---
author: Latent Space
date: '2026-09-22'
guest: ''
layout: post.njk
source: https://www.youtube.com/watch?v=2xBSGluFkG0
speaker: Latent Space
tags:
- predictive-models
- descriptive-models
- ai-for-science
- scorable-tasks
- scientific-discovery
title: 预测模型与描述性模型：科学发现中的外推能力与人工智能的潜力
summary: 文章探讨了预测模型与描述性模型在科学研究中的区别，强调描述性模型需要具备外推能力来捕捉现实本质。同时，文章介绍了人工智能赋能科学（AI for Science）的最新趋势，特别是将科学问题映射为“可评分任务”，以及利用通用大语言模型（LLM）在科学发现中的应用，并展望了未来构建“全能实验室”的愿景。
insight: ''
draft: true
series: ''
category: ai-ml
area: tech-engineering
project: []
people: []
companies_orgs: []
products_models: []
media_books: []
status: evergreen
---
### 预测模型与描述性模型
**Host**: 你所说的是引入显式先验吗？也就是基于人类的直觉，或者在这个情境下，基于大语言模型（LLM）的直觉？
Original English
**Host**: Are you talking about introducing explicit priors that you know based upon some human intuition or maybe in this case LLM intuition?
**John Platt**: 当你谈到多重假设检验时，存在预测模型（predictive models）和描述性模型（descriptive models）。所谓预测模型，就像是给定一些输入和一些输出，你只想构建一段代码，尝试在某个数据集或统计模型上达到最低的错误率。而描述性模型，实际上才是科学试图去探索的方向——它应该具备外推（extrapolate）的能力，因为它内部捕捉到了某种物理机制，或者对现实本质的某种描述。这样你才能利用它来进行推断外推。是的，牛顿当年由苹果联想到了万有引力，但引力的本质并不是关于苹果的，对吧？如果你用一个17世纪的机器学习模型，它可能会学到“苹果会往下掉”；但面对行星时呢？它会说：“我不知道，我没有任何关于行星的数据，谁知道行星会怎么运动。”这两者之间的界限其实是有些模糊的。因为当物理学家或科学家进行研究时，他们会运用直觉，甚至比单纯直觉更深层的东西——本质上是他们所掌握的关于这个世界的坚实事实基础，然后他们确保所构建的任何模型都能与这些已知事实保持一致。
Original English
**John Platt**: When you talk about multiple hypothesis testing, right, there's predictive models and there's descriptive models. A predictive model is like, let's say you just have some inputs and you have some outputs and you just want to build a piece of code that tries to just have the lowest error rate on some data set, statistical model. A descriptive model is actually what science is trying to get to, which is: okay, it should be able to extrapolate because it has sort of the physics or the actual description of reality that's captured within it, and then you can use it to extrapolate. Yes, Newton thought of apples and gravity, but gravity isn't actually about apples, right? If you take the 17th century machine learning model, like "Oh, apples will fall, but how about planets?" You know, "I don't know, I have no data about planets, so who knows what they do, right?" The distinction between those is a little bit blurry, right, because when a physicist or scientist comes, they use their intuition, or maybe even more than intuition. Like essentially there's maybe a solid pile of facts that they know about the world, and then they make sure that whatever model they build is sort of consistent with what's known.
### 嘉宾介绍：Google Fellow John Platt 的传奇经历
**Host**: 和我的搭档 R.J. 一起，我们今天非常荣幸能邀请到 John Platt。John 是 Google Fellow，也是 Google Research 应用科学（Applied Science）部门的负责人。他有着极其丰富且传奇的经历。我想几分钟前我们聊天时，你把自己形容为一个“超级极客”（mega nerd）。
Original English
**Host**: My co-host R.J., it's a pleasure to have John Platt with us today. John is a Google fellow and head of applied science at Google research. He has really a fun background. I guess you described yourself when we were talking a few minutes ago as a mega nerd.
**John Platt**: 噢，是吉咖极客（giga nerd）。
Original English
**John Platt**: Oh, giga nerd.
**Host**: 哈哈，吉咖极客，吉咖极客！他对万事万物都抱有极大的热情，而且表现得淋漓尽致。如果我说错了什么，你随时纠正我：你14岁上大学，18岁在加州理工学院（Caltech）开始攻读博士学位。你的导师或联合导师是 John Hopfield，对吧？
Original English
**Host**: Giga nerd. Giga nerd. He's excited in absolutely everything. And it really shows. Yeah. You will correct me if I'm wrong about any of this stuff, but so you started college at 14 and started your PhD at 18 at Caltech. You were advised or co-advised by John Hopfield, right?
**John Platt**: 噢，是的，没错。
Original English
**John Platt**: Oh, yeah. Yeah.
**R.J.**: 是的，就是两三年前刚获得诺贝尔奖的那位 John Hopfield。
Original English
**R.J.**: Yeah. Yeah. Who just won a Nobel Prize in you know two or three years ago.
**Host**: 是的。John 创造并奠定了数个教科书级别的经典算法。其中一个被称为 Platt 缩放（Platt Scaling），另一个是序列最小优化算法（Sequential Minimal Optimization, SMO）——这是训练支持向量机（SVM）的教科书标准算法。即便在今天，如果你使用 scikit-learn，它依然在底层运行。此外，John 还发现并命名了两颗小行星；在2006年获得了奥斯卡科学技术奖。如果你曾经看过皮克斯（Pixar）的动画电影，你就见识过 John 的算法与成果。John 的埃尔德什-培根数（Erdős–Bacon number）是6——两边各自是3和3。接下来我要跳过你职业生涯中大概20年的辉煌历程（笑），直接跳到你在 Google 的工作——在 Google 应用科学部门，你主导过核聚变、量子计算、气候建模以及诸多其他前沿课题的研究。这些概述基本准确吧？
Original English
**Host**: Yes. So John created several responsible for several textbook algorithms. One known as Platt scaling, another one sequential minimal optimization, which is the textbook algorithm for training SVMs. Even today if you use sklearn it's there. John has discovered and named two asteroids, has an Oscar for technical developments from 2006. So if you've ever watched a Pixar movie you've seen John's algorithms and work. John has an Erdős–Bacon number of six or three—three and three from either side—and I'm going to skip over like 20 years of your career [laughter], but then jumping to Google at working at Google sciences you worked on fusion, quantum computing, climate modeling and many other topics, is that more or less right?
**John Platt**: 没错，大致是这样。
Original English
**John Platt**: That that's right, yeah.
**Host**: 太棒了。今天我有什么重要履历遗漏了吗？
Original English
**Host**: Okay okay cool, did I miss anything important for today?
**John Platt**: 没有，我的意思是我还做过很多应用数学、信号处理以及各种好玩有趣的研究。
Original English
**John Platt**: No, I mean I've also done, you know, lots of applied math and signal processing and all sorts of fun things like that.
**Host**: 是的，没错。我想你的维基百科页面上还记载着关于专利以及与 iPhone 之间一段很有意思的故事。
Original English
**Host**: Yeah. Yeah. I think you also your your Wikipedia has a fun story about patents and the iPhone too.
**John Platt**: 其实是 iPod。
Original English
**John Platt**: The iPod.
**Host**: 是 iPod，对对对。非常欢迎你做客我们的节目！
Original English
**Host**: iPod. Yeah. Yeah. Yeah. Welcome.
**John Platt**: 谢谢你们，非常感谢你们邀请我。
Original English
**John Platt**: Thank you. Thank you for having me.
### 科学人工智能与“可评分任务”：ERA 的核心机制
**R.J.**: 能跟我们聊聊 ERA 吗？我想这个首字母缩写词应该是这么念的吧。
Original English
**R.J.**: Can you tell us about the ERA—is the I think the way that the acronym is pronounced.
**Host**: 我知道在 Google 内部以及外部，其实有很多形形色色、略微相关的研究。那么关于 ERA 的具体细节，你能跟我们分享些什么？它到底有什么独到之处？
Original English
**Host**: Um and and I know that there's a lot of different semi-related stuff out there both within and outside of Google. So what can you tell us a little bit about the details of ERA and what makes it special?
**John Platt**: 我们在 Google Research 开展“AI for Science”（AI赋能科学）相关的工作已经有十多年了。大约十年前，主要方式是非常传统的——使用现在大家所谓的经典机器学习模型，比如卷积神经网络（CNN）等，来构建专门针对特定科学问题的定制模型。但大约两年前，伴随着这几年通用大语言模型（LLM）的爆发，我们对它们感到极其兴奋，并且一直在思考能用它们做些什么。当然，业内许多人也一直在尝试和探索到底什么是正确的切入方向。我们机缘巧合地碰到了这样一种映射方法：换言之，我们发现许多截然不同的科学问题，其实都可以映射为我们所称的“可评分任务”（scorable tasks）。也就是说，你通常可以把一个科学问题表述为：“天哪，我真的很想要一段代码，它能够最大化某个评分函数。”令人惊讶的是，通过映射到这样一个框架中，你能够在海量不同的科学难题上取得巨大突破。其中一点是，许多科学家往往把大量时间花在构建模型上——这些模型可能是统计模型，也可能是基于物理规律的模型。如果是机器学习中的统计模型，你的评分函数通常就是：“我有一批数据集，我希望提高模型与这批数据集的拟合优度。”一会儿我们可以专门聊聊过拟合的问题，但是……
Original English
**John Platt**: Well, we've been doing sort of AI for science in Google research for more than 10 years now. And around 10 years ago it was very much using—I don't know what you call it now, maybe classical machine learning models, you know things like convolutional nets or whatever to—and they were specific models to build to solve specific science problems. But about 2 years ago, we got very excited about these more general LLMs that have popped up in the last few years and we were wondering what can be done with them. And of course a lot of people have been playing and trying to figure out what the right what the right thing to do is. And we kind of stumbled into this mapping. In other words, we found that many different scientific problems can be mapped into something we call scorable tasks. So you can often phrase a scientific problem as a: gosh, I really would like to have a piece of code that, you know, maximizes some score. And it's surprising the number of different sort of scientific problems you can make a lot of progress on by mapping into that framework. Well, one thing is a lot of scientists spend a lot of time sort of building models. They might be statistical models or they might be, you know, physically based models. And if it's a statistical model like in machine learning your scoring function is: well I have some data set and I'd like to have the fit of the model and the data set go up. And we can talk about overfitting in a minute, but...
**Host**: 这正好是我们准备问的问题之一！（笑）
Original English
**Host**: That was one of our questions. [laughter]
**John Platt**: 这实际上非常耐人寻味。机器学习只是这类“可评分任务”的一个子集，对吧？但你还可以做很多其他事情，特别是 ERA 论文的第一作者 Michael Brenner，他在这方面造诣极高——他现在借助这一工具，甚至可以在一个晚上就敲定一篇科学论文。举个例子，应用数学中有一个概念叫作“渐近展开”（asymptotic expansions），它所探讨的是常微分方程（ODE）或偏微分方程（PDE）在某个参数（比如含有一个 ε 参数）趋近于零时的行为规律。事实证明，你可以把这个问题转化为一个经验任务（empirical task）：本质上就是让系统提出一些在渐近意义上正确的候选解，然后你验证该渐近解在诸如 ε = 1e-4 时是否精确拟合。你不仅检验这种数值拟合程度，同时还要求底层的核心 AI——Gemini，去进行数学推理以推导求解，并同步最大化与验证数据的拟合分数。所以实际上你可以使用很多巧妙的策略。因为在底层修改代码、做出决策的并不是随机游走过程，而是一个具备高度智能、通晓众多领域知识的 AI。正因为核心内环是一个拥有海量先验知识的 AI，你才能够解决大量极其深刻有趣的科学问题。这就是其中的诀窍所在。因此，我们一直在各处奔走，尝试将各种科学问题映射为可评分任务并加以解决。这真的非常有趣，我也很乐意具体讲讲我参与过的那几个项目。
Original English
**John Platt**: Uh, and then that's actually very that's actually very interesting. So machine learning is kind of a subset of this sort of scorable task, right? But you could do other kind of things like especially Michael Brenner who's the lead author on the ERA paper. He's very very skilled because he likes to sort of knock out a scientific paper in an evening now with a tool. So there's something in applied math called asymptotic expansions, which is you're asking how does an ordinary differential or partial differential equation—but say ordinary differential equation—behave? There's some parameter has an epsilon in it and you're trying to say how does it behave as epsilon goes to zero. And it turns out you can turn that into an empirical task by essentially asking that it proposes some solutions that are asymptotically correct and you check to see if the asymptotic [clears throat] solution is correct for like epsilon equals 1e-4 or something and then you check that fit. But then you ask Gemini, which is the core AI underneath it, to do the mathematical reasoning, try to solve the problem while also maximizing the fit to the data. So you can actually—there's a lot of sort of tricks you can do because it's not that the underlying thing that's altering the code or the underlying thing that's sort of making the decisions is not a random process. It's an AI itself that is smart and knows about things and knows a lot about the world. You can solve a lot of interesting problems because that sort of core inner loop is an AI that has huge amounts of prior knowledge. So that's sort of the trick. So we've been running around trying to map lots of scientific problems into scorable tasks and trying to solve them and it's really been kind of fun and I'm happy to talk about the ones that I've been involved in at least.
