out like, well, here's a blender you should buy.

当然。

Absolutely.

是的。

Yeah.

这在模型所知的内容上是双重的。

It's an ex of both in terms of what the model knows.

模型并不知道广告是否存在。

The model doesn't know whether an ad is there or not.

如果你问它，嘿，那句话怎么说？

If you ask it like, hey, what does that saying?

它会说，我其实不知道，但如果你想要提问，可以按一些按钮把它添加到模型中。

It'll say, I actually don't know, But you can actually press some buttons and add it to the model if you want to ask question.

完全独立。

Totally separate.

完全正确，广告位显示的任何内容，模型都不知道这些内容存在。

Totally, whatever's being displayed in the ad space, the model has no idea that's in

没错。

That's right.

我认为在视觉上也是如此，这样用户能迅速分辨出哪些是模型给出的答案，而底部横幅中的内容则明显不同，通常你不会混淆这两者。当然，我们会不断优化这种体验，但目标是让系统和模型保持高度隔离，并将这种广告内容放在下游。

And I think in both visually as well so that the user can very quickly say hey this is the answer that I got from the model and then the bottom banner which has that in it which is very clearly distinct so usually also you don't confuse that of course we will learn how that experience evolves but the goal is to both keep the system, the models very, very separate and add this kind of downstream of.

好的，是的，我认为这是一个非常重要的区别，因为有些人似乎试图暗示广告部分和模型部分之间存在某种勾结，但你说的是模型完全独立。

Okay, yeah, that's I think that's a very important distinction because I think some people have sort of kind of tried to sort of spin that there's some sort of collusion between the ad part and the model part but you're saying the model is completely separate.

是的。

Yeah.

这很有趣。

So it's interesting.

所以当我正在对话时，可能会出现某些内容，我可以点击并说：好的，告诉模型，我看到了这个，这样它就知道发生了什么。

So as I'm having a conversation, something might come up and I can then click and say, okay, tell the model, hey, saw this, then it knows what's going on.

没错。

That's right.

是的，你必须在第一次体验中明确按下‘向ChatGPT询问这个广告’的按钮。

Yeah, you have to go, like in the first experience you'll explicitly have to press a button that's Ask ChatGPT about this ad.

这就像你从互联网上取了一个链接并就其提问一样。

And that would be as if you took a link from the internet and asked a question about it.

几乎是一样的。

It's almost the same.

我们不希望让这个体验变得更复杂。

We don't want to make that experience harder.

但如果你说，这个广告在讲什么？

But if you say, Hey, what is this ad talking about?

它会说：我不知道。

It'll say, I don't know.

现在很容易开始说：哦，对，我们要做一件了不起的事。

It's easy to start now and say, Oh yeah, we're going do the great thing.

我们会把它做对。

We'll do it right.

但十年后，当整个部门都负责广告收入时，你可能会想：我们真的需要在模型和广告之间设一道墙吗？

But ten years later, when there's an entire division in charge of ad revenue, you might say like, well, do we need the wall between the model and the ads?

是的。

Yeah.

对。

Yeah.

我认为这个问题可能有多个角度。

I think maybe there's multiple, multiple angles to this.

首先，我认为我们的核心业务是赢得用户的信任，为他们提出的问题提供出色的回答，这是面向消费者的产品方面。

So one, we are in the business of I think if we have to say what is our core business is like to win users trust and give amazing answers to the question that they're asking, that's in the consumer product side.

而在企业端，一切都是可信赖的，因为你把最重要的数据托付给了我们。

And of course, on the enterprise side, trusted everything, which is like you're entrusting us with your most important data.

我们当然需要维护这种信任。

We need to, of course, maintain that.

因此，我认为由于我们的雄心和愿景如此宏大，信任是其中的核心。

So because I think like that the ambition and the vision is so expansive, think trust is the central point of it.

我们希望拥有能为你提供帮助的设备。

We want to have devices which are helpful for you.

如果我们真想成为你最得力的个人助手，你就需要能够分享你最重要的信息，同时确信这些信息会以你对待它们的方式被处理——所以我认为我们的商业模式就是信任。

If we truly want to be your best personal assistant then you need to be able to do share your most important information but know that it will be dealt in a way which is like how you would treat it yourself So I think our business model is trust.

这与许多其他场景非常不同。

This is very different than many other scenarios.

如果你只是做类似交易性的事情，比如你提出问题，然后得到答案，事情就结束了，就像一个搜索查询。

If you're just doing like kind of a transactional stuff like again, you give the question and answer came back and that's the end of it like a search query.

