**Jared Kaplan：** 大家好，我是 Jared Kaplan。今天我会简短聊一下 scaling（规模扩展）以及通往人类级别 AI 的道路。不过我猜在座各位对这些概念应该都不陌生，所以我会讲快一些，之后我们会和 Diana 做一个炉边对话式的问答环节。
**Jared Kaplan:** Hey everyone. I'm Jared Kaplan. I'm going to talk briefly about scaling and the road to human level AI, but my guess is for this audience, a lot of these ideas are pretty familiar, so I'll keep it short and then we're going to do a sort of fireside chat Q&A with Diana.
**Jared Kaplan：** 其实我进入 AI 领域才大约六年。在那之前，我漫长的职业生涯绝大部分时间都是做理论物理学家，在学术界工作。那我是怎么转到 AI 的呢？说来话长，但我尽量简短。我最初为什么选择物理？基本上是因为我妈妈是一位科幻作家，我想搞清楚我们能不能造出超光速引擎，而物理学是实现这个目标的途径。同时我也对理解宇宙充满热情——事物是如何运作的？我们周围一切背后最宏大的规律从何而来？比如，宇宙是确定性的吗？我们有自由意志吗？这些问题我都非常非常感兴趣。
**Jared Kaplan:** I actually have only been working on AI for about six years. Before that I had a long career, the vast majority of my career as a theoretical physicist, working in academia. And so how did I get to AI? Well, I want to be brief. Why did I start in physics? It was basically because my mom was a science fiction writer and I wanted to figure out if we could build a faster than light drive and physics was the way to do that. I also was very excited about just understanding the universe. How do things work? How do the biggest trends that underly sort of everything that we see around us, where does that all come from? For example, is the universe deterministic? Do we have free will? I was very, very interested in all of those questions.
**Jared Kaplan：** 幸运的是，在做物理的过程中，我遇到了很多非常有趣、思想很深刻的人，其中包括现在 Anthropic 的很多联合创始人，我们现在每天都在一起工作。我对他们做的事情很感兴趣，一直保持关注。而当我在物理学的不同方向之间辗转——大型强子对撞机物理、粒子物理、宇宙学、弦理论等等——我开始有点沮丧，有点无聊。我觉得我们的进展不够快。
**Jared Kaplan:** But fortunately, along the way, during my career as a physicist, I met a lot of very, very interesting, very deep people, including many of the founders of Anthropic that I now work with all of the time. And I was really interested in what they were doing and I kept track of it. And as I moved from different among different subject areas in physics from large hadron collider physics, particle physics, cosmology, string theory, and on I got a little bit frustrated, a little bit bored. I didn't feel like we were making progress quickly enough.
**Jared Kaplan：** 很多朋友告诉我 AI 正在变成一件大事。但我不信。我非常怀疑。我想，AI 这东西人们已经研究了 50 年了，SVM（支持向量机）也没什么激动人心的——那是我们在 2005 年、2009 年读书的时候唯一了解的东西。但后来我被说服了，也许 AI 会是一个值得投入的令人兴奋的领域。而且我很幸运认识了对的人，后面的事情就顺理成章了。
**Jared Kaplan:** And a lot of my friends were telling me that AI was becoming a really big deal. And I didn't believe them. I was really skeptical. I thought, well, AI, people have been working on it for 50 years. SVMs aren't that exciting. That was all we knew about back in 2005, 2009 when I was in school. But I got convinced that maybe AI would be an exciting field to work on. And I got very lucky to know the right people and the rest is history.
**Jared Kaplan：** 接下来我讲讲当代 AI 模型是怎么工作的，以及 scaling 是怎样让它们越来越好的。当代 AI 模型（比如 Claude、ChatGPT 等）的训练基本上有两个核心阶段。第一个阶段是预训练（pre-training），在这个阶段我们训练 AI 模型去模仿人类书写的数据和文本，并理解数据背后的关联模式。
**Jared Kaplan:** So I'm going to talk a little bit about how our contemporary AI models work and how scaling is leading them to get better and better. There are really two fundamental phases to the training of contemporary AI models like Claude, ChatGPT, etc. The first phase is pre-training and that's where we train AI models to imitate human written data, human written text and understand the correlations underlying that data.
**Jared Kaplan：** 这些图非常复古了。这实际上是最初 GPT-3 模型的 playground 截图。你可以看到，作为一个期刊讨论会的演讲者，你可能会期待我说某些特定的话。"elephant"（大象）这个词出现在那个句子里真的非常非常不合理。预训练做的事情就是教模型在大规模文本语料库中——现在还包括多模态数据——什么词最可能跟在什么词后面。
**Jared Kaplan:** And these figures are very very retro. This is actually from the playground of the original GPT-3 model. And you can see that as a speaker at a journal club, you're probably elephant me to say certain things. The word elephant in that sentence is really really unlikely. What pre-training does is teach models what words are likely to follow other words in large corpora of text and now with contemporary models multimodal data.
**Jared Kaplan：** 当代 AI 模型训练的第二个阶段是强化学习（reinforcement learning）。这也是一张非常复古的幻灯片，展示的是我们最早为 Claude zero 或者说 Claude negative one 使用的界面——那是 2022 年的远古时代，我们当时在收集反馈数据。你看到的基本上就是一个和非常早期版本的 Claude 对话的界面，你需要根据自己的判断——或者众包标注者的判断——挑选 Claude 的哪个回复更好。利用这个信号，我们优化模型，强化那些被判定为好的行为——有帮助的、诚实的、无害的行为，同时抑制那些不好的行为。
**Jared Kaplan:** The second phase of training for contemporary AI models is reinforcement learning. This is another very retro slide. It shows the original interface we used for sort of Claude zero or Claude negative one back in the ancient days of 2022 when we were collecting feedback data. And what you see here is basically the interface for having a conversation with very very early versions of Claude and picking which response from Claude was better according to you, according to crowdworkers, etc. And using that signal, we optimize, we reinforce the behaviors that are chosen to be good, that are chosen to be helpful, honest, and harmless. And we discourage the behaviors that are bad.
**Jared Kaplan：** 所以训练这些模型真正要做的就是两件事：学习预测下一个词，然后通过强化学习来学会完成有用的任务。而事实证明，这两个训练阶段都存在 scaling laws（缩放定律）。
**Jared Kaplan:** So really all there is to training these models is learning to predict the next word and then doing reinforcement learning to learn to do useful tasks. And it turns out that there are scaling laws for both of these phases of training.
**Jared Kaplan：** 这张图是我们大约五六年前做的，它展示了当你增大预训练阶段的规模时，模型的性能会可预测地越来越好。这个发现的起因是我在问一个最蠢的问题——作为物理学家，这就是你被训练要做的事。你要看全局，然后问非常笨的问题。我听说在 2010 年代，大家都说大数据很重要，所以我就想知道，数据到底应该有多大？它到底有多重要？能帮多少忙？同样，很多人注意到更大的 AI 模型表现更好。所以我们就问了一个问题：这些模型到底好了多少？
**Jared Kaplan:** So this is a figure that we made five or six years ago now and it shows how as you scale up the pre-training phase of AI, you predictably get better and better performance for our models. And this was something that came about because I was just sort of asking the dumbest possible question. As a physicist, that's what you're trained to do. You sort of look at the big picture and you ask really dumb things. I'd heard it was very popular in the 2010s to say that big data was important and so I just wanted to know how big should the data be? How important is it? How much does it help? Similarly, a lot of people were noticing that larger AI models performed better. And so we just asked the question, how much better do these models perform?
