#64 - OpenAI 正在为提示词之后的未来构建什么？- Tara Seshan, Ty Geri

但现在有了像 Codex 或者 ChatGPT Work 这样的工具，我就能做出我一直想要的那个小应用了。

But now with, like, Codex or which ChatGPT Work, I can, build the little app I really wanted.

维护它也不需要花什么钱。

It doesn't cost me anything to maintain.

我不用投入大量的时间和精力。

Like, I'm not spending a ton of time and effort.

它完全符合我想做的事情。

It's exactly suited to the thing that I want to do.

就像在一个厨房里，你拥有所有你想要的单一功能厨房设备，但你又不用负责管理它们。

It's like being in a kitchen where you had the perfect like single purpose kitchen device for everything you've ever wanted to do but you don't have the responsibility to manage it all.

这真是个理想的场景。

Like it's it's an ideal scenario.

或者你也不用去存放它们。

Or you don't to store it.

你不需要额外的台面空间之类的。

You don't have that extra counter space or something.

对。

Exactly.

就像你想象一个厨房，里面每件事都有专用的工具，比如完美的柠檬刨丝器，专门用来刨柠檬。

Like imagine you had a kitchen where you could have a single purpose kitchen device for everything you had to do, like the perfect, you know, lemon zester for your lemon.

为了

For the

塔拉喜欢的那种柠檬。

way Tara likes to have lemons.

为了塔拉喜欢的，但你从来不用把它收进抽屉里。

For the Tara likes to but have you never had to store it in any drawers.

没有任何麻烦。

There was no problems with that.

你从来不需要，比如

You never had to like

事后清理。

Clean up after.

用完之后还得收拾。

Clean up after.

我现在用软件的情况差不多就是这样。

That's kind of how my life is right now with software.

那挺好的。

It's great.

不，这个比喻真的很棒。

No, that's a great, it's a great, metaphor.

我特别喜欢。

I really like it.

那我们也聊聊这个吧。

So let's talk a little bit about this too.

Ty，你有没有让人帮你建过一些专门针对你做的事情的东西？

Ty, are there any things that you've had it build for you that are specific to things that you do?

我也有一个例子可以分享。

I also have an example of one I can share.

对。

Yeah.

我觉得有几件事吧。

I mean, think there there are a couple things.

第一，我很多时间都在看数据。

So one, I spend a lot of my time looking at metrics.

好的。

Okay.

其实我觉得挺难达到那种收益递减的点，就是了解用户怎么用产品。

And it's actually like, I find it's kind of hard to get to the point of like diminishing returns on understanding how, you know, people are using the product Sure.

你知道，哪里有问题，怎么改进。

You know, where are we seeing friction and how can we improve.

所以我就跟它合作，基本上就是创建了一套技能，每天运行，就像我每天会问的那些关于产品的问题，然后找异常或者找可以改进的地方。

And so I've I've worked with it, just basically, you know, I've created like a set of skills, that run every day that sort of just try to ask like the questions that I would ask every single day about the product, and and look for for anomalies or look for things that we can fix and and make better.

这就像，如果说到去掉琐碎工作的乐趣和拥有真正思考伙伴的乐趣，我真的觉得我处于真正思考伙伴的乐趣这一边。

And that's something that it's like, if we talk about like the the sort of like the joy of taking away menial labor versus the joy of like having a real thought partner, I I really feel like it's I'm I'm at the corner of like the the joy of a real thought partner.

对。

Yeah.

这是在工作方面。

That's at work.

在个人生活里，做网站也特别有意思。

In my personal life, it's also really fun to make sites.

你可以做出各种有趣的网站。

And you can make all sorts of exciting sites.

比如我有个朋友做了一个超棒的婚礼邀请网站，互动性特别强，你可以选择控制新娘或新郎，然后在场地里走动，去不同的站点了解信息。

Like I had a friend that made like an awesome wedding invite site that was like super interactive, like it let you like control either the bride or the groom, and you could like walk around the venue and like visit different stations for information about.

他们是在手机上做的，完全按自己的想法来，就像那种可塑的软件，也许可以说，可塑的婚礼邀请函。

Like they did it on their phone and made it exactly what they wanted, like this like malleable, people wanted malleable software, maybe like, you know, malleable wedding invites.

这真的带来很多快乐。

It's just, it's so much joy.

我来分享一个我做的，我觉得特别有意思。

I'll share one that I did, so that I thought was really interesting.

所以我有这个，这是当ChatGPT Atlas消失的时候。

So I have this is this is when ChatGPT Atlas went away.

所以ChatGPT Atlas真正酷的一点是，它可以让你保存你的提示词。

So one of the actually really cool things about ChatGPT Atlas is it let you save your prompts.

你知道吗？

You know?

所以我想，好吧，ChatGPT Atlas要消失了。

And so I thought, okay, ChatGPT Atlas is going away.

我有很多保存的提示词，我每天用，或者至少每周用好几次，大概有10个我经常用的。

I had a lot of prompt save that I use every day or at least multiple times a week probably like 10 that I use regularly.

所以我想，我得找个办法来做这件事。

And so I thought well I need to find a way to do it.

所以我用的是，我在用一个叫Raycast的产品，它基本上是一个快捷方式工具。

And so I was using like one of the I'm using this product called Raycast which is essentially a shortcut thing.