### 超越纯统计：从数学推导到遥感与温室气体监测
**R.J.**: 是的，我很想听听那些更具特色的案例。统计建模大家都比较熟悉，你刚才提到的渐近展开也很有启发。那还有哪些其他引人注目的非统计任务案例呢？
Original English
**R.J.**: Yeah, I would love to hear about some of the more—so, it's a statistical one is what everyone listening will probably know about. What you just mentioned makes sense. What are some of the other interesting ones?
**John Platt**: 好的，让我想想那些非统计类的任务。我们手头确实有一些非常有趣的课题。比如我们最近刚在 arXiv 上发表了一篇论文（其实也可能发布在 GitHub 上了），它涉及遥感领域中经常遇到的权衡问题。卫星在地球上空运行，这里面始终存在着物理权衡：卫星重访地球同一地点的频次、空间分辨率（像素大小）以及光谱分辨率（包含多少个波段）之间无法兼得。最理想的状态下，你当然希望对地球进行全天候不间断监测，比如每5分钟拍摄一帧高光谱分辨率、地面分辨率达到10厘米的高清图像。但这在物理上是不可能实现的。不过举例来说，为了监测大气中的二氧化碳（CO2）浓度，你可以利用某一颗卫星的数据——比如 OCO-2 或者 OCO-3（OCO-3 实际上是搭载在国际空间站上的）。利用它……
Original English
**John Platt**: Uh let's see that that are not statistical. Let's see, cuz we have interesting ones like one that we just put a paper up on arXiv is—or actually I think it might be on GitHub. It's you often run into this in remote sensing because there's always a trade-off. There's satellites flying above the Earth and there's a trade-off between how frequently they can revisit a spot on the Earth, what their spatial resolution is, how big the pixels are, and their spectral resolution, so how many bands they have. And ideally you'd like to have monitoring of the earth that's constant and, you know, a frame every five minutes at hyperspectral resolution at at whatever 10 centimeters. You can't get that. But, for example, to monitor CO2, the atmospheric concentration of CO2, you can take data from one satellite, that's for example it's OCO-2 or OCO-3—OCO-3 is actually attached to the International Space Station, but so it...
### 从跨卫星超分辨率到科学问题的形式化映射
**Speaker A**: ……这能为你提供一条非常精确且分辨率相当高的二氧化碳（CO2）测量数据带。实际上你可以尝试去利用这一点，因为很多测量数据都位于红外波段。像 GOES 这类气象卫星拥有部分红外波段，它基本上每 5 分钟就会拍摄一张图像，但它的像素非常大，而且光谱分辨率也没有那么高——毕竟它最初的设计并不是用来检测 CO2 的。所以，你只需要让一个模型去通过一种数据估计另一种数据，同时把其他数据也输入进去，比如当前的气象条件、长期的地表反照率（albedo）等。通过这种方式，我们构建出了一个非常优秀的模型，它几乎可以实现跨卫星的信息融合超分辨率（informed super-resolution）。这是一个很典型的实际应用案例。
Original English
**Speaker A**: ...gets you like a little strip of CO2 measurements that are highly accurate and pretty high resolution. You can actually try to do—because a lot of it is in the infrared, weather satellites like GOES have some infrared bands and take a picture essentially every 5 minutes, but the pixels are very large and it doesn't have such great spectral resolution in terms of—it wasn't designed to find CO2. So you just ask one to estimate the other and you shovel other data in, like what's the current weather, what's sort of the long-term albedo. And so came up with this very nice model that can do almost like an informed super-resolution of one satellite to another. So that's like one example.
**Speaker B**: 对，明白了。所以任何能够被映射到这个框架中的科学问题都可以这样处理。也就是说，输入到系统的核心工作就是完成这种问题的形式化映射，而系统的输出则是代码，是这样吗？
Original English
**Speaker B**: Yeah. Yeah. Okay. So any scientific problem that you can map into this framework. So the input to Aera is sort of this mapping and the output is code, is that...
**Speaker A**: 可以这么说。不过在我们的产品设计中，实际上的输入方式非常自然：你只需要开始和它对话。因为很多时候，如何完成这种科学问题的形式化映射并不是显而易见的，尽管像迈克尔·布伦纳（Michael Brenner）这样的领域专家非常精通此道。因此，我们专门编写了一个智能体（Agent），它通过与你对话来帮助你厘清并定义出可量化的评分任务（scorable task）应该是什么样。所以从一开始，就有一个 Gemini 实例在那里协助你编写代码。这实际上可以说是一个中间成果：你开始向它描述你的问题，它在后台尝试生成一个 Python Notebook，其中包含一个能够产出具体分数的评分函数。随后，它开始以一种非常巧妙的方式对这个 Notebook 进行代码变异与演进——因为底层是 Gemini，它会不断提出新的代码方案，尝试最大化那个评估分数。
Original English
**Speaker A**: Well, sort of. I mean the input is—the way we've got it set up in the product is you just start talking, right? And because a lot of times it's non-obvious how to do this mapping, although experts like Michael Brenner know how to do it, we actually wrote an agent that actually helps you. It sort of talks to you to try to help you define what your scorable task should be. So already there's sort of an instance of Gemini sitting there trying to help you write code. That's actually sort of almost like an intermediate result: you start talking to it about your problem and it tries to produce essentially a Python notebook underneath that has a function with a scorable—which essentially produces a score. And then it starts to mutate that notebook in a clever way, because again it's Gemini, and it will try to sort of keep proposing code that tries to maximize the score.
**Speaker B**: 那么，这种机制与常规的、用于优化 Notebook 的通用 Agent 系统相比，核心区别究竟在哪里？
Original English
**Speaker B**: So what is different about this than just a general agentic system that can sort of optimize notebooks?
### 基于 MCTS 与 UCB 算法的代码搜索机制
**Speaker A**: 用现代 2026 年的术语来说——我们在 2024 年和 2025 年就在研发这项技术了，但用当下的概念来描述——它本质上是一个高度专业化的运行控制框架（specialized harness），其底层运行的算法正是蒙特卡洛树搜索（MCTS）。具体而言，系统在内存中维护着成百上千个潜在的 Notebook 候选实例，然后依据特定策略选取其中一个。至于具体如何选取，我稍后可以详细解释。在选中某个 Notebook 后，Gemini 会自我审视并思考：“我能做些什么来让这个 Notebook 表现得更好？”接着它会生成一个新版本的代码并执行测试，测试完成后再将其放回候选池中。因此你可以想象，整个候选池在逻辑上呈现为树状结构，每个候选节点都可以派生出子节点。而挑选节点的依据，则是强化学习领域一个非常经典的算法——上置信界算法（Upper Confidence Bound, UCB）。这是一种乐观选择算法（optimistic algorithm），它会尝试预估代码变异后处于 95% 置信分位数的可能收益，并优先挑选乐观上界最高的那一个候选。换句话说，系统绝不是贪婪地每次都挑当前表现最好的 Notebook，而是去预测“当前性能均值加上两倍标准差（μ + 2σ）”的最大值。正因如此，它始终在整个搜索空间中积极探索。
Original English
**Speaker A**: Right now it's essentially its own—in modern 2026 parlance, we actually worked on this in '24 and '25, but in modern parlance it's kind of a specialized harness that runs an algorithm which in Aera was Monte Carlo tree search. So essentially it's keeping hundreds or thousands of possible instances of notebooks and then it selects one—and I can explain how it selects one—and it decides, "Well, what can I do?" Gemini asks itself what can I do to make that notebook be better, and then it will make a new one and test it and then put it back into the candidate pool. So you can imagine the candidate pool is actually tree-structured because every candidate possibly has some children. And what you do is you pick based on something called—it's actually a fairly standard algorithm from reinforcement learning called upper confidence bound (UCB). So you essentially pick—it's an optimistic algorithm. So it tries to estimate, say, what's the 95th percentile outcome of mutation, and it tries to estimate that and it picks the one with the highest bound, the highest optimistic bound. So in other words, it doesn't always pick the best performing notebook. It tries to predict what's the current performance plus two sigma of its guess. And so it's always trying, so it hunts around.
**Speaker B**: 所以本质上是为了保证高召回率（high recall），对吧？
Original English
**Speaker B**: So high recall basically.
**Speaker A**: 没错，高召回率。系统在不断权衡并下注，以便以最高效的方式推动探索进展，而这往往并不等同于贪婪地只选择眼前得分最高的那个候选方案。有时候它挑中的是排名第五的方案。此外，我们还尝试过跨分支交叉重组的机制，也就是从两个候选方案中提取优秀构想并将它们融合在一起，以此为基础派生出第三个候选方案。
Original English
**Speaker A**: High recall. It's trying to make its bet so that it most efficiently tries to make progress, which isn't always greedily doing the best candidate. Sometimes it's the fifth best. We've also played around where it kind of recombines—it sort of takes ideas from two candidates and smashes them together and tries to make a third candidate out of that.
### 初始候选池生成与学术论文先验
**Speaker B**: 那么系统最初是如何生成第一批初始候选池种子的？
Original English
**Speaker B**: How does it seed the initial candidate pool?
**Speaker A**: 这一点非常令人惊叹：底层搭载的 Gemini 模型本身写代码的能力就极其出色。你直接给它一段关于问题的自然语言文字描述，让它写一个解决方案，它就能写出来。输入并不仅仅是一个简单的“这是你的评分函数，开始跑吧”，而是包含对目标函数的文字描述。甚至在很多任务中，我们还会直接附带：“这里有五篇以往学者尝试解决该问题的学术论文。” Gemini 非常聪明，它会亲自去阅读并理解这些论文，然后直接写出第一版尝试性的代码。虽然初版代码可能并不完美，有时甚至会报错、产生 bug 并返回负无穷大（-∞）的分数，但随后它就会在此基础上持续对代码进行突变和优化，不断思考如何让代码变得更好。
Original English
**Speaker A**: Well, that's the amazing thing is that underneath, Gemini is actually good at writing code. I mean, you just ask it, "Write me a thing," because you have a textual description of the problem. It isn't just, "Oh, here's your scoring function, start." You say a textual description of the function, and you might give it—in fact, we have under some things like, "Here are five papers that people tried to solve this problem with." And it's kind of smart. It actually goes and reads the papers and will actually take a first stab at code. It might not be great, or sometimes it has bugs and it returns essentially minus infinity, but it will then try to mutate the code and say, "Oh, it'll try to make it be better." So...
**Speaker A**: 这真的很酷。你完全不需要提前给它提供基础模板代码。当然，如果你愿意的话，提供一些初始 starter 代码也是可以的，但这并不是必须的。
Original English
**Speaker A**: ...it's pretty cool. You don't actually have to give it—I mean, you can if you want give it some starter code, but you don't have to.
### 并行搜索分支与跨迭代上下文学习
**Speaker B**: 那么在每次迭代过程中，你们会同时启动多少个 Agent？或者说，在搜索树的每一个迭代步上，你们会同时衍生出多少个不同的分支？
Original English
**Speaker B**: How many agents are you spinning up? I guess maybe not agents, or how many different tree branches are you spinning up at each iteration?
**Speaker A**: 在每一次迭代中展开的分支数，实际上涉及到一个核心权衡（trade-off）。从直觉上看，我们当然希望尽可能多地进行并行计算；但是，如果并行任务开得过多，各个分支就无法及时从前序迭代的成果中吸取经验。因此在目前的实现中，我们的默认设置大约是 10 路并行。也就是说，每次树搜索会同时生长 10 个叶子节点。
Original English
**Speaker A**: Oh, at every iteration. Well, there's a trade-off. You'd like to do a lot of parallel work, but if you do too much parallel work, you can't learn from previous things. So right now we use about—the default is 10 parallel. So you try to grow 10 leaves at a time.
**Speaker B**: 好的。
Original English
**Speaker B**: Okay.
**Speaker A**: 从实践效果来看，10 个分支左右是一个非常平衡的折中方案。
Original English
**Speaker A**: That seems to be about the right trade-off.
**Speaker B**: 当你提到“并行过多就无法从之前的迭代中学习”时，这具体是指 Orchestrator（协调调度器）层面的机制吗？你前面提到存在某种步骤，能够超越单纯的分数指标来进行方案的重组或决策。人类在做机器学习项目时，往往不会仅仅盯着单一的优化指标，很多时候还存在正交指标，甚至可以通过观察训练曲线来获取直觉判断，或者通过深入分析特定样本的坏例来寻找灵感。系统是否也会进行类似这种深度的内省分析（introspection）？
Original English
**Speaker B**: When you say you can't learn from previous iterations, that means that the orchestrator or is there some sort of—what's the—you said that there is some step which is able to like recombine or make decisions beyond just like the score. I mean, so yeah, I guess maybe one of the questions is, as a human when you are doing some sort of ML project, you don't just look at like, "Oh, there's this one metric that we're trying to optimize." Oftentimes there's like orthogonal metrics, sometimes even insights such as just watching training curves can sometimes give you intuition about what's going on, or like looking into specific examples. Does it do any sort of introspection like this? Is there...