我觉得这样也可以，但这不是一种长期的关系。

I think that's that's okay, but it is not a long term relationship.

我认为，如果你考虑内容发现，本质上只是推送东西，而信任并不是其核心组成部分。

I think if you think of content discovery, certainly, mean, like it is just pushing things and trust is not a core component of that.

对我们来说，整个产品——无论是企业级、消费级，还是未来的任何设备——都围绕着信任展开。

For us, I think the whole product, whether it's enterprise, whether it's consumer, whether it's anything, devices in the future, they're all centered around trust.

因此，对我们而言，这是至关重要的。

So, for us, it's kind of imperative.

对其他人来说，这可能是可选的。

For others, could be optional.

我认为不同公司以不同的方面著称，而我们确实希望以信任著称。

And I think different companies are known for different things and we do want to be known for trust.

所以，我认为将这一点与这个问题联系起来：我们会偏离吗？

So I think like connecting this to the question, which is, will we drift?

当激励机制被设定为在这一点上做到最好时，你可能会偏离，而这就是我们希望实现的目标——其他一切皆为此愿景服务，但最高原则是信任。

I think you can drift when the incentive is set up to be the best at this and this is the goal that we want to achieve everything else is is there to support that vision but the uber principle is trust

开放平台有大量用户使用免费版本。

Opening has a very huge number of people using the free tier.

是的

Yeah.

还有付费订阅用户、使用这些服务的人，以及专业用户等等。

And then there's also paid subscribers and people who do that and pro users, whatever.

对

Yeah.

广告在整个平台上的表现会怎样？

How are ads gonna play out across the platform?

是的，广告会展示给使用免费版和Go版的用户。

Yeah, so ads are shown to people who are on the free and the go tier.

嗯

Mhmm.

对于Pro版、Plus版和企业版，都没有广告。

And for Pro and Plus and for enterprise, there are no ads.

我认为这一点很重要，因为公司所处的环境实际上包含多个使命，这些使命共同推动AI普及，当企业使用时，这是一种非常特定的场景。

And I think that's an important thing, like the context in which the company operates is actually like multiple missions which come all together to bring AI to everyone, which is when enterprise use it, that's a very specific contact.

那边没有广告，围绕这一点有一个非常强大的商业模式。

There is no ads over there and there is a specific business model around that which is very powerful.

对于那些想要最好、最高限制和最先进功能的订阅用户，我认为这也行得通。

For subscribers who want like, you know, the best, like, know, highest limits and very advanced features, I think that also works.

但对于许多消费者来说，最好的方式是提供高限制和免费使用，同时加入真正有用的广告。

But for a lot of people, a lot of consumers, the best way to do that is to have high limits and free usage and then add ads which are actually useful.

是的。

Yeah.

我听说过一些人说，这部分目标是为了避免让免费版变得功能极度受限。

I've heard people talk about, you know, part of the goal of this is to avoid making the free tier just like the most limited thing available.

绝对如此，我认为这对许多真实用户来说是最令人沮丧的事情之一——你问了五个问题，然后就停止了，其他公司都是这样，对吧？

Absolutely, I think that is the most, I think, frustrating things for a lot of real users, just like you ask five questions and then it just stops in other businesses, right?

我认为我们应该大力扩展这一点，而且这符合我们整体的目标：更高的用户限制更好。

I think like we want to grow that a lot more and I think it fits with our overall goal that higher user limits is better.

那我们如何为它提供资金并保持务实呢？

And how do we fund that and be practical about it?

所以，稍微深入一下幕后，这些决策是如何做出的？

So going a little behind the scenes, how are these decisions made?

比如，有哪些人参与讨论这个问题？

Like who's in the room talking about this?

是的，我认为这也是一个很好的机会，来谈谈整体的企业文化。我认为不同公司有不同的文化，这会导致不同的产品。而我们的公司有着研究团队的基因，对吧？

Yeah, I think this is a good opportunity also to talk a little bit about like overall, like there is a company culture and I think different companies have different cultures which results in different products and our company has like this DNA of a research team, right?

因此，我认为我们有更严谨的辩论，更深入地理解应该如何制定这些原则，激励机制如何运作，以及这个模型如何才能避免日后被扭曲。

So we have much more, I think, like rigorous debates, rigorous understanding of how should we make these principles, how does incentives work, how does model of this going to work in a way that it doesn't get corrupted later on.