我想，你知道吗，我真正想要的是，在Mac菜单上有一个下拉菜单，我拉下来，然后它把提示词分成不同的类别，然后我就可以直接选择那个提示词。

I thought, you know, what I what I really want is I just want like a drop down on the Mac menu that I drop it down and then it puts them in different categories and then I can just select that prompt.

然后不管我用哪个浏览器还是哪个软件，它都能直接放进去。

And then whatever browser or whatever piece of software I do, it can just drop it in there.

我就想，天哪，真希望有这种东西。

And I thought, man, I wish there was something like that.

我到处找，想看看有没有现成的。

And I'm looking around trying to find it.

能不能在Raycast里做？

Could I do it in Raycast?

结果发现，哦，可以。

It's like, oh, yeah.

它有API，能实现这个功能。

There's an API and you can do it.

然后我就想，不如直接让ChatGPT帮我做一个。

And then I was like, I should just ask, you know, ChatGPT if it could make it for me.

对吧？

Right?

它做出来了，而且大概十五分钟就搞定了，做出来的正是我想要的。

It made it and made it in like fifteen minutes and it made exactly what I wanted.

它做出来了那个东西，然后它还提出了好几个后续的建议，比如‘好，这是第二阶段’。

It made the thing and then it was like and then it had multiple levels of things that it suggested is like, okay, here's phase two.

我就问，第二阶段是什么？

And I was like, what's phase two?

它说，咱们做个Markdown的玩意儿吧。

It was like, let's make a markdown thing.

你可以把它保存到某个地方，比如Google Drive，或者苹果的iCloud Drive，这样就能从桌面同步到笔记本上。

You can save it in some place like Google Drive or maybe like Apple, the Apple Drive, Apple so that you could have it shared to from your desktop to your laptop.

我就说，哦，这主意不错。

And I was like, oh, that's a great idea.

那就做第二阶段吧。

Go ahead and do phase two.

大概十分钟就搞定了，它也做到了。

Was like ten minutes and it did that as well.

所以，不到一个小时，我就有了这个完美的软件。

So so within like less than an hour, I had this perfect piece of software.

它就像是一个软件扩展，但我不会去应用商店卖它什么的。

It's like this software extension that, that like is not, I'm not gonna go have it sell it in the app store or something.

它就是专门为我做的。

It's just exactly for me.

就像厨房里的工具，你知道的，对。

It's like the kitchen tool that, you know, Yeah.

所以这很棒。

And so it's great.

我其实这周开始做了一些事情，让我很惊讶，我每天都被它能做的事情震惊到。

I I I actually started doing something this week that I'm amazed that, like, I I'm continuously every day amazed at at what it can do.

我觉得，这很大程度上取决于你给ChatGPT Work和Codex提供的上下文越多，它的能力就越强。

And I think, like, it's a lot about, like, you know, the more context you bring into ChatGPT Work and Codex, the more capable it is.

所以，你知道，如果你连接了你的邮箱和日历，甚至对于Plus订阅用户及以上，你还可以连接你的财务信息到ChatGPT Work。

So, you know, you if you connect your email and you connect your calendar, and you can connect even, you know, for for plus subscribers and up your finances to ChatGPT Work.

而且它几乎可以贯穿你生活的方方面面。

And it can just like operate across your entire life.

而且，现在每次我订机票、订酒店，或者订任何东西，我都会让 ChatGPT Work 去做一件事。

And like, I now I asked ChatGPT Work every time I book a flight or a hotel or like book anything.

我希望你每天检查一下价格有没有降，如果降了，看看能不能重新预订。

I want you to just check every day to see if the prices dropped and if I can rebook.

对。

Yeah.

它真的每天都做这件事，它会检查，比如酒店价格降了一点，你可以重新预订，你就直接重新订，简直就像魔法一样。

And it does it everyday and it like it checks like you know hotel dropped but a little bit you can rebook, you just rebook and it's like it's magic.

我的意思是，这真的就像有一个私人助理在为你办事。

I mean it's really having this like personal assistant that is is working on your behalf.

那太酷了。

That's very cool.

非常酷。

Very cool.

我想聊聊这东西的一些挑战，因为一旦你开始用它，很快就会意识到它能做多少事，然后你也会开始消耗大量的token。

I want to get to some one of the challenges of this stuff because once you start using it quickly you can realize like all the things it can do and then you can also start burning a lot of tokens as well.

所以我想问问这个挑战，因为我可以再给你举个例子。我之前跟Shoki，还有你的公关团队说过，我拿一个东西，让它每十分钟扫描一次网页，抓取内容，然后发到我们团队的Slack频道里，这样大家就不用一直刷新页面了。

So I want to ask about this challenge because I'll give you another example that I had where I had and I was telling Shoki your PR your PR team about this that I was taking something and having it scan something on the web every ten minutes, pull it, and then put it into a Slack channel for my team so that they didn't have to go and keep, like, refreshing, you know, a page.

效果非常好。

It works great.

它真的做到了。

It, like, it did it.

每次都会发一条Slack消息。

It just made a a Slack post every time.

后来我从专业版换到了Plus版，结果几个小时内token就用完了，因为它运行得太频繁了。

And then and I switched from, like, the pro tier to the plus tier, and I, like, ran out of tokens in, like, a couple hours because it's running, you know, so often.

所以你看，智能体确实很耗token。

And so, you know, agents can be really token heavy.

对吧？

Right?

所以，还有一件事，我和你们团队另一位PR成员聊过这个问题，劳伦，向劳伦致敬，就是，嘿，我该这么做吗，然后也许用其中一个经验？

And so one of the things another one of the PR team members I was talking to about this problem on your team, Lauren, shout out to Lauren, which is like, hey, should I do this and then maybe use like one of the lessons?