**Speaker A**: 它确实掌握着完整的演进历史。但我刚才强调不能把过多任务完全并行的核心原因在于：如果你同时启动 10 个平行的搜索分支，第 1 号分支在执行时是无法看到第 2 到第 10 号分支正在尝试什么的。假若你一口气并发跑 1000 个分支，那就会耗费海量的计算资源，却得不到跨分支协同学习的收益。反之，当你跑完一个相对较小的批次时，系统就能沉淀下一段完整的历史记录。当然，我们必须对这段历史进行合理的修剪（prune），以防止上下文窗口发生爆炸；但模型能够完整获取它在编写代码时的思考过程以及这些代码运行后的实际表现和反馈结果。正因如此，它能够真正从先前的尝试中不断学习和迭代。
Original English
**Speaker A**: Well, it has the history of—but I meant why you can't do too many things in parallel is if you have 10 parallel searches at once, then number one can't actually see what numbers two through 10 are doing. So if you do a thousand at once, then you're using a huge amount of computation without a lot of cross-learning. Whereas once you finish a little batch, you get the history. Obviously you have to prune it so it doesn't blow up the context, but you get the history of what it was thinking about as it was kind of writing the code and the results of the code. So it can learn from its previous attempts.
**Speaker B**: 明白了。那么它能够跨越不同分支进行知识学习吗？
Original English
**Speaker B**: Okay. And does it learn across?
**Speaker A**: 当然可以，完全可以。本质上这就像是一个全局共享的上下文（shared context）。因此，它是在沿着探索路径不断推进思考的，绝不是 1000 个彼此孤立、完全无关的独立分支在各自盲目运行。
Original English
**Speaker A**: Oh yes. Yes. Essentially it's like one essentially shared context. Yes. So it is sort of thinking as it goes along. It's not like it's a thousand different completely independent branches.
**Speaker B**: 看来你们把 Gemini 的长上下文处理能力发挥到了极致。
Original English
**Speaker B**: You're really pushing Gemini's long-context abilities.
**Speaker A**: 确实如此。在这过程中必须配合非常周密精细的上下文状态管理机制。
Original English
**Speaker A**: That's right. And you have to do the right management and stuff.
### 目标评分函数的定义、对齐与防作弊循环
**Speaker B**: 太棒了，这非常巧妙。我想回到 R.J. 之前提出的那个问题：核心关键在于，第一阶段的首要目标是明确界定出你究竟想要优化的具体分数或评估指标，对吧？我也非常认同，很多时候这恰恰是整个问题中最具挑战性的部分。因此我觉得很有意思的一点是，我不确定自己是否会完全放心地把这部分工作托付给 Agent。在这个环节中，由人类参与其中（Human-in-the-loop）似乎显得更加不可或缺。
Original English
**Speaker B**: Yeah. Yeah. Okay. That's cool. I mean, going back to R.J.'s question, the key point here being the first key goal is to I guess identify what specific score that you are trying to optimize, right? Sometimes that is, I agree, kind of the hardest part of the problem. And so I find it interesting that I'm not sure I'd always trust my agent to do that part. That part seems like the more human task in the loop.
**Speaker A**: 确实如此，而且在这个环节必须格外谨慎。实际上，我们所做的很多工作都带有一种“元”（meta）的属性，一切都在非常高层抽象的维度上运行。你必须确保评分机制不存在逻辑漏洞。现实中非常普遍的一种情况是：你设计了一个评分函数，或者 Agent 提出了一个，亦或是你们共同讨论确定了一个，但在后续的代码迭代演进过程中，模型会钻空子、找到走捷径“作弊”（cheat）的漏洞——让你忍不住惊呼：“天呐，我根本不是这个意思！”因此，你必须反复测试、探索，在外部构建一个交互循环，在循环中不断修正：“不对不对，我的初衷不是那样”；或者在给它的 Prompt 指令中明确加上限制规则：“记住了，千万不能通过这种方式取巧”等等。
Original English
**Speaker A**: It is. And often you have to be careful. In fact, a lot of what you do is kind of very meta. I guess everything we do is very sort of high-level. You have to make sure that there's no—one common thing is you come up with a scoring function, or the agent does, or you do it together, and then the iteration finds a way to cheat or a hole, like "Oh no, I didn't mean that." And so you have to go through and often sort of play and have a loop around it where you kind of iterate like, "No, no, I didn't mean that," or you have to tell it in its instructions, "Okay, you know, don't do this." So...
### 科学创新的抽象层级：从底层繁琐代码到高阶评分函数
**研究员**：是的，这中间通常需要经历许多次迭代。因此，即使有了智能体（Agent）的协助，你也不一定能在第一天就构建出完全正确的评分函数（Scoring Function）。事实上，这种工作模式非常精妙。我想说的是，在过去——也就是 2024 年或更早的时候——很多很多研究生会把大量时间耗费在编写科学软件上；光是把代码写出来并跑通，就要付出极其巨大的心力。这就导致你往往只能尝试少数几种方法，或者试几样彼此高度相关的方案，然后就不得不停下来，因为你还得赶着写论文、赶着推进下一个实验。
Original English
**Researcher**: Yeah, there's often iterations. And so, even with agent help, you don't necessarily get the right scoring function from day one. And in fact, it's really neat because, I mean, in the old days—i.e., 2024 or something—you know, a lot of grad students would spend a lot of time doing scientific software, and it's just so much effort to write code at all that you kind of try maybe a few things, or a few things that are very related, and then you sort of stop because you have to write your paper, you have to do your next experiment.
**研究员**：而智能体系统在底层表现出一种近乎不知疲倦的韧性，它会不断尝试、不断尝试、再不断尝试。因此，使用它的人现在几乎可以把所有时间都投入到真正恰当的层面上——几乎就是纯粹的科学创造力层面。比如去深思：设计一个代价函数（Cost Function）究竟意味着什么？你明白我的意思吗？这几乎切中了科学问题的本质。你不再需要深陷于具体琐碎的泥潭里，比如“噢，我得把这个 CSV 文件导进来”，或者“我得把这个数据库调通”之类的杂事。你现在是在对自己的实际科学问题进行近乎深层哲学的思考，而不是困在对数据库等杂七杂八底层琐事的焦虑中。事实上，智能体能做的一件很酷的事，就是主动向你推荐数据集，比如提示你：“噢，你有没有考虑过引入这个数据集并做一个 Join 操作？”它会给出很多诸如可关联数据集之类的建议，这真的非常棒。回到你刚才提到的那个话题——关于智能体往往热衷于走捷径、搞奖励黑客（Reward Hacking）……
Original English
**Researcher**: This thing is kind of underneath kind of relentless because it keeps trying and keeps trying, keeps trying. And so the people who use it are now spending all their time almost at the right level, almost at the scientific creativity level. What does it mean to have a cost function? You know what I mean? And so that's almost like the essence of the scientific problem. You're not so much now in the details of, "Oh, I have to import this CSV file," or "I have to get this database to work," or whatever. You're now sort of thinking almost like deeply philosophically about your actual scientific problem, not down in the grungy goop of worrying about, you know, databases. In fact, one cool thing the agent can do is actually suggest datasets to you like, "Oh, have you thought about maybe pulling in this dataset and doing a join?" And so it'll make suggestions about datasets you can join with, which is kind of cool. Going back to what you said a second ago, in terms of agents love to hack things and the reward hack...
### 奖励黑客与“许愿神灯”的魔咒
**主持人**：在这方面，你有没有什么好玩的趣事，或者那种智能体行为彻底滑稽跑偏的精彩案例可以分享？
Original English
**Interviewer**: Are there—do you have any fun stories or interesting stories about, you know, where things were comedically run off the rails?
**研究员**：天哪，我一下子有点短路了。我知道其他人确实遇到过不少这种情况。我手头可能没有足够的细节来生动描绘出其中的喜剧效果，但它确实经常会让你……大吃一惊。
Original English
**Researcher**: Boy, I'm blanking. I know other folks have run into it. I don't know if I have enough details to sort of express the comedy of, but it does—you kind of get...
**主持人**：大吃一惊，是的。
Original English
**Interviewer**: Surprised. Yeah.
**研究员**：是的。我现在脑子里一时想不出特别具体的细节，抱歉，突然卡壳了。
Original English
**Researcher**: Yeah. I don't know if I have any really concrete—sorry, I'm blanking.
**主持人**：没关系，完全理解。我总喜欢把机器学习比作神灯精灵的故事，就像《猴爪》（Monkey's Paw）之前的那种古老寓言一样——［笑声］小心你许下的愿望，因为你真的会如愿以偿。
Original English
**Interviewer**: No, it's fine. Yeah. I always like to think of machine learning as sort of like the old genie stories before Monkey's Paw, like [laughter] careful what you wish for because you're going to get it.
### 代码空间进化与大模型的常识引导
**研究员**：没错，完全是这样。这种情况在这种系统里非常普遍，所以你必须格外谨慎。但另一方面，它本身也具备相当丰富的知识储备。妙处在于，Gemini 对海量领域的了解，某种程度上远超任何单一个体所能掌握的知识。因此，它至少懂很多东西，尤其是当你为它指定论文时——比如告诉它：“这里有五篇论文，它们都在尝试以某种方式解决这个问题。”所以在某种程度上，它确实带有一点“神灯精灵”的特质，但同时它又能在很大程度上做出合乎常理的事情。
Original English
**Researcher**: Yeah, that's right. And that happens very much with this. So, you have to be careful. But on the other hand, it has some knowledge. The nice thing is that sort of Gemini knows a lot about many things, sort of more than any one person can do. So it at least knows, especially if you point papers—like, "Here are five papers that tried to do this in some way." So to some extent it does have that genie feel, but to some extent it also sort of does sane things.
**研究员**：这也是为什么——回想一下所谓“进化编程”（Evolutionary Coding）的整个概念，它其实从上世纪 70 年代就已经存在了，人人都曾热衷于此，想着：“噢，让我们对 Lisp 代码进行随机变异来解决问题吧。”然而它之所以一直没有真正流行开来，是因为在代码空间里进行纯粹的随机变异几乎毫无价值。这就像 DNA 突变一样，绝大多数突变都是有害的。但在我们这里，情况不同了：我们实际上能够发现有价值的突变路径，因为大模型在底层具备理解力，它知道哪些探索梯度是有意思、值得尝试的。这正是这套系统之所以能够奏效的原因——其底层的演化循环本身就是一个 AI。因此，尽管它确实可能会发生过拟合，并产生你刚才提到的那些令人啼笑皆非的“神灯悖论”问题，但它同时也具备相当程度的合理性，因为底层模型理解世界。
Original English
**Researcher**: This is why—remember the whole idea of evolutionary coding. It's been around since the '70s. Everyone's loved to do that, like, "Oh, let's mutate Lisp code or whatever to do things." But the reason why it just hasn't taken off is that random mutation in code space is pretty much worthless. I mean, just like DNA, most things are harmful. So here, we can actually find—it sort of knows underneath and knows interesting gradients to try, which is why the thing works: that the underlying loop itself is an AI. So yes, it can maybe overfit and have funny sort of genie problems like you allude to, but it also has some amount of sanity because it sort of...
**主持人**：因为它对世界有所了解，它内部融入了世界知识。
Original English
**Interviewer**: It knows about the world and it has world knowledge in it.
### 从 Gemini 2.0 到 2.5：模型代际跃升带来的质变
**主持人**：在你们发表的论文里，你们使用的是 Gemini 2.5。我想大家也都知道，Gemini 模型的发展步伐非常迅猛。你们内部是否有量化指标，或者说因为你们一直在持续使用和改进这个工具，你们在内部是否观察到这种工具效能的“相变”（Phase Transition）？在过去一两年里，这种性能提升究竟有多显著？
Original English
**Interviewer**: So the paper, though, you were doing Gemini 2.5, and I think Gemini has advanced quite a bit. Do you have metrics, or—this is a tool that you're continuously using and it sounds like you're improving—and I'm wondering, internally have you seen almost like a phase transition in how effective this tooling has been? How dramatic has the improvement been over the last, like, I guess year or two?
