探索人工智能交互中的潜在空间

欢迎收听AI Hustle播客。

Welcome to the AI Hustle Podcast.

今天这期节目，我们要聊一个来自Thinking Machines的全新产品。

Today on the show, we're talking about a brand new product that is coming out of Thinking Machines.

这是一个能在说话的同时还在听的AI模型。

It is a AI model that essentially listens while it talks.

就像人类一样，我们之前用Eleven Labs这类AI语音助手时，总遇到延迟问题。

So similar to a human, a lot of times we've had this latency issue with things like eleven Labs or other of these kind of AI voice receptionists.

这种技术适用的行业很多，但问题在于，你一说完，它得先接收所有数据、处理、思考回应，再尽快回你，中间总有个卡顿。

There's so many different industries that this applies to, but there's a latency where you speak and it has to try to take in all the data, compute it, think of its response, and spit it out to you as fast possible.

这个延迟的空档挺明显的。

And there's kind of a little latency gap.

他们现在的解决办法是，让AI在说话的同时也能持续接收信息，像人一样实现真正的双向实时互动。

So they're trying to solve this by essentially having the AI model listen and actively take in information while it's also talking, and be able to do this kind of simultaneously like a human would.

接下来我们会详细聊聊这些。

We're gonna get into all of this.

这也很有意思，因为这是来自米里亚姆·马雷达蒂的Thinking Machines公司，她筹到了不少钱，但之前推出的產品并不多。

It's fascinating also because this is coming out of Thinking Machine, which is from Miriam Maradi who's raised a ton of money and hasn't had a ton of products come out.

总之，今天要聊的内容挺多的。

So anyways, a lot to get into here.

不过在开始之前，我想提一下，如果你还没加入我们的AI Hustle Skool社群，我们非常欢迎你成为一员。

But before we do, I wanted to mention, if you haven't already joined our AI Hustle Skool community, we would love to have you as a member.

每周我们都会录制一段额外的视频，详细解析我们用来推动和扩展业务的工具、策略和方法。

Every single week we record a bonus video where we break down a tutorial on the tools, strategies, and things we're using to grow and scale our businesses with AI.

我们会向你展示我们正在实际操作的产品和项目。

We show you the actual products and projects we're actively working on.

我们还会公开我们的收入数据。

We show you revenue numbers.

我们会做深度剖析。

We show you deep dives.

所有那些我们不能公开分享的内容，都会告诉你我们到底在用什么工具。

Everything we can't share publicly, we show you the tools we're using.

所以这是一个很棒的地方。

And so it's an amazing place.

你可以去加入。

You can go join.

每月只要19美元。

It's only $19 a month.

我们努力让价格对每个人都很友好。

We try to make this affordable for everybody.

我们会保持这个价格不变。

We'll keep this price low.

如果你现在锁定这个价格，将来即使涨价，也不会影响到你。

And if you lock in this price, if we raise it in the future, it won't be raised on you.

这里已经有两百多位成员了。

There's over 200 members in here.

很多人正在积极搭建项目，互相分享进展。

So lots of people that are actively building and sharing what they're working on.

我们非常希望你能加入我们。

And we'd love to have you be one of them.

我会在描述里放上链接。

So I'll leave a link in the description to that.

杰米，我们来聊聊Thinking Machine Labs这个项目吧，你觉得这个软件最吸引你的地方是什么？

Jamie, kicking over to this story with Thinking Machine Labs, I guess, what was the most interesting part of this software to you?

我知道你个人也在做相关工具，还帮企业搭建类似AI接待员这样的系统。

Because I know you personally are working on, and you have tools, and you help businesses with kind of these AI receptionist kind of things.

那对你来说，第一反应是什么？

So for you, what are what are your first thoughts on this?

我觉得这个想法很棒。

I mean, I think this is a great idea.

他们确实解决了个实际问题，因为就算AI接待员的响应速度再快，说话和回应之间还是会有那么一点点不自然的延迟。

I think that's a real problem that they need to solve because, you know, even have you if even if you have really quick latency with, let's say, an AI receptionist, there's still a little bit of unnatural lag that happens between the time you finish speaking and then it starts talking back to you.

有时候你只是自然地停顿了一下，还没说完，它却已经抢着回应了。

And sometimes if you even if you pause, but it's a natural pause, like I just stopped talking for a moment, It will start responding even though you haven't finished your thoughts.