我该试试吗？

Should I try?

有没有办法告诉它用Luna？

Is there a way that I could tell it to use like Luna?

因为OpenAI最近降低了Luna的价格，变得非常合理。

Because OpenAI recently cut the prices for Luna, made it very reasonable.

但是，那还是有点，我一直在想办法怎么做，怎么告诉它，好吧，用这个模型，这个更便宜的模型。

And so, but then that's a little, I've still been trying to figure out how to do that, how to tell it, okay, use this model instead, this less expensive model.

当人们开始使用这些东西并发现它们的能力时，他们就会遇到这个问题，就是，哦，你可以，那个专业版是每月200美元，普通版是Plus版。

When people start, you know, working with these things and finding the capabilities, that's the thing that they start running into is like, oh, you can, so that pro tier is $200 a month tier, like the regular plus tier.

当我从专业版转到Plus版时，突然就烧掉太多token了。

When I went from pro to plus, all of a sudden I was burning too many tokens.

对吧？

Right?

我知道，这也是另一个关于易用性的问题，你肯定不希望用户还得操心这些。

I know that's another thing when you think about usability, you know, you would prefer users not to have to think about that.

比如，如果它能根据你当前的套餐自动选择模型，哪怕任务跑得慢一点，我想你肯定希望模型能自动做这个决定。

Like if the if it could if it could use the whatever tier you're on and the job might run a little slower, I'm assuming you would love for the model to automatically sort of make that decision.

我这是在替你把话说出来了。

So I'm putting words in your mouth.

但你觉得，随着人们开始更多使用这个产品，你会怎么管理这个问题呢？

But like, how would you think about managing that as people start using the product more?

对。

Yeah.

我觉得，你指出的其实是产品的一个缺陷，就是说，如果用户得自己决定哪个模型最适合哪个场景，用这个还是那个，那确实是个问题。

Mean, I think, again, I I think you're pointing out like a failure in in the product, which is, you know, if you if you have to have the if the user has to have the burden of deciding like what model is the best for, you know, which use case and, you know, using this model or that model.

我觉得这也是我们正在纠结的事情。

And I think this is something that, you know, we're wrestling with.

好的。

Okay.

这甚至可能回到那个权衡的问题，就是我们怎么把最强大的功能带给尽可能多的人，但同时还要解决这个简化的问题。

And it maybe even goes back to that that trade off of like how do we bring, you know, the most powerful things to as many people as possible, but still, you know, wrestle with this with this issue of, you know, like how do we make it simple?

所以，我想提一点，我们的研究团队非常出色，他们不断推动性能和效率的边界。

And so, you know, one thing I'll call out is that and our research team is is incredible, and they and they continuously push the frontier of performance and efficiency.

所以我们能够持续以更低的成本提供更强的能力，真的是每个月都在进步。

And so we're continuously able to give more capability at a more affordable cost, really month by month.

如果你回头看，如果我们谈论一年前有什么可用，你会发现很多以前属于高端功能的东西，现在突然变得更容易获得了。是的。

If you look back, if we talked about what's available a year it's like you look at all these features that used to be sort of like higher tier features and suddenly they're more available Yeah.

每美元每token的智能，或者抱歉，每token每美元的智能，已经大幅提升了。

Intelligence per dollar per token or sorry, the intelligence per token per dollar has only dramatically gone up.

你每花一美元能买到的智能量，所以从某种程度上说，我们的研究团队已经做出了这样的承诺，并且继续让这成为可能。

Like the amount of intelligence you can buy per dollar that you spend is So in some ways, our research team has made such a commitment to making that possible and continue to make that possible.

比如，即使是最高智能级别的模型，在token消耗方面也非常高效，这是我们非常非常在意并推动的事情。

Like, SOLE, even at the highest tier of intelligence, is such an efficient model when it comes to its token spend and that's something we really, really care about pushing.

但我认为，知道何时使用正确模型的这种用户体验，是我们正在深入思考的问题。

But I think this user experience of knowing when to use the right model is something that we're thinking a lot about.

好的。

Okay.

希望我们能把它做得更好。

And hopefully we'll be making it better.

因为理想的情况是，界面、工具、智能体来决定，比如，你在这个层级，所以我们就用这个模型。

Because like the ideal scenario would be like the interface, the harness, the agent makes a decision of like, okay, you're on this tier and so we're gonna use this model.

甚至可能用上一代的模型，它足够强大来完成这件事。

Maybe even a model that's sort of this model, the last generation model is powerful enough to do this thing.

对。

Yeah.

而且智能的成本要低得多。

And the intelligence is a lot less expensive.

是的。

Yes.

但它也无法预测你下周要做什么，对吧。

But it also can't predict what you're gonna be doing next week, right, either.

这也是挑战之一。

It's also one of the challenges.

但如果它知道你会在一小时内用完这些令牌，它可能会决定把你切换到另一个模型。

But if it knows that like, okay you're gonna burn through all these tokens you know in an hour then it's gonna maybe make a choice to move you to a different model.

所以，模型选择这部分，是不是也是你们思考产品和设计产品时考虑的一部分？

Is that so that's that's part of the that sort of model choice piece is that part of the way that you all think about the product too and design the product?

这可能是更大的问题。

That's maybe the bigger question.