**研究员**：噢，过去这一两年……是的，简直令人惊叹！换句话说，哪怕仅仅跨越半个版本——在本质上，我认为如果基于 Gemini 2.0，这套系统根本就不可能实现。［笑声］
Original English
**Researcher**: Oh, well, I mean year or two—yeah. Amazing. In other words, every—even every half version of—I mean essentially I think it would have been impossible under Gemini 2.0. [laughter]
**主持人**：是的，我也这么认为。换言之，在 2.0 下它根本跑不通。
Original English
**Interviewer**: Yeah, I think so. In other words, it wouldn't have worked.
**主持人**：所以你们是从 2.5 开始尝试的，而那仅仅是一个起点……
Original English
**Interviewer**: So, you started at 2.5 and that was like just the...
**研究员**：噢不，其实我们在这些方向上的实验已经持续了相当长一段时间了。早期阶段各种尝试就是行不通，后来突然间它开始奏效了，而到了现在，效果简直令人惊艳。因此，Gemini 主版本迭代所带来的飞跃，真的是惊人且震撼的。
Original English
**Researcher**: Oh, no. We've been trying to experiment with these things actually for a while. And things just weren't working, and then they started to work, and then now they're just amazing. So, the progress on Gemini major versions has just been stunningly amazing.
**主持人**：对，很多人最近都有类似的切身体会：那些曾经看似完全不可能的事情，在极短的时间内突然间就变得不可思议地好用。
Original English
**Interviewer**: Yeah. I think this is an experience a lot of people have been having where things that just seemed impossible are suddenly becoming magically useful really quickly.
### ERA 与 Antigravity 的结合及开源生态
**研究员**：因此，如果有人说——我甚至经常对科学家们这么说，因为总有些人会讲：“噢，我试过 2.0 之类的版本，觉得不好用。”我会告诉他们：“噢天哪，那已经是老黄历了，那是一年前的事了，在 AI 领域那简直像是一个世纪以前的事！”确实是这样。事实上，我们在其中一篇预印本论文中，已经将 ERA 与 Antigravity 进行了整合。
Original English
**Researcher**: And so, if people are—I even say this to scientists, cuz there's some people like, "Oh, I tried whatever 2.0, and I didn't like it." Oh yeah, that was a long time ago, that was a year ago, that was like eternity ago, right? Yes, and in fact, we even have one of the preprints where we've sort of combined...
**主持人**：把 ERA 和 Antigravity 结合起来。Antigravity 的整套实验脚手架也相当厉害，在那个架构下，你可以引入海量的学术论文，它能帮你自动编写大量的代码……
Original English
**Interviewer**: ERA with Antigravity, and that whole harness of Antigravity is pretty amazing too. That's the one where you can sort of pull in lots of papers and it can write lots of code for you, and...
**研究员**：是的。
Original English
**Researcher**: So, yeah.
**主持人**：那个结合版本现在对外公开发布了吗？还是说……
Original English
**Interviewer**: Is that publicly available, or is that...
**研究员**：额，Antigravity 本身？是的，是的。
Original English
**Researcher**: Uh, the Antigravity? Yeah, yeah.
**主持人**：Antigravity 当然是公开的。
Original English
**Interviewer**: Well, Antigravity is certainly publicly.
**研究员**：噢，抱歉抱歉。你是说 ERA 加上 Antigravity 的组合？
Original English
**Researcher**: Oh, sorry. Sorry, yeah. The ERA plus Antigravity.
**研究员**：额，那个目前还没有公开。
Original English
**Researcher**: Uh, not yet.
**主持人**：好的，目前还没有，明白了。
Original English
**Interviewer**: Okay. Okay, not yet. Okay.
### 多重假设检验与状态空间爆炸的困境
**主持人**：我觉得这个领域真的特别引人入胜。就像你说的，某种形式的代码突变自从计算机科学诞生之初就已经存在了。这里的典型难题在于过拟合，或者说多重假设检验（Multiple Hypothesis Testing）问题。我认为多重假设检验可能更能准确地描述这个困境：在这个场景下，你基本上是在不断产生新的假设——“我现在的假设是这个算法能跑通，接下来的假设是另一个版本有效”——这就带来了状态空间呈指数级爆炸的巨大风险。因为突然之间，我好像在优化一整套超级超参数（Hyper-hyperparameters），探索的状态空间发生了爆炸式扩张，导致系统似乎极易针对某个具体问题发生过拟合。对此你是怎么看的？因为另一方面，从我个人的实践经验来看——我甚至尝试过 ERA 的开源版本，把它接入了 Claude 并且现在就在后台运行着，所以我目前还无法断言它的实际表现究竟如何。
Original English
**Interviewer**: I find this area really fascinating cuz, like you said, there's been some form of code mutation out there since the dawn of computer science, basically. The canonical problem is sort of the overfitting or multiple hypothesis testing problem, I think, which is maybe a little bit better match to the problem where you're basically—my hypothesis now that this algorithm works, now my hypothesis is this—and so you run the risk that sort of it has exponentially exploded, right? Because now suddenly I have like these—it's like hyper-hyperparameters that I'm optimizing, and so you have this explosion of state space that you're exploring, and so that it seems much easier to sort of overfit to a problem. What are your thoughts about that? Because on the other hand, empirically my experience, I even tried the sort of open-source version of ERA. I kind of strapped it into Claude and it's running right now, so I can't tell you how well it's working.
**研究员**：好啊，我对结果很期待。
Original English
**Researcher**: Okay, I'm curious.
**主持人**：是的，之后我会告诉你进展。但我确实很好奇，这关于整个 AI for Science 领域也是一直萦绕在我心头的一个疑问：从科学研究的第一线实际经验来看，你对此有什么体会？
Original English
**Interviewer**: Yeah, I'll let you know. But I'm just curious to know—this is a question that's been on my mind about just general AI for science, and so what are your experiences with this sort of on the ground?
### 科学的本质：预测模型 vs 描述模型
**研究员**：我认为你的问题里实际上包含了两个相互交织的问题。因为当你谈到多重假设检验时，我们要明确：模型分为“预测模型”（Predictive Models）和“描述模型”（Descriptive Models）。你能看出来我在机器学习和统计学领域做了很多年吧。
Original English
**Researcher**: I guess there's two questions sort of embedded in your question, I think, right? Because when you talk about multiple hypothesis testing, right, there's predictive models and there's descriptive models, right? You know, can you tell I've been doing machine learning for a long time in statistics?
**主持人**：这是一个非常深刻的切入点！你介意详细展开讲讲吗？我不确定听众里的每个人都对这个概念很熟悉。
Original English
**Interviewer**: That's actually a really good point though. Do you mind explaining that, expanding that? I'm not sure that's something that everyone in the audience would be familiar with.
**研究员**：特别是在当今时代，我认为人们往往倾向于将这两者混为一谈。
Original English
**Researcher**: Especially in modern days, I think people are trying to sort of obscure the two.
**主持人**：如果一个人是从大语言模型（LLM）开始接触 AI 的，我甚至怀疑这种区分对他来说是否还有实质意义。
Original English
**Interviewer**: If you started with LLMs, I'm not sure that distinction would be meaningful.
**研究员**：说得太对了。所谓的“预测模型”，就好比你有一组输入数据，还有一组输出数据……
Original English
**Researcher**: That's right. So, a predictive model is like—let's say you just have some inputs and you have some outputs...
**主持人**：而你只是想写一段代码，让它在某个特定数据集上达到最低的误差率。
Original English
**Interviewer**: And you just—I just want to build a piece of code that tries to just have the lowest error rate on some dataset.
**研究员**：没错，那充其量只是一个纯粹的统计模型，对吧？而“描述模型”才是科学研究真正试图追求的目标。也就是说，模型必须具备外推（Extrapolate）的能力，因为它内部捕捉到了物理法则，或者说是对现实本质的某种真实写照；这样你才能依靠它来进行有效的外推。因为这就像……你知道，牛顿当年由苹果落地联想到了万有引力，但重力法则本身绝不是仅仅关乎苹果的，对吧？如果牛顿当年只是把这当成一个纯粹的机器拟合任务，如果你只是……
Original English
**Researcher**: That's just a statistical model, right? A descriptive model is actually what science is trying to get to, which is: okay, it should be able to extrapolate because it has sort of the physics or some actual description of reality that's captured within it. And then you can use it to extrapolate. Because it's sort of like—you know, yes, Newton thought apples and gravity, but gravity isn't actually about apples, right? If he had just fit—if he had taken it as a machine, you...
### 外推能力与物理先验的融入
**Speaker A**：你知道，就像17世纪的机器学习模型一样，它会说：“噢，苹果确实会往下掉，但行星呢？”你看，它会说：“我不知道啊，我完全没有任何关于行星的数据，所以谁知道它们会怎么运动呢，对吧？”
Original English
**Speaker A**: ...know the 17th century machine learning model like, "Oh, apples will fall, but how about planets?" You know, "I don't know, I have no data about planets, so [laughter] who knows what they do, right?"
**Speaker B**：所以，当你提到“外推”（extrapolative）能力时，我很好奇。当你谈到外推时，我可以用几种不同的方式去理解：其中一种是，你提到了物理模型或世界模型。你所指的是引入显式的先验知识——这些先验可能基于某种人类的直觉，或者在这种情况下是基于大语言模型（LLM）的直觉；还是说，你指的是物理规律实际上是由模型本身在底层学习出来的？或者说底层世界的运行规律本身就处于模型掌控之下？
Original English
**Speaker B**: So when you say extrapol... I'm curious. Okay, so when you say extrapolative, there's different ways I could think about this. One of them is, you said a model of physics or model of the world. Are you talking about introducing explicit priors that based upon some human intuition or maybe in this case LLM intuition, or are you talking about the physics is actually learned by the model, or the underlying process of the world is under the model?
**Speaker A**：这两种方式之间的界限其实是有点模糊的。对吧？因为当一个物理学家或科学家进行研究时，他们会运用自己的直觉；或者甚至不仅仅是直觉，实质上他们脑海中掌握着一套关于客观世界极其扎实确凿的事实基础。接着，他们会确保自己构建的任何模型在某种程度上都与这些已知事实保持自洽与一致。目前在 ERA 中，它在很大程度上依靠的是大语言模型的直觉——这基本上就是我之前所说的在底层拥有良好梯度的含义。特别是当你让它参考现有的学术论文时，它会尽力构建在底层逻辑上合理且站得住脚的模型。因为如果你给它明确的引导，例如提示它“一定要把这项技术和这项机制整合进去”，或者让它“去参考这几篇特定论文”，你就能得到相应的成果。因此，你完全可以为模型引入倾向于某些特定建模选择的偏置（bias），而且它自身也具备这种偏置，因为在预训练阶段，它内部就已经沉淀并积累了属于自己的那套关于世界的知识体系。
Original English
**Speaker A**: The distinction between those is a little bit blurry, right? Because when a physicist or scientist comes, they use their intuition, or maybe even more than intuition, essentially there's a solid pile of facts that they know about the world, and then they make sure that whatever model they build is sort of consistent with what's known. Currently in ERA, it is sort of LLM intuition. Essentially that's what I was trying to say about having a good gradient underneath. Especially if you point it at existing papers, it will try to build models that are kind of sane underneath. Because again, especially if you give it guidance like, "Oh, be sure to incorporate this and this," or "Look at these papers," you get these things. So you can introduce a bias towards certain model choices, and it will have a bias because its own little sort of world knowledge is accumulated inside of itself in pre-training.
### 从学术文献到代码复现的物理建模范式
**Speaker B**：你能给我们举几个具体的例子，说明这实际看起来是什么样子的吗？它是以某种方式进行建模——例如，如果你在处理偏微分方程（PDE）相关的问题，学术界有一些人们常用的形式化方法，比如神经算子（Neural Operators），或者在某种意义上将微分方程直接编码嵌入进去；又或者像流体和偏微分方程建模系统中人们所称的物理信息神经网络（Physics-Informed Neural Networks, PINNs）；再或者比如在分子系统中，文献中经常会用到等变性（equivariance）的概念。模型是在主动吸纳这些在现有文献中发展出来的成熟技巧呢，还是在为某种物理先验直接添加权重参数？当它引入这类物理或科学领域的知识时，实际表现究竟是什么样的？
Original English
**Speaker B**: Can you give us examples of what that might look like? Is it modeling something in a way where it's actually... there's different, for example, if you're doing something with partial differential equations, there's these formalisms people have, like neural operators for example, or where you can embed, you can encode a differential equation in some sense. Or I think that's called physics-informed neural networks or something is like one for fluid and kind of PDE type modeling systems. Or for let's say molecular systems, there's oftentimes this idea about equivariance. Is the model picking up on these tricks which have been developed in the literature, or are they adding some weights for some physical prior or something? Yeah, kind of what does that look like when it introduces like a physics... when it introduces some sort of knowledge like that?
**Speaker A**：我手头并没有足够详尽的统计数据，去断言说“噢，它在73%的情况下都会这样做”。但是，特别是当你为它指定现有的学术论文作为参考时，它会非常努力地尝试——事实上是非常擅长——将论文中所描述的方法迁移并适配到当前面对的具体问题上。实际上，它在这方面的表现令人惊叹。很多时候，你完全可以直接复现或者逆向工程出一篇学术论文。这正是迈克尔·布伦纳（Michael Brenner）非常喜欢做的事情。迈克尔常常会说：“噢，那篇论文听起来挺有意思的。”
Original English
**Speaker A**: I don't know if I have enough data to sort of say, "Oh, you know, 73% of the time it does this." But especially when you point it at existing papers, it will try to—in fact, it will do very well at adapting the methods that are described in the papers for the problem. In fact, we'll do an amazing job. You can actually often just recreate or reverse engineer a paper. That's again what Michael Brenner actually likes to do. He'll say, "Oh, that sounds like an interesting paper."