我们为此进行了大量辩论，最终形成了这些原则，以及一套包含数百场跨部门圆桌会议的评估体系，参与人员遍布公司各个部门，而不仅仅是广告团队，每个人都提供了反馈，共同制定了这些原则。

So we have had both a lot of debates on that which actually resulted in these principles, which resulted in this rubric with like hundreds of roundtables with like folks around the company on different areas, not just like working on ads, but everyone on every different part of the company, giving feedback to create, these principles.

然后我们将这些内容简化为一个非常简洁的评估标准。

Then we convert those to a very simple rubric.

我认为，这个评估标准的核心是：用户信任最重要，其次是用户价值，再其次是广告主价值，最后才是收入。

I think like the rubric is user trust is the most important thing, user trust more than user value, which is then more important than advertiser value, which is more important than revenue.

虽然这看起来非常直接，但实际上这是一个非常深入、非常复杂的决策，我们可以稍微深入探讨一下为什么用户信任比用户价值更重要。

And I think while this seems very straightforward, it's actually a very, very, very, I think, in-depth decision, so we can go into a little bit just like user trust more than user value.

一个很好的例子是，如果我给你看了一则非常不错的广告，你喜欢它，点击了它，还购买了东西，但后来你开始怀疑：这个应用是不是在监听我？麦克风是不是开着的？

A good example of that is if I showed you a really good ad, but and you liked it, you clicked on it, you bought something, but later on you ask the question, was this app listening to me and is the mic on?

这不符合用户信任。

That's not user trust.

也许它确实提供了一些价值，但对我们来说，我们绝不能接受这种情况。

Probably did provide some value, so for us, our goal is like we cannot have that.

用户必须相信、理解并掌控正在发生的一切。

The users need to believe and understand and control what's happening.

所以这只是一个例子，但一旦你确立了这个评分标准，即使从基层开始，团队也会这样思考。

So that's just one example, but I think once you said that right, Rubric Up, then even bottom up, the team thinks like that.

当然，由于我们在不同层级做出不同决策，公司在讨论隐私和安全时有着非常严谨的流程，有明确的讨论渠道；而在高层做决策时，我们也会回归到这个简单的评分标准，因为虽然它简单，但其实非常深入，而且在面对这类问题时极具区分力——比如，广告可以做得很好，但用户却不知道这些数据从何而来，这可以吗？

But of course, like, you know, as we have different decisions at different level, I think we have a pretty rigorous process on how we discuss privacy within the company, how we discuss safety within the company and there are very clear forums for that and then of course like as we make decisions at leadership level, one go back to the simple rubrics because although the rubric is simple, it's actually pretty in-depth and actually is very discriminating if you think about these kind of questions like should the ad be so good but the users don't know where this data came from?

如果很诡异但效果很好，这样可以吗？

Is creepy okay if it is good?

不可以。

It's not.

所以我认为这可能由此而来。

So I think maybe it follows from there.

我这边会看到什么？

What am I gonna see on my end?

我有什么样的控制权？

What kind of controls do I have?

个性化功能将如何运作？

How's personalization gonna work?

所以我认为，真正提供优秀广告的一个重要部分是允许个性化。

So I think like a big part of actually delivering really good ads is allow personalization.

比如，当我表示我想去优胜美地旅行时，系统就会向我展示露营装备，因为那是我喜欢的活动。

So there is, so when I say I want to go on a trip to Yosemite, and then that shows me camping gear because that's what I like to do.

当然，另一方面，如何赢得用户信任呢？那就是：你是怎么知道这些的？

Of course, I think the flip side of that, how do you gain user trust is, like, how did you know about this?

你是怎么了解到这些的？

How did you learn about this?

首先是透明度方面，你可以看到我们用于广告的关于你的数据有哪些。

So one is transparency aspect, which is like you can see what is the data that we have on you, which is being used for ads.

其次是控制功能，让你能够查看哪些过去的聊天数据可以被使用。

The second is the controls, which let you see, say, which part of the data from your past chats can be used.

当然，敏感的聊天内容是绝对不会被使用的。

Of course, sensitive chats, those are never used.

而且你可以清除你的数据，这一点其他人都做不到，这简直是个疯狂的概念。

And but you can clear your data, which actually nobody else does, which is kind of a crazy concept.

你可以清除你的数据，这样我们就不会知道，也不会再使用这些数据。

It's like you can clear your data so we don't know and we won't use that.

如果你在意的话，你可以选择不使用我过去的聊天记录，或者完全关闭个性化功能。

You could say don't use my past chats if that's what you care about or you could say turn off personalization fully.