到目前为止，这些AI接待员语音助手确实进步很大，但离真正的人类水平还差一点。就像我现在说话的时候，你其实是在一边听一边思考、理解我的意思。

So, you know, I think there's up to this point, this AI, you know, receptionist AI voice agents have gotten really good, but they're still not quite to that that human level yet where, you know, as I'm talking right now, you could be you are thinking and processing what I'm saying.

传统的AI系统得先把你说的话转成文字，再分析上下文，最后才回复，整个流程是分步骤的。

A traditional agent will has to first turn all of your what your sentence into a text, then figure out the context and respond.

而这个系统想做得更像自然对话，让反应更流畅。

And it's it's kind of like a a chain process where this is this is trying to make it more conversational.

总之，我觉得这个想法很棒。

Anyways, I think it's a great idea.

我还觉得挺有意思，因为咱们以前在这档播客里聊过Thinking Machines Lab，挺搞笑的是，他们网站很久以来都特别神秘。

I also think it's interesting because we've talked about Thinking Machines Lab on this podcast before and how it's kind of funny because their website was very cryptic for a long time.

是的。

Yeah.

去年七月，他们融了20亿美元，估值达到120亿美元，可那时候连产品都没有。

Last year, they raised $2,000,000,000 in last July, giving them a $12,000,000,000 valuation, yet they had no product.

所以从我的角度看，这一切基本上都是靠米里亚姆·马雷达的个人声望和信誉撑起来的。

So it was basically, from my perspective, all based around Miriam Maradi's name, her credibility, and things like that.

说这么多，其实就是他们终于推出了一款产品，或者至少他们管这叫一个研究项目。

All that to say is we finally have a product, or at least they're calling it a research project.

他们还没正式发布产品，但我觉得这真能解决一个大问题。

They haven't officially released a product yet, but I think it's going to actually solve a really big problem.

我觉得，这个估值说不定真有道理。

And I think, you know, the valuation maybe will be legit.

你觉得怎么样，杰登？

So what are your thoughts on it, Jaeden?

是的。

Yeah.

我觉得这个生意里最有趣的一点是，我想聊聊一些技术细节，因为真的特别吸引人。

I mean, I I do think one of the interesting parts about this whole business I I wanna get into this some of the technical stuff here because I do think it's really fascinating.

不过说到这整件事，我们之前聊过Thinking Machines，也说过他们融了那么多钱，全靠玛丽亚·莫拉迪的名气。

But, like, as far as the whole business because we've talked about Thinking Machines before, and we're like, you know, they've raised so much money, and it's all kind of off of Maria Moradi.

不过有一点值得提的是，单靠她在OpenAI的股份，她可能早就成了亿万富翁了。

One interesting thing about her, though, that I will say is she is probably a multi multi billionaire just from her OpenAI stake.

所以当她去筹措几十亿美元的时候，人们就这么把钱给了她，这纯粹是靠名气和信誉在撑着。

So when she goes and raises a couple billion dollars, people like, just giving it to someone, I guess, that is just pure, you know, kinda clout and credibility.

最近有一篇文章提到，OpenAI的另一位联合创始人伊利亚·苏茨克韦尔，曾出庭为自己驱逐萨姆·阿尔特曼的行为辩护。

There was recently an article that I talked about how Ilya Seskever, one of the other co founders of OpenAI, he went up on trial to kind of defend why he kicked out Sam Altman.

总之，埃隆·马斯克和萨姆·阿尔特曼之间的那些纷争，闹得沸沸扬扬。

Anyways, there's all the drama between Elon Musk and Sam Altman trial.

但我最感兴趣的是，因为这是法律程序中的证据开示，我们才能看到这么多内幕细节，这种机会可不多见。

My favorite thing about the trial, though, is that we're getting all of the, like, nitty gritty details because of discovery and because it's, a legal case that you don't get anywhere else.

而在这些披露中，他承认自己持有大约七十亿美元的OpenAI股份。

And in all of that, he had to reveal that he has about a $7,000,000,000 OpenAI stake.

他是和米里亚姆·拉迪一起的原始联合创始人之一。

So he's one of the cofounders with Miriam Radi, one of the originals.