**Speaker A**：我们实际上针对一个特定问题做过这样的尝试，这其实算是某种巧妙的极客技巧（hack）。当时是我向迈克尔提出了这个构想：麻省理工学院（MIT）有一位教授编写了一套代码，主要解决这样一个问题——假设你在建筑屋顶上拥有固定的底面积，并且希望最大化一天之中所能捕获的太阳辐射能量。显然，你可以在垂直方向上向上搭建结构，从而捕获更多的阳光；你可以让模型去设计一个包含反射镜、支架或太阳能电池板的装置结构，让这些部件以你喜欢的任意角度或尺寸进行堆叠组合，通常会设定一个最大高度限制，然后让程序在那个庞大的设计空间中展开自主探索。我相信迈克尔——我们可以去向他求证核实——我相信系统当时甚至都没有真正安装原作者的物理仿真器环境，我记得代码仅仅是直接把论文里的核心算法与逻辑代码复现了出来。
Original English
**Speaker A**: Oh, we did this actually for the... we had this thing. It was actually kind of a hack. I suggested this to Michael where there's this one MIT professor who came up with some code to do essentially if you have a rooftop with a fixed area and you want to sort of maximize the amount of solar power you capture over a day, sort of solar energy. You can build up, which of course captures more sunlight, and you can sort of build... you can have it design a widget involving mirrors or struts or solar panels to sort of stack at whatever angles or sizes you like, and usually with a maximum height, and then try to let it sort of explore that design space. And I believe Michael—we can ask him—I believe it actually just... I don't think he actually installed the simulator. I think the code just reproduced the code...
**Speaker A**：因为它内部集成了一个专门的代码智能体（coding agent）。因此，它直接根据论文复现了代码，并把所有机制彻底理清楚了。所以答案是肯定的，只要你为它指明别人已经做过的既有工作作为指引，它就非常擅长消化和借鉴这些方法。在某些领域，比如我目前还没见过它主动去尝试真正的等变性建模（equivariant modeling）——如果你了解克莱布施-戈尔丹系数（Clebsch-Gordan coefficients）的话，就会知道那套数学体系可能会变得异常繁琐棘手，但也是非常精妙有趣的。所以我还不确定它是否能从零推导出真正的等变模型，但它掌握的知识确实极其渊博。
Original English
**Speaker A**: ...because it has like a coding agent inside of it. So just reproduced the code from the paper and sort of figured it all out. So yes, it's very good especially if given a pointer to what other people have done. It's very good at kind of like... oh, I haven't seen it try equivariant modeling. That that can get very hairy if you know about Clebsch-Gordan coefficients. It's pretty fun. Yes. So I don't know if it'll do the true equivariant stuff, but it actually knows a lot.
**Speaker A**：我其实还清晰地记得 Gemini 2.5 刚发布时的情景。我知道这并不完全局限于 ERA 的范畴，但当 2.5 推出的那一天，我依稀记得自己心里当时就感叹：“天哪，这是一个全新的世界。”因为那天我对它说：“嘿，Gemini 2.5，你能帮我写一段提升决策树（boosted decision tree）的代码吗？”然后它当场就写出来了。
Original English
**Speaker A**: Um, I remember actually when Gemini 2.5 came out—I know this is not exactly about ERA, but I remember sort of thinking, "Oh, this is a new world." When like the day 2.5 came out, cuz I said, "Hey, Gemini 2.5, can you write me some boosted decision tree code?" And it did.
**Speaker B**：而且代码顺利运行了？
Original English
**Speaker B**: And it worked.
**Speaker A**：它就那样直接运行成功了，毫无差错。于是我说：“是的，这确实是一个崭新的世界。”所以，回到你最初提出的那个问题，我认为答案是肯定的：如果你为它提供明确的方向性指导，告诉它“将这类物理机制纳入模型至关重要”，它就完全能够做到这一点。因此，至少在我们目前的观察中，它未必能完全无中生有地发现一套全新的物理规律——就像如果你完全不了解克莱布施-戈尔丹系数或相关的基本事实，你明白我的意思吧，它不会凭空从零开始独立发现一种前所未见的物理模型架构。但只要你向它传授自然界中已知的关键约束条件，它就绝对能够在这些约束下演化推进。我不知道这是否回答了你的问题。
Original English
**Speaker A**: It just did. Yes. And I said, you know, yeah, this is a new world. So yes, I think to loop back to your question, I think yes, if you give it sort of guidance about, "Oh, you know, it's important to put this kind of thing in," it will. And so it won't necessarily, at least not that we've seen, discover completely new... like if you didn't know about Clebsch-Gordan coefficients, I don't know, you know what I mean? Didn't know about something, it won't completely discover a new kind of physical model from scratch, but it will certainly if you tell it about interesting constraints about the world that are known, it will certainly evolve. I don't know if I answered your question.
### 人类科学家的独特价值与参差不齐的能力前沿
**Speaker B**：这么说来，在未来一到两年里，人类科学家依然还有立足之地。
Original English
**Speaker B**: So, so there's still room for humans for the next year or two.
**Speaker A**：噢，事实上，回到这个话题，我认为人类绝对拥有不可替代的巨大空间。因为怎么说呢，尽管我们拥有像 Co-Scientist 这样的工具，它能够辅助科学家生成科学假说，但说实话，我至今仍然没有在 AI 身上看到真正的人类创造力、哲学思考深度以及那种严谨缜密的推演求证。你绝对依然离不开人类。我完全看不到人类会退出历史舞台的迹象。AI 确实经常会提出一些稀奇古怪、出人意料的建议；我自己也确实在地球化学领域的一个有趣难题上实际使用过 Co-Scientist，并且借助它了解到了一种我之前从未意识到会出现在岩浆当中的新型离子。因此，它确实能告诉你一些引人入胜的新知识，你也能从中学到很多东西，但我认为它根本无法替代人类真正的创造力。
Original English
**Speaker A**: Oh, in fact, going back, I think there's totally room for humans because I don't know. I mean, we have Co-Scientist that tries to help you come up with sort of hypothesis generation, but really I still haven't seen sort of the creativity and the philosophy and the sort of the careful rigor. You totally need the humans. I don't see humans going away. They can make strange suggestions, and I've used Co-Scientist actually for an interesting problem in geochemistry, and I learned about a new kind of ion I didn't realize happened in magma. But and so it'll tell you interesting things and you'll learn stuff, but I don't think it sort of substitutes for human creativity.
**Speaker B**：你看，回顾一下我们刚才讨论的技术演进历程，短短两年前从 Gemini 2.0 到 2.5，大家就已经见证了一次巨大的跨越；而如今又过去了一两年，我们已经发展到了 3.5 版本，正如你所说的，它现在的表现要好得多。通常当我们观察任何增长曲线图时，如果某段曲线看起来呈现出指数级增长的态势，它要么实际上正处于 S 型曲线（Sigmoid）的中后段，要么正处于大爆发腾飞的起点阶段，对吧？我想，现实中的所有指数增长最终都会不可避免地转变为 S 型曲线。
Original English
**Speaker B**: You know, going back, we were just talking about two years... 2 point... you had Gemini 2.0 to 2.5, and this was like you already saw a leap. And now it's been another year or two, and now we're at 3.5, and you're saying this is working much better. I mean, whenever you look at a graph, you know, if something looks like an exponential, it can either... you can either be in a sigmoid or you can be at the beginning of a takeoff, right? I guess every exponential turns into a sigmoid eventually.
**Speaker A**：所有的指数增长最终都会转变成 S 型曲线……
Original English
**Speaker A**: Every exponential turns into...
**Speaker B**：但关键的核心问题在于，我们此刻究竟身处这条曲线的哪一个位置？
Original English
**Speaker B**: But the question is like, where are we on that?
**Speaker A**：我想，我个人非常坚信所谓的“参差前沿”（jagged frontier）这个概念……
Original English
**Speaker A**: I mean, I guess I'm a big believer in sort of the whole jagged...
**Speaker A**：没错，就是能力极不均衡的参差前沿。至少从我亲眼所见的现状来看——虽然我无法断言未来两三年究竟会发生什么——但不可否认的是，在参差不齐的能力版图上，某些尖峰确实表现得极为突出：比如在编写代码的能力方面、在搜集与整合广博知识方面，以及在检索和发现关联事物方面，模型展现出了巨大的峰值，这是极其了不起且令人振奋的突破，对广大科研工作者来说也是巨大的福音。但截至目前，在逻辑严密性方面，它的表现相对较弱。我们可以探讨像国际数学奥林匹克或者常规数学领域的复杂推理，但在哲学思辨和原创性创造力方面，我认为它依然存在明显的短板。也许未来一切能力都会像某些人预言的那样迅速膨胀并跨越临界点，但至少现在，我依然目睹着非常强烈的参差不齐特性。因此我可以预见，也许它在编程、模型拟合以及提供灵感建议方面会变得越来越擅长；但就目前而言，事实并非如此——到目前为止，科学探索依然绝对需要人类科学家的参与。
Original English
**Speaker A**: Yeah, the jagged frontier. And so certainly at least what I see—I mean I don't know what's going to happen in a couple years, but yes, there's some big spikes out in jaggedness in terms of coding ability and just gathering knowledge and finding related things, and that's huge and wonderful, which I think is great for scientists. So far, it's kind of less in terms of rigor. And we can talk about things like the International Math Olympiad and math in general, but in terms of sort of philosophy and creativity, I think it's still kind of not. And maybe it'll... maybe everything will, some people are saying everything is going to inflate and pass, but I'm still seeing a lot of very strong jaggedness. So I can see, well, maybe it'll get to be extra good at coding, and extra good at fitting models, and extra good at making suggestions and things, but I don't know. So far no, so far you need the humans.
### 多重假设检验与科学发现的本质
**Speaker B**：是的。我想把话题拉回到关于多重假设检验（multiple hypothesis testing）的问题上来。
Original English
**Speaker B**: Yeah, I want to get back to the question about the multiple hypothesis testing.
**Speaker A**：抱歉，刚才完全沉浸在这个很有意思的旁支话题里了。所谓的多重假设检验，是指当你构建了一个描述性模型，声明客观世界就是按照这种规律运作的；同时你掌握着海量的数据，就如同拿起了十亿支飞镖朝着目标投掷过去，然后……
Original English
**Speaker A**: Sorry, totally love the tangent. Multiple hypothesis testing is when you have a descriptive model and you're saying this is the way the world works, and you have a bunch of data and you take a billion darts and you throw and...
### 科学发现中的预测模型、描述性模型与过拟合风险
**Guest**: 所以必须保持审慎。统计学中有一个概念叫做假发现率（False Discovery Rate, FDR），对吧？因此核心问题在于：当前发现的究竟是描述性模型，还是某种预测性模型？归根结底，科学家的职责就是确保系统给出的结论具有真实的描述性意义。到目前为止，我们还没有仅凭这些现有组件构建出一个能够真正发现全新物理规律或开辟全新科学前沿的系统；但它确实算得上是一件超级工具（Power Tool），能够辅助你去发现全新的科学。所以，我或许是在尝试以另一种方式回应你之前提出的问题——我认为这触及了问题的核心所在。是的。
Original English
**Guest**: So you have to be careful. There's something called false discovery rate, right? And so the question is: is this finding descriptive models or is this finding sort of predictive models? And fundamentally, the scientist is there to make sure that whatever it's saying is descriptive. We haven't been able to make a system so far out of these pieces that can really sort of discover completely new physics or completely new science, but this is sort of a power tool to help you discover completely new science. So, maybe I'm trying to unask your question about—I think this gets at the heart of it. Yes.
**Host**: 是的。但如果按照你所说的，撇开纯粹的——好吧，我们先不谈这个。既然它并不是试图独立构建出对物理世界的描述性模型，那部分工作仍然属于科学家自身。那么，面对那种最普通、最典型的过拟合问题（Plain Old Overfitting）又该如何处理呢？
Original English
**Host**: Yes. But then you're saying what about just pure over—okay, let's set aside. It's not trying to figure out a descriptive model of the world. That's still up to the scientist. But what about just plain old overfitting?