当然，另一个极端就是我不想看广告，这也是一种控制方式，而升级到专业版或高级版以完全屏蔽广告，正是提供这种选择。

Of course, is the other extreme which is like I don't want ads, that's a form of control and that's where I think upgrading to the pro or plus to completely stop ads is also there.

所以我认为，从整个光谱来看，如果你非常在意这一点，觉得这不是正确的商业模式，那么专业版和高级版才是正确的商业模式。

So I think all the way from the spectrum, was like, I really care about this, I don't think this is the right business model, like Pro and Plus is the right business model.

比如，嘿，我不记得我们昨天聊了什么，我就把历史记录清了吧。

For like, Hey, I don't know what we were talking about yesterday, I'll just clear my history.

很好。

Great.

就这么做。

Do that.

或者，嘿，我更倾向于关闭广告使用，但不希望我的过往对话被使用。

Or it's like, Hey, I'm more comfortable with like, you know, clicks off on the ads being used, but not my past conversations.

你其实也可以这么做。

You could actually do that as well.

当然，人们会逐渐了解这些功能如何改善他们的体验。

Of course, people will learn hopefully like how it improves their experience.

我们在使用这些数据时设定了非常高的标准，但最终用户需要了解并能掌控这些设置。

We have a very high bar and how we use these things but in the end the users need to know and be able to control that.

那么，我预期会看到多少广告？这些广告会出现的频率是怎样的？

What will be, you know, the kind of the expectation for how many ads I'm gonna see or how often these would come up?

也许在那种情况下，Uber 的原则仍然适用，那就是是否存在一个有用的好广告？

Maybe the Uber principle still goes back to, in that context, is there a good ad to show which is useful?

如果不是，我们宁愿不向你展示任何内容。

If it is not, we'd rather not show you anything.

事实上，随着我们推出这个测试，你会发现广告非常少，因为我们既想保持谨慎，又想学习如何恰当地插入广告。

In fact, like, you know, as we roll out this test, you'll see that there'll be very few ads because, like, you know, we want to be both conservative and we want to learn how to, where to insert those.

但原则更侧重于：它是否有用？是否有帮助？是否能提升用户的体验？我们是否能展示真正优质的产品？

But the principle is a little bit more around, is it useful, is it helpful, does it add to what the user is doing and can we actually show a really good product as well.

同时也要保持内容的质量、广告的质量以及相关性的高水平。

Keep the quality of the content very high as well, keep the quality of the ad very high as well, keep the relevance very high.

如果我们找不到合适的匹配，那也没关系。

If we can't find a good match, it's fine.

你不需要展示广告。

You don't need to show an ad.

你提到敏感对话，你们怎么判断什么是敏感的，什么不是？

You mentioned that sensitive conversations, how do you know what something is sensitive or not?

这实际上是OpenAI的一大优势，无论是我们的自发工作还是公司内部的大量研究，都深入定义了什么是敏感内容，比如健康、政治、暴力等不同领域，都有非常详尽的定义。

So that's actually one of the big strengths of OpenAI, think both for our organic work and a lot of research in the company has gone into defining what sensitive is like it is health, politics, like violence, like many different kind of verticals, very, very in-depth definitions of that.

当然，还有一些最先进的模型能够预测和理解对话内容，并判断其是否属于敏感内容。

Then of course, some of the best models to actually predict and understand the conversation and saying, marking it as sensitive or not.

我认为，在我的职业生涯中，从未见过任何产品能达到如此高的精准度，而我们通过采纳这些政策，在这里成功实现了这一点。

I think like I've actually never seen such high precision in any product so far in my career, but we have been able to build over here by taking in those policies.

有一个团队非常严谨地制定这些政策，然后与内外部合作伙伴分享以供审查，当然，还有来自预测系统的执行机制，它会指出：‘这个内容匹配了某项政策，因此不要这样做。’

There's a team which works on defining those policies very, very rigorously and then actually also sharing them with internal external partners for review and then of course the enforcement that comes from the prediction system that actually says, hey, like this is matching this policy, so don't do that.

我们之前稍微讨论过这一点，但我还想再回头谈谈设计问题。

We've talked a little bit about this but I'd like to touch back on this again, design.

那么，设计未来会朝什么方向发展？

So where is that gonna head?

它们会是什么样子？

What are they gonna look like?

在设计这个产品时，我认为我们确立了一个非常明确的原则：回答必须与模型分离。

As you were designing this product, I think like of course we set up a very clear principle that the answers are separate from the model.