你可以想象，米里亚姆·拉迪可能加入得稍晚一点，但怎么说也是元老级人物了，她的股份估计在四十亿到七十亿美元之间，尤其现在OpenAI还在不断增值。

So you can imagine Miriam Radi, maybe she may have joined, like, a little bit later, and maybe she was a little bit more of an OG, But, I mean, she's gotta have somewhere between 4 and $7,000,000,000 in equity or a stake in OpenAI, especially as it's kind of growing and stuff.

所以她现在正忙着搞她的Thinking Machines实验室。

So she's over there doing her Thinking Machines Lab.

她正在融资，估值大概二十亿美元。

She's, you know, raising it, like, a $2,000,000,000 valuation.

她为这个项目融资二十亿美元。

She's raising $2,000,000,000 for it.

我想把估值抬得更高一点。

Let me get a higher valuation.

但她手上还握着一笔巨额财富，跟OpenAI紧密相关。

But she also has, like, a massive fortune just tied to opening eye.

归根结底，她主要的收入来源就在这里。

And at the end of the day, like, that's where she can make a lot of her money.

总之，我觉得这整个故事里，这部分特别有意思。

So, anyways, I do think that's a very interesting part of the whole story.

至于他们发布的技术本身，我其实非常欣赏。

But as far as the the tech of what they've released, I actually love it.

我当初特别好奇她会推出什么，毕竟如果她只是想做个跟ChatGPT竞争的大模型，那倒也不奇怪，但感觉她好像不是这个意思。

I think I was, you know, I was really curious what she was gonna come out with, because obviously, if she was just trying to make another competing LLM to chat GPT, which they very well might, who knows, but like, it didn't seem like it was I don't know.

你知道吧，你离开OpenAI，然后搞个跟它一模一样的复制版。

You know, you go leave OpenAI and just make another OpenAI clone.

这地方很厉害，因为它完全是全新的东西，这也正是我预期她和Ilya会做的——打造一些前所未有的平台和解决方案。

This is cool because it's something completely unique, which is kinda what I expected her and even Ilya's working on something, and that they're making these kind of completely unique platforms and solutions that didn't exist before.

他们这次具体做的这个，叫作全双工。

And so, specifically what they have done with this, it's called a full duplex.

这是个技术术语。

That's the technical term.

就像你提到的，现在的语音助手都是靠一种叫顺序循环的机制运行的。

And like you mentioned, voice assistants right now, they run on this, like, what's called a sequential loop.

我在说话。

So I'm talking.

它在听。

It listens.

它把我说的话转成文字。

It transcribes it.

然后它再回传回来。

It sends it back.

但当你和真人真正聊天时，你会发现完全不一样。

But when you have an actual conversation with someone, you'll know it's it's quite a bit different.

对吧？

Right?

比如想象一场激烈的争论。

Like, especially imagine like a heated argument.

我想的是播客，但也许你和某人吵得不可开交。

I'm thinking of like a podcast, but maybe you have a heated argument with someone you know.

你俩你一句我一句，对方刚说完，你就立刻打断说‘不对’，然后接着往下说。

And like, you kind of talk over each other, and they say something, and you say no, and then whatever.

对吧？

Right?

或者像一场特别激烈的谈判，大家语速飞快、你一言我一语地抢着说——这对AI来说简直难到不可能。

Or like a negotiation that's, like, really active, or people, like, talking really fast, and they're all talking over each other, that is incredibly hard slash impossible for an AI model.

我只是在想，我自己用ChatGPT的语音模式聊天时，那语音听起来确实挺自然的。

I'm just thinking about myself having conversations with ChatGPT on their voice mode, which, of course, the voice mode, it sounds natural.

听着挺舒服的。

It sounds good.

但问题在于，ChatGPT的语音模式总是给我一大段回复，其实它前半段就已经把问题答了。

But the problem with ChatGPT voice mode is it's, like, giving me a long response, and it kind of already answered the question in the first half of its response.

所以我心想，行吧。

And so I'm like, okay.

反正我也不需要这么多内容。

Anyways, like, I don't really need that.

别啰嗦了，直接说重点吧。

Like, move on.

再多讲讲这个。

Tell me more about this.

我一开口，它就卡住，然后说‘呃呃’。

As soon as I start talking, it's like it like freezes and it's like, uh-uh.