**Guest**: 没错。确实如此，你必须异常小心，因为这确实是一件大功率的电动工具（Power Tool）。甚至可以说——虽然可能不太中听——它稍有不慎就会切断你的手指。你能理解我的意思吗？你必须[笑]非常谨慎，必须保持极其严谨的态度。事实上，现在的环境下你反而需要比以往更加小心、更加严密，才不至于自欺欺人。你真的、真的必须极其细致地设置那些隐蔽的留出测试集（Hidden Holdout Sets），绝不能提前窥探它们。你必须保持超乎寻常的严谨，以确保自己不会彻底翻车，因为它完完全全就是一件大杀器级别的强力工具。
Original English
**Guest**: Yeah. Yes. You have to be very careful because it's a power tool. It can—I shouldn't probably say this, but it can slice your fingers off. You know what I mean? You have to be [laughter] very careful and you have to be very rigorous. In fact, now you have to be more careful, more rigorous to not fool yourself. You really, really need to be just excruciatingly careful about having, you know, very hidden holdout sets you don't look at. You have to be just super, super rigorous to make sure that you don't completely—because it is a total power tool.
**Host**: 所以对于“如何才能不切断自己的手指”这个问题，答案本质上就是：你依然需要沿用那些经典的验证技术，只是在运用它们时必须更加如履薄冰。
Original English
**Host**: So the question to how do I not slice my fingers off is: you need to use the same techniques but be very careful with them.
**Guest**: 没错。
Original English
**Guest**: Yes.
**Host**: 这真是一个非常明确透彻的回答[笑]。
Original English
**Host**: That's a very clear answer that [laughter]—
### 审美品味与工程严谨性的平衡
**Host**: 好的，明白了。在我的印象中，之前似乎还没有哪位受邀嘉宾提到过这一点。这确实是一项至关重要的技能，甚至可以说是新时代最关键的核心能力之一。当下人们经常大谈特谈所谓的“品味”（Taste）。
Original English
**Host**: Yeah. Okay. Yeah. I actually don't think I've heard any guests say that. Yeah. It's like it is, I think, a very important skill, maybe one of the most important skills in the new—people talk a lot about taste.
**Guest**: 是的。不过，也许这本身就是品味的一种变体，但是……
Original English
**Guest**: Yeah. But and maybe this is a variation of taste, but—
**Host**: 品味如今往往被视作硬币的另一面。
Original English
**Host**: The taste is like the other side right now.
**Guest**: 严谨性就是与之对应的另一面。正是如此，确实如此。事实上，如果一定要说的话……
Original English
**Guest**: It's like the rigor. It's like the—yes. Uh, yes. In fact, if anything—
**Host**: 对，那么品味与严谨性之间，究竟哪一个更重要呢？
Original English
**Host**: Yeah. Taste or rigor, which one's more important?
**Guest**: 这个很难一概而论。在我看来，至少按照我个人的理解，一方面科研人员本质上也是软件开发者，这两个领域在很多方向上有着巨大的重合度。我观察到软件工程师群体普遍存在不少焦虑与担忧，大家会想：“天呐，以后我该何去何从？写代码似乎正变得越来越自动化了。”因此我认为分化出了两种主要倾向：许多人选择倒向创造力那一端，认为“那我就去充当灵感的源头，由我去探索新科学、构思新产品，专注于发挥极致的创造力”。我也坚信这种创造力绝不会消失。但同时，另一部分人则走向了严谨性这一侧，他们关注的是：“我要确保系统不会崩溃，确保架构具备良好的可扩展性，确保计算结果万无一失。”我认为这两者都是必不可少的。你需要在这两方面分别具备卓越才能的人才，但他们并不一定非得是同一个人。更广泛地讲，随着整个软件工程的演进，这种平衡甚至已经超出了纯科学的范畴。未来必定是有人贡献天马行空的创造力，有人构筑坚如磐石的严谨性，这两者将成为不可动摇的双重锚点。
Original English
**Guest**: Well, I don't know. I think people—at least the way I am viewing it—I mean the aspect that researchers are software developers, which there's a lot of overlap in a lot of fields. I'm seeing that software engineers are—it's almost like obviously there's a lot of concern like, 'Oh no, what am I going to do? Coding seems to be getting automatic.' So I think there's sort of both. I think there's a lot of people get pulled into, 'Well, I'll be the creative source, so I'll try to figure out new science, I'll try to figure out new products, I'll try to sort of really be very, very creative.' And again, I'm a strong believer that I don't think that's going to go away. There's also people sort of pulled towards rigor like, 'Oh, I want to make sure this doesn't crash. I want to make sure this scales. I want to make sure this isn't wrong.' I think you need both. And I think you need people who are really good at both. But they don't necessarily have to be the same people. But yes, I think you need—I think this is even broader than science, just as sort of software engineering evolves. Yeah, it'll be, you know, the people who will bring the creativity and the people who will bring the rigor, and I think those will be sort of anchors.
### 竞赛机制、古德哈特定律与奖励黑客
**Host**: 我还注意到业界存在一些相关的探索。比如斯坦福大学针对科学问题推出了一个非常酷的排行榜，类似针对 Agent 的基准测试排行榜。不知道你是否了解这个项目？在我看来，让不同的智能体在公开排行榜上同台竞技，是一个非常有意思的想法。粗看之下，ERA 所做的事情本质上其实也有点类似于排行榜机制，只不过它是在内部运行，并不断对各种想法进行重组融合。对此你怎么看？你们团队目前是否也在探索类似的方向？这种机制究竟存在哪些潜在弊端或者独特优势？
Original English
**Host**: There's some other things that I've seen are related work out there. There's a really cool leaderboard for, you know, like Claw leaderboard, agent leaderboard for scientific problems from Stanford. I don't know if you're familiar with it. It seems like a really interesting idea to me to have, you know, sort of different agents kind of competing on the leaderboard. So, it seems like if you squint a little bit, what ERA is doing is kind of a leaderboard, but it's internal and it's recombining ideas. Whereas what are your thoughts about this, and is that like a thing that you guys are working on, and is there problems with that or advantages to that?
**Guest**: 说来耐人寻味，整个 ERA 项目最初的缘起，其实正是因为——可能很多人并不知道——Kaggle 其实是属于谷歌旗下的。
Original English
**Guest**: Ironically, you know, the whole ERA project actually started because—people may not realize—Kaggle is actually part of Google.
**Host**: 噢，确实如此。
Original English
**Host**: Oh, yeah.
**Guest**: 当时我们面临的课题被称为“AutoKaggle”（自动 Kaggle 竞赛）问题。初衷非常纯粹：我们能否构建一套算法系统，让它自动在 Kaggle 各类数据竞赛中斩获优胜？这也是 ERA 系统之所以呈现出今天这种形态的由来，整个项目就是从这个愿景起步的。而这最终又绕回到了过拟合的问题上，对吧？如果你曾经亲自参加过 Kaggle 竞赛就会深有体会。
Original English
**Guest**: Uh, and so it was called the AutoKaggle problem. So it was actually like that's what it was: let's try to have a system that can sort of, you know, win at Kaggle competitions. So that's sort of why it sort of has the shape. That's sort of how the project started. And it goes back to sort of overfitting, right? If you've ever actually competed in a Kaggle competition.
**Host**: 我确实参加过 Kaggle 竞赛，至少完整打过一场。那真是一种极为奇特的现象，因为在比赛中过拟合简直到了泛滥成灾的地步。参赛者们往往能以匪夷所思的方式对特定数据集进行极度过拟合，令人叹为观止。
Original English
**Host**: I have done Kaggle competitions, and—or I've done one. It is a really interesting phenomenon because there's this overfitting is like rampant. Yeah. Yeah. And it's really impressive how people can overfit to certain datasets in a way that is—yeah.
**Guest**: 我们甚至做过一个非常有趣的项目，如果你感兴趣我可以多讲讲——那个项目旨在减少飞机尾迹云（Contrails）的温室效应。如果你想了解的话我很乐意展开。当时我们举办了一场预测飞机尾迹云的 Kaggle 竞赛，结果有选手的成绩居然超过了我们自己的基线模型。然而深入排查后才发现，原来是我们发布的标注标签中存在半个像素（Half-pixel）的坐标偏差。这牵涉到底层定义是将 (0,0) 原点设定在像素的左下角还是几何中心的问题——你知道我的意思吧？参赛选手敏锐地捕捉到了这个漏洞并加以利用，从而榨取出了微弱的额外提分。因为当你在做数据增强、生成合成数据并进行旋转变换时，必须严格补偿那半个像素的偏移量。所以[笑声]，事实证明人类自身在这种竞赛环境下，行为模式和那些大语言模型如出一辙，都会千方百计地去利用规则漏洞进行“奖励黑客”（Reward Hacking）。
Original English
**Guest**: Or we even had—we have a fun project I can talk about more if you like, that tries to mitigate contrails. I'm happy to talk about that if you want. Uh, and we had a contrail Kaggle competition and people actually beat us, but they found that we had a half-pixel error in our labels, and they—cuz it had to do with the center versus the lower-left, like where is (0,0)? Is it in the lower-left of the pixel or is it in the center? You know what I mean? So, they found that and exploited that and squeezed whatever a little bit extra stuff. Because it turns out when you make artificial data and you rotate it, you have to make sure that you take into account that half-pixel offset. So they [laughter]—so yes, people or people themselves will act like these LLMs and try to sort of reward hack upon these things.
**Host**: 这又令人联想到了古德哈特定律（Goodhart's Law）。正所谓：任何一项指标，一旦被选作考核的目标，它便不再是一个优秀的指标了。确实是这样。
Original English
**Host**: Oh, it sort of goes back to, what is it, Goodhart's law. You know, Goodhart's law: any metric that becomes a target is no longer good as a metric. Yeah.
**Guest**: 确实如此。竞赛本身固然有价值，但你必须极其谨慎，必须建立层层递进的严密验证防线。比如，你可以在明确知道系统在针对该指标进行优化的前提下去开展工作，但心里必须时刻清醒：此时古德哈特定律已经悄然生效，稍有不慎就会走入歧途。这也是为什么整个 AI 领域建立的各种 Benchmark 排行榜总是很快就会被耗尽失效的原因，因为古德哈特定律会对你创建的每一个排行榜无差别地起作用。所以归根到底，你必须随时后撤一步，保持极度严密的审视。
Original English
**Guest**: Uh, and so that's the—I mean it's good, and it's just you have to be very, very careful and you have to, again, you have to have like layers of rigor. Like, okay, but we'll do this and we'll optimize for this, but you have to realize, okay, that's just now Goodhart's law applies and you have to be careful. And so that's a lot of reasons why the whole AI field has been kind of constantly exhausting these things, because again Goodhart's law applies individually to every leaderboard you make. So again, it's sort of you just have to step back and be very, very careful.
### ERA 在真实竞赛中的表现与考验
**Host**: 这个观点对我非常有启发。随着我们采访了越来越多的嘉宾，如何驾驭大语言模型及 AI 科研所带来的复杂性，已经成为一个反复出现的核心主题。
Original English
**Host**: No, that that's really useful to my thinking. This is, you know, as we've had guests on, it's been a recurrent theme of how do you manage all this complexity that's introduced by LLMs and AI science.
**Host**: 我接下来的追问还是关于 Kaggle 上的过拟合。在你们立项攻克 AutoKaggle 之后，这个系统的实际胜率或成功频率究竟如何？我的推测是，你们很可能把这个自动化系统直接拉去跑了所有的 Kaggle 竞赛，实际表现到底怎么样？
Original English
**Host**: I think my followup question was about overfitting in Kaggle. Yeah, it is. If you had an AutoKaggle problem, and then the question is given AutoKaggle, how often was it successful? I mean, I assume you probably just ran this on like all of your Kaggle competitions or something.
**Guest**: 我们在各种练习赛性质的“游乐场竞赛”（Playground Competitions）中进行了广泛测试，它在这些游乐场赛事中的表现非常出色。随后我们也让它实际报名参加了各类正式比赛。事实表明，过去几年里各种排行榜和算法竞赛的规模已经呈现井喷之势，早已远远超出了 Kaggle 一家的范畴。我们在不少比赛中都取得了极佳的战绩。其中特别令我们引以为傲的是美国疾控中心（CDC）组织的一场专业赛事：任务是提前一周预测全美各个州和特区下一周的新冠（COVID）与流感（Flu）确诊病例数。ERA 在这场提前一周的预测挑战中交出了极其亮眼的成绩单。
Original English
**Guest**: Well, we tried it on various, like what they call playground competitions, and it did very, very well in the playground competitions. We've entered into different competitions. Some of them, it turns out in the last few years just the number of leaderboards and competitions and whatnot have just exploded far beyond Kaggle. So we've done very well, and some of them—like one thing we're super proud of is the CDC set up this competition where you try to predict next week's the number of COVID and flu cases that will happen in every state and territory in the US. You try to predict a week in advance, and ERA did super well on that.
**Host**: 这太有意思了，因为某种意义上，当年谷歌正是通过 Google 流感趋势（Google Flu Trends）项目，最早开创了利用大数据追踪疾病传播动态的先河。时隔二十年左右，你们以这样一种方式完成了一个历史性的闭环轮回。
Original English
**Host**: It's funny because in some sense Google invented the concept of using data to track disease progression with Google Flu. So it's kind of funny that you were sort of going full circle 20 years later or something like that.