然后问题是，这在产品中实际是什么样子的？

And then the question was like, how does that actually look in the product?

在这个光谱上，一方面是如何让它看起来自然，不显得突兀？

On that spectrum, on one side is like, how do you make it look native so that it's not jarring?

另一方面，还有一个问题，就是如何让它明显区分开来？

And on the other side, there is a question which is like, hey, how can it be very clearly separated out?

我认为这两者都无可争议，而且它们各自都有价值。

And I don't think you can debate with on both of them and there is values in both of them.

我们一开始就想设计一个实验，以便在过程中不断学习。

We started we wanted to kind of set up the experiment in a way which is that we can learn as we go.

我们采取了保守的方案，但仍牢记这一原则。

We take the conservative option and still keep that principle in mind.

随着我们通过构建产品和获取数据不断学习，再逐步调整。

As we learn through building the product to getting the data, evolve it.

但核心理念仍然是：如何保持答案与广告之间清晰分离，并确保有明确的可理解性和视觉区分。

But the idea still is how can we maintain that principle of the answers being very clearly separated from the ads and having a very clear understandability and visual distinction.

我认为我们会不断优化这些形式，随着时间推移，它们会变得更有用、更出色，但这一原则是恒定的，在我们现有的选项中，我们选择了明显分离且保守的方案。

I do think we will evolve the formats and I think they will get even more useful and better over time, but that principle is constant and within the options that we had we started with something which is clearly separated out in conservative.

所以你从技术层面解释了这种分离是如何实现的，模型不会看到它，同时也提到了一些防护机制。

So you explained kind of at a technical level how there's a separation, how the model doesn't see it, but also for guardrails and stuff.

我觉得你之前提到过，但假如我说，‘我有点担心这次旅行’。

And I think you mentioned this before, but if I'm talking about, you know, you know, saying like, hey, I'm afraid of this trip.

那就会有人回应：‘要不要考虑买份人寿保险？’

It's like, well, hey, how about some life insurance?

这种事是不会发生的。

You know, that's not going to happen.

但你们是怎么设置防护机制的？又是如何判断哪些广告合适、哪些不合适呢？

But how do you guys put in guardrails and how do you decide what's appropriate ads and what's not?

这里面可能包含两个问题：一是什么样的广告才算合适，二是什么样的场景才是合理的。

So maybe there's two questions in there, like what's appropriate ads or not and which context is a reasonable one.

第二个问题是，你们设置了哪些控制措施，以确保随着时间推移，这种情况不会失控。

And the second is what are the controls in place so that like, you know, over time this doesn't, this doesn't add it all.

我认为，公布我们的原则并在内部明确我们的评估标准，最初就是为了建立这套机制。

So I think maybe a part of announcing our principles and being very clear internally for our rubric was to actually set that up in the first place.

之后，公司内部的许多治理和决策方式都会自然而然地遵循这些原则。

Then automatically a lot of the governance within the company, how we make decisions follows from that onwards.

所以，比如说，我想对产品做这个改动。

So it's like, hey, I want to make this change to the product.

这符合我们的哪条原则吗？

Do this fit with this principle?

它符合我们已经建立的评估标准吗？

Do they fit with this rubric that we have already set up?

这是第一道筛选。

That's the first pass.

我认为，敏感场景我们也同样高度重视。

I think the sensitive context is something that we take very seriously as well.

一些很简单的事情，比如围绕健康、政治或其他不合适的场景展开的对话，这些数据根本不会被用来投放广告，甚至不会用于广告匹配。

Very simple things like, you know, conversation around health or politics or other contexts where where there has don't fit and that data is not going to be used for making ads like even even matching ads.

你直接过滤掉它。

You just filter it out.

所以第一层其实是看广告是否适合这里，如果合适并且能带来帮助和补充，那就加上。我觉得这又回到了我们的原则：你既不希望因为展示过多广告而损害用户，也不希望损害商家，因为你不希望广告主为随机的展示付费，也不希望用户看到太多广告，你只想展示那一条最合适的广告。作为一家有望成为顶尖AI公司的企业，我相信我们在这方面一定能做得很好。

So the first layer is really see does an ad belong here if it does and it can be helpful and additive then add it I think like this goes back to the principles like actually you don't ruin neither for users nor for businesses by showing many ads because you don't want advertisers to pay randomly for impressions, you don't want users to see too many ads, you want to share the one right ad and being one of like, you know, the best AI companies, I think that's hopefully something we'll do really well.