**Jared Kaplan：** 我们非常幸运，发现 AI 训练背后存在一些非常非常精确和令人惊讶的东西。这真的让我们震撼——存在这些整洁的趋势线，其精确程度堪比物理学或天文学中的任何发现。这给了我们很大的信心，让我们相信 AI 会以一种非常可预测的方式持续变得越来越聪明。因为你可以在这些图中看到，早在 2019 年，我们就已经在跨越了计算量、数据集大小和神经网络规模的很多很多个数量级上观察了这些趋势。当你看到一个规律在非常多个数量级上都成立时，你自然会预期它很可能会继续成立相当长的时间。所以这基本上是我认为推动 AI 进步的根本性因素之一。
**Jared Kaplan:** And we got really lucky. We found that there's actually something very very very precise and surprising underlying AI training. This really blew us away that there are these nice trends that are as precise as anything that you see in physics or astronomy. And these gave us a lot of conviction to believe that AI was just going to keep getting smarter and smarter in a very predictable way. Because as you can see in these figures already back in 2019, we were looking across many many many orders of magnitude in compute, in data set size, in neural network size. And so we expected once you see something is true over many many many orders of magnitude you expect it's probably going to continue to be true for a long time further. So this has sort of been one of the fundamental things that I think underlies improvements in AI.
**Jared Kaplan：** 另一个其实也很早就开始出现的因素——虽然它在最近几年才变得真正有影响力——就是你在强化学习阶段也能看到 scaling laws。大约四年前有一位研究员决定研究 AlphaGo 的 scaling laws，基本上是把两个非常高调的 AI 成果——GPT-3 的预训练 scaling 和 AlphaGo——结合在一起。这位研究员叫 Andy Jones，他当时是独自一人工作，我记得可能只有一块 GPU——那是远古时代了。
**Jared Kaplan:** The other is actually also something that started to appear quite a long time ago although it's become really really impactful in the last couple of years is that you can see scaling laws in the reinforcement learning phase of AI training. So a researcher about four years ago decided to study scaling laws for AlphaGo. Basically putting together two very very high-profile AI successes, GPT-3 and scaling for pre-training and AlphaGo. This was just a researcher Andy Jones working on his own with like his own I think maybe single GPU back in these sort of ancient days.
**Jared Kaplan：** 他研究不起 AlphaGo，太贵了，但他可以研究一个更简单的棋类游戏叫 Hex。所以他画了你们看到的这张图。ELO 评分在当时还没有那么广为人知，但 ELO 评分本质上就是国际象棋等级分。它们描述的是一个棋手在国际象棋中击败另一个棋手的概率。现在 ELO 被用来评估 AI 模型——看人类多频繁地偏好一个 AI 模型而不是另一个。但当时它就是经典的国际象棋评分应用。
**Jared Kaplan:** And so he couldn't study AlphaGo, that was expensive, but he could study a simpler game called Hex. So he made this plot that you see here. Now, ELO scores, I think, weren't as well known back then, but all ELO scores are, of course, is chess ratings. They basically describe how likely it is for one player to beat another in a game of chess. They're used now to benchmark AI models to see sort of how often does a human prefer one AI model to another. But back then this is just sort of the classic application of ELO scores as chess ratings.
**Jared Kaplan：** 他观察了当你训练不同的模型去玩 Hex 这个游戏——一个比围棋简单一些的棋盘游戏——它们的表现如何。结果他看到了这些惊人的直线。在科学中，发现非常简单的趋势本身就是一种能力。我认为这个发现当时被忽视了。人们没有足够早地关注强化学习中这种 scaling 行为，但最终它得到了印证。
**Jared Kaplan:** And he looked at as you train different models to play this game of Hex, which is a very simple board game, a bit simpler than Go, how do they do? And he saw these remarkable straight lines. So it's sort of a skill in science to notice very very simple trends and this was one I think it went unnoticed. I think people didn't focus on this sort of kind of scaling behavior in RL soon enough but eventually it came to pass.
**Jared Kaplan：** 所以我们看到的基本上是：无论在预训练还是强化学习中增大计算量，都能获得越来越好的性能。我认为这才是驱动 AI 进步的根本因素。不是 AI 研究员特别聪明或者突然变聪明了，而是我们找到了一种非常简单的系统性方法来让 AI 变得更好，然后我们在不断转动这个把手。
**Jared Kaplan:** So we see that basically you can scale up the compute in both pre-training and RL and get better and better performance. And I think that's sort of the fundamental thing that is driving AI progress. It's not that AI researchers are really smart or they suddenly got smart. It's that we found a very very simple way of making AI better systematically and we're turning that crank.
**Jared Kaplan：** 那么这解锁了什么样的能力呢？我倾向于从两个维度来思考 AI 能力。我觉得不太有趣但仍然很重要的维度是 AI 的灵活性——AI 适应我们所处环境的能力。比如把 AlphaGo 放在这张图上，它会在 X 轴下方很远的地方，因为虽然 AlphaGo 超级聪明——下围棋比任何棋手都强——但它只能在围棋棋盘这个宇宙里运作。自从大语言模型出现以来，我们在让 AI 能够处理人类所能处理的各种模态方面取得了稳步进展。目前我觉得还没有能闻到气味的 AI 模型，但那大概也快了。沿着 Y 轴往上走，你就能得到在现实世界中能做越来越多相关事情的 AI 系统。
**Jared Kaplan:** So what kinds of capabilities is this unlocking? I tend to think of AI capabilities on two axes. I think the less interesting axis, but it's still very important is basically the flexibility of AI, the ability of AI to meet us where we are. So if you put say AlphaGo on this figure, it would be very very far below the X-axis because although AlphaGo was super intelligent, it was better than any Go player at playing Go, it was only able to operate in the universe of a Go board. But we've made steady progress since the advent of large language models making AI that can deal with many many many all of the modalities that people can deal with. We don't have AI models I think that have a sense of smell. But that's probably coming. And so as you go up the y-axis here you get to AI systems that can do more and more relevant things in the world.
**Jared Kaplan：** 不过我觉得更有趣的维度是 X 轴——一个人完成 AI 模型能做的那类任务需要多长时间。这个指标随着我们提升 AI 能力在稳步增长。这就是任务的时间跨度（time horizon），一个叫 Metr 的机构对此做了非常系统的研究，发现了又一个 scaling 趋势。他们发现，AI 模型能完成的任务时长大约每 7 个月翻一倍。
**Jared Kaplan:** I think the more interesting axis though is sort of the x-axis here which is how long it would take a person to do the kinds of tasks that AI models can do and that's something that has been increasing steadily as we increase the capability of AI. This is sort of the time horizon for tasks and an organization Metr studied this very systematically and found yet another scaling trend. They found that if you look at the length of tasks that AI models can do, it's doubling roughly every 7 months.