**Guest**: 所以那次比赛的表现相当亮眼。但在我们报名的另外一些比赛中，效果就没有那么理想了。这往往是因为人们非常——再次印证了那个道理——你在这些比赛中的最终名次，很大程度上取决于你投入了多少悉心照料（TLC, Tender Loving Care），以及你愿意为了压榨出极限的……
Original English
**Guest**: So that did very well. Other ones where we've entered, we weren't quite as good, often because people are very—again, you know, sometimes how well you do in these competitions is a measure of how much sort of TLC you put into it and how much you're willing to squeeze the—
### 优化瓶颈与人类在环交互
**Speaker B**: 最后的 0.001……所以，我的意思是 ERA 确实表现不错。它让你非常接近目标，但我们没能消除最后那大约 30 个位次左右的差距，因为没有人去把那最后可能 0.01 的微小差距给抠出来削减掉。确实是这样。
Original English
**Speaker B**: ...last 0.001. And so, it was—I mean ERA did well. You got pretty close, but we didn't close the jump in the last whatever 30 places or whatever because no one was there to shave the last, you know, 0.01 off the thing. Yeah.
**Speaker A**: 整个过程是非常迭代式的吗？比如你去做这个任务，我在论文的图表里看到，当它发现某些新东西时，就会出现阶跃式的提升变化，然后曲线变平缓，接着又是这样。所以这在很大程度上是一种人机协同（human-in-the-loop）的过程吗？比如：“好吧，你在这个问题上卡住了，试试这种思路”，是类似这样吗？
Original English
**Speaker A**: Is it very iterative like you get to it, you know, it sort of—I saw the charts in the paper and, you know, you sort of get these step changes as it discovers something, and then and then flat, and then... So is it very much human-in-the-loop, like: "Okay, you've stalled on the problem, like try this kind of thing"? Okay.
**Speaker B**: 是的。在外层循环中——我认为那差不多是更有趣、更具创造性的部分。确实如此。比如：“来看看这篇论文”，或者“哦，你现在的做法不对”，你明白我的意思吗？所以，这几乎就像带了一个极度积极主动、而且不需要睡觉的研究生一样。你大致告诉它一些想法，引导它往各个方向去探索。
Original English
**Speaker B**: Yes. At the outer loop which is—I think that's almost like the more fun creative part. Uh, so yes: "Oh, here, have a look at this paper. Oh, you're doing something bad," or you know what I mean? So, it's almost like having a hyper-eager grad student or something who doesn't sleep. And you sort of tell it things and you sort of guide it around.
**Speaker A**: 通常一个人多久会介入干预一次？相比之下，外层循环的具体形态是怎样的？
Original English
**Speaker A**: How often does someone intervene versus like what does the outer loop look like?
**Speaker B**: 它可能会自主运行几个小时，然后回过头来给你呈现一些样本和示例。然后你就可以——只要你想，你就可以继续尝试、继续去测试和探索它。
Original English
**Speaker B**: Uh, it might run for a few hours and and come back and give you uh some examples, and then you would, you know, you can—as far as you like, you can sort of keep trying and keep poking at it.
**Speaker A**: 这么说来，它的设计初衷在很大程度上就是人类在环（human-in-the-loop）了。
Original English
**Speaker A**: So that's—it's very much designed to be human-in-the-loop then.
**Speaker B**: 没错。是的。
Original English
**Speaker B**: Yes. Yes.
**Speaker A**: 这很有意思，因为我试过的许多其他工具，往往都非常偏向于单次触发（one-shot）的模式。
Original English
**Speaker A**: Interesting, because a lot of the other tools I've tried tend to be very the one-shot.
**Speaker B**: 嗯，我想这取决于你怎么定义，对吧？显然你会先跟它对话并启动它，它会自主运行若干个小时，然后带着结果返回。但接下来，那就是人类创造力介入发挥作用的地方了——你这时是在运行外层循环。在循环中，这取决于你想不想睡觉，但每隔几小时你就会过去看一下，给它新的尝试方向，继续推进……
Original English
**Speaker B**: Well, I guess it depends on your definition, right? I mean it's obviously you talk to it and you start it, and it'll go for some number of hours and then come back. But of course then that's where the human creativity kicks in, and then you're sort of doing the outer loop where you sort of—you know, depends if you want to sleep, but you know, every few hours you go and you give it another try and you...
### 计算预算与资源权衡
**Speaker A**: 你们给这个系统分配了什么样的预算？[笑] 就像是不小心一下子烧掉了一百万美元那种情况吗？
Original English
**Speaker A**: What kind of budget are you giving this thing? [laughter] Like you blew through a million dollars accidentally kind of thing.
**Speaker B**: 我实际上并不确切知道具体数字，因为我们调用的是对 Gemini 的内部接口。所以实际上我确实不太清楚。
Original English
**Speaker B**: I don't actually know because we're using sort of, you know, internal calls to uh to Gemini. So actually I don't actually know.
**Speaker A**: 不过你看，这里面有 Token 预算的考量，但同时还有另外一面，那就是我正在解决的问题本身在计算上是非常昂贵的。
Original English
**Speaker A**: So but look, there's token budget, but then there's also like: I'm solving a problem that is computationally expensive.
**Speaker B**: 哦对，那也是一方面。系统底层确实是这样，因为评分函数本身可能就需要进行蒙特卡洛估计（Monte Carlo estimation）之类的计算。是的。所以你最终实际上可能会消耗大量的计算资源，仅仅是为了完成评估，或者是仿真模拟——比如如果里面包含一个模拟器，它就必须运行一次模拟。所以是的，你完全可能会消耗相当数量的 CPU 或 GPU 算力。
Original English
**Speaker B**: Oh yes, that also. It essentially underneath it, cuz the scoring function itself might have, you know, Monte Carlo estimation or whatever. Yes. So you actually end up—you can actually end up using a lot of compute just to even do—or simulation, like if you have a simulator inside, has to run a simulation. So yeah, you can spend a fair amount of just CPU or GPU.
**Speaker A**: 我自己拿 ERA 和 Claude 做的小实验，是为某些分类问题构建神经网络。显然，如果有足够的数据，更大的网络表现会更好，但它们的训练成本也更昂贵。于是你就会开始遇到一个权衡问题：如果我有一个固定的预算，我该如何管理这笔预算，才能把资金花在最有效的解决方案上？
Original English
**Speaker A**: So my little experiment with ERA—ERA and Claude—is to build a neural network for some classification problems. And so obviously like if you have enough data, then larger networks work better, but they're more expensive to train, and you start to run into a question of how do I manage my budget if I have a fixed budget so that I'm spending my dollars on the most effective solutions.
**Speaker B**: 确实如此。我认为这仍然是我们需要去探索解决的问题。但这当然与你带一个研究生并没有什么不同：假设研究生试图训练一个极其庞大的神经网络或处理非常庞大的数据集，他们自己也必须面对类似的问题，比如“是否存在缩放定律（scaling law）？我能够外推预测吗？”所以这并不是一个新问题，两者面临的是同样的困境；但在这个系统上问题可能显得更加紧迫，因为它是如此不知疲倦地持续运转。正因为不知疲倦，它遇到这种预算与算力瓶颈的速度要比一个研究生快得多。
Original English
**Speaker B**: That's right. And I think that's still something we need to figure out. But it's of course it's no different than if you have a grad student and they're trying to train a very, very large neural network or a very, very large data set. They themselves have to—there's some thing like: "Oh, is there a scaling law? Can I extrapolate?" So it's the same problem, but maybe more urgent because it just—it runs into this problem because it's so relentless. It runs into the problem much quicker than a grad student could.
### 飞机凝结尾迹与气候影响
**Speaker A**: 你们所优化的其中一个课题是飞机凝结尾迹（contrails）。你能详细谈谈这个项目吗？
Original English
**Speaker A**: One of the things that you optimized was contrails. Can you talk a little bit about that?
**Speaker B**: 好的，那我就花一两分钟时间来谈谈凝结尾迹这个问题。
Original English
**Speaker B**: Uh well, let me maybe spend a minute or two talking about the contrails problem.
**Speaker A**: 太好了，请讲。需要背景说明的是，这里指的是 contrails（凝结尾迹），而不是 chemtrails（化学尾迹——那纯粹是一个阴谋论）。
Original English
**Speaker A**: Yes, please. For context, contrails, not chemtrails, which is a conspiracy theory.
**Speaker B**: [笑] 是的。不过即便如此，你也同样应该不喜欢凝结尾迹，只是原因完全不同而已。所谓凝结尾迹——如果你曾经抬头见过喷气式飞机后方形成的那些白色云带，它们就被称为凝结水汽尾迹，或者简称为凝结尾迹。事实证明，至少根据现有的科学估算，全球所有由人类活动引起的全球变暖中，大约有 1% 是由飞机凝结尾迹造成的。为什么会这样呢？我可以谈谈这背后的物理原理。
事实证明，这里实际上存在两种截然相反、相互抵消的效应。凝结尾迹有时——如果你观察过的话，它们划过天空，随后很快就消散了。那种短暂消散的尾迹其实不会产生什么实际影响。但有时它们会在空中持续停留很长时间。你会在天空中看到几乎像华夫饼格子网状交织的景象，这些被称为“持久性凝结尾迹”（persistent contrails）。
它们会产生两种效应。由于它们是轻薄的白云，因此它们会反射太阳光；但显然，这只发生在白天。而事实表明，所有物体都会向外发射所谓的黑体辐射（black body radiation）。地球在约 300 开尔文（Kelvin）的环境温度下发射辐射，这种辐射处于大约 10 微米波段的远红外区。而在这些波长波段下，凝结尾迹的反照率（albedo）非常低，它们在物理上几乎就等同于完全吸收的黑体。因此，它们会吸收一部分向外太空散发的地表红外辐射，然后同时向两个方向（地表与太空）重新辐射。本质上，它们把地球散发出的部分热量重新阻挡并反射回地表。因此，它就像一层厚厚的毛毯一样将热量牢牢困住。
Original English
**Speaker B**: [laughter] Yes. Although you should also dislike contrails, but maybe not for the same reason. So, contrails are—if you've ever seen those white clouds formed behind jets, those are called condensation trails or contrails. And it turns out they add at least according to the estimates that people have, about 1% of all anthropogenic global warming is caused by contrails. Why is that? I can just talk about the maybe the physics of that.
So, it turns out that there's actually two countervailing effects. Contrails are—well, sometimes if you've ever seen them they they streak and then they kind of go away. Those don't really do anything, but sometimes they last for a long time. You'll just see in the sky just like almost like a waffle of just persistent contrails they're called. And there's two effects that they have. Those are thin white clouds. So they reflect sunlight, but that only of course happens during the day. It turns out all objects emit something called black body radiation. And the Earth does at whatever the temperature is, about 300 Kelvin. It's in the far infrared around 10 microns. And at those wavelengths, contrails have very low albedo. They're almost essentially black. And so they'll absorb a little bit of the outgoing infrared radiation and then remit it both directions. So essentially they'll reflect some of the outgoing heat. So it'll trap heat like a blanket.
### 冰过饱和区与飞行路径规避
**Speaker B**: 而且因为这个保温过程是全天 24 小时不间断发生的，[清嗓子] 所以整体上它们倾向于产生净变暖效应。事实表明，这带来了一个惊人的数字——尽管目前仍然存在一定的不确定性，但是由凝结尾迹转化演变而成的凝结尾迹卷云（contrail cirrus），可能会在天空中覆盖相当可观的面积；尤其是在欧洲这样空中航线极为密集的地区，实际的天空中有几个百分点的比例都被额外产生的凝结尾迹所覆盖。因此，在欧洲这些地区，仅在局部它就增加了大约每平方米 1 瓦特（1 W/m²）的辐射强迫（radiative forcing）。这意味着什么？给你一个对比的直观概念：由人类全部活动所造成的全球平均增温辐射强迫，大约也就是每平方米 3 瓦特。所以在飞机交通密度极高的局部区域，它所产生的局部增温效应是极其显著的。
那么，面对这种情况我们能做些什么呢？事实证明，凝结尾迹是由大气层中处于“冰过饱和”（ice super-saturated）状态的区域所引发的。这些区域的物理特性有点像制作冰糖的过程。比如在做冰糖时，你会配制出一杯糖分过饱和的水溶液。此时只要往里面落入一丁点微小的糖晶体，所有的过饱和糖分就会瞬间迅速结晶析出。与此完全类似，在凝结尾迹形成的环境中存在着这样的特殊区域：它们往往呈扁平的薄煎饼状，纵向厚度通常只有几百米。如果喷气式客机恰好穿行飞过这片区域，发动机排出的喷气尾气中含有微量的水分，这些水分就会迅速凝结成微小液滴并冻结为冰晶。在这样恶劣的过饱和区域里，飞机排气每释放出 1 克的液态水、冰或烟尘颗粒，就会从周围空气中直接诱导析出并吸附大约 10 千克的水分！也就是说，这里存在着高达 10,000 比 1 的极其惊人的结晶放大倍率。
因此，这是一个非常巨大的气候问题。那么你能够采取的应对措施是：你可以设法找出这些冰过饱和区域的具体分布位置——当然，肉眼是完全看不见这些区域的——然后通知并引导飞机从这些区域的下方穿行规避。而飞机在操作上基本上只需要下降两个所谓的飞行高度层（flight levels）。因此，采取航线规避来绕开这些不利区域，实际上只会额外消耗极少量的航空燃油。
为此，我们构建了一套系统，专门分析卫星遥感图像，并尝试检测出凝结尾迹究竟在哪里产生。这样我们就拥有了一个基本上全天候连续运行的实时监测系统。随后，我们尝试构建预测模型，因为现有的常规气象预报模型精度不足，无法准确预测这些冰过饱和区的位置。所以我们构建了一个专用的定制模型——同样是类似于卷积神经网络（CNN）或者 UNET 架构的模型——来预测它们究竟会在何时何地出现。这样我们就可以将生成的动态地图提供给航线规划软件，使航空公司能够合理规避这些区域，从而以极低的经济成本大幅削减航空业对全球气候变化带来的负面影响。
Original English
**Speaker B**: Um and so because that happens 24 hours a day, [clears throat] they tend to be warming. And it turns out it's a surprising—again, there's some uncertainty about it, but you know, contrail cirrus—cirrus that sort of comes from contrails—might cover, especially in places like Europe which has a lot of airline traffic, a few percent of the actual sky is covered by sort of additional contrails. So it adds, in those places like Europe, it adds about one watt per square meter of forcing locally at least. Which means that—just to give you a sense, all of anthropogenic warming sort of averaged across the whole globe is about three watts per square meter. So in places of high airplane traffic, it can be a lot of warming locally.