不好意思。

Sorry.

我没听清。

I missed that.

你刚说啥？

What did you say?

就是，对。

Like, it Yep.

你知道吗，它老是突然中断，搞得特别尴尬——我正说着话，它本该听完我的话再继续说完它的句子，然后马上接着回应我，但它就是卡住了。

You know, it just like cuts itself off and it's like has this awkward moment where if I was just talking and I had kind of maybe it would finish its sentence and listen to me at the same time, and then immediately start talking to me throughout it.

这让我想到我们现在用的Claude、CoWork，甚至OpenAI也在做类似的事：你可以在它回复中途发新信息，它会动态调整回答内容。不过对于短回答来说这不重要，因为它本来就很快。

And it's kind of reminds me of what we're actually getting out of something like Claude, co work, or maybe even OpenAI is kinda doing this too, where mid response, you can actually ping it with new information as it's kind of writing out its response, and it will change dynamically the response, which for quick little responses isn't that important because it does them pretty fast.

但像Cloud Cowork这种，它要花五分钟帮你完成一个复杂的多步骤项目，比如建个网站，这时候就关键了。

But when it's doing like something like Cloud Cowork, it's doing like this massive multi step project or building a website or a webpage for you and, you it takes five minutes.

中途你发条消息，它就能立刻调整方向，改变内容。

And halfway through, you send it a message, and then it can kind of pivot and change.

我不知道。

I don't know.

感觉是同一个概念，但他们用技术手段把响应时间缩短到了不是几秒或几分钟。

It feels like the same concept, but they've done this in a technical way where it's not seconds or minutes.

而是毫秒级的响应和延迟，这太惊人了。

It is milliseconds that they're able to get the turnaround and the latency down, which is incredible.

是的。

Yeah.

你觉得这会带来什么影响？

I mean, what do you think the implications are for this?

因为现在，我觉得很多人还是对比如AI接线员这种想法有抵触。

Because, you know, right now, I think I think people are still turned off by the idea of, like, for example, AI receptionists.

这需要很强的能力。

So it takes a lot of ability.

你得特别有创意地去呈现它，让人觉得这东西真适合他们的业务，尤其是大家还记着两年前打电话给技术支持时，遇到的那种特别机械的自动应答系统。

It takes a lot of You have to be very creative in how you present that to somebody as far as something that they would want for their business, especially because people are thinking back two years to the last time they called tech support somewhere and they had to talk to like a very robotic answering service.

他们会觉得，我绝对不要这种东西。

And they're like, I absolutely don't want that.

虽然已经改变了很多，但这次真的要再上一个台阶。

It's changed a lot already, but this is kind of going to take it to the next level.

那你觉得，这种技术要多久才能被大众接受，或者至少让企业觉得用起来安心？

So, you know, how do we what are your thoughts as far as, like, how long will this take to be adopted by by the masses or at least by by companies so that they feel comfortable with this?

嗯，我觉得未来几个月会先推出一个有限的测试版本。

So, yeah, I mean, I think there is a limited research preview that's coming in the next few months.

他们计划在今年晚些时候全面上线。

They're gonna have a wider release later this year.

我觉得，只要这个关键环节还没解决，我就特别好奇，OpenAI 是会直接照搬这套方案，还是其他公司会跟他们合作来实现它。

I think until this piece of the puzzle gets solved, and I'd be so curious if OpenAI just clones this, which pot very possibly they can, or if people integrate with them to help them do it.

我不太清楚这背后的技术到底有多复杂。

I'm not sure what the you how technical it actually gets.

但我觉得，只要这个环节没搞定，大家还是会觉得它太机械了。

But I think until this piece of the puzzle is solved, people are still gonna feel like it's quite robotic.

因为你说话的时候，它也在说，然后突然停顿，卡一下，那种尴尬感特别明显，完全就是机器人，谁受得了这个。

Because if you're talking and it's talking, and then it pauses, and you have this glitch, it's just this awkward moment, and it just feels like a robot and we don't like it.

归根结底，AI接待员这个想法一定会实现，到时候人们根本不会在意它说话是不是完美、行为是不是完美。

I think at the end of the day, the AI receptionist idea will be fully realized, people won't mind or care if it is sounds perfect, acts perfect.

