BuildSpeak每日 builder 文摘
今日归档生词本关于
🎙 播客Training Data· 2026 年 5 月 1 日· 5,551 词 · 约 28 分钟

OpenAI's Greg Brockman: Why Human Attention Is the New Bottleneck

SPACE 播放 / 暂停·←→ 上一句 / 下一句
Speaker 100:02 - 00:24
So Greg, thank you for coming back here. I don't think we ever charge you for rent. So maybe I'll send you an invoice later. But Greg, you've been part of like two really spectacular companies, Stripe as employee number four and then the first CTO. I just recently heard that they process 1,600,000,000, sorry, 1.6% of the global GDP.
Speaker 100:02 - 00:24
所以,Greg,感谢你再次来到这里。我想我们好像从来没收过你的房租。所以也许我晚点会给你寄张 invoice(发票)。不过,Greg,你参与过两家非常了不起的公司:Stripe,作为第四号员工,后来又成为第一任 CTO;我最近刚听说,他们处理的交易额占全球 GDP 的 1,600,000,000——抱歉,是 1.6%。
Speaker 100:24 - 00:25
You must be proud of that.
Speaker 100:24 - 00:25
你一定为此感到自豪。
Speaker 200:25 - 00:26
That's amazing.
Speaker 200:25 - 00:26
这太惊人了。
Speaker 100:26 - 00:34
You must be even more proud of the fact that OpenAI has almost a billion or maybe more than a billion, in weekly active users at this point.
Speaker 100:26 - 00:34
你一定更为这样一个事实感到自豪:到现在,OpenAI 的 weekly active users(周活跃用户)几乎已经达到 10 亿,甚至可能超过了 10 亿。
Speaker 200:34 - 00:37
I mean, it's all it's all very exciting. It shows you what technology can do.
Speaker 200:34 - 00:37
我的意思是,这一切都非常令人兴奋。这让你看到 technology(技术)能够做到什么。
Speaker 100:37 - 00:44
And, you're not just co founder and president, but you're also chief builder at OpenAI. I heard that that was one of your titles.
Speaker 100:37 - 00:44
而且,你不只是 OpenAI 的 co-founder(联合创始人)和 president(总裁),你还是 chief builder(首席建设者)。我听说那是你的头衔之一。
Speaker 200:44 - 00:48
I'm not sure if it's ever an official title, but I I've been called many things. Let's just say that.
Speaker 200:44 - 00:48
我不确定那是否曾经是个正式头衔,不过别人确实叫过我很多种称呼。就这么说吧。
Speaker 100:49 - 01:05
Well, you have a audience of great builders here, so we'll start from all the way at the bottom of the stack. OpenAI has multiple stacks of the business, one of which is compute. And you guys have been very aggressive, very aggressive on securing compute. Why is that?
Speaker 100:49 - 01:05
好,这里有一群非常优秀的 builders(建设者、创业者、创造者)在听你分享,所以我们就从最底层的 stack(技术栈)开始说起。OpenAI 的业务有多个 stack,其中之一是 compute(算力)。而你们在 확보 compute 方面一直非常激进,非常激进。这是为什么?
Speaker 201:05 - 01:32
Well, in many ways, we have a very simple business. We buy, rent, build, compute, and we resell it at a margin. That's it. As long as the margin's positive, then you want to scale it because the demand for solving problems, the demand for intelligence, that's unlimited. And the AIs that we have right now really are able to rise to the challenge of effectively any kind of problem that you want to throw at them.
Speaker 201:05 - 01:32
嗯,从很多方面来说,我们的业务其实非常简单。我们购买、租用、建设 compute(算力),然后在此基础上加上利润转售。就是这样。只要利润率是正的,你就会想把规模做大,因为解决问题的需求、对 intelligence(智能)的需求,是没有上限的。而且我们现在拥有的 AI,确实已经能够应对几乎任何你想抛给它们的问题。
Speaker 101:32 - 01:33
Do you have enough compute?
Speaker 101:32 - 01:33
你们有足够的 compute(算力)吗?
Speaker 201:33 - 01:36
No. Really? Yeah, definitely not.
Speaker 201:33 - 01:36
没有。真的吗?对,绝对不够。
Speaker 101:36 - 01:45
I was just with Matt Garman and he says the GPU compute availability in twenty twenty six rounds to zero. Don't you guys have all of it?
Speaker 101:36 - 01:45
我刚刚还和 Matt Garman 在一起,他说到 2026 年,GPU compute(GPU 算力)的可用性基本上约等于零。你们不是已经拿走了几乎全部吗?
Speaker 201:45 - 02:04
I mean, we have we we would love more. We're we're constantly out there hunting for more, honestly. And I'll tell you, like when we first launched, when we launched ChatGPT, I remember being on a call with my team and they were like, all right, how much compute should we buy? And I said, all of it. And they're like, no, no, seriously, like, come on, how much should we buy?
Speaker 201:45 - 02:04
我的意思是,我们——我们当然希望有更多。说实话,我们一直都在外面不断寻找更多算力。我跟你说,就像我们刚发布的时候,当我们推出 ChatGPT 时,我记得我和团队在开一个电话会,他们问我:好,那我们该买多少 compute(算力)?我说,全部都买。他们说,不不,认真点,拜托,到底该买多少?
Speaker 202:04 - 02:10
I'm like, no matter how fast we try to ramp compute, I guarantee we're not going be able to keep up with demand. And that has been true ever since.
Speaker 202:04 - 02:10
我当时想的是,不管我们多快地去提升 compute(算力)规模,我都敢保证,我们都不可能跟上需求。而这一点从那以后一直都是真的。
Speaker 102:11 - 02:34
That's fascinating. Moving up from compute, since I don't know if much of this audience can help you with securing more compute, because most of them are founders of startups. About architecture and scaling laws, what are the, where are we in the scaling laws? Are they still doubling each year? Are you changing architecture?
Speaker 102:11 - 02:34
这很有意思。从 compute(算力)这个话题往上走一步,因为我不确定今天在场的很多听众是否能帮你们拿到更多算力,毕竟他们大多是 startup(创业公司)的创始人。关于 architecture(架构)和 scaling laws(缩放定律),现在处在什么阶段?它们还在每年翻倍吗?你们在改变 architecture(架构)吗?
Speaker 102:36 - 02:39
What are you guys pushing on the frontier on the research side?
Speaker 102:36 - 02:39
你们在研究层面的前沿方向上,正在推进哪些东西?
Speaker 202:39 - 03:04
Well, I would say, first of all, the scaling laws are a deep and very beautiful mystery. They feel deeply fundamental. It's like this scientific truth that just like you think about physics and Newton's laws and things like that, there's somehow this truth of the universe. And they're empirical. We don't necessarily have all the theory to explain exactly why it works.
Speaker 202:39 - 03:04
嗯,我首先会说,scaling laws 是一个深刻而且非常优美的谜。它们给人的感觉非常基础、非常根本。就像你想到 physics(物理学)、Newton's laws(牛顿定律)之类的东西时,会觉得那里存在某种宇宙真理一样;这里似乎也有这样一种真理。而且它们是 empirical(经验性的)。我们未必已经拥有全部理论,能够精确解释它为什么有效。
Speaker 203:04 - 03:27
But to me, the most beautiful thing is that neural networks were really designed in the 1940s, before they were computers. And somehow, we've been able to take the exact ideas that were developed back then and apply increasing amounts of computation. And as you pour more compute into the models, they get correspondingly more capable. It just keeps going. There's no wall.