每次我们制作一期节目，都会有一些人在评论区说：不要广告，不要广告，不要广告。

Every time we do an episode, we get a few people who go in the comments are like, no ads, no ads, no ads.

现在是你直接向这些人回应的机会。

Now's your chance to talk to those people directly.

是的，我觉得在某种意义上，当人们说不要广告时，我能理解他们的感受，这种想法并不错，因为我认为行业的发展方式确实让人对广告产生了一些疑虑。因此，我们确实有责任制定更好的原则，获得更清晰、更明确的规则来指导我们如何做这件事。

Yeah, I think like in some sense, when people say no ads, I feel like there is a perception and it's not wrong that, because I think like maybe how the industry has evolved that there is some suspicion around how So this I do think it is kind of incumbent on us to come up with better principles, get better clarity, better rules on how we're going to do this.

我认为，整个在线广告行业，如果从历史来看，才大约二十年，而许多其他行业已经有数百年的历史了。

I think there is, like, we, like, again, this whole ad industry, if you think about the online ad industry, it's, like, maybe twenty years old compared to many other industries which are hundreds of years old.

我觉得我们可能才处于这个行业的第三局，正在说：好吧，我们已经从人们提出的各种问题和挑战中学到了很多。

I think maybe we are in the third inning of this where we are saying, okay, we have learned from all these questions and problems that people have.

我认为当人们说‘不要广告’时，我确实相信他们对隐私的担忧是合理的，我们有责任通过更好的透明度、更强的控制权，以及打造令人愉悦的体验，来赢得他们的信任。

I think when people say no, as I do believe that they have valid questions and concerns around privacy, it's on us to do a really good job to earn their trust through better transparency, through better control, through building that is also delightful.

我认为仍然会有人持怀疑态度，而我认为我们也有办法提升，因为我认为这也是一个合理的选择。

I think there'll still be skeptics and then I think we have a way to upgrade because I think that's a valid choice as well.

但通过良好的原则来实现优质的广告，我认为是可能的。

But enabling really good ads with good principles, I think it's possible.

我认为其中很大一部分在于拥有强大的人工智能来驱动这些广告，使它们真正有用。

I think a big part of it is having really strong AI to power these ads also so that they are actually useful.

因此，这个产品能够惠及如此多的人，且限制更高。

Then as a result, this product to so many people without with with higher limits.

你们的一些竞争对手一直在拿广告这个想法开玩笑。

Some of your competitors have been having a little bit of fun at the idea of ads.

是的，不同公司有不同的使命。

Yeah, think like different companies have different missions.

我们的使命是将人工智能带给全人类，当然我们所处的环境也不同。

Our mission is to take AI to all of humanity and of course we have different contexts.

我们有企业业务，有订阅业务，还有一个非常庞大的消费者群体在使用我们的产品。

We have the enterprise business, we have our subscription business, and we have a very, very huge consumer base using our product.

所以我认为在这样的背景下，需要为每一个群体提供服务。

So I think within that context, need to serve each one of them.

我们会拥有一个非常强大的企业业务，那里不会有任何广告。

We will have a really robust enterprise business and there will be no ads over there.

然后我们会有一个非常强大的消费者业务，广告将帮助我们在这一领域实现增长。

And then we'll have a very robust consumer business and ads will help us grow within that.

所以我认为，如果这不是你的使命，也许这并不合适，但我们的使命是在所有这些场景中构建产品，我们相信它们都是相互关联的，我们通过打造最好的AI并将其普及给每个人。

So I think if that's not your mission, maybe it doesn't make sense, but our mission is to actually build in all of these contexts and we believe they're all actually related and and how we build the best AI and then actually take it to everyone.

我认为优势在于，我们的业务线拥有不同的垂直领域，我们不仅仅是一家广告公司，有些公司纯粹是广告公司，他们的动机实际上不同。

And I think the good part is that we have different verticals in the business line, it's not just an ads company, there are some companies which are purely just ads companies and their incentives are actually different.

是的。

Yeah.

但我认为我们对这个问题有着更全面的视角。

But I think we have a much more holistic view on this.

而且当你不服务数亿免费用户时，说‘我们不必做这个’就容易多了。

And also when you're not serving hundreds of millions of free users, it's easier to sort of say, we don't have to do this.

我不认为这是一种抽象设定的愿景。

I don't think it's like a vision vision, which is which is set in abstract.

这真正是一种关于人工智能如何真正帮助人们的愿景。

This is truly a vision that is like, how does AI actually help people?