只要它能把一个AI接待员该做的事全做到了就行。

It it accomplishes everything an AI reception an AI reception could accomplish.

也许它还拥有和真人一样的权限，比如能给你退款什么的。

Maybe it has all of the same authorizations to issue refunds or whatever.

对我来说，当我跟公司网站上的聊天机器人打交道时，只要它能给我退款、帮我解决问题，或者 whatever，只要它真能搞定我的事，我才不在乎是不是真人。

Like, for me, when I talk to chatbots on a company's website, if it can issue me my refund or help me with my problem or, you know, whatever the issue is, if it has authorization to actually solve my problem for me, I don't care if it's a human.

事实上，我更希望它不是真人。

In fact, I'd rather it not be a human.

我宁愿直接跟一个能解决我问题的机器人聊。

I'd rather just talk to a bot that can solve my problem.

但问题是，我们很多时候跟这些机器人对话，它们总是一堆问题问个没完。

But the problem is, a lot of times we talk to these bots and they ask all these questions.

它们总说，我帮你转接人工吧。

They're like, let me route you to a human.

是的。

Yeah.

然后就特别烦人。

And then it's just annoying.

感觉白浪费了一番功夫。

Just feel like you kinda wasted your time.

就像有个前台在先把你筛一遍。

Like, there was a screener kind of like screening you.

所以我觉得这可能是问题的一部分。

So I think that's probably part of the issue.

还有延迟问题，这正是他们现在要解决的。

And then there's like the whole latency problem, which is what they are basically solving here.

我觉得一旦这功能上线，再配上合适的权限，这些机器人会非常出色。

And I think once this rolls out and with the right authorization, that these these bots will be awesome.

另一点我觉得挺有意思的是，你之前提过好几次，很多人其实对人工智能这个概念不太感冒。

The other thing that I do think is interesting, because you mentioned this a couple times of how people don't really like the concept of AI.

甚至当你提到接待员的时候，他们脑子里想到的都是那些糟糕的经历。

And even when you talk about a receptionist, they they have these bad experiences they think back to.

我觉得要让人们对这个东西更有好感，关键不在于用AI取代人，比如用AI取代接待员，而在于强调：你现在能做以前做不到的事了。

One thing that I I think would give people a better, more positive connotation to all of this is less of replacing a person with the AI, so like replacing your receptionist with the AI, and more of, hey, you can do more than you did before.

以前你的客服人员，回复客户请求通常得等六到十二个小时。

Like, you know, before your customer support people, they would get around, you know, within six to twelve hours to customer support requests.

有时候忙起来，积压好几天都处理不完。

Or maybe you're getting super bogged down, you get days behind.

那种体验真的很差。

Like, you know, that's a very bad experience.

还有些时候，你根本记不得回电话，事情就这么不了了之。

Or maybe you're really bad at calling people back and it just doesn't happen.

所以我觉得，更好的说法应该是：以前你做不到，现在你终于能做到了。

So I think pitching things as like, you you were unable to do this before and now you're able.

而不是说，我现在就能把那个人换掉，或者做得快一点。

Versus like, you can you know, now I just can replace that person or do it a little faster.

稍微好一点，哪有彻底解决你之前一直掉链子的问题来得有力呢。

Like, a little better isn't as good as like, you are dropping the ball in this whole area and now this is gonna fix that problem.

所以，我也说不准。

So, I don't know.

我觉得这种视角的转变，可能更容易让人接受。

That's a kind of a a perspective shift that I think might be a better pitch for people.

确实如此。

For sure.

我还有一个问题想问你，因为我刚想到，最近我发现11 Labs已经不只做语音助手了。

I have one I have one more question for you because I just thought of this, but, you know, someone who is has seemed to expand beyond just, like, voice agents recently that I've noticed has been 11 labs.

他们已经开始加入一些真正的智能代理功能。

You know, they have they've adopted on some actual agent functionality.

他们还做起了音乐制作。

They have, you know, a little music production.

他们真的想把业务从语音扩展到整个音频领域。

They're they're really trying to expand on beyond voice just to audio in general.

你觉得呢？

Do you think?

我觉得这跟Thinking Machines的Full Duplex产品简直是绝配。

I mean, I feel like that would be a perfect pairing with Thinking Machines product here, full Duplex.