Speaker 203:04 - 03:27
但对我来说,最美的一点在于,neural networks(神经网络)其实早在 1940 年代就已经被设计出来了,那甚至还在 computers(计算机)出现之前。不知怎么地,我们竟然能够把当年发展出来的那些原始想法直接拿来,施加越来越多的 computation(算力/计算)。而当你向这些模型投入更多 compute(算力)时,它们的能力也会相应增强。这个过程一直在继续,没有墙,没有天花板。
Speaker 203:27 - 03:29
And I think that's a beautiful thing.
Speaker 203:27 - 03:29
我觉得这是一件很美的事。
Speaker 103:29 - 03:49
That's pretty beautiful. Are there more research or more algorithms that are in the works? Because in the past, we had neural networks, to your point, in the 1940s, but we didn't have the compute for it. Now that we have the compute for it, are we just pushing the same things, or are there new architectures and new ideas coming up?
Speaker 103:29 - 03:49
这确实相当美。那现在还有更多研究、更多算法正在推进中吗?因为正如你所说,过去我们在 1940 年代就有了 neural networks,但当时没有足够的 compute。现在既然我们有了 compute,我们是在继续推动同样的东西,还是说正在出现新的 architectures(架构)和新的想法?
Speaker 203:49 - 04:14
Yeah, so I would think of it as we absolutely have new ideas that are constantly powering what we do. It's very simplified to say, well, let's take a neural network from the 1940s and it in a gigawatt data center, right? We have made tons of innovations and we constantly are improving things. And sometimes these are micro tweaks. You just realize that the way you've been formatting data was not quite right, and that can actually be a very big deal.
Speaker 203:49 - 04:14
对,所以我会这样看:我们当然有新的想法,而且这些想法一直在持续推动我们所做的事情。要是把它简化成“拿一个 1940 年代的 neural network,把它放进一个 gigawatt data center(吉瓦级数据中心)里”,那就太过简化了,对吧?我们已经做出了大量创新,而且一直在不断改进。有时候这些改进只是非常细小的 micro tweaks(微调);你只是突然意识到,自己一直以来格式化数据的方式并不完全正确,而这实际上可能是非常重要的事。
Speaker 204:14 - 04:35
Sometimes it's larger. You think about the shift from the LSTM to the transformer. And I don't think the transformer is everyone's moved past the transformer as described in the other 2018 paper. So there's constant innovation happening. And I think of places that have been perhaps the most invested in long term research on how to improve the architectures, how to improve the fundamental algorithms, and how to get the paradigm shifts.
Speaker 204:14 - 04:35
有时候改动会更大。比如你可以想想从 LSTM 到 transformer 的转变。而且我并不认为 transformer 就停留在那篇 2018 paper(论文)里所描述的那个 transformer;大家其实早就已经在那个版本之上继续往前走了。所以,创新一直都在发生。我会特别关注那些可能在长期研究上投入最深的地方:如何改进 architectures,如何改进基础 algorithms(算法),以及如何实现 paradigm shifts(范式转变)。
Speaker 204:35 - 04:40
I think OpenAI has been leading the pack there. And that's something we continue to invest, and I see lots of fruit on the horizon.
Speaker 204:35 - 04:40
我认为 OpenAI 在这方面一直处于领跑位置。这也是我们会持续投入的事情,而且我已经看到地平线上有很多成果即将出现。
Speaker 104:40 - 04:54
Got it. And on the models, does OpenAI have a formal definition for AGI? Are we closed? Are we not closed? Pat and Sonya published this thing that we are at AGI functionally.
Speaker 104:40 - 04:54
明白了。那说到模型,OpenAI 对 AGI 有正式定义吗?我们算接近了吗?还是还没有接近?Pat 和 Sonya 发表过一个观点,说我们在功能上已经到了 AGI。
Speaker 104:54 - 04:56
Do you agree with that? Do you not agree with that?
Speaker 104:54 - 04:56
你同意这个说法吗?你不同意这个说法吗?
Speaker 204:56 - 05:16
Well, we do have a formal definition, but to some extent, one thing I have learned is that everyone has their own intuitions about what AGI is, and maybe you can view it as, according to my view of where we are, I think we're about 80% of the way there, in that we have models that are smart. They're very capable. They're able to, if you give
Speaker 204:56 - 05:16
嗯,我们确实有一个正式定义,但在某种程度上,我学到的一件事是,每个人对 AGI 都有自己直觉式的理解。也许你可以这样看:按照我对我们当前所处位置的判断,我觉得我们大概已经走完了 80% 的路,因为我们已经有了聪明的 model(模型)。它们能力很强。它们能够——如果你给它
Speaker 105:16 - 05:17
Are they smarter than you?
Speaker 105:16 - 05:17
它们比你更聪明吗?
Speaker 205:18 - 05:35
I mean, they're certainly more capable than I am at writing software, right? If you give it all the context, then yes, I think that they are just so capable. It's really remarkable. Like, does anyone here feel better at writing software than GPD 5.4? Oh.
Speaker 205:18 - 05:35
我的意思是,至少在写软件这件事上,它们当然比我更有能力,对吧?如果你把所有上下文都提供给它,那么是的,我认为它们的能力就是如此之强。这真的非常了不起。比如,这里有人觉得自己写软件比 GPD 5.4 更好吗?哦。
Speaker 205:35 - 05:58
Oh, it's got one. It's got one. All right, writing kernels. So even there, we're seeing massive gains from Exactly. And for some of our internal results, there we're really seeing if you pour the right kinds of If you have the right setup for your problem, then you're able to get really massive results out of very low level, even low level tasks.
Speaker 205:35 - 05:58
哦,有一个。还真有一个。好吧,写 kernel(内核)的话。所以即使在那方面,我们也看到了巨大的提升,完全正确。而且从我们的一些内部结果来看,我们确实看到,如果你投入正确类型的——如果你为你的问题搭建了正确的 setup(配置/环境),那么即使是在 very low level(非常底层)的任务上,甚至低层任务上,你也能够得到非常巨大的结果。
Speaker 205:58 - 06:44
And just to give you one example of how things have been trending, one of my systems engineers also very similar, he was like, hey, I haven't been able to get value out of the models for GPT-five, for 5.1, for 5.2 as well. For 5.3, Anil Lark, had prepared this design document for a very complicated systems optimization he was about to do. He handed it over to the model, went to sleep, waking up intending to give this to his team to work on for the next week. And when he woke up, was done, that the model had actually implemented the initial spec, had seen that it was slow, had added instrumentation, had actually run the code, used a profiler to figure out where things were slow, and iterated multiple times until it had gotten to an optimized result. And like, that is incredible.
Speaker 205:58 - 06:44
再举一个例子来说说事情是如何一路发展到现在的:我的一位 systems engineer(系统工程师)也有非常类似的经历。他当时说,嘿,我一直没法从这些 model(模型)里获得价值,对 GPT-five、5.1、5.2 都是这样。到了 5.3,Anil Lark 准备了一份 design document(设计文档),内容是一个他正要着手做的、非常复杂的系统优化。他把它交给了 model(模型),然后去睡觉了;他原本打算醒来后把这个交给自己的团队,让他们在接下来一周里完成。但等他醒来时,事情已经做完了——model(模型)实际上已经实现了初始 spec(规格说明),还发现它运行得很慢,于是加上了 instrumentation(监测埋点/仪表化),真的运行了代码,用 profiler(性能分析器)找出了慢在哪里,并且反复迭代了很多次,直到得到一个优化后的结果。这个真的令人难以置信。
Speaker 206:44 - 06:45
That's where we are.