是的，我的意思是，我们绝不想牺牲用户控制权，但我觉得我们确实想提供一个更顺畅、更理想的路径，让事情以最高效的方式完成。

Yeah, I mean, think we never wanna sacrifice like user control, but I think we do wanna give a more seamless, like happy path for for sort of getting things done in the most efficient way possible.

我觉得我们还没完全解决这个问题，但这是我们一直在思考的。

I don't I don't think we've quite cracked that yet, but this is something that we are we're at we're we're thinking about a lot.

我们如何尽可能拥有最智能的默认设置，也就是说，根据你正在尝试做的事情，我们如何引导你或默认你做出正确的选择，同时如果用户基于他们的任务想要覆盖这些决定，我们还要给他们所需的控制权。

How do we have the most intelligent defaults possible in the sense that like based on what you're trying to do how do we nudge you or default you to the right choice while giving users like the control they need if they want to based on their task to override those decisions.

是的，非常好。

Yeah, very good.

你们知道吗，三月份的时候，OpenAI的总裁格雷格·布罗克曼谈到了超级应用，对吧？

How about you know in March, Greg Brockman, president of OpenAI talked about, the super app, right?

我记得第一次是在一群记者面前，他甚至提到了这个。

The first time I was I remember being in a with a group of journalists and you know and he even mentioned it.

他谈到了超级应用，之前一直有传闻，然后他就直接承认了，是的。

He talked about the super app, and it was the first like, it had sort of been rumored, and then he just sort of admitted, yes.

我们正在开发一个超级应用。

We're working on a super app.

我们会把所有这些东西整合在一起。

We're gonna bring all these things together.

它会是一个统一的东西。

It's gonna be one thing.

但据我所知，OpenAI并没有明确说过我们现在就是超级应用，对吧？

But as I understand it, OpenAI has never necessarily declared, like, we are at the super app now, right?

我们不太喜欢超级应用这个词。

We don't like, I don't like the term the super app.

我觉得“超级应用”这个词不太准确。

I don't think that the super app term quite makes sense.

我们其实也聊过这个。

I think we also talk about that.

当然，Thibault 也公开说过，“超级应用”并不是我们内部用来描述我们正在做的事情的词。我们真正想的是，怎么把智能代理的能力带给尽可能多的人，也就是 ChatGPT 上超过十亿的用户。

Think Thibault, of course, has like also said publicly that like super app is not really an internal word that we use to talk about what we're It's trying to about how do we bring the power of agents to as many people as possible, the over a billion people on ChatGPT?

我们怎么在给高级用户提供最大能力、让他们能站在智能和能力前沿的同时，做出这些复杂的权衡？

How do we do so whilst making these complex trade offs between giving the power users the power to do the maximum set of things at the frontier of intelligence and capability?

我们怎么把这件事做好，让它尽可能成功？

How do we bring this and make this as successful as possible?

还有，我们怎么提供合适的产品体验和基础设施，来把这两者连接起来？

And how do we have the right product experience and infra to bridge those things?

我觉得这可能是一种非常平淡的表述方式，听起来没有“超级应用”那么激动人心。

That I think is the maybe very prosaic way of framing this thing that sounds as exciting than super app.

作为产品建设的指导方针，“超级应用”这个框架不如“我们到底在服务谁”来得有用。

In terms of a product building guidepost, the frame of the super app is not as useful versus oh, who are we trying to serve?

我们怎么才能做到这一点？

How do we make this possible?

我们怎么把最好的模型和能力带给用户，同时不阻碍模型发挥出真正的优势？

How do we bring the best models and capabilities to our users and get out of the way of the model being really excellent?

对。

Yeah.

不过我觉得，如今ChatGPT应用在移动端、网页端和桌面端，都能使用全套最强大的功能，这些功能都建立在同一个智能体框架上，并由我们的前沿模型驱动。

I do think though that like today, the ChatGPT app across mobile, web, and desktop has access to the full suite of most powerful features built on the same agentic harness and powered by our frontier models.

这其实是一路走来的过程，我们一直在努力确保，你只要打开ChatGPT，我们就能为你提供最强大、最实用的体验和个人助手。

And that is, you know, has been sort of a journey over time, making sure that, you know, you just come to ChatGPT, and when you're in ChatGPT, we can get you the most powerful and capable experience and personal assistant as possible.

我觉得我们还在这个过程中，对吧？

I think we're still on that journey, right?

你提到的一些事情也表明我们还在……

And some of the things that you brought up shows that we're

面临挑战。

The challenges.

我们还有一段路要走。

We have a way to go.

对，对。

Yeah, yeah.

但我觉得方向是对的，至少从这些功能的使用情况来看，我有信心我们走在正确的道路上。

But I think the direction and at least what we can see from from adoption of these features, I feel confident that we're in the right direction.

因为我觉得如果我们方向不对，就不会看到越来越多的人以新的方式使用ChatGPT，这正好说明我们想贴近用户的需求。

Because I think if we if we weren't, we wouldn't be seeing more and more people using ChatGPT in new ways, which is just, you know, we wanna meet our users where they are.

所以，超级应用应该是人们每天都会用的应用。

And so, like, the super app should be the app that people use every day.

我觉得，嗯，就是这样。

I think that's, like Yeah.

我是这么看的。

The the way I see it.

它就是那个你打开的东西，不管是桌面端、本地云、移动端还是网页端，你都不用担心自己是在跟对的东西还是错的东西对话。

It's just, the thing that you open, whether it's desktop local like, local cloud, mobile, web, you shouldn't have to worry that you're you're talking to the right thing or the wrong thing.