你觉得他们会跟这类公司合作，还是打算自己搞？我只是在想，他们该怎么推广这项新成果，真正把它变成赚钱的业务。

Do you think they're gonna partner with someone like that or do you think they're gonna try to I'm just trying to think of how they can market this and actually make money off this new development they've come

出来。

up with.

如果这技术特别难、特别专业，解决了别人搞不定的难题，那像11 Labs这样的公司肯定会跟他们合作。

If it's incredibly hard and technical and like they solve something that's very difficult for other people to solve, then I would see companies like eleven Labs 100% partnering with them on this.

但如果只是他们偶然发现了个挺酷的东西，而其实别人稍加努力也能做到——这在AI领域太常见了。

If it's just like they figured out something cool, but they figured out at first and everyone else can kinda figure it out, which is what it feels like happens so often in AI.

对吧？

Right?

就像当OpenAI说，我们有了一个推理模型，正在跑一大堆提示的时候。

Like as soon as OpenAI is like, we have a reasoning model and it's just running through a whole bunch of prompts.

大家都能大概看出来这推理是怎么运作的。

Everyone could kind of see how the reasoning worked.

他们反过来把原理给拆解出来了。

They reverse engineered it.

中国的所有实验室都照着复制了一遍。

All the Chinese labs cloned it.

现在我们所有人都有了推理模型，因为这根本不是什么特别难搞的技术。

And and then now we all have the reasoning models because they weren't like a crazy technical thing to figure out.

如果真是这样，我觉得OpenAI和谷歌这类公司大概会自己搞一套出来。

If that's the case, then I think you'll probably see companies like OpenAI and Google roll their own version of it.

就算不是这样，OpenAI和谷歌也可能自己做，因为他们本来就不太喜欢跟别人合作。

You might see OpenAI and Google roll their own version regardless, because they don't really like integrating or or working with other people.

他们更愿意自己琢磨怎么搞定它。

They can just try to figure out how to figure it out.

不过说实话，对很多小公司来说，OpenAI 和 Eleven Labs 现在已经算得上是大玩家了。

But, yeah, for a lot of smaller companies, I mean, OpenAI Eleven Labs is getting to be one of the bigger companies now.

但对于不是 OpenAI 和 Google 的公司来说，我觉得他们很可能会跟这两家合作，好尽快推出产品。

But for for ones that aren't OpenAI and Google, I think you'll see a lot of partnerships with with them to be able to get stuff down.

他们的响应速度能控制在 0.4 秒以内，比 OpenAI 和 Google 还快。

I mean, they're they're able to get the response time in point four seconds, which is faster than OpenAI and Google.

而且他们确实解决了一个实实在在的问题。

And I I mean, they solve a real problem.

但我想问的是，当他们说‘未来几个月内推出有限研究预览，今年晚些时候全面上线’的时候，

But my question is, when they say, look, limited research preview is coming in the next few months, wider rollout later this year.

这不就是说，到今年晚些时候，Google 和 OpenAI 早就该搞明白、早就该推出类似的东西了吗？

It's like, yeah, is later this year is Google and OpenAI not gonna have figured this out and already rolled this out?

所以我觉得关键问题就在于时间差。

So I think that'll be the big question is timing.

没错。

For sure.

是的

Yeah.

不是

No.

谢谢你的分享

Thanks for your input.

喂

Hey.

如果你喜欢这期节目，我们非常感谢你能给个评分和评论

If you enjoyed this episode, we'd really appreciate a rating and review.

希望你正在听

I hope you're listening.

这些反馈能帮我们触达更多人，我们真的很感激

Those help us reach more people and we really appreciate them.

也别忘了加入我们的AI Hustle Skool社群

Also be sure to check out our AI Hustle Skool community.

我们每周都会在那儿发布一些额外内容，教你怎么实际使用这些工具。

We release bonus content over there each week about how you can actually use some of these tools.

我们会持续跟进最新推出的工具，并为你提供详细教程。

We try to stay up to date on the latest ones that are released and give you guys tutorials.

不妨试试看，看看它们能不能真帮你赚钱。

Try them out, see if they can actually make you money.

你一定要来了解一下我们的Skool社区。

You're gonna wanna check out our Skool community.

我们非常欢迎你加入我们。

We'd love to have you be a part of it.

感谢收听，我们下次再见。

Thanks for listening, we'll see you next time.