Speaker 206:44 - 06:45
这就是我们现在所处的位置。
Speaker 106:46 - 07:18
And so what would you advise all startups here to do? Because the models keep getting more and more capable. I've asked this when Sam was here in the past. If you're building today, do you need to rebuild in two years when a new model comes out because all the functionality and all the capabilities all change around you? Do you need to make sure that you're not in OpenAI's way because just going to run over startups because the models are so much more capable?
Speaker 106:46 - 07:18
那么,你会建议在场所有 startup(初创公司)怎么做?因为这些 model(模型)变得越来越有能力了。我以前 Sam 在这里的时候也问过这个问题。如果你今天在做产品,那么两年后当一个新 model(模型)出来时,你是否需要重做,因为你周围的所有功能和所有能力都会随之改变?你是否需要确保自己不要挡在 OpenAI 的路上,因为随着 model(模型)能力越来越强,它会直接碾过这些 startup(初创公司)?
Speaker 107:19 - 07:26
How would you recommend a set of startup founders to build in this environment?
Speaker 107:19 - 07:26
你会建议一批 startup 创始人在这样的环境下做些什么、如何去 build(构建)?
Speaker 207:26 - 07:53
Well, of all, would say to lean in. The tools right now have become incredibly useful. And if you look even over the course of December, I think that we went from these agentic coding tools being like, they're like writing 20% of your code to writing 80% of your code, which means they go from being kind of a sideshow to being the main thing that you're doing. And I think we're doing that across all of the work that people do with computers, all computer work this year. And you can look at the recent progress on Codecs.
Speaker 207:26 - 07:53
嗯,首先,我会说要主动拥抱它。现在这些工具已经变得极其有用。而且如果你看一下,哪怕只是整个 12 月这段时间,我觉得我们已经从这些 agentic coding tools(agent 式编程工具)大概只能写 20% 的代码,发展到能写 80% 的代码了。这意味着,它们已经从某种配角式的东西,变成了你正在做的主要事情。我认为,今年这种变化会发生在所有人用电脑做的工作上,也就是所有 computer work(电脑工作)上。你也可以看看 Codecs 最近的进展。
Speaker 207:53 - 08:19
It's really changing from a tool for software engineers to a tool for anyone who's doing work with a computer. And just over the past week, we've released a bunch of features that just make it so much more powerful and capable. And one thing we just announced today is a new tool called Chronicle that plugs into Codecs, where it actually can see everything you're doing with your computer and can form memories of what's going on. And so you ask it a question. It instantly knows what you're talking about.
Speaker 207:53 - 08:19
它真的正在从一个给 software engineers(软件工程师)用的工具,变成一个给任何使用电脑工作的人都能用的工具。而且就在过去一周里,我们发布了一堆功能,让它变得强大和能干得多。还有一件我们今天刚刚宣布的事,是一个叫 Chronicle 的新工具,它可以接入 Codecs,实际上它能够看到你在电脑上做的一切,并且能对正在发生的事情形成记忆。所以你问它一个问题,它立刻就知道你在说什么。
Speaker 208:19 - 08:32
You're like, what was I doing five minutes ago? It knows. You're like, oh, what was this person talking about? It knows. To me, it was this real wake up call to realize you spend so much of your effort right now just explaining to computer what's going on.
Speaker 208:19 - 08:32
你会说,我五分钟前在做什么?它知道。你会说,哦,这个人刚才在说什么?它知道。对我来说,这是一个真正的警醒时刻:我意识到,你现在花了大量精力,只是在向电脑解释到底发生了什么。
Speaker 208:32 - 08:46
Like, why are you explaining to your computer what's going on? That makes no sense. And so I think what's going to happen over upcoming years is the models are going to get much more capable. We'll have better harnesses. We'll be able to solve harder and harder problems, come up with new knowledge, all of these things.
Speaker 208:32 - 08:46
比如说,你为什么要向你的电脑解释发生了什么?这根本说不通。所以我觉得,未来几年会发生的是,这些模型会变得强大得多。我们会有更好的 harnesses(支撑/调度框架)。我们将能够解决越来越难的问题,产出新的知识,以及做到所有这些事情。
Speaker 208:46 - 09:01
But there is a one time shift that's happening now, which is really about context. It's really about, is your AI able to you have all these meetings. You didn't include the AI. That's not very nice to the AI. You're asking it to help you with things, and it has no information.
Speaker 208:46 - 09:01
但现在正在发生一个一次性的转变,它真正关乎的是 context(上下文)。真正的问题是:你的 AI 能不能获得足够的上下文。你开了这么多会,却没有把 AI 包含进去。这对 AI 可不太友好。你让它帮你做事,但它根本没有信息。
Speaker 209:01 - 09:24
So I think really leaning into, how do you make sure the AI even has enough information in theory to solve the problem and then trust the models are going to really get there and improve. So I think it will be a constant cycle of improvement and iteration and leaning into the tools and talking to your friends and figuring out how they're using it, but that there is this investment, that's a one time investment that now is the time to make.
Speaker 209:01 - 09:24
所以我认为,真正应该投入的是:你要如何确保 AI 至少在理论上已经拥有足够的信息来解决这个问题;然后再去相信,这些模型最终会真正达到那个水平,并持续改进。所以我觉得,这会是一个持续改进、持续迭代、不断拥抱这些工具、和朋友交流、弄清他们如何使用这些工具的循环;但与此同时,确实有这样一项投资,而且这是一次性的投资,而现在就是进行这项投资的时候。
Speaker 109:24 - 09:35
And in terms of, let's say you set that all up, is OpenAI using Codecs differently than you think everybody else outside is using it?
Speaker 109:24 - 09:35
还有,换个角度说,假设你把这些都搭好了,OpenAI 使用 Codecs 的方式,和你认为外部其他所有人使用它的方式相比,有什么不同吗?
Speaker 209:35 - 10:02
Well, I think one of the amazing things about being at OpenAI is you do get to live in the future, right? You do get to really see the shape of what's emerging, and we can co design, right? We can really change the models, harness, everything together in order to better serve the needs that we see. And a lot of the approach we've been taking is, so we started with software engineering, and we set some clear guidelines, for example, saying that we still want a human to be accountable for all code that gets merged. Right?
Speaker 209:35 - 10:02
嗯,我认为在 OpenAI 工作最令人惊叹的一点之一,就是你确实能活在未来,对吧?你确实能够真正看清正在浮现事物的轮廓,而且我们还能共同设计,对吧?我们真的可以一起调整 model(模型)、harness 以及其他一切,从而更好地服务于我们所看到的需求。我们的很多做法都是这样展开的:我们先从 software engineering(软件工程)开始,并设定了一些明确的准则,比如说,我们仍然希望所有被 merge(合并)的代码都由人类来承担责任,对吧?
Speaker 210:02 - 10:19
So at the end of the day, is it a good thing to merge this piece of code? Is it well structured? Is it going to make our code base more maintainable? We want to make sure there's a human who is signing off to say yes. And I think that thoughtfulness of not just saying, oh, because it's blindly used this or Oh, we don't want to use this at all.