而且现在它就是一个应用了。

And and it is now one app.

也许这才是最重要的。

Maybe that's the most important thing.

因为，你知道，四个月前，你用的，或者我用的，是ChatGPT、Atlas和Codex。

Because even, you know, four months ago, you're using, or I was using ChatGPT, Atlas, and Codex.

对。

Mhmm.

你知道，现在只有一个东西了。

You know, now there's only one one thing.

你在桌面应用里用浏览器吗？

Do you use the browser in the desktop app?

Chrome扩展？

The Chrome extension?

但你用的是

But you use

就是那个Chrome扩展，还有应用内置的浏览器。

the Chrome extension and then just like the in app browser inside the

应用内置的浏览器？

The in app browser?

对。

Yes.

就像，我听说很多人都在用那个浏览器做很多浏览操作。

Like, it's still I've heard people saying that they're using it for a lot of like they're using a lot of their browsing right in the browser.

只是有些功能它还不支持，比如书签。

It just there's some things it doesn't do you know bookmarks.

对。

Yes.

你知道的。

You know

我们还有很多功能想加进去，但确实，那就是我们的目标。

other We have a list of many things we're trying to build into it but but yeah that is that is the aspiration.

我对这个想法挺开放的，如果你有一个大屏幕，可以把整个屏幕都用上，一半放ChatGPT，另一半放浏览器。

I'm sort of open to that idea of if you have a big screen and you can have the whole screen there and maybe on half of it is ChatGPT on one side and the other side is the browser.

对，那会挺不错的，对。

Yeah, that would be pretty Yeah,

现在就能做到，而且我们正在做很多改进。

can do that today and we're making a bunch of improvements to that

好的。

All right.

也是。

As well.

我一定会留意的。

I'll keep an eye on that for sure.

那你最期待它能做什么？

How about what's the thing that you can't wait for this to be able to do?

你知道它能做的事情还有很多很多。

You know there's so many more things that it can do.

跟几个月前相比，它能做的事情多了太多了。

There's so many more things that it can do from just a few months ago.

我们大概一年前聊过，那时候它就像个不存在的东西，现在却一直在不断获得越来越多的功能。

We talked about a year ago like it's just like something that didn't exist or fully like now is sort of gaining so many more capabilities all the time.

你希望用ChatGPT Work做什么，或者你期待它能做什么？

What's the thing that you wish that you could do with ChatGPT Work or you're looking forward to being able to do with ChatGPT Work broadly.

我知道不能透露产品发布的消息，

I know no product announcements,

对，不能透露产品发布的消息。

but Yeah, no product announcements.

我觉得，大家可能公开讨论过，但我最期待的一点是更强的持久性。

I think, I think folks have talked about this publicly, but one of the things I'm most excited for is like more persistence.

我觉得能够越来越长时间地做我的工作，而且我越来越多地从更高的层面去引导它，这真的很棒。

I think that being able to do my work over longer and longer periods where I am increasingly steering it from like a higher and higher level has been amazing.

从自动补全到让它构建一个功能，再到让它构建一个应用，再到像斜杠目标那样，如果你用过斜杠，现在事情能运行很长时间来完成工作，这个转变我非常喜欢，我很看好这个方向。

The transition from like auto complete to like have it build a feature, to have it build an app, to like slash goal, if you've used to slash now things running over really long periods to get stuff done has been, I love the direction of where that's going.

我个人很喜欢这个方向，因为它让我的工作日变得好多了，也更有趣了。

I love that direction for me personally because it makes my work day just so much better and so much more fun.

但我也很喜欢这种转变，我喜欢那种工作像多人游戏一样的感觉，我和朋友们还有智能体一起协作，共同完成任务。

But I also love that transition in terms of like, I love the idea that like work feels like a multiplayer game where I'm working with like my friends and agents to like get stuff done together.

而且我觉得这个未来越来越触手可及了。

And I just feel like that future is like more and more in our grasp.

Tara，你能举个例子吗？比如一个你试过的斜杠目标之类的。

Tara, can you give an example of something like a slash goal that you could share that tried?

最近就有一个。

A very recent one.

我想想最近做过的斜杠目标。

I'm trying to do like a recent slash goal that I might have done.

哦，我在尝试做一个原型，用来表达我对侧边导航改动的一些想法。

Oh, I'm trying to build a prototype to like express a bunch of ideas of like what I wanna do to to some like side nav changes.

所以就是直接斜杠目标，然后说，这里有一个模拟图。

And so just like slash goal, here's like a here's a mock.

就直接去做，看看会发生什么。

Just go do it and like see what happens.

嗯。

Mhmm.

对。

Yeah.

好的。

Okay.

非常好。

Very good.

是的，我非常赞同坚持这一点，而且从主动性的角度来看，我真的很期待，而且我觉得，你之前也提到过，我们感觉模型变得越来越主动了。

Yeah, would, big plus one to persistence and I think like from a proactivity perspective, I'm really looking forward to and I feel like, and you even mentioned this before about how we're feeling that the models are getting a little bit more proactive.

我们感受到这一点的方式，可能是通过定时任务。

And and and the way we're feeling that is, you know, maybe through scheduled tasks.

也许我们是通过打开应用时，模型在首页突然给出建议来感受到的。

Maybe we're feeling that through the model suddenly suggesting things when you open the app on the homepage.