**Jared Kaplan：** 这意味着，通过增加预训练和强化学习的计算量所注入到 AI 中的日益增长的智能，正在转化为可预测的、有用的任务能力，包括越来越长时间跨度的任务。所以你可以推测这个趋势的走向。AI 2027 的作者们就这样做了。这种图景暗示，在未来几年内，我们可能会达到这样一个节点：AI 模型能完成的任务不再只需要几分钟或几小时，而是几天、几周、几个月甚至几年。
**Jared Kaplan:** And so what this means is that the increasing intelligence that is being baked into AI by scaling compute for pre-training and RL is leading to predictable useful tasks that the AI models can do, including longer and longer horizon tasks. And so you can sort of speculate about where this is heading. And in AI 2027 folks did. And this kind of picture suggests that over the next few years we may reach a point where AI models can do tasks that don't just take us minutes or hours but days, weeks, months, years etc.
**Jared Kaplan：** 最终，我们设想 AI 模型——也许是数百万个 AI 模型协同工作——将能够完成整个人类组织所能做的工作。它们将能够完成目前整个科学界所做的那类工作。数学或理论物理的一个好处是，只靠思考就能取得进展。所以你可以想象，AI 系统协同工作，在几天或几周内取得理论物理学界 50 年才能取得的那种进展。
**Jared Kaplan:** Eventually, we imagine AI models or millions of AI models perhaps working together will be able to do the work that whole human organizations can do. They'll be able to do the kind of work that the entire scientific community currently does. One of the nice things about math or theoretical physics is that you can make progress just by thinking. And so you can imagine AI systems working together to make the kind of progress that the theoretical physics community makes in say 50 years in a matter of days, weeks etc.
**Jared Kaplan：** 那么如果 scaling 能带我们走很远，还剩下什么问题呢？我认为要广泛地实现人类级别的 AI，需要解决的东西其实相对简单。我认为最重要的要素之一是相关的组织知识。我们需要训练的 AI 模型不能只是一张白纸来迎接你，它得能学会在公司、组织、政府中工作，就像一个已经在那里工作了好几年、拥有相应背景知识的人一样。所以我认为 AI 模型需要能够运用知识。
**Jared Kaplan:** So what is left if this sort of picture of scaling can take us very far? What is left? I think that what may be left in order to unlock kind of human level AI broadly construed is relatively simple. One of the most important ingredients I think is relevant organizational knowledge. So we need to train AI models that don't just greet you with a blank slate but can learn to work within companies, organizations, governments as though they have the kind of context that someone who's been working there for years has. So I think AI models need to be able to work with knowledge.
**Jared Kaplan：** 它们还需要记忆（memory）。记忆和知识有什么区别呢？我把它们区分开来，是因为当你做一个耗时很长的任务时，你需要追踪你在这个具体任务上的进展，需要构建相关的记忆并能够使用它们。这是我们已经开始在 Claude 4 中构建的功能，我认为它会变得越来越重要。
**Jared Kaplan:** They also need memory. What is memory if not knowledge? I distinguish it in the sense that as you do a task that takes you a very very long time, you need to keep track of your progress on that specific task, you need to build relevant memories and you need to be able to use them. And that's something that we've begun to build into Claude 4 and I think will become increasingly important.
**Jared Kaplan：** 第三个我认为我们需要改进并且正在取得进展的要素是监督（oversight）。也就是 AI 模型理解细粒度细微差别、解决困难模糊任务的能力。现在我们在训练 AI 模型写能通过测试的代码或正确回答数学题方面看到了爆发式的进展，因为什么是对的什么是错的非常清晰，所以很容易应用强化学习让 AI 模型在这类任务上越来越好。但我们需要、也正在开发的是能帮我们生成更细致的奖励信号（reward signal）的 AI 模型，这样我们就能利用强化学习来做更多的事情——比如讲好笑话、写好诗、以及在研究中有好品味。
**Jared Kaplan:** A third ingredient that I think that we need to get better at and we're making progress on is oversight. The ability of AI models to understand sort of fine grained nuances to solve hard fuzzy tasks. So it's easy right now and you see an explosion of progress for us to train AI models that can say write code that passes tests or that answer math questions correctly because it's very crisp what's correct and what's incorrect. So it's very easy to apply reinforcement learning to make AI models do better and better at those kinds of tasks. But what we need and are developing are AI models that help us to generate much more nuanced reward signals so that we can leverage reinforcement learning to do things like tell good jokes, write good poems, and have good taste in research.
**Jared Kaplan：** 其他我们还需要的要素相对简单一些。我们显然需要训练 AI 模型完成越来越复杂的任务。我们需要沿着 Y 轴往上走——从文本模型到多模态模型再到机器人技术。我预计在未来几年，当 scaling 应用到这些不同领域时，我们会继续看到持续的收益。
**Jared Kaplan:** The other ingredients that we need, I think, are simpler. We obviously need to be able to train AI models to do more and more complex tasks. We need to work our way up the y-axis from text models to multimodal models to robotics. And I expect that over the next few years, we'll see increasing continued gains from scale when applied to these different domains.
**Jared Kaplan：** 那么我们应该如何为这样的未来和这些可能性做准备呢？有几件事我一直在推荐。第一个是我认为去构建那些"还差一点就能用"的东西是个很好的主意。这可能永远都是个好建议——我们总是应该有雄心壮志。但我认为特别是现在，AI 模型正在非常非常快地变好，而且我认为这会持续下去。这意味着如果你构建了一个产品，它因为 Claude 4 还不够聪明而差一点才能用，你可以期待 Claude 5 的到来会让这个产品真正运转起来并创造巨大价值。所以我一直推荐的就是去 AI 能力的边界上做实验，因为这些边界正在快速移动。
**Jared Kaplan:** And so how should we sort of prepare for this future, these possibilities? I think there are a few things that I always recommend. One is I think it's really a good idea to build things that don't quite work yet. This is probably always a good idea. We always want to have ambition, but I think specifically AI models right now are getting better very very quickly. And I think that's going to continue. That means that if you build a product that doesn't quite work because Claude 4 is still a little bit too dumb, you could expect that there'll be a Claude 5 coming that will make that product work and deliver a lot of value. So I think that's something that I always recommend is sort of experiment on the boundaries of what AI can do because those boundaries are moving rapidly.
**Jared Kaplan：** 第二点是 AI 将会帮助我们集成 AI。我认为 AI 目前的主要瓶颈之一其实就是它发展太快了，我们还没来得及把它集成到产品、公司以及我们做的其他所有事情中，包括科学研究。所以我认为为了加速这个过程，利用 AI 来帮助 AI 集成将会非常有价值。
**Jared Kaplan:** The next point I think is that AI is going to be helpful for integrating AI. I think that one of the main bottlenecks for AI is really just that it's developing so quickly that we haven't had time to integrate it into products, companies, other things, everything else that we do, into science. And so I think that in order to sort of speed that process up, I think leveraging AI for AI integration is going to be very valuable.
**Jared Kaplan：** 最后一点，我觉得对在座各位来说可能很显而易见，但我认为找到 AI 能够被非常快速采纳的领域是关键。我们正在看到 AI 在编程领域的集成呈爆发式增长。软件工程之所以是 AI 的一个绝佳应用场景有很多原因，但我认为一个大问题是：下一个是什么？除了软件工程，还有什么领域能增长得这么快？我当然不知道答案，但希望你们能找到。好了，演讲就到这里，我想邀请 Diana 上台来做个对话。
**Jared Kaplan:** And then finally, I mean, I think this is sort of obvious for this crowd, but I think figuring out where adoption of AI could happen very very quickly is key. We're seeing an explosion of AI integration for coding. And there are a lot of reasons why software engineering is a great place for AI, but I think the big question is sort of what's next? What beyond software engineering can grow that quickly? I don't know the answer, of course. But hopefully you guys will figure it out. So that's it for the talk. I want to invite Diana on stage for a chat.