Speaker 210:02 - 10:19
所以归根结底,这段代码合并进去是不是一件好事?它的结构是否合理?它会不会让我们的 code base(代码库)更容易维护?我们希望确保有一个人明确签字认可,说可以。我认为这种审慎很重要,而不是简单地说,哦,因为可以就盲目使用这个,或者哦,我们完全不想用这个。
Speaker 210:19 - 10:54
Like, I think neither extreme is quite right. And then we are also going vertical by vertical within OpenAI to adopt these tools within finance, within sales, within IT. And there we have a small dedicated team who's really deeply understanding the domain, working with the people who are the experts in it in order to build skills, in order to modify the codec's UI, whatever it is that is needed in order to get it to be good. And then that's something we can then, once we have it in good shape, we will externalize and that we're able to ship that to all of you. And so we are starting to work with certain customers as well.
Speaker 210:19 - 10:54
比如,我觉得这两个极端都不太对。与此同时,我们也在 OpenAI 内部按 vertical(垂直领域)逐个推进,在 finance、sales、IT 中采用这些工具。在这些领域里,我们有一个小型专门团队,会非常深入地理解该领域,并与真正懂这个领域的人合作,去构建能力、修改 codec 的 UI,或者做任何为了把它打磨好所需要做的事。然后,一旦我们把它做到足够成熟,我们就会 externalize(对外开放)它,并把它交付给你们所有人。因此,我们现在也开始与一些客户合作。
Speaker 210:54 - 11:11
So for people who want to be very AI forward and want to be part of defining this revolution, that there's a place for that, and I'd love to talk afterwards. But yeah, think that just this desire to say, hey, we really want to be AI forward, really live in the future, and experience what it will be like for everyone else one year, two years, three years down the road.
Speaker 210:54 - 11:11
所以,对于那些想要非常 AI forward、想成为这场革命定义过程一部分的人来说,这里有属于你们的位置,我也很愿意之后再聊。不过,是的,我认为这种“我们真的想要 AI forward,真正活在未来,并提前体验一两年、三年后其他所有人将会经历的样子”的愿望,本身就很重要。
Speaker 111:11 - 11:39
Do you guys structure your company differently or the engineering teams differently because of living in the future? I mean, if you have to go way back, when my father learned computer science, he was just himself. Then we had these long software releases that became waterfall. And then when the web happened and the cloud happened, we had these two pizza teams and we had scrum. Now that we have these coding agents, how do you structure around everything differently?
Speaker 111:11 - 11:39
你们会因为这种“活在未来”的状态,而以不同方式来组织公司或 engineering team(工程团队)吗?我的意思是,如果把时间往回拉很远,在我父亲学习 computer science 的时候,他基本上就是单打独斗。后来我们有了很长的软件发布周期,变成了 waterfall。再后来 web 出现了,cloud 出现了,我们有了 two pizza teams,也有了 scrum。现在既然我们有了这些 coding agents(编码 agent),你们会怎样围绕这一切重新调整组织结构?
Speaker 211:39 - 11:55
I think we're still figuring it out. And there's certain places where you really see it. For example, the cost of building a prototype is cheap now. It's so cheap. And if you want to build a dashboard that used to be like, oh, would take someone like a week to do it, and you just do it now.
Speaker 211:39 - 11:55
我觉得我们仍然在摸索。而且在某些地方,你真的能看到这种变化。比如,现在构建 prototype(原型)的成本已经很低了,低得惊人。如果你想做一个 dashboard,以前可能会觉得,哦,这得花一个人大概一周时间,而现在你直接就能做出来。
Speaker 211:55 - 12:35
And so actually, a lot of the bottleneck has shifted to things like sharing. And so we actually have some internal work on this as well that, again, we will be externalizing of how do you make it really easy for anyone in your enterprise to build a dashboard, a widget, a bot, whatever the thing is, then share it with others. And then that starts to really put pressure on having good governance. You want your IT organization to be able to see all these different threads of execution that are happening, all the little things that are being shared around, have some control over data provenance to really make sure that, Okay, a good example of this is I think people are now starting to take their internal knowledge dumps, turn them into wikis. We have a really cool one of these internally.
Speaker 211:55 - 12:35
所以实际上,很多 bottleneck(瓶颈)已经转移到了分享这类事情上。因此,我们内部其实也在做一些这方面的工作,而且同样会对外开放:怎样让企业里的任何人都能非常轻松地构建一个 dashboard、一个 widget、一个 bot,或者任何别的东西,然后再分享给其他人。接着,这就真的开始对良好的 governance(治理)提出压力了。你会希望你的 IT 组织能够看见所有这些正在发生的不同执行链路、所有那些被四处分享的小东西,并且对 data provenance(数据来源)有一定控制,以真正确保——一个很好的例子是,我觉得现在人们开始把内部知识 dump 整理出来,转成 wiki。我们内部就有一个非常酷的这样的系统。
Speaker 212:35 - 13:03
And the thing you immediately think about is, well, if someone has a document in the internal knowledge base that was accidentally permissioned incorrectly, and they realize, oh no, I didn't want this information to be accessible, how do they fix that? So normally, it's they go into the doc, they change the permissions, but now there's these derived artifacts. And so you need to make sure you have some way of tracking through the system to say, well, this output document came from this source one. The source one is no longer accessible to this audience. Let's go in and validate that as well.
Speaker 212:35 - 13:03
你立刻会想到的问题是:如果某人放在内部知识库里的一个文档,权限设置一开始不小心配错了,而他们后来意识到,哦不,我并不希望这些信息可被这样访问,那他们该怎么修复?通常的做法是,他们进入这个 doc,修改权限;但现在还存在这些 derived artifacts(派生产物)。所以你必须确保,你有某种方式能沿着整个系统追踪下去,说,这份输出文档来自那份源文档;而那份源文档现在已经不再对这批受众开放了;那么我们也要进一步进去验证这一点。
Speaker 213:03 - 13:16
And so you have to start really building your technical architecture with awareness of the way that people are going to use this information. And it really changes how teams relate to each other because it really changes where the bottlenecks are and what's hard.
Speaker 213:03 - 13:16
所以,你必须真正开始在搭建技术架构时,就意识到人们将如何使用这些信息。这也确实会改变团队之间的协作方式,因为它真的改变了瓶颈在哪里,以及哪些事情是困难的。
Speaker 113:17 - 13:25
Do you think team sizes are going be a lot smaller? Are we going to have still human software engineers in a decade?
Speaker 113:17 - 13:25
你觉得团队规模会小很多吗?十年后我们还会有真人 software engineers(软件工程师)吗?
Speaker 213:25 - 13:46
Well, a decade is a long time from now. And the ceiling on this technology is really hard to internalize. I think that it is clear that what a company is will change in a lot of ways. I think that we're going to have this ability for solopreneurs to build very incredible businesses. And so anyone who has a vision, I think, will be able to realize it.
Speaker 213:25 - 13:46
嗯,十年其实是很久之后的事了。而且,这项技术的上限真的很难被真正内化理解。我认为,很明显,公司这种组织形态会在很多方面发生变化。我觉得,我们将拥有一种能力,让 solopreneurs(单打独斗的创业者)也能建立起非常惊人的企业。所以我认为,任何有愿景的人,都将能够把它实现出来。
Speaker 213:46 - 14:15
I think the jobs that you all have will become way easier in a lot of ways, way more fun. Now, it might be more competitive too, because everyone's going have these amazing tools. And so really figuring out what is your niche, what is your unique angle, is probably going to become the most important core. But a lot of how we run organizations right now, and there's almost only one way to organize large groups of people, where you have teams, you have management structures, and you have scopes, you have these hierarchies and all these things, maybe that can change. Maybe you can be much more flat, small teams that can really just do incredible things.