如果存在一种精英主义观点，认为只有付得起钱的人才能使用它，而其他人不能使用，我认为这本身就是人工智能如何对人们产生价值的一个重大分岔点。

And if there is the elitist view that some people get to use it and some don't get to use it based on who can pay, I think that itself is a pretty big fork in the road in terms of how AI can be valuable to people.

我认为我们的立场是，每个人都需要能够接触到最好的人工智能。

And I think our position is pretty much like everybody needs to have access to the best AI.

我有一些朋友经营小企业，他们总是试图想办法推广自己。

I have friends that have small businesses and they are always trying to figure out how to promote themselves and do that.

从这个角度来看，你能解释一下，对于那些真正想接触新受众的人，这会是什么样子吗？是的。

Could you explain from that point of view, like what it's going be like for people who are actually trying to reach new audiences Yeah.

通过

Through

我认为这是个非常好的问题。

I think that's such a good question.

通常，我有几个朋友创办了一家电商公司卖鞋子，他们几乎自己包办了一切，就像创始人一样。

Like, usually, I have a few friends who started this e commerce company selling shoes, and they did almost everything on their own, like the founders.

就是去工厂，搞定生产，安排物流。

It's just like, go to the factory, get this done, get the logistics done.

但到了做广告时，他们不得不雇了三个效果营销人员来处理，因为这太繁琐、太需要数据分析了。

But when it came to ads, they actually had to hire like three performance marketers to do the work because it's so, cumbersome, so analytical.

如果你没做对，可能会浪费大量金钱。

If you don't do it right, you could end up wasting a lot of money.

所以我确实认为，未来的愿景应该是，就像你现在提问那样简单，你可以说：‘我的目标是在中西部多卖些鞋子’，然后就搞定。

So I do think the vision has to be where almost as easy as you were prompting nowadays for questions, you could say my goal is sell these shoes more in Midwest and go.

然后系统会回复你：‘我做了一些实验，根据你的定价，我觉得这个出价最合适。’

And then it comes back as like, hey, I tried some experiments and I think this is the right bid given your price point.

这才是你应该思考的方式：你愿意在这个项目上多花点钱吗？

This is the right way to think that do you want to spend more money on this?

然后你继续这个对话，几乎就像在为它做代理，但如今，小企业必须雇佣效果营销人员，而这在某种程度上可能是最大的成本之一——毕竟，通过这些人员投放广告的成本，实际上就是其中最大的开销之一，这当然会让事情变得更贵。

And then you continue that conversation and almost become an agent for that but today literally a small business has to hire performance marketers which could be one of the biggest costs in some sense actually like you know just that cost of running ads through through that is actually one of the biggest costs in there which of course then makes things more expensive.

所以我认为，未来的愿景应该是像开车一样简单，你只需要告诉系统你业务上的需求，描述你要做什么，而不需要操心具体怎么操作、要开多少个广告活动、花多少钱等等。你只需要说：我想花这么多钱。

So I think the vision would be that it is as easy as just steering and telling what you need from your business or describing the what, but not having to think about how it will work and how many campaigns and how much dollars and everything else was like, Hey, I want to spend this much.

我想尽可能地扩大我的业务。

I want to grow my business as much.

这些就是限制条件，广告会根据这些限制自动生成并投放。

These are the constraints and ads are created and run to match your constraints in some sense.

是的，这种想法非常有趣，因为竞价机制曾经是革命性的。

Yeah, it's a very interesting way to think about it because auctions were revolutionary.

你只需要进去，说我想发布这些内容，然后为它付费即可。

The idea that you just go in there, I wanna put this words out there and pay for that to do it.

但这也催生了一个完整的生态系统，需要各种专业知识和配套服务。

But that created an entire ecosystem of all the sort of expertise and stuff that you have to do.

对于小企业来说，想要在这个领域参与竞争真的很难。

And it's really hard for small businesses to try to play in that space.

是的，我认为随着竞争加剧，那些有更多时间和资金去优化、分析数据并投放最佳广告的人才能从中获益，而如果像我这样不懂这些，比如我举过一个叫‘All Words’的实际品牌例子。

Yeah, I think as the competition on that increase, I think like the people who had more time and money to spend on optimizing that and analyzing the data and then running the best possible ad got the benefit from that versus like if I didn't know that like, hey, actually, I think I gave an example of an actual brand which is all words.

它与大型鞋类品牌竞争，但不知怎的，他们发现科技公司的每位设计师都会爱上我的鞋子。

It competed with really big brands on shoes, but somehow they found that every designer in the tech company is going to love my shoe.