**Diana：** 太棒了，刚才关于 scaling laws 的演讲非常精彩。Anthropic 最近刚发布了 Claude 4，已经上线了。我很好奇，随着这些模型不断迭代，未来 12 个月内这会如何改变我们能做的事情？
**Diana:** That was an awesome talk about all the scaling laws and recently Anthropic just launched Claude 4 which is just available. Curious, how does it change what is possible as all these model releases keep compounding for the next 12 months?
**Jared Kaplan：** 如果 12 个月内都没有更好的模型出来，那我觉得我们就有麻烦了。关于 Claude 4，说几点。我觉得从 Claude 3.7 Sonnet 开始，用 3.7 来写代码就已经非常令人兴奋了。但我觉得大家都注意到一个问题：3.7 有点太急切了。有时候它特别想让你的测试通过，会做一些你其实并不想要的事情。比如到处加 try-except 之类的。
**Jared Kaplan:** I think that we'll be in trouble if it's 12 months before an even better model comes out. But I guess a few things with Claude 4. I think that with Claude 3.7 Sonnet it was already really exciting to use 3.7 for coding. But I think something that everyone noticed was that 3.7 was a little bit too eager. Sometimes it just really wanted to make your tests pass. And it would do things that you don't really want. There are a lot of like try excepts, things like that.
**Jared Kaplan：** 所以 Claude 4 的一个改进是，我们提升了模型作为 agent 的能力——特别是在编程方面，但也体现在搜索和各种其他应用中。同时也改善了它的监督能力，就是我在演讲中提到的那种 oversight，使它能更好地遵循你的指令，并且有望提高代码质量。
**Jared Kaplan:** So with Claude 4, I think that we've been able to improve the model's ability to act as an agent specifically for coding, but in a lot of other ways for search, for all kinds of other applications. But also improve its supervision, the sort of oversight that I mentioned in my talk, so that it follows your directions and hopefully improves in code quality.
**Jared Kaplan：** 另一个我们着力改进的方面是提升它保存和存储记忆的能力。我们希望大家能充分利用这个功能，因为 Claude 4 在处理非常复杂的任务时可能会用完上下文窗口（context window），但它也可以把记忆保存为文件或记录，然后检索这些记忆，从而跨越多个上下文窗口持续工作。
**Jared Kaplan:** I think the other thing that we've worked on is improving its ability to save and store memories and we hope to see people leveraging that because Claude 4 can blow through its context window with a very complex task but can also store memories as files or records, retrieve them in order to sort of keep doing work across many many many context windows.
**Jared Kaplan：** 最后我想说的是，scaling laws 描绘的是一幅渐进式进步的图景。所以我认为你们会看到 Claude 在每次发布时都会在很多不同方面稳步变好。但我认为 scaling 真正暗示的是一条通向人类级别 AI 或 AGI（通用人工智能）的平滑曲线。
**Jared Kaplan:** But I guess finally I think the picture that scaling laws paint is one of incremental progress. And so I think that what you'll see with Claude is that steadily it gets better in lots of different ways with each release. But I think that scaling really suggests a kind of smooth curve towards what I expect is kind of human level AI or AGI.
**Diana：** 有没有什么特别的功能是你觉得在座的开发者们会特别兴奋的？有什么内测版本或者抢先消息可以透露一下，让大家知道新的 API 里什么会让人爱不释手？
**Diana:** Is there some special feature that a lot of the audience here are going to get excited? Some beta that you can, some alpha leak you can give everyone on what you think people are going to fall in love with the new APIs.
**Jared Kaplan：** 我觉得最让我兴奋的就是记忆功能解锁越来越长时间跨度的任务。我认为随着时间推移，我们会看到 Claude 成为一个能承担越来越大块工作的协作者。
**Jared Kaplan:** I think the thing that I'm most excited about is sort of memory unlocking longer and longer horizon tasks. I think that as time goes on we're going to see Claude as a collaborator that can sort of take on larger and larger chunks of work.
**Diana：** 这正好呼应了你之前说的，未来的模型能处理越来越大的任务。目前的模型已经能做到以小时为单位的任务了。
**Diana:** This is to your point of all these future models being able to take bigger and bigger tasks right now. At this point, they're able to do tasks in the hours.
**Jared Kaplan：** 是的，我觉得差不多。这个衡量标准不太精确，但如果你看软件工程任务的话，Metr 确实基准测试了人类完成各种任务需要多长时间，我觉得现在的时间尺度确实是以小时计的。
**Jared Kaplan:** Yeah, I think so. I think it's a very imprecise measure, but I think that right now if you look at sort of software engineering tasks, I think Metr literally benchmarked how long it would take people to do various tasks and yeah, I think it's a time scale of hours.
**Jared Kaplan：** 更广泛地来说，当人们和 AI 一起工作时，AI 的怀疑者们会很正确地指出 AI 会犯很多蠢的错误——它能做出绝对令人惊艳的事情，但也能犯最基本的错。我觉得 AI 智能和人类智能在形态上有一个根本区别：有很多事情我做不了，但至少我能判断别人做得对不对。对 AI 来说，它的判断能力和生成能力非常接近——这意味着人类在和 AI 协作时可以扮演一个重要角色，就是像管理者一样去做理智性检查（sanity check）。
**Jared Kaplan:** I think just generally broadly as people work with AI, I think that the people who are skeptics of AI will say correctly that AI makes lots of stupid mistakes. It can do things that are absolutely brilliant and surprise you, but it can also make basic errors. I think one of the sort of basic features of AI that's different about the shape of AI intelligence compared to human intelligence is that there are a lot of things that I can't do but I can at least judge whether they were done correctly. I think for AI the judgment versus the generative capability is much closer which means that I think that a major role people can play in interacting with AI is kind of as managers to sort of sanity check the work.
**Diana：** 这一点很有意思。我们在 YC 往期批次中观察到，去年很多公司在卖产品时还是把它当做副驾驶（co-pilot）来卖——比如客户支持的副驾驶，在发送回复之前仍然需要人类最后审批。但有一件事在今年春季批次中变了：我觉得现在很多 AI 模型已经有能力端到端地完成任务了，正如你说的，这很了不起。创始人们现在直接在卖整个工作流程的替代方案。你怎么看这个趋势对你希望大家去构建的东西的影响？
**Diana:** Which is fascinating because one of the things we observe through the batches in YC last year a lot of companies when they were out and selling products they were selling it more still as a co-pilot where you would have a co-pilot let's say for customer support where you still need the last human approval before they would send the reply for a customer but one thing that has changed just in the spring batch I think a lot of the AI models are very capable to do task end to end to your point which is remarkable. Founders are selling now directly replacements of full workflows. How have you seen this translate to what you hope the audience will build?