Speaker 213:46 - 14:15
我觉得,你们现在从事的工作在很多方面都会变得容易得多,也有趣得多。当然,它也可能会变得更有竞争性,因为每个人都会拥有这些惊人的工具。所以,真正弄清楚你的 niche(细分定位)是什么、你的独特切入角度是什么,可能会成为最重要的核心。不过,我们现在运营组织的很多方式——而且大规模组织人群几乎只有一种方式,就是你有团队、有管理结构、有职责范围,有这些层级和所有这些东西——也许是可以改变的。也许你可以变得更加扁平化,用小团队真正做出不可思议的事情。
Speaker 214:15 - 14:28
We're seeing it right now in mathematics, where these individuals on the internet are using GPT-five point four Pro to solve these unsolved math problems. Normally, we need a math team, and they're just doing it.
Speaker 214:15 - 14:28
我们现在已经在数学领域看到了这种情况:网上有些个人正在使用 GPT-five point four Pro 来解决这些尚未解决的数学问题。通常这需要一个数学团队来做,而他们现在单靠自己就在做了。
Speaker 114:28 - 14:34
Yeah. My son's a math nerd. I just told him that maybe we should be studying something else besides math.
Speaker 114:28 - 14:34
对。我儿子是个数学迷。我刚刚还跟他说,也许除了数学,我们应该去学点别的东西。
Speaker 214:34 - 14:51
But I well, but see, this is the question. Right? If you look at something like AlphaGo, Move 37, this move that just, like, changed humanity's understanding of the game. But the thing that was surprising is it made the game more interesting and important for humans. And maybe that will be true for these other domains too.
Speaker 214:34 - 14:51
不过我——嗯,但你看,这才是问题所在,对吧?如果你看 AlphaGo 的 Move 37,这一步棋几乎改变了人类对这项游戏的理解。但令人惊讶的是,它反而让这项游戏对人类来说变得更有趣,也更重要。也许在其他领域里也会是这样。
Speaker 114:51 - 15:05
True. What about common failure modes when you're building with production GenSync workflows? What do you see as the common things that founders get wrong and they're building incorrectly these days?
Speaker 114:51 - 15:05
确实。那么,在用 production(生产环境)GenSync workflows(工作流)进行构建时,常见的 failure modes(失败模式)有哪些?你看到创业者现在最常犯的错误、以及他们这些天构建方式不对的常见问题,通常是什么?
Speaker 215:05 - 15:29
Well, I think that these models, they have such power and really understanding how to operate them well takes thought. And so we've been investing a lot in primitives, security primitives, observability, having, again, good governance, things like that. But just to give you one anecdote that I think is evocative, I asked so I was working with my codex. I asked it to install some package that someone at OpenAI had written. It ran into an error.
Speaker 215:05 - 15:29
嗯,我认为这些模型拥有非常强大的能力,而真正理解如何把它们用好,是需要认真思考的。所以我们一直在 primitives(基础原语)、security primitives(安全原语)、observability(可观测性),以及良好的 governance(治理)之类的方面投入很多。不过我想讲一个我觉得很能说明问题的小故事:我当时在和我的 codex 一起工作,我让它安装一个由 OpenAI 某个人写的 package。结果它遇到了一个错误。
Speaker 215:29 - 15:38
I was like, oh, ping that person on Slack and asked them for help. So I pinged the person on Slack. Two minutes later, it said, this is taking too long. I've escalated to the person's manager. And it actually pinged the person's manager.
Speaker 215:29 - 15:38
我就说,哦,在 Slack 上 ping 那个人,找他帮忙。于是它就在 Slack 上 ping 了那个人。两分钟后,它说,这样太慢了。我已经把这件事 escalated(升级上报)给这个人的 manager 了。然后它居然真的去 ping 了那个人的 manager。
Speaker 215:40 - 15:56
And you realize it's like, on the one hand, it's kind of a reasonable thing for the model to do. It's being proactive. It's trying to solve my problem. It's like not just sitting around waiting to be told what to do. But on the other hand, like, you know, maybe it should have taken a little bit longer, maybe it should have checked with me.
Speaker 215:40 - 15:56
这时你会意识到,一方面,这么做对模型来说某种程度上其实挺合理的。它很主动,它在试图解决我的问题,它不是那种坐着等别人告诉它该做什么。但另一方面,你也会觉得,也许它应该再多等一会儿,也许它应该先问问我。
Speaker 215:56 - 16:11
And so I think that really thinking about these questions where we're still building up the EQ of the model and that in some places it's getting very good. For example, clicking approve, approve, approve is kind of where we've been. And humans are not very good at that either, right?
Speaker 215:56 - 16:11
所以我认为,真正去思考这类问题非常重要;我们现在仍在逐步建立模型的 EQ(情绪智力),而且在某些方面它已经做得很好了。比如,一路点 approve、approve、approve,基本就是我们过去的做法。而人类在这件事上其实也不怎么擅长,对吧?
Speaker 116:11 - 16:12
They default.
Speaker 116:11 - 16:12
他们会直接按默认选项走。
Speaker 216:13 - 16:26
They just default. And so now we're starting to have AIs that can actually take care of flagging, is this a high risk action? Hey, this one should be escalated. This one's okay to auto approve. And it really makes you realize that human attention is going to be this incredibly scarce resource, right?
Speaker 216:13 - 16:26
他们就是会默认处理。所以现在我们开始拥有一些 AI,能够真正负责做标记:这是不是一个高风险动作?嘿,这个应该 escalated(升级上报)。这个则可以 auto approve(自动批准)。而这也会让你真正意识到,人类的注意力将会成为一种极其稀缺的资源,对吧?
Speaker 216:26 - 16:36
The doing of things now is easy. The, is this a good thing? Is this what I wanted? Is this aligned with my values, with my desires? That is going to become the single most important bottleneck.
Speaker 216:26 - 16:36
现在,“去做事”本身已经很容易了。真正困难的是:这是不是一件好事?这是我想要的吗?这和我的 values(价值观)、我的 desires(意愿)一致吗?这将会成为最重要、也几乎是唯一的瓶颈。
Speaker 216:36 - 16:43
And so I think building systems that take that into account and really think about the human factor, that's the most important thing to do now.
Speaker 216:36 - 16:43
所以我认为,现在最重要的是构建能够把这些因素纳入考虑的系统,真正去思考 human factor(人的因素)。
Speaker 116:43 - 17:06
Another human factor is security. How would you advise people to think about security in this world of AI? And I just heard about breaches left and right with Vercel recently. These models are incredibly powerful at finding security holes. So how would you recommend people here use the models to find those security issues?
Speaker 116:43 - 17:06
另一个人的因素是安全。你会建议大家在这个 AI 世界里如何思考安全问题?而且我最近刚听说 Vercel 到处都在出现 breach(安全漏洞/入侵)事件。这些模型在发现 security hole(安全漏洞)方面强得惊人。那么你会建议在场的人怎样使用这些模型来找出那些安全问题?