但对我来说，这仅仅是个开始。

But to me, it's just that's just the start.

我是说，我真心觉得，如果我们想象一下那个能真正帮你在个人生活和工作中达成目标的私人助理，它们应该会对如何提供帮助充满好奇，并且能多想一步。

Because I like, I really think, like, you know, if we we imagine, like, our our our our true personal assistant that can help you achieve your goals both personally and and and at work, they should be curious about how they can help and think one step ahead.

我对我们与这些模型的互动方式变得越来越主动、不再只是被动响应、不再光等着我来提问感到特别兴奋。

And like, I'm really excited about the way we interact with these models becoming more and more proactive and less just reactive, not just waiting on me.

来问个问题。

To come ask a question.

没错。

Exactly.

我觉得我们现在已经开始这么做了。

And I think, you know, we're starting to do this.

所以你在 ChatGPT 的对话里会看到，模型开始主动建议一些它可以帮你完成的定时任务。

So you'll see that in conversations in ChatGPT, the model will start to suggest scheduled tasks that it can do.

嘿，要我帮你做这个吗？

Hey, do you want to this for you?

嘿，你想让我一周后再帮你看看这个吗？

Hey, do you want me to check on this in a week from now?

然后当你打开应用时，如果连接了插件，就能得到这些个性化的建议。

And then when you open the app, if you have plug ins connected, can get these personalized.

接下来该做什么？

What to do next?

但我认为我们其实还处在起步阶段，我觉得我们出色的研究团队正在非常努力地研究如何让模型更主动地帮助用户。

But I think we're really at the beginning and I think, like, our amazing research team is is is working really hard about how do we make the models more proactively helpful for our users.

而且我认为这将会解决一个问题，就是大家不需要再自己去琢磨，到底能用这些模型帮我做什么。

And I think that's going to, one, bring down this problem of, everybody needing to figure out, okay, what can I use these models to do to help me out?

对。

Yeah.

而且更多时候会是模型主动说，嘿，我可以这样帮你。

And it can be more about the model saying like, hey, here's how I can help you out.

我能再给你分享一个我特别期待的例子吗？

Can I give you another one I'm really excited about?

对。

Yeah.

这个想法没那么高大上，其实挺实际的，我就是希望它能跟我用的每一个第三方工具都兼容。

It's this is like less, you know, sort of like aspirationally high minded and it's really prosaic, which is I just want it to work with every single third party tool I use.

比如你那个Google Task就让我很头疼。

Like your Google Task thing kills me.

我之前就特别想让它跟Google Task配合。

I like wanted to work with Google Task.

我希望所有这些工具都能正常工作。

I want all those things to work.

我希望它们都能连起来，而且连接要稳定、可靠、体验要好。

I want them all to be connected and those connections to be excellent and really reliable and fantastic.

这些功能好不好用，完全取决于你给它提供的数据，数据好它才能帮你完成生活中的各种操作。

Like these things are as good as like all the data that you give it such that it can perform all these actions in your life.

电脑操作这个功能太厉害了。

Computer use is sick.

浏览器使用也很厉害。

Browser use is sick.

所有这些都很棒。

All these things are great.

我还是希望它能和Google Tasks配合使用，而且能和每一个系统兼容，不管是那些特别老式的系统，比如UCSF让我登录来管理病人护理的那种，还是那些很酷的新东西，比如我们用的Linear这样的好应用。

I still want it to work with Google Tasks out And of the so like it couldn't work with every single system, whether those are like really old school ones, like whatever, I don't know, my UCSF wants me to log into to manage my patient care all the way to like really cool new things, like awesome apps that we use like Linear, for example.

它应该能和所有东西配合。

It should just work with everything.

应该能和你所有的工具配合。

Should work with all your tools.

它就应该能正常工作。

It should just work.

当然，其中一部分是我们可以在插件平台上做的事情。

And of course, some of that is is things we can do to our plug in platform.

我们非常关心那个平台上的开发者。

We really care about developers on that platform.

对。

Yeah.

其他事情就像第三方工具，直接内置在我们的平台上。

Other things are like third party tools just like build into our platform.

现在用Codex真的很容易，他们可以很容易做到，尤其是在Codex里。

It's really easy now with Codex and they can do it really easily especially in Codex.

比如我们已经让在Codex里开发插件变得非常好用。

Like we've made it developing plugins really good in Codex itself.

哦，不错。

Oh, nice.

所以，是的，我就是想和我的第三方工具一起工作。

And so, yeah, I just wanted to work with my third party tools.

听起来你想

It sounds like you wanted to

和我的第三方工具一起工作，我确实想。

work with my third I do.

第三方工具，当然。

Party For sure.

还有，多个账户也很重要。

And and like multiple accounts too.

比如我想记住工作用的Gmail和个人用的Gmail。

Like I wanna remember my work Gmail and my personal Gmail.

你还有什么其他功能需求？

What are your other feature requests?

你尽管说，

You got us

你想要什么，

What do what do you want

一次说？

at a time?

我们想知道。

Wanna know.

对。

Yeah.

非常好。

Very good.

对。

Yeah.

那些都很棒。

Those are those are great.

我还有一个想法，你刚才也稍微提到了一点，就是当你现在打开ChatGPT，进入ChatGPT Work的时候，会有三个选项，比如“主动”之类的。

My other one would be, and you sort of touched on this a little bit, the, When you go to ChatGPT now and you go to ChatGPT Work, there'll be like three things where it says proactive

建议测试。

Suggest tests.