**Jared Kaplan：** 我觉得有很多可能性。基本上这取决于什么水平的成功率或表现是可接受的。有些任务做到 70% 正确就够了，有些则需要 99.9% 才能部署。我觉得说实话，为那些 70-80% 就够用的场景去构建可能更有意思，因为这样你才能真正触及 AI 能力的前沿。
**Jared Kaplan:** I think there are a lot of possibilities. Basically, it's a question of what level of success or performance is acceptable. There are some tasks where getting it sort of 70% right is good enough and others where you need 99.9% to deploy. I think that honestly I think it's probably a lot more fun to build for use cases where 70-80% is good enough because then you can really get to the frontier of what AI is capable of.
**Jared Kaplan：** 但我们同时也在提升可靠性。所以我觉得我们会看到越来越多这样的任务。我认为现在人机协作是最有意思的领域，因为对于最复杂的任务来说，你真的需要人在回路中。但我也确实认为，从长远来看，会有越来越多的任务可以完全自动化。
**Jared Kaplan:** But I think that we're sort of pushing up the reliability as well. So I think that we will see more and more of these tasks. I think that right now human AI collaboration is going to be the sort of most interesting place because I think that for the most advanced tasks you're really going to need humans in the loop. But I do think in the longer term there will be more and more tasks that can be fully automated.
**Diana：** 你能多说一点你认为人机协作循环的世界会是什么样的吗？Dario 写过一篇《Machines of Love and Grace》的文章，描绘了一幅非常乐观的图景。具体来说我们怎么才能到达那里？
**Diana:** Can you say more about what you think the world is going to look like with this human to AI loop collaboration? Because there's the essay from Dario with Machines of Love and Grace that he paints this picture that's very optimistic and what are the details of how we get there?
**Jared Kaplan：** 我觉得我们已经看到了一些。至少当我和做生物医学研究的人交流时，如果有恰当的编排（orchestration），用当前的前沿 AI 模型已经可以为药物研发等方向产出有趣有价值的洞见。所以我觉得这已经在发生了。
**Jared Kaplan:** I think that we already see some of that happening. So at least when I talk to folks who work in say biomedical research with the right sort of orchestration I think it's possible to take frontier AI models now and produce interesting valuable insights for say drug discovery. So I think that's already starting to happen.
**Jared Kaplan：** 我经常思考的一个方面是，智能分为需要深度的和需要广度的。比如在数学中，你可以花十年时间去证明一个定理，像黎曼猜想或费马大定理。这属于解决一个非常具体的、非常难的问题。
**Jared Kaplan:** I guess an aspect of it that I think about is that there's sort of intelligence that requires a lot of depth and intelligence that requires a lot of breadth. So for example in math you can sort of work on trying to prove one theorem for a decade like the Riemann hypothesis or Fermat's last theorem. I think that's sort of solving one very specific very hard problem.
**Jared Kaplan：** 但在很多科学领域——可能更多是在生物学，也许有趣的是在心理学或历史学中——整合来自很多不同领域的大量信息片段才是关键所在。AI 模型在预训练阶段吸收了人类文明的全部知识。所以我怀疑，利用 AI 的这个特点——它知道的东西远比任何一个人类专家多——把许多不同专业领域的知识结合在一起（比如生物学中的跨领域研究）来产出洞见，这里面有很多可以收获的果实。
**Jared Kaplan:** I think there's a lot of areas of science, probably more so in biology, maybe interestingly in psychology or history, where putting together a very very large number of pieces of information across many many different areas is kind of where it's at. And I think that AI models during the pre-training phase kind of imbibe all of human civilization's knowledge. And so I suspect that there's a lot of fruit to be picked in using that sort of feature of AI that it knows much much more than any one human expert and therefore you can kind of elicit insights putting together many different areas of expertise say across biology for research.
**Jared Kaplan：** 我们在让 AI 擅长更深度的任务方面取得了很大进展——比如困难的编程问题、困难的数学问题。但我觉得在那些需要整合知识的领域——也许没有哪个人类专家能同时拥有这些知识——这种智能特别有用，存在一个特别大的红利窗口。所以我预期会看到更多利用 AI 知识广度的应用。
**Jared Kaplan:** So I think that we're making a lot of progress on making AI better at deeper tasks like hard coding problems, hard math problems, but I suspect that there's a particular overhang in areas where putting together knowledge that maybe no one human expert would have where that kind of intelligence is very useful. So I think that's something that I'd expect to see more of, sort of leveraging AI's breadth of knowledge.
**Jared Kaplan：** 至于具体会怎么落地，我真的不知道。预测未来真的非常困难。Scaling laws 给了你一种预测未来的方式——告诉你这个趋势会继续。长期来看我预期很多趋势会持续——经济、GDP，这些指标是非常可靠的未来指标。但具体到细节上事物将如何实现，我觉得真的很难说。
**Jared Kaplan:** In terms of how exactly it will roll out, I really don't know. It's really really hard to predict the future. Scaling laws give you one way of predicting the future which says this trend is going to continue. I think a lot of trends that we see over the long haul I expect will continue. I mean the economy, the GDP, these kinds of trends are really reliable indicators of the future. But I think in terms of in detail how will things be implemented, I think it's really really hard to say.
**Diana：** 你觉得有没有什么特定领域，可以让更多的创业者用这些新模型去构建产品？编程任务已经做了很多了，但还有哪些领域有大量的绿地（green field），是当前模型刚刚解锁的？
**Diana:** Are there specific areas that you think a lot more builders could go into and build with these new models? I mean there's a lot that has been done let's say for coding tasks but what are some tasks that have a lot more green field that are just getting unlocked right now with the current models?
**Jared Kaplan：** 我的背景是研究而不是商业，所以我不确定我能说出什么特别深刻的东西。但我觉得总体来说，任何需要大量技能并且主要是坐在电脑前和数据打交道的任务都有机会。我觉得金融领域——那些大量使用 Excel 表格的人。我觉得法律也应该有机会，虽然法律可能监管更多，需要更多专业资质作为背书。但我觉得这些领域可能都还是绿地。
**Jared Kaplan:** I come from a research background rather than business so I don't know that I have anything very deep to say but I think that in general any place where it requires a lot of skill and it's a task that mostly involves sitting in front of a computer interacting with data. I think finance, people who use Excel spreadsheets a lot. I think I expect law although maybe law is more regulated requires more expertise as a stamp of approval. But I think all of these areas are probably green field.
**Jared Kaplan：** 我刚才还提到了另一个方向：如何把 AI 集成到现有业务中。当年电力出现的时候，有一个很长的采纳周期，最初最简单的用电方式未必是最好的。你不能只是用电动机替代蒸汽机，你需要从根本上重新设计工厂的运作方式。类似地，我觉得利用 AI 来尽可能快地将 AI 集成到经济各环节中，这里面蕴含着巨大的杠杆效应。
**Jared Kaplan:** I think another that I sort of mentioned is how do we integrate AI into existing businesses? I think that when electricity came along, there was some long adoption cycle and the very first simplest ways of say using electricity weren't necessarily the best. You wanted to not just replace a steam engine with an electric motor. You wanted to sort of remake the way that factories work. And I think that probably leveraging AI to integrate AI into parts of the economy as quickly as possible. I expect there's just a lot of leverage there.
**Diana：** 另一个问题。你有扎实的物理学训练，你是最早真正观察到 scaling laws 这个趋势的人之一，这可能正得益于你作为物理学家看到了自然界中自然发生的这些指数关系。这种训练背景如何帮助你在 AI 领域做出世界顶尖的研究？
**Diana:** Now other question is you have extensive training as a physicist and you were one of the first to really observe this trend with scaling laws and it probably comes from being a physicist and seeing all these exponentials that happen naturally in nature. How has that training come about with being able to perform the best research in the world with AI?