Speaker 217:06 - 17:38
Well, I think there's a couple levels to the answer. I do think that this is I think that the Internet has been a place where security has been just like a a ratcheting important concern over time. You think about where it started going through the '90s with viruses and worms and malware and those things, and we've moved past that. I think we are also moving now to a much more ultimately secure regime, but it does require kind of an internet wide effort to get there. And so a lot of this honestly is just, again, leaning into the technology, having these models that can scan your code base, that can actually be used for end to end red teaming.
Speaker 217:06 - 17:38
我觉得,这个问题的答案有几个层面。我确实认为,随着时间推移,Internet 一直是一个让安全逐步变得越来越重要的地方。你回想一下它在 90 年代的发展起点——病毒、蠕虫、malware(恶意软件)之类的问题——而我们已经走过了那个阶段。我也认为,我们现在正在迈向一种最终会更加安全的体系,但要达到那一步,确实需要某种覆盖整个 internet 的共同努力。所以老实说,这里面很大一部分,还是再次要拥抱这项技术:拥有这些能够扫描你的 code base(代码库)的模型,能够真正被用于端到端 red teaming(红队测试)。
Speaker 217:38 - 18:05
There's a lot that can be done with them. And a lot of how we're thinking about further models and improvements there is really leaning into how do we actually leverage trusted access programs? How do we leverage the community of people who really care about being defenders and making the internet more secure? And I think that's something where everyone has a role to play and can participate. But the number one thing is just recognizing that these models are very powerful, but they're not magic, right?
Speaker 217:38 - 18:05
它们能做的事情很多。而且我们在思考后续模型以及这方面改进时,很大程度上也确实是在深入考虑:我们到底该如何利用 trusted access programs(可信访问项目)?如何借助那些真正关心成为防守者、让 internet 更安全的人群社区?我认为这是每个人都有角色可以发挥、也都能参与的事情。但最重要的一点,就是要认识到这些模型非常强大,但它们不是魔法,对吧?
Speaker 218:05 - 18:39
That they are just like a part of the overall resilience ecosystem. And I think that we as a society, and I think every company again, really contributes to this, have something to build in terms of how do we incorporate these in a way that results in more assurance and more sort of certainty on the impacts of whether it's this particular patch that you're taking, whether it's thinking about how do you make sure that you're just sort of rolling in updates quickly as they're being released. So I think that there's a lot of work to be done, but I have a lot of optimism for where this is going.
Speaker 218:05 - 18:39
它们只是整个 resilience ecosystem(韧性生态系统)的一部分。我认为,作为社会整体,以及我想每一家公司——再次强调——其实都在为此作出贡献,我们还有很多东西需要建设:我们该如何以一种能够带来更多 assurance(保障)和更多确定性的方式来纳入这些工具,无论是评估你正在采用的某个 patch(补丁)的影响,还是思考如何确保你能在 update(更新)发布时尽快把它们滚动部署进去。所以我认为还有很多工作要做,但我对它的发展方向非常乐观。
Speaker 118:40 - 19:02
Let's switch to speed. It seems like things are moving faster and faster and faster in the world of accelerating change. We were talking about it when you were walking up here around how you're trying to keep up with things. How do you keep up with all the accelerating change? How would you recommend everybody here keep up with everything that's changing?
Speaker 118:40 - 19:02
我们切换到速度这个话题。看起来,在这个加速变化的世界里,一切都在变得越来越快、越来越快、越来越快。刚才你走上台的时候,我们还在聊你是怎么努力跟上这些变化的。你是怎么跟上所有这些加速变化的?你会建议在场的每一个人如何跟上所有正在发生的变化?
Speaker 219:02 - 19:16
Well, I think this is the new normal. And I think to some extent, it's not really because of AI. I think it's just been the trend of technology for the past two decades. There's more people doing things. It's easier to do things than ever.
Speaker 219:02 - 19:16
我觉得,这就是新的常态。而且我认为在某种程度上,这其实不完全是因为 AI。我觉得这只是过去二十年 technology(技术)发展的一种趋势。做事情的人更多了。做事情也比以往任何时候都更容易了。
Speaker 219:16 - 19:37
Barrier to entry goes down. It means it's also much more easy to build value, to have great successes. And so I think that really trying to keep your ear to the ground and understand what's changing and to some extent, always starts with the same thing, which is play with the technology yourself. It's very different to hear AI described versus to use it. But the beautiful thing about AI is it's so intuitive.
Speaker 219:16 - 19:37
准入门槛在下降。这也意味着,创造价值、取得巨大成功都变得容易得多了。所以我认为,真正要做的是始终保持贴近现实,理解什么在变化;而且在某种程度上,这始终都要从同一件事开始,那就是你自己去玩一玩这项技术。听别人描述 AI,和你亲自去使用它,是完全不同的。但 AI 美妙的地方就在于,它非常直观。
Speaker 219:37 - 20:03
That's the whole point, is rather than have the machine be something you have to contort yourself to, the machine contorts itself to you. It's doing work for you. And it should be something where you ask it and does something. And so I think that just really trying to just get your finger on the pulse of what's changing, what's possible, where the models lag, that is, I think, the core skill that is going to really determine a lot of the success of companies in the future.
Speaker 219:37 - 20:03
这正是重点:不是你必须把自己拧成某种形状去适应机器,而是机器把自己调整来适应你。它是在替你做事。它应该是一种你提出要求,它就去完成事情的工具。所以我认为,真正努力去把握什么在变化、什么是可能的、模型在哪些地方还落后——这就是我所认为的核心技能,而它将真正决定未来很多公司的成功。
Speaker 120:03 - 20:24
And then on the flip side of that, you guys have held back models to work with security agents. So it's like the opposite of going as fast as possible. So you're doing things responsibly, too. So how do you think about the balance, because you're in a competitive environment. You want to ship as quickly as possible, and yet you're trying to do the right thing as well.
Speaker 120:03 - 20:24
但从另一个角度看,你们也曾为了与 security agents 相关的工作而暂缓发布一些模型。所以这就像是在做和“越快越好”相反的事。也就是说,你们也在负责任地推进。那么你们是怎么考虑这种平衡的?毕竟你们身处一个竞争激烈的环境中,既想尽快 ship(发布)产品,同时也想把事情做对。
Speaker 220:24 - 21:06
Yeah. I think at a values level, like what OpenAI is about, we really want to put the power of AI in people's hands. We want to empower people to build the future with the tools that are being created, but we need to do that in a thoughtful way, that we really think about both sides of here are the benefits, here are the risks, how do you maximize the benefits, how do you mitigate those risks. And I think that in cybersecurity and in biosecurity, those are areas where we're very thoughtful. We've been building we've been working on these kinds of both mitigations and trusted access programs for quite a long time, and that what we see coming is models that are going to be increasingly powerful and capable in a continuous way across all dimensions of capability.
Speaker 220:24 - 21:06
对。我觉得从价值观层面来说,这也是 OpenAI 的核心:我们真的希望把 AI 的力量交到人们手中。我们希望让人们能够使用这些正在被创造出来的工具去构建未来,但我们也必须以一种审慎的方式来做这件事,真正同时考虑两面:这边是收益,这边是风险,如何把收益最大化,如何缓解这些风险。我认为,在 cybersecurity(网络安全)和 biosecurity(生物安全)这些领域,我们一直都非常谨慎。长期以来,我们一直在建设这类缓解措施以及 trusted access programs(可信访问项目);而且我们看到的趋势是,未来的模型将在各种能力维度上持续变得更强、更有能力。
Speaker 221:07 - 21:20
And we announced last week the expansion of our Trusted Access for Cyber program. By the way, has anyone here applied? No one? Oh, I see one hand, two hands. Okay, more of you should apply.