推荐一些东西。

Suggest things.

我希望它能变得越来越个性化，比如，嘿，参考一下我每天的早晨安排，然后说“你还没做这件事，你准备好发那封邮件了吗？”之类的。我希望它能更针对我个人。我知道这很有挑战性，但我觉得这会非常强大。

I would love for that to keep getting more and more personal of like, hey and like reference maybe my my daily my morning sort of thing and say you still haven't done this one yet do you are you ready to send that email or something like that I'd loan for to get like even more specific to me I know that's you know challenging but that could be really I think that would be so powerful.

就像你还没做完这件事。

It's like you still haven't done this thing yet.

去ChatGPT或ChatGPT Work里做，我就想，哦对，糟糕。

Go to ChatGPT or ChatGPT Work to do and I'm like, oh yeah, shoot.

我需要，是的，我只需要批准一下。

I need to, yes, all's I need to do is approve that.

对吧？

Right?

点一下，点批准，然后它就会帮我发邮件去完成某件事。

Click it, hit approve and then it sends the email for me to finish something.

那太棒了。

That would be amazing.

我喜欢这个。

I love it.

嗯，是不是

Well Are

你之前提到过，抱歉现在才问你问题。

there you mentioned earlier sorry to now ask you questions.

但你之前提到过，有很多记者或者甚至可能报道AI或科技的人，都还没用过智能体。

But you had mentioned earlier that there are lots of journalists or folks who even maybe cover AI or technology who haven't used agents yet.

那怎样才能让他们去用智能体呢？

What would get them to use agents?

我知道。

I know.

我知道。

I know.

这确实是真的。

This is this is true.

这太有意思了。

It's so interesting.

对吧？

Right?

不只是记者，还有很多在企业里工作、已经用AI好几年的人，我经常问他们，你们在用哪些智能体？

Like, not just journalists, but, like, people that that are, like, working enterprise that have been using AI for years and, like, I'll you know, the question I'm always asking people, like, so what what agents are you using?

你让它们做什么？

What are having it do?

类似这样的问题。

Things like that.

然后他们就说，是啊，我其实不太用智能体。

And and they're like, yeah, I'm not really into agent.

我还没真正试过智能体。

I haven't really tried agents yet.

这其实也涉及我问他们时的一个问题，就是，是什么在阻碍你？

This does get to actually one of the questions I had when I asked them, like, are the what are what's holding you back?

你想看到什么？

You know, what do you want to see?

大多数障碍其实都归结为隐私和安全问题。

Most of the obstacle actually comes down to like privacy and security.

对。

Yep.

他们对连接自己的Gmail或Slack这类事情有顾虑。

They they have reservations about connecting their Gmail or their Slack or that kind of thing.

所以我很想听听，你们对那些有这类顾虑的人是怎么说的，比如他们会担心自己的私人数据会不会被用来训练模型。

So I I'd love to hear like, do what do you all say to people that have sort of those reservations of like, is now my private data gonna be used to train models?

诸如此类的问题。

All of those kinds of things.

我们为企业客户提供了一整套经典的企业级保障措施，关于我们如何使用他们的数据。

There is like a whole set of things that we do for enterprise that are like classic enterprise guarantees about how we use their data.

我们一直有这方面的政策。

We've always had like a policy around this.

每个购买ChatGPT的企业客户，我们都和他们签有相关协议。

Like every enterprise who buys ChatGPT, like we have that agreement with them.

当然，对于使用我们商业计划的用户，也有一系列经典的企业级保障。

Certainly for folks on our business plan as well, like there's a set of classic enterprise things.

比如你的数据存在哪里？

Like where is your data stored?

有哪些安全限制？

Like what are all the security restrictions?

还有谁能查看和分享你在这里的信息？

Or like the around who can view and share information that you have here?

可以连接哪些插件？

What plug ins can be connected?

我们还深度投入了管理员相关的功能，比如他们对整个组织的控制、可见性和监控能力，数据如何流动，采取了哪些操作，整套功能都有。

There's also just a deep investment we've made in, admins and, like, what control and visibility and observability they have over their organization, the data, how it flows, actions that are taken, like that whole suite.

其中一些就是企业产品应该具备的基本功能，必须做得很好。

Some of that is just bread and butter things that should just be like excellent about any enterprise product.

所以我觉得更有意思的是，我们如何帮助企业处理那些不那么确定的问题，比如这个人有没有权限，而是如何利用模型本身来帮助他们评估这些数据共享甚至安全相关的问题。

And so like, I think that the more interesting things that we can do are how do we enable enterprises, like, for some of these less, like, like, deterministic things, like this person has access or this person doesn't have access, but how can we help them evaluate some of these, like, questions or these, like, like, data sharing or even security questions using the models themselves.

我们有一些非常棒的网络安全模型。

We have some really amazing, like, cyber models.

我们内部确实有非常优秀的人才在思考这些问题。

We have really amazing folks who think about this internally.

模型越来越擅长提出和理解这些问题了。

The models are increasingly good at being able to ask and understand these questions.

所以我不光对我们已经完成并持续投入的核心企业功能感到兴奋，也在想我们还能怎样用新的方式帮企业解决这个问题。

And so I'm not only excited about there are lots of like core enterprise things that we have done and continue to invest in, but like how can we also build products here that help solve this problem for enterprises in new ways.

对。

Yeah.