**Jared Kaplan：** 我觉得物理学训练中最有用的一点是去寻找最宏观的大图景趋势，然后尽量让它们变得尽可能精确。我记得我遇到过一些非常聪明的 AI 研究员，他们会说"学习在指数收敛"之类的话，然后我就会问一些非常笨的问题——你确定是指数吗？会不会只是幂律（power law）？是二次的吗？这个东西到底是怎么收敛的？这真是一种非常笨、非常简单的问题，但基本上我觉得在试图尽可能精确地描述你看到的大趋势方面，有很多唾手可得的成果——可能现在仍然有——因为精确化后你就拥有了很多工具。
**Jared Kaplan:** I think the thing that was useful from a physics point of view is looking for the biggest picture, most macro trends and then trying to make them as precise as possible. So I remember meeting kind of brilliant AI researchers who would say things like learning is converging exponentially and I would just ask really dumb questions like are you sure it's an exponential? Could it just be a power law? Is it quadratic? Like exactly how is this thing converging? And it's a really dumb kind of simple question to ask, but basically I think there was a lot of fruit to be picked and probably still is in trying to make the big trends that you see as precise as possible because that gives you a lot of tools.
**Jared Kaplan：** 它让你能问"到底什么才算推动了进展"。我觉得对 scaling laws 来说，圣杯就是找到一个更好的斜率，因为这意味着当你投入更多算力时，你相对于其他 AI 开发者的优势会越来越大。但在你把看到的趋势精确化之前，你并不真正知道"击败它"意味着什么、能击败多少、以及如何系统性地验证你是否实现了目标。所以这些就是我用的工具。并不是直接把量子场论之类的东西应用到 AI 上——那就太具体了。
**Jared Kaplan:** It allows you to ask like what does it really mean to move the needle? I think with scaling laws, the holy grail is finding a better slope to the scaling law because that means that as you put in more compute, you're going to get a bigger and bigger advantage over other AI developers. But until you've sort of made precise what the trend is that you see, you sort of don't know exactly what it means to beat it and how much you can beat it by and how to know systematically whether you're achieving that end. So, I think those were kind of the tools that I used. It wasn't necessarily like literally applying say quantum field theory to AI. I think that's a little bit too specific.
**Diana：** 有没有一些具体的物理学启发法，比如重整化（renormalization）、对称性（symmetry），在持续观察和度量这个趋势时特别好用？
**Diana:** Well, are there specific physics heuristics like renormalization, symmetry that came in very handy to really keep observing this trend or measuring it?
**Jared Kaplan：** 如果你观察 AI 模型你会发现它们很大。神经网络很大。它们有数十亿、现在甚至数万亿个参数。这意味着它们由大矩阵构成。研究这种近似——取神经网络非常大的极限，特别是构成神经网络的矩阵非常大的极限——这实际上挺有用的。这在物理和数学中是一种众所周知的近似方法，已经被应用到 AI 中了。
**Jared Kaplan:** Something that you'll observe if you look at AI models is that they're big. Neural networks are big. They have billions now trillions of parameters. That means that they're made out of big matrices. And basically studying approximations where you take the limit that neural networks are very big and specifically that the matrices that compose neural networks are big. That's actually been kind of useful and that's something that actually was a well-known approximation in physics and in math. That's something that's been applied.
**Jared Kaplan：** 但我觉得总的来说，问非常天真的笨问题才是最能让你走远的。AI 从当前训练方式的角度来说，其实只有大约 10 到 15 年的历史。这意味着它是一个极其年轻的领域。很多最基本的问题还没有被回答——比如可解释性（interpretability）的问题，AI 模型到底是怎么工作的。所以我觉得在这个层面上有太多东西可以学了，而不是去应用非常花哨的技术。
**Jared Kaplan:** But I think generally it's really asking very naive dumb questions that gets you very far. I think AI is really in a certain sense only like maybe 10, 15 years old in terms of the current incarnation of how we're training AI models. That means that it's an incredibly new field. A lot of the most basic questions haven't been answered like questions of interpretability, how AI models really work. And so I think there's really a lot to learn at that level rather than applying very very fancy techniques.
**Diana：** 有没有你会用来做可解释性研究的具体物理学工具？
**Diana:** Are there specific tools in physics that you apply for interpretability?
**Jared Kaplan：** 我觉得可解释性更像生物学、更像神经科学。所以那些才是相关的工具。其中有一些数学的成分，但我觉得它更像是试图理解大脑的特征。AI 相比神经科学的一个好处是，你可以真正测量 AI 中的一切。你做不到测量大脑中每个神经元、每个突触的活动，但在 AI 中你可以。所以在逆向工程 AI 模型的工作原理方面，有多得多的数据可以用。
**Jared Kaplan:** I would say that interpretability is a lot more like biology. It's a lot more like neuroscience. So I think those are kind of the tools. There is some more mathematics there. But I think it's more like trying to understand the features of the brain. The benefit that you get with AI over neuroscience is that you can really measure everything in AI. You can't measure the activity of every neuron, every synapse in a brain, but you can do that in AI. So there's much much much more data for reverse engineering how AI models work.
**Diana：** 关于 scaling laws 的另一个问题：它们已经在超过五个数量级上成立了，这太疯狂了。接下来是一个有点逆向的问题——什么样的经验证据会让你相信曲线正在发生变化，也许我们已经偏离了曲线？
**Diana:** Now one aspect about scaling laws, they've held for over five orders of magnitude, which is wild. This is a bit of a contrarian question, but what empirical sign would convince you that the curves are changing, that maybe we're getting off the curve?
**Jared Kaplan：** 这是一个非常难的问题。因为我主要是用 scaling laws 来诊断 AI 训练是否出了问题。
**Jared Kaplan:** I think it's a really hard question, right? Because I mostly use scaling laws to diagnose whether AI training is broken or not.
**Diana：** 嗯。
**Diana:** Mhm.
**Jared Kaplan：** 一旦你发现了一个非常令人信服的趋势，去研究它在哪里失效就变得非常有意思。但我的第一反应是，如果 scaling laws 失效了，那是因为我们在 AI 训练中搞砸了什么。也许我们选错了神经网络架构，或者训练中有某个我们没看到的瓶颈，或者我们使用的算法在精度上有问题。所以要让我相信 scaling 在这些经验定律层面真的不再有效，需要非常多的证据——因为在我过去五年的经验中，太多次看似 scaling 失效了，结果其实是我们的做法不对。
**Jared Kaplan:** So I think that once you see something and you find it's a very compelling trend, it becomes very very interesting to examine where it's failing. But I think that my first inclination is to think if scaling laws are failing, it's because we've screwed up AI training in some way. Maybe we got the architecture of the neural network wrong or there's some bottleneck in training that we don't see or there's some problem with precision in the algorithms that we're using. So I think it would take a lot to convince me at least that scaling was really no longer working at the level of the sort of these empirical laws because so many times in my experience over the last 5 years when it seemed like scaling was broken it was because we were doing it wrong.