Speaker 221:07 - 21:20
我们上周还宣布扩展了 Trusted Access for Cyber program。顺便问一下,这里有人申请过吗?一个都没有?哦,我看到一只手,两只手。好,更多人应该去申请。
Speaker 221:20 - 22:10
It's great. We really need help because it's very important that people who are trustworthy and responsible and really want to push these models are participating in this because that is how that's going to pay dividends for everyone. We're going to have more to announce over upcoming weeks on how we're expanding the program, and also when we release models to everyone, kind of the mitigations that we have and how we're going to tune those to be both to really balance, to really try to bring these capabilities as broadly as possible while also making sure that the ones that you know, that we're thinking about the risks and and able to to have some observability over them and to ensure that this is maximally positive in terms of deployment. So I think the short answer is, like, it's core to our mission. We care a lot about the impacts of what we're doing, not just building the technology in isolation, but it is a whole community and a whole world effort to really get to where we need to be.
Speaker 221:20 - 22:10
这很好。我们确实需要帮助,因为非常重要的一点是:那些值得信赖、负责任、并且真心想推动这些模型发展的人参与进来;因为这样做最终会让所有人受益。接下来几周,我们还会公布更多关于如何扩展这个项目的消息;另外,当我们把模型发布给所有人时,我们也会说明我们有哪些 mitigations(缓解措施),以及我们将如何调整这些措施,去真正做好平衡——一方面尽可能广泛地带来这些能力,另一方面也确保对那些我们所知、我们正在考虑的风险保持一定的 observability(可观测性),并确保这项部署尽可能带来正面效果。所以简短地说,这就是我们使命的核心。我们非常在意自己所做之事带来的影响,而不只是孤立地构建技术;这实际上需要整个社区、乃至整个世界共同努力,才能真正走到我们需要到达的地方。
Speaker 122:11 - 22:25
Now moving up from the models to the application layer, which is what a lot of people here are building, How do you how does OpenAI decide what in the application layer you're you're gonna build and what you're gonna leave out?
Speaker 122:11 - 22:25
现在把视角从模型上移到 application layer(应用层),也就是在场很多人正在构建的东西。OpenAI 是如何决定在应用层要做什么、又要留出什么不做的?
Speaker 222:26 - 22:35
Well, people have probably seen the word focus being applied to OpenAI quite a lot recently, possibly for the first Always time smiling. In a while.
Speaker 222:26 - 22:35
嗯,大家最近可能已经很多次看到“focus(聚焦)”这个词被用在 OpenAI 身上了,也许还是第一次——她一直在笑——过了挺久之后又这样。
Speaker 122:36 - 22:37
It's been applied to her, too.
Speaker 122:36 - 22:37
这个词也被用在她身上了。
Speaker 222:38 - 23:06
And it's hard because the field of AI is one of opportunity, right? It's like anything you can imagine is going to be great. No question, it's going to be great. And we, as a company, as a single company, no matter how much compute we build, no matter how many people we have, are only going to be able to do so much. And so a lot of we've been thinking about things is what is the sort of most focused strategy that covers the parts of the space?
Speaker 222:38 - 23:06
这很难,因为 AI 这个领域本身就充满机会,对吧?感觉就像任何你能想象到的东西都会很棒。毫无疑问,都会很棒。而我们作为一家公司,终究只是一家公司,不管我们建设多少 compute(算力),不管我们有多少人,我们能做的都只有那么多。所以我们一直在思考的一个重点是:怎样制定一种最聚焦的战略,去覆盖这个空间中的那些部分。
Speaker 223:08 - 23:29
Maybe it's an eightytwenty or just like the parts of the space that we think we can have most impact on. I think there it's very clear. Right now we're going through this agentic transition, and so products that are and it's not just about enterprise versus consumer, right? So it's like clear we are being very serious about enterprise. Like we're selling to to big companies and and building a whole muscle and sales motion there.
Speaker 223:08 - 23:29
也许这是个 eighty-twenty,或者说,只是我们认为自己能够产生最大影响的那部分领域。我觉得这一点非常清楚。现在我们正在经历这一场 agentic 转型,所以产品会——而且这不只是 enterprise(企业)和 consumer(消费端)之间的区别,对吧?所以很明显,我们现在对 enterprise 是非常认真的。比如,我们正在向大公司销售,并且在那边建立一整套能力和 sales motion(销售推进机制)。
Speaker 223:29 - 23:54
But consumer what consumer is is going to change. Right? It's kind of a very broad term that buckets in multiple things. But the the slice of consumer that's about not just productivity, but about goals, about achieving your goal, about even knowing what is your goal, being able to elicit that and having an AI that can proactively do that, it's all kind of the same thing. Like in the end, we're trying to build an AGI that you can talk to, that has all this context that you can use in your personal life, your work life.
Speaker 223:29 - 23:54
但 consumer 会是什么,这件事本身是会改变的,对吧?这是一个非常宽泛的词,里面装着很多不同的东西。但 consumer 里有一部分,不只是关于 productivity(生产力),而是关于 goals(目标)、关于实现你的目标,甚至关于弄清楚你的目标到底是什么,能够把这些引导出来,并且拥有一个能主动去做这些事的 AI——这些本质上都属于同一件事。说到底,我们想构建的是一个你可以与之对话的 AGI,它拥有所有这些 context(上下文),你可以把它用在个人生活里,也用在工作生活里。
Speaker 223:54 - 24:22
It's trustworthy, that you can go to it for advice and give you useful information, maybe health information or maybe about finances or about if you're trying to figure out what to do with your career. All of these things, they all kind of ladder into one thing. And it's meant we had to make some very painful decisions about what not to do. But I think I would just say that that's the aperture that we look at things through and that things that accrue to that singular vision of what we want to build you should expect us to pursue.
Speaker 223:54 - 24:22
它值得信赖,你可以向它寻求建议,它也能给你有用的信息,也许是健康信息,也许是财务方面的信息,或者是当你在思考自己的职业发展该怎么走时提供帮助。所有这些事情,最终都在层层汇聚到同一个方向上。这也意味着,我们不得不做出一些非常痛苦的决定,决定哪些事不做。但我想说,这就是我们看待问题的 aperture(观察口径),而凡是能够积累到、服务于我们想构建的那个单一愿景的事情,你都可以预期我们会去做。
Speaker 124:22 - 24:32
Got it. Do you think we'll be coding with command lines and agents in a few years, or is it going to be completely changed?
Speaker 124:22 - 24:32
明白了。你觉得几年之后,我们还会用 command line(命令行)和 agent 来写代码吗,还是说这一切都会被彻底改变?
Speaker 224:32 - 24:51
I think that we're in a very unnatural state right now for how we work. Like we all sit behind this box and kind of type away, and it's very clear our bodies were not designed for this. We got our carpal tunnel and our hunched shoulders and all these things. And I don't think we want that. I don't think any of us wanted think that we want Want
Speaker 224:32 - 24:51
我觉得,我们现在的工作方式其实处在一种非常不自然的状态。就像我们都坐在这个盒子一样的东西后面不停打字,而很明显,我们的身体并不是为这种状态设计的。于是就有了 carpal tunnel(腕管综合征)、耸塌的肩膀,以及这些各种各样的问题。而我不觉得这是我们想要的。我不觉得我们任何人真心想要——想要——
Speaker 124:51 - 24:52
more free time?