我觉得除了企业视角之外，同样重要的是用户始终拥有控制权。

I think also, outside of the enterprise, lens, what's so important is that users are always in control.

而且你知道，比如说一个用户想连接他的邮箱，对。

And, you know, I'd say for for for a user that wants to connect, for example, their email Yeah.

他们完全可以控制ChatGPT如何与该插件交互。

They have full control over, you know, ways ChatGPT can interact with that plugin.

ChatGPT可以设为只读，也可以总是询问。

The ChatGPT can be read only, and it can have also ask always.

对吧？

Right?

在它对你的任何外部插件做任何操作之前，它会先问你。

Before it does anything with any of your external plugins, it will ask you.

而且我觉得，这些选项和控制的存在，正是为了让用户掌握主动权，让他们能够体验这些功能，同时保持在他们感到舒适的水平上。

And I think, like, these are these options and and control exist precisely to make it so that the users are are at the steering wheel and that they're able to experience these capabilities, but at the at the sort of level of comfort that they have.

而且希望随着时间的推移，你能逐渐信任这个系统，因为你看到它问的问题是对的，而且我们不会深入得太深。

And and hopefully over time, you know, build trust with with the system as as you see that, oh, you know, it's asking the right questions and, you know, we're not diving too deep.

所以我觉得，无论是在企业里面对管理员，还是对我们所有的自助服务层级，一个重要的部分就是如何确保用户始终感到自己在掌控之中。

And so I think like a big part about this, you know, both in the enterprise with admins and both for all of our self serve tiers, is like how can we make sure that the users always feel in control?

这真的是我们构建产品的方式。

And that's like really how we're building the product.

那么，对于担心个人数据的人，比如担心它会用我的个人数据训练，然后有人能通过逆向工程提示词来找出我的私人数据，这样说是否公平？

And is it fair to say like on that personal data and stuff for the people who have concerns about, oh, it's going to train on my personal data and then somebody's going to be able to sort of reverse engineer the prompt and figure out my private data.

你知道，这是你愿意分享的事情吗？

You know, that's something that you feel comfortable, you know, sharing.

就是说，你不用担心这个。

Like, that's you don't have to worry.

对。

Yeah.

其实有多层安全、隐私和匿名化的保护。

There are multiple levels of sort of security and privacy and anonymization.

而且说到控制权，用户完全可以自己决定，他们和ChatGPT分享的任何数据能不能用来训练其他模型。

And again, in terms of control, users are completely in control over whether or not any data they share with ChatGPT can be used to train other models.

这完全由用户自己掌控。

It's fully user's control.

在设置里面吗？

In the settings?

对。

Yeah.

是的。

Yeah.

非常好。

Very good.

Tara和Ty，我知道我们可以继续聊很久。

Tara and Ty, we could keep doing this, I know, for a lot longer.

非常感谢你们抽出时间。

I really appreciate your time.

了解ChatGPT Work的更多内容真是太棒了。

It's been great learning about more about ChatGPT Work.

我知道观众们一定会非常兴奋地了解这些。

I know the audience is going to be really excited to learn it.

我希望通过这个，他们会去尝试一下，因为有很多事情它都能帮上忙。

And I hope that from this, of them will give it a try because there's a lot of stuff that it can help them do.

如果我们能让他们做一件事，比如选个任务，去处理停车罚单，或者用它解决停车罚单，或者去建个网站。

If we can tell them to do one thing, like go pick a task, go file your parking ticket or resolve your parking ticket with it or go make a site.

你好像建了个网站，而且真的很喜欢网站。

You made a site, it sounds like, and really love sites.

比如说，去做个网站。

Like, go make a site.

还有别的什么你会直接推荐他们试试的吗？

Like, are there other, like, things you would just tell them to try out of the box?

对。

Yeah.

我觉得如果你能试试桌面端的电脑使用功能，你会惊讶它居然能帮你办理登机手续。

I mean, I think if you can try computer use, on the desktop experience, it just blows your mind that it can just check into a flight for you.

简单的事情，比如整理邮件，基本上你在电脑上能做的事它都能做。

Know, simple things like triage email and, you know, do pretty much do anything that you would do on your computer.

对我来说，如果你想体验前沿技术，电脑使用功能就是最好的选择。

I think like that to me is like if you want to experience like the frontier, computer use is where it's at.

如果你刚开始用，可以让它帮你过一遍邮件，挑出那些你还没回复但需要你回复的。

If you're just getting started like have it go through your email for you and pick what are the things that I haven't responded to that you know need a response from me.

像这样的功能就很棒。

Things like that are great ones to

或者让它把邮箱里的垃圾邮件都删掉。

Or start delete all the spam in my email.

你知道的。

That's you know.

对。

Yeah.

比如告诉我哪些是垃圾邮件之类的。

Like tell me any of these like spam things.

嗯，我喜欢Ty说的重新预订酒店那个。

Well I like Ty's like hotel rebooking.

那个也特别好。

That's also a really good one.

我打算试试那个。

I'm gonna do that.

对。

Yeah.

你确实应该这么做。

You should definitely do that.

听起来很棒。

That sounds awesome.

非常

Very

好。

good.

谢谢你们两位。

Thank you both.

是的。

Yeah.

祝你们继续把产品做得更好。

Good luck as you keep making the product better.

谢谢。

Thank you.

把你的功能需求发给我们。

Send us your feature request.

继续分享你的反馈。

Keep keep sharing the feedback.

这很好。

It's good.

太棒了。

Excellent.

谢谢。

Thank you.