**Diana：** 有意思。那我接着聊一个非常具体的相关话题：要沿着这条曲线继续走下去需要大量算力。当算力越来越稀缺时会怎样？你会在精度阶梯上走多远？比如你会探索 FP4 吗？会探索三值表示（ternary representations）吗？你怎么看？
**Diana:** Interesting. So I guess going into something very specific that goes hand in hand is a lot of the compute power required to keep going on this curve. What happens as compute becomes more and more scarce, how far down do you go into the precision ladder, like do you explore things like FP4, do you explore things like ternary representations, what are your thoughts around that?
**Jared Kaplan：** 是的，我觉得现在 AI 确实非常低效，因为 AI 有很大的价值。解锁最强大的前沿模型有巨大的价值。所以像 Anthropic 和其他公司正在尽可能快地推进——既要让 AI 训练更高效，也要让 AI 推理（inference）更高效，同时还要解锁前沿能力。但目前很多注意力确实集中在解锁前沿上。
**Jared Kaplan:** Yeah, I mean I think that right now AI is really inefficient because there's a lot of value in AI. So there's a lot of value in unlocking the most capable frontier model. And so companies like Anthropic and others are moving as quickly as we can to both make AI training more efficient and AI inference more efficient as well as unlocking frontier capabilities. But a lot of the focus really is on unlocking the frontier.
**Jared Kaplan：** 我觉得随着时间推移，AI 变得越来越普及后，我们会大幅降低推理和训练的成本。现在我们看到的是，在算法、计算规模扩展和推理效率方面，每年大约有 3 到 10 倍的提升。
**Jared Kaplan:** I think that over time as AI becomes more and more widespread, I think that we're going to really drive down the cost of inference and training dramatically from where we are right now. I mean right now we're seeing sort of 3x to 10x gains algorithmically and in sort of scaling up compute and in inference efficiency per year.
**Jared Kaplan：** 有一个段子是说我们要让计算机回归二进制了。所以我觉得我们确实会看到精度越来越低，这是让推理更高效的众多途径之一。但目前 AI 的发展非常非常不在均衡状态——AI 在快速进步，事情在快速变化，我们还没有完全实现当前模型的全部潜力，但同时又在不断解锁更多能力。
**Jared Kaplan:** I guess like the joke is that we're going to get computers back into binary. So I think that we will see much much lower precision as one of the many avenues to make inference more efficient over time. But sort of we're very very very out of equilibrium with AI development right now. AI is improving very rapidly. Things are changing very rapidly. We haven't fully realized the potential of current models, but we're unlocking more and more capabilities.
**Jared Kaplan：** 所以我觉得均衡态——也就是 AI 不再快速变化的状态——应该是 AI 变得极其便宜的状态。但很难说我们是否真的会到达那里。AI 可能会持续快速进步，智能的提升会释放出更多价值，所以我们可能会持续聚焦于此，而不是去搞 FP2 精度。
**Jared Kaplan:** So I think that what the equilibrium situation looks like where AI isn't changing that quickly, I think is one where AI is extremely inexpensive, but it's sort of hard to know if we're even going to get there. Like AI may just keep getting better so quickly that improvements in intelligence unlock so much more and so we may continue to focus on that rather than say getting precision down to FP2.
**Diana：** 这其实就是 Jevons 悖论（Jevons paradox）——智能变得越好，人们反而会要更多，而不是说成本降低就满足了，这就是这个悖论的讽刺之处，对吧？
**Diana:** Which is very much the Jevons paradox, as intelligence becomes better and better people are going to want it more, not that it is driving the cost down, which is this irony right?
**Jared Kaplan：** 完全同意。我们确实看到了这个现象——在某些节点上 AI 变得足够易用了。话虽如此，我觉得随着 AI 系统变得越来越强大、能做我们越来越多的工作，为前沿能力付费会是值得的。
**Jared Kaplan:** Yeah absolutely. I mean I think that yeah that's certainly something that we've seen that there are certain points where AI becomes accessible enough. That said, I think as AI systems become more and more capable and can do more and more of the work that we do, it's going to be worth it to pay for frontier capabilities.
**Jared Kaplan：** 我一直在思考的一个问题是：所有的价值都集中在前沿，还是那些更便宜但没那么强的系统也有很大价值？我觉得时间跨度的视角也许是一种思考方式。你可以用较弱的模型完成很多简单的小任务，但能够用一个 AI 模型端到端地完成一个非常复杂的任务显然方便得多，而不是需要我们人类去编排一个笨得多的模型，把任务切成很小的片段再拼起来。所以我确实倾向于认为很多价值会来自最强的模型，但我也可能是错的。这可能取决于具体情况，也可能真正取决于 AI 集成者能否足够高效地利用 AI。
**Jared Kaplan:** I think it's a question that I've always had is kind of like is all of the value at the frontier or is there a lot of value with kind of cheaper systems that aren't quite as capable? And I think the sort of time horizon picture is maybe one way of thinking about this. I think that you can do a lot of very simple bite-sized tasks, but I think it's just much more convenient to be able to use an AI model that can do a very complex task end to end rather than requiring us as humans to sort of orchestrate a much dumber model to break the task down into very very small slices and put them together. So, I do kind of expect that a lot of the value is going to come from the most capable models, but I might be wrong. It might depend and it might really depend on the capabilities of AI integrators to sort of leverage AI really efficiently.
**Diana：** 你会给在座各位什么建议？大家都处于职业生涯的早期，有很大的潜力。在未来这些模型会变得越来越厉害的世界里，怎样才能保持自己的竞争力？大家应该擅长什么、学习什么，才能持续做出好的工作？
**Diana:** What advice would you give this audience which there everyone is early in the career with lots of potential in terms of how do you stay relevant in the future where all these models are going to become so awesome. What should everyone be really good at and study and to still do really good work?
**Jared Kaplan：** 我觉得正如我提到的，理解这些模型是怎么工作的、能够高效地利用和集成它们很有价值。在前沿上构建东西也很有价值。我不确定还能说什么了，不如我们把话筒交给观众来提问吧。
**Jared Kaplan:** I think as I mentioned there's a lot of value in understanding how these models work and being able to really efficiently leverage them and integrate them and I think there's a lot of value in kind of like building at the frontier. I don't know, we could turn it over to the audience for questions.
**Diana：** 好的，让我们把时间留给观众提问。
**Diana:** Let's turn it out to the audience for some questions.
**Audience：** 我有一个关于 scaling laws 的问题。你展示了很多 scaling laws 是线性的——当计算量指数增长时，在 scaling loss 上我们看到的是线性进步。但在你最后一张幻灯片上，你展示的是你预期会突然出现指数级增长——在我们节省的时间方面。我想问的是，你认为为什么在这张图上我们突然变成了指数级而不再是线性的？谢谢。
**Audience:** I had a quick question on the scaling laws. You show that a lot of the scaling laws are like linear, that like the more we have exponential compute going up but then like we have linear progress in the scaling loss but then on your last slide you show that you expect then suddenly like an exponential growth in like how much time we save. I want to ask you like why do you think that suddenly on this chart we're exponential and not linear anymore? Thank you.