Speaker 124:51 - 24:52
更多空闲时间?
Speaker 224:52 - 25:04
We want more time. But it's not even about free time necessarily. It's like you want to spend more time with your loved ones? Yes. You want to spend more time talking to people and coming up with brilliant visions or just what you're excited about or just understanding yourself.
Speaker 224:52 - 25:04
我们想要更多时间。但这甚至不一定是 free time(空闲时间)的问题。更像是:你想花更多时间和你爱的人在一起吗?当然想。你想花更多时间与人交谈、提出精彩的愿景,或者只是去做那些让你兴奋的事,或者只是更好地理解你自己。
Speaker 225:04 - 25:32
So it's kind of like, do you want to be a CEO of an organization of 100,000 agents? That actually seems pretty good. And I think that we're all going to be able to get so much more done. But the mechanics of it are going to feel as different as going from having to write out things by hand with a quill or something to being able to just send a text message and have people go and working on your behalf on your goals.
Speaker 225:04 - 25:32
所以有点像,你想成为一个由 100,000 个 agents 组成的组织的 CEO 吗?这听起来其实相当不错。而且我认为,我们所有人都将能够完成多得多的事情。但其中的 mechanics(运作方式)会像这样发生变化:就像从必须用羽毛笔之类的东西手写一切,变成你只需要发一条 text message(短信),然后就会有人代表你、围绕你的目标去开展工作,那种程度的不同。
Speaker 125:32 - 25:49
All right. We talked about compute. We talked about model and security and agents and app layer. Let's talk about Frontier. When are the models going to be good enough to push the frontiers of science, physical AI?
Speaker 125:32 - 25:49
好的。我们聊过算力(compute),也聊过模型、安全、agent(智能体)和应用层。我们来谈谈 Frontier。模型什么时候才会足够强,能够推动科学前沿,以及 physical AI(物理 AI)的边界?
Speaker 125:49 - 26:14
We had Jen Fang here. It seems like LLMs have been a great scaling law for digital intelligence. It hasn't been as strong for robotics, for physical intelligence, for aspects of biology and science where the problems are probably a lot harder to verify or takes a long time to verify. Well, how are you keeping track of science and physical AI in the world?
Speaker 125:49 - 26:14
我们之前请到了 Jen Fang。看起来,LLM(大语言模型)一直是 digital intelligence(数字智能)里非常有效的 scaling law(规模定律),但对 robotics(机器人)、physical intelligence(物理智能),以及 biology(生物学)和 science(科学)中一些问题来说,效果似乎没那么强——这些问题大概更难验证,或者需要很长时间才能验证。那么,你们是如何跟踪当今世界里的 science 和 physical AI 进展的?
Speaker 226:14 - 26:56
Well, science is one domain that we're really leaning into, and we see line of sight to really incredible progress. And we're starting to have some signs of life, and I think it's always important to ground in what is happening today when trying to predict what will happen six months, a year from now. So for example, we had a physics result where our AI came up with this very beautiful formula that physicists who've been working on this for quite some time thought was totally impossible, thought it was maybe an unsolvable problem. And it's pretty significant. It's real serious physicists who really view this as a step towards really being able to get to some sort of answer for quantum gravity and all these things.
Speaker 226:14 - 26:56
science 的确是我们非常重点投入的一个领域,而且我们已经能看到通向真正惊人进展的清晰路径。我们也开始看到一些初步的生命迹象(signs of life)。我觉得,在试图预测六个月或一年后会发生什么时,始终立足于今天正在发生的事情是很重要的。比如说,我们最近在 physics 上有一个成果:我们的 AI 提出了一个非常优美的公式,而研究这个问题相当长时间的 physicists(物理学家)原本认为这完全不可能,甚至觉得这也许是个无解的问题。而且这件事相当重要。这是真正严肃的 physicists,他们确实把这看作是朝着 quantum gravity(量子引力)以及相关问题得到某种答案迈出的一步。
Speaker 226:56 - 27:20
It's not there, but it's a step. That's much bigger than where we were just a couple of months ago. And so it makes you really wonder a year from now, how far will we have traveled? Now, things like biology, that they are different from physics and math, that they are, you've got to leave your beautiful simulated world and deal with messy reality. But I think we've been learning how to deal with messy reality in other domains.
Speaker 226:56 - 27:20
当然,还没有到终点,但这是一大步。和仅仅几个月前相比,这已经是巨大的进展。所以这会让你不禁去想:一年之后,我们究竟会走到多远?当然,biology 这类领域和 physics、math(数学)不同——你必须离开那个优美的模拟世界,去面对混乱的现实。但我认为,我们已经在其他领域里学会了如何处理这种混乱的现实。
Speaker 227:20 - 27:45
Software engineering is a perfect example where we've really realized that just building the thing that solves programming competitions, that's not enough. You need something that's seen real world messy code bases, humans interrupting in different ways, this adversarial banging at it. And so I think that on science, I expect we're going to see a real renaissance. Maybe we'll see some big results this year. Next year, I think, is going to be a totally wild, wild time.
Speaker 227:20 - 27:45
Software engineering(软件工程)就是一个完美的例子。我们已经真正意识到,仅仅造出一个能解决编程竞赛问题的系统是不够的。你需要的是一个见过真实世界里混乱代码库(code bases)、能应对人类以各种方式打断、还能承受这种带有对抗性的反复敲打的东西。所以我认为,在 science 上,我们将会看到一次真正的 renaissance(复兴)。也许今年我们就会看到一些重大成果。而我觉得明年将会是一个完全疯狂、非常疯狂的时期。
Speaker 127:46 - 28:03
We live in interesting times. I promise that I get you out on time because you're a busy man. Before we let you leave we've got one minute on the shot clock what since you have no time, but soon you will have lots of time, what do you and Anna do for fun?
Speaker 127:46 - 28:03
我们正生活在一个有趣的时代。我保证会准时放你走,因为你是个大忙人。在让你离开之前,我们的 shot clock(倒计时)还剩一分钟。既然你现在没有时间,但很快你可能会有很多时间,那么你和 Anna 平时会做些什么来放松、找乐子?
Speaker 228:03 - 28:18
Fun? Mean, same as anyone. Like to watch movies, go on hikes, those kinds of things. Not as much time for it as maybe we'll hopefully have post AGI. But you've got to enjoy the ride along the way.
Speaker 228:03 - 28:18
找乐子?嗯,和其他人差不多。喜欢看电影、去徒步(hikes)之类的。只是现在可能没有那么多时间去做这些——希望到了 post AGI(AGI 之后)的阶段会有更多时间吧。不过,无论如何,你还是得享受这一路上的过程。
Speaker 128:18 - 28:21
Thank you, Greg, for joining us. Thank you, everyone.
Speaker 128:18 - 28:21
谢谢你,Greg,今天来参加我们的活动。也谢谢大家。
原文 ↗https://www.youtube.com/playlist?list=PLOhHNjZItNnMm5tdW61JpnyxeYH5NDDx8
BuildSpeak — 关于本项目BUILT IN PUBLIC · 跟随 builders 而非 influencers