找到这个细分市场，然后真正创造出针对这一群体的广告素材和信息，让他们取得了成功。

And finding that niche and then actually being able to create your creatives, your message to focus on that made them win.

在硅谷，你会看到到处都是这种鸟，就是因为这个原因。

In this, like, you go in Silicon Valley, you'll see all birds everywhere because of that.

所以我认为，这种方式并不是每个人都能做到的。

So I think like but that I think is not accessible to everyone.

如果你非常善于分析，并且拥有一整支团队专门思考这些问题，你就可以做到。

If you are very analytical and you have a whole team of people who think about that, you could do that.

理论上，如果我们能找到最适合产品的细分市场并全力投入，最好的产品就能诞生。

Theoretically, the best products can come to life if we can find out where the right niche distribution for it is and really go for that.

关于这一点，还有另一个故事：有一家公司生产了一种素食即食拉面，我特别喜欢，因为吃它时我不用感到内疚。

I think another story on this is like that there is this company which creates a vegan instant ramen, which I love because I don't have to feel bad about eating it.

但这个概念太奇怪了，素食即食拉面——如果我不知道这家公司，单凭想象，我会觉得这根本不可能存在。

But it's such a weird concept, vegan instant ramen, like if I was just thinking about it without knowing about this company, was like, this can't exist.

谁会想要这个？

Who will want this?

但我认为一个好的产品能帮你找到并发现这些小众受众，然后你再打造出真正优秀的产品。

But I think a really good product can help you find and discover those niche audience, then you build a really good product.

所以我认为，从这个角度来看，它丰富了每个人的生活，因为它促进了这些产品的创造与销售。

So I think it enriches everyone's life from that perspective, like enabling creation of those products, selling of those products.

也许它不会成为一家数十亿美元的公司，但那已经很棒了。

Maybe it's not a multi billion dollar company, but that's great.

它仍然是一家规模可观的中小型企业，在成长，而且正是这些关心特定问题的人在推动它。

That's still, it's like a really big SMB, which is growing and it's sort of these people who care about that very specific problem.

所以我认为，如果你能为这些小众市场创造产品，它真的会丰富人们的生活。

So I think it really enriches people's life if you are able to create products for these niches.

当我们以更具自主性的方式使用这些工具时，未来会是什么样子？

What does this look like in the future where we're using things in a more agentic way?

十年后，广告还会是怎么运作的？

How do ads even work ten years from now?

我认为下一步将是更真实的对话式广告，让你真正理解这个产品是什么。

I think next step would be more actual conversational ad where you could truly kind of understand what this product is about.

下一个版本是，它能否在后台工作，自动聚合最佳折扣、最佳优惠和产品最佳版本？

The next version would be can it work behind the scenes and actually aggregate the best discounts and best deals and the best version of the product?

比如，如果我知道我喜欢拉面，假设ChadGPT已经理解了我的这个偏好，它就能帮我找到我甚至不知道存在的产品，然后在后台告诉我：哦，我找到了一款素食拉面，这可能很有价值。当然，这里有一个市场，有人可以说：帮那些有类似需求的人发现这些产品，因为发现是双向的，对吧？

Like, for example, if I know that I like Ramen and let's say somehow ChadGPT has understood that preference of mine, then it could find that for me I didn't even know that that product exists and then behind the scenes it could actually say oh actually I found a vegan ramen maybe that's something that's valuable and of course there is a marketplace where somebody could say hey help people who are like this to discover because because discovery goes from both directions, right?

当然，我正在寻找某样东西，同时别人也希望我注意到某些东西，这两者之间会形成匹配。

Like, of course, I'm searching for something and then people want me to discover something and there's a match between those.

所以我认为，未来它会更加自主，但至少在当前的模式下，我觉得我们应该从这里开始，不断改进，让它更相关、更可控、更易理解、更值得信赖。

So so I think it will be more agentic, but in the future, but at least the current modality, I think we start from there, improve it and make it relevant, make it controllable, understandable, trustworthy.

随着这些系统的发展，原生的、自然的产品也会随之进化。

And as I think their systems evolve, the native the organic products evolve, this will evolve as well with that.

很好。

Excellent.

谢谢您，阿萨德，解释得这么清楚，我很期待看到接下来的发展

Well, Assad, thank you for explaining this, and, look forward to seeing what's gonna

接下来会发生什么。

happen next.

太棒了。

Awesome.

谢谢您邀请我。

Thanks for having me.