**Jared Kaplan：** 这是一个非常好的问题，我也不完全确定。Metr 的发现主要是一个经验性发现。我通常这样思考：要完成越来越复杂、越来越长时间跨度的任务，你真正需要的是某种自我纠错能力。你需要能发现自己的问题——你制定一个计划然后开始执行，但大家都知道我们的计划往往不靠谱，我们会遭遇现实，会搞错。
**Jared Kaplan:** Yeah, this is a really good question and I don't know. I mean the Metr finding was kind of an empirical finding. The way that I tend to think about this is that in order to do more and more complex longer horizon tasks what you really need is some ability to self-correct. You need to be able to sort of identify that you've, you make a plan and then you start executing on the plan. But everyone knows that our plans are kind of worthless and we encounter reality. We get things wrong.
**Jared Kaplan：** 所以我觉得决定模型能完成多长时间跨度任务的关键因素是它发现自己做错了并纠正的能力。而这并不需要特别多的信息量——不一定需要智能的巨大跃升就能多发现一两次错误并知道如何纠正。如果你修正了一个错误，你可能就大约把任务的时间跨度翻了一倍——因为原来你会卡在某个地方，现在你能走到两倍远的地方才卡住。
**Jared Kaplan:** And so I think that a lot of what determines the horizon length of what models can accomplish is their ability to notice that they're doing something wrong and correct it. And I think that's not sort of like a lot of bits of information. It doesn't necessarily require a huge change in intelligence to sort of notice one or two more times that you've made a mistake and how to correct that mistake. But if you sort of fix your mistake, maybe you sort of on the order double the horizon length of the task because like instead of getting stuck here, you get stuck twice as far out.
**Jared Kaplan：** 所以我的理解大概是：通过相对适度的能力提升——在理解任务和自我纠错方面——你就能解锁越来越长的时间跨度。但这些毕竟只是语言描述。我觉得经验趋势本身可能是最有意思的。也许我们可以为这个趋势建立更详细的模型来解释它为什么成立，但说实话，你的猜测和我的一样好。
**Jared Kaplan:** So I think that's sort of the picture that I have that you can kind of unlock longer and longer horizons with relatively modest improvements in your kind of ability to understand the task and self-correct. But that just kind of like those are just words. I think the empirical trend is maybe the most interesting thing. And maybe we can build more detailed models for why that trend is true, but it's sort of your guess is as good as mine.
**Audience：** 我也有一个问题。非常荣幸。关于增加时间跨度的问题——我对神经网络的心智模型很简单：如果你想让它做什么，你就用相应的数据来训练它。所以如果你想增加时间跨度，你需要逐步获得验证信号（verification signal）。我觉得一种方式是通过产品来获取——比如 Claude agent，然后用验证信号来逐步改进模型。我的问题是，这种方式在编程领域非常好用，因为你有一个足够好的产品可以部署，然后获取验证信号。但其他领域呢？在其他领域我们是不是只能靠不断扩大数据标注规模直到 AGI？有没有更好的方法？
**Audience:** So I also have a question over here. So it's an honor. Basically in terms of increasing the time horizon, I feel like my mental model of neural networks is very simple. If you want them to do something, you train on such data. So if you want to increase the time horizon you have to slowly get for example verification signals. Now I think one way to do this is via product. So like for example Claude agent and then you use the verification signal to incrementally improve the model. Now my question is basically this works really nicely for for example coding where you have a product that is sufficiently good such that you can deploy it and then get the verification signal but what about other domains? Like in other domains are we just scaling data labelers to AGI or is there a better approach?
**Jared Kaplan：** 好问题。当怀疑者问我为什么觉得我们能通过 scaling 达到广泛的人类级别 AI 时，基本上就是因为你说的这个原因。确实存在一条非常需要大量运营投入的路径——你只要不断为 AI 模型构建越来越多、越来越复杂、越来越长时间跨度的任务，然后转动把手用 RL 在这些更复杂的任务上训练。我觉得这是 AI 进步的最坏情况。考虑到 AI 领域目前的投资水平和我认为 AI 正在创造的价值水平，如果有必要的话人们会这么做的。
**Jared Kaplan:** Yeah, it's a good question. I mean, so when sort of skeptics ask me why do I think we will be able to sort of scale and get something like broadly human level AI, it's basically because of what you said. There is some sort of very kind of operationally intensive path where you just sort of build more and more different tasks for AI models to do that are more and more complex, more and more long horizon and you just sort of turn the crank and train with RL on those more complicated tasks. So I sort of feel like that's the worst case for AI progress. And I mean given the level of investment in AI and I think the sort of level of value that I think is being created with AI, I think people will do that if necessary.
**Jared Kaplan：** 话虽如此，有很多方法可以简化这个过程。最好的方法是让一个 AI 模型来监督 Claude 的训练——比如你在训练 Claude，另一个 AI 模型提供监督，而且不只是在最后才说"这个极其复杂的任务你做对了吗？"——比如说"你有没有成为教授并拿到终身教职"，这需要六七年时间，你不可能把它当做一个端到端任务等六七年后才给一个对或错的信号，那太荒谬了，太低效了。
**Jared Kaplan:** That said, I think there are a lot of ways of sort of making it simpler. The best is to have an AI model that is trained to oversee and supervise what Claude, like you have Claude say which you're training to be Claude, when you have another AI model that's sort of providing supervision and is not just saying did you do this incredibly complicated task correctly. Like did you become a faculty member and get tenure, will that take six or seven years, is that like an end-to-end task where at the end you sort of either get tenure or not over seven years? That's ridiculous. That's very inefficient.
**Jared Kaplan：** 更好的方式是提供更精细的监督——告诉你这部分做得好，那部分做得不好。我认为随着我们能越来越多地以这种方式使用 AI，我们很可能会让针对非常长时间跨度任务的训练变得更高效。而且我们在某种程度上已经在这样做了。
**Jared Kaplan:** But instead can provide more detailed supervision that says you're doing this well, you're doing this poorly. I think that sort of as we're able to use AI more and more in that kind of way, we'll probably be able to make training for very long horizon tasks more efficient and I think we're already doing this to some extent.
**Diana：** 最后一个问题。
**Diana:** We'll do one last question.
**Audience：** 好的，我想在刚才的基础上追问一下。当你开发这些任务然后用 RL 来训练时，你会尝试用大语言模型来创建这些用于 RL 的任务吗，还是仍然在用人工？
**Audience:** Yeah, I wanted to build on top of that. When you're basically developing like these tasks and then training them with RL, would you like try creating these tasks using large language models, like the tasks you use for RL, or are you still using humans?
**Jared Kaplan：** 好问题。我会说是混合的。我们显然在尽可能地用 AI 来构建任务——比如用代码来生成任务。我们也让人类来创建任务。所以基本上是两者的某种混合。我认为随着 AI 变得越来越好，我们有望越来越多地利用 AI。但当然，这些任务的难度前沿也在同步提升。所以我觉得人类仍然会参与其中。
**Jared Kaplan:** Great question. So I would say a mix. I mean obviously we're building the tasks as much as possible using AI to sort of like say generate tasks with code. We do also ask humans to create tasks. So it's basically some mixture of those things. I think that as AI gets better and better, hopefully we're able to leverage AI more and more, but of course the frontier of the difficulty of these tasks also increases. So I think humans are still going to be involved.
**Diana：** 好的，谢谢。
**Diana:** Okay. Thank you.
**Jared Kaplan：** 好的。让我们给所有人鼓个掌。
**Jared Kaplan:** All right. Let's give it a round of applause.
**Diana：** 非常感谢。谢谢大家。
**Diana:** Thank you so much. Thanks.