So what can you do? Well, it turns out contrails are caused by areas in the atmosphere that are ice super-saturated. They're a little bit like rock candy. So like when you have rock candy, you get a water solution that has too much sugar in it. And any little, you know, little bit of sugar in it will just crystallize all the sugar out. Just like in this contrail, there these regions—they tend to be kind of pancake shaped, only a few hundred meters tall. And if you fly through it, the jet exhaust has a little bit of moisture in it which will turn into droplets and then freeze. And then for every gram, if you're in this bad region, for every gram of water, ice or soot you put out, it's about 10 kg of water gets sucked out. So there's this enormous 10,000-to-1 curing ratio.
So it's a big problem. So what you can do is you can figure out where these regions—they're invisible, of course—these regions of ice saturated are, and then tell the plane to go underneath, and you only have to drop essentially what they call two flight levels. So it actually costs a little bit of fuel, but not very much to kind of avoid these sort of bad regions. So we built a system that sort of looks at satellite images and tries to detect where contrails are. So we have essentially a continuous monitoring system and then try to build a model—because it turns out the weather models are not quite accurate enough to find these places of ice super-saturation. So we build a custom model, again like a convolutional net or a UNET or something, essentially to predict where they're going to happen, so that then we give maps to a flight planning software so they can dodge it, and inexpensively reduce the climate impact of aviation by a lot.
**Speaker A**: 能够对它进行有效预测，这背后的物理原理是什么？对吧，这纯粹是因为我从卫星图像上看到了它，所以认为明天它还会出现在那里，毕竟飞机的航线目的地是一样的？还是有别的机制……
Original English
**Speaker A**: What's the physics behind why you can predict that? Right. Is it just I see it in the satellite and then tomorrow I think it'll be there because planes go to the same place, or...
**Speaker B**: 哦，并不是这样。那是因为你真正尝试检测和预测的是这些冰过饱和区域。
Original English
**Speaker B**: Oh, no. It's because you're trying to detect these regions of ice super-saturation.
### 飞机尾迹的持续性与成因机理
**Guest**: ……因为它们具有极强的持久性。从本质上来说，它们……
Original English
**Guest**: ...because they're very, very persistent. Essentially they're...
**Host**: 噢，它们非常持久。
Original English
**Host**: Oh, they're persistent.
**Guest**: 噢，是的，确实如此。虽然目前还没有人能给出完全确切的定论，但它们确实可以在空中持续存在好几天。从机理上讲，目前学界普遍认为这是由于温暖湿润的空气被注入到了对流层顶的交界处——也就是平流层的底部边缘。一旦水汽和湿度上升到那个高度，就会在原地滞留很长时间，然后才极其缓慢地逐渐消散。
Original English
**Guest**: Oh, yeah. Yeah. I mean, no one knows exactly, but they could last for days. Essentially, they're caused by—they think—sort of warm moist air being injected just at the boundary of the tropopause, between the... just at the bottom of the stratosphere. And then when humidity gets up there, it sort of sticks there for a long time and then gradually dissipates.
**Host**: 明白了。所以说，那里就像是大气层中存在某种……
Original English
**Host**: Got it. So it's... there's just a sort of...
**Guest**: 它们就像是大气层中的不良区域，是你绝对不想驾机飞越的地方。
Original English
**Guest**: They're like bad spots in the atmosphere you don't want to fly through.
**Host**: 确实如此。那么，一旦你们确定了这些区域的位置，这个预测结果大概至少能在接下来的几天内保持有效吧？
Original English
**Host**: Right. Okay. And so once you've established that, it's probably good for a couple days at least.
**Guest**: 呃，其实你必须不断进行动态预测，实时掌握它的具体位置。
Original English
**Guest**: Uh well, you have to keep predicting where that is.
### 反事实问题与辐射效应评估
**Host**: 好的。你们目前使用的模型，之前提到过似乎是卷积神经网络（CNN）或者类似的技术架构？
Original English
**Host**: Yeah. And the models you're using, you mentioned like CNNs or something like that.
**Guest**: 没错。我们目前还没有把那些替换成 ERA 级别的先进模型。不过这里面引出了一个非常引人深思的核心问题：你其实非常想精确获知，某一条具体的飞机尾迹究竟造成了多少增温效应？它到底给全球变暖增加了多少负担？这一点至关重要，举例来说，你可能希望精准找出那些增温效应最严重的尾迹区域，因为飞机绕开这些区域是需要额外消耗燃油的，对于航空公司而言避让飞行会产生实际的经济成本。因此大家会说：“天哪，我真的很想确切了解它到底造成了多大影响。”
但这在本质上属于统计学所说的“反事实问题”（Counterfactual Problem）。也就是说：现实中你确实产生了一条凝结尾迹，并引发了一定量的红外辐射变化。如果我们观测手段足够严密，这个实际发生的值是可以测量出来的；但真正棘手的问题在于——如果当时那里根本没有产生这条尾迹，情况又会是怎样？要估算这一点极其困难，因为你根本无法跨入那个“尾迹未曾发生”的平行宇宙。
Original English
**Guest**: That's right. And we haven't replaced those with ERA-level models yet. But there was a very interesting problem that came up, which is you sort of want to know, well, just how much warming did this contrail make, and how much did it add to global warming? Because, for example, you might want to find the biggest ones, because there's some fuel cost, and maybe it costs a bit of money for the airplanes to avoid it. So you say, "Well, gee, I'd like to kind of know how much it did." But that's actually what they call a counterfactual problem, like: okay, you made a contrail, and a certain amount of infrared radiation happened. So we can measure that if you're careful, but what would have happened if there hadn't been a contrail there? That's a very difficult thing to estimate because you can't access the universe where the...
**Host**: 事情没有发生的那个宇宙。
Original English
**Host**: ...didn't happen.
### 利用反事实模型与 ERA 突破估算瓶颈
**Guest**: 没错，因此你必须构建所谓的“反事实模型”（Counterfactual Models）。我不知道你的听众是否了解，这类反事实模型在实际拟合与构建时是非常复杂棘手的。大家应该还记得，这里涉及两种关键辐射机制：一种是对太阳光的反射作用，另一种则是长波红外辐射的吸收与截留。事实证明，准确测量尾迹反射太阳光的净效应反而要困难得多，我们团队甚至在这个难题上足足卡了两年之久。
当时我们手头已经有一个表现尚可的反事实模型，用来估算射出长波辐射（Outgoing Longwave Radiation），但在应对反射阳光的模型构建上却屡屡碰壁。后来 ERA 实际上帮了大忙，它协助我们探索并找到了一种能够全面检索所有混杂因素（Confounders）的模型结构，从而理清了究竟该如何对其进行有效估算。因为我们甚至在人造数据集上编写了严格的基准测试代码——在合成数据中你可以人为注入模拟的飞机尾迹，由于这些尾迹是我们主动注入的，所以我们完全掌握其真实的物理辐射效应基准。在此之前，我们自己尝试构建的各种模型甚至连我们自己的测试套件都通不过，但 ERA 生成的模型却成功通过了验证，彻底帮我们解开了这个僵局。
所以，我们目前正在撰写相关的学术论文。关于射出长波辐射的研究我们已经有一篇成果，而关于这个难题的突破性解法，虽然论文尚未正式投稿，但我们此前已经在欧洲地球科学联盟大会（EGU）上对其进行了专题报告。
Original English
**Guest**: So you have to make these things called counterfactual models. And those are—I don't know if your listeners know—counterfactual models are actually pretty tricky to fit and make. And remember there were two... the reflecting of the sunlight, and then there's the infrared. It turns out measuring what the effect of reflecting sunlight is is actually more difficult, and we were actually stuck on it for two years. We had a model that worked okay, a counterfactual model for the outgoing longwave radiation, but not for the reflected sunlight. ERA actually helped us find a model that sort of searched all the confounders and sort of figured out like, "Oh, how can we estimate it?" Because, again, we even had test code on sort of artificial datasets, where there's sort of injected contrails and we sort of figure out, "Oh, well, we know how much it was because we injected it." And so, again, our own attempts didn't even pass our own tests, but ERA—this thing actually did, and sort of unstuck this problem.
So, yeah, we're in the middle of writing up a paper. We have a paper about the outgoing long-range radiation, but we have a paper that's not submitted yet, but we've talked about it at EGU, I think, where we actually solve this problem.
**Host**: 那么 ERA 最终给出的这些模型，结构上究竟是一套极其庞大冗杂的代码怪物，还是说其实相当精简优雅，只是此前需要特定维度的物理洞察力才能构建出来？
Original English
**Host**: And the models that ERA comes up with, are they just like a big monstrosity of code, or are they like pretty basic and it's just you needed the intuition to develop that?
**Guest**: 是的，就这个具体案例而言，它其实完全属于后者。本质上，它帮助我们识别出了核心的特征逻辑——那其实是一个结构非常精炼的模型，引入了一定数量的关键混杂变量，只是此前我们从未尝试过那样的变量组合方式，而最终测试表明它的效果极其出色。所以说，它是从全局中提炼出了最优解，事后回溯审视时显得十分清晰合理。我认为这绝对是一次巨大的胜利。
Original English
**Guest**: Yes, it's actually in this particular case it was actually more of the latter, that it essentially sort of helped identify what the... it was a very simple model with some number of confounders that we just hadn't tried that combination before and it worked very, very well. So yeah, it actually sort of came up with the... and it was seen in retrospect. So that was, I think, a big win.
### 生物圈碳通量与未来的巨大不确定性
**Host**: 这确实非常引人入胜。我知道你在气候科学领域做了大量深入的研究工作，除了这项课题，你还开展过哪些方面的探索？
Original English
**Host**: Yeah, that's interesting. I know you've done a lot of work in climate. What other stuff have you done?
**Guest**: 我想我之前应该提到过二氧化碳（CO2）通量估算的研究吧。那同样是一个极具趣味性的方向，因为直到今天它依然是一个充满未知的重大科学难题。评估大气中二氧化碳之所以如此具有挑战性，根本原因在于我们至今无法准确厘清生物圈内部及其与外部环境之间的真实碳通量（Carbon Flux）。
当然，宏观层面的大框架我们是清楚的：人类向大气层排放了海量的二氧化碳，其中一部分通过无机化学平衡以及浮游植物的生物过程被海洋吸收，还有很大一部分则被陆地生态系统吸收固存。然而，关于当下陆地吸收量具体是多少，现有的科学误差范围已经相当宽泛；而如果把时间线拉长到未来 50 年甚至 2100 年的气候模型预测，这个误差范围更是大得惊人。因为科学界目前根本无法确知，随着全球温度持续攀升以及二氧化碳浓度不断升高，全球生物圈究竟会产生怎样的非线性反馈响应。
正因如此，我们完全无法准确判定未来到底有多少二氧化碳能被自然界重新吸收。仅仅由于对吸收机制认知的不确定性，到 2100 年模型中二氧化碳浓度的误差区间就高达 300 ppm。需要特别提醒大家的是，目前全人类大气中的二氧化碳总浓度大概也就只有 440 到 450 ppm 左右！因此高达 300 ppm 的误差范围是极其悬殊且惊人的。未来的实际走势，可能只是不太理想，也可能会演变成一场触目惊心的大灾难。既然存在如此巨大的不确定性，如果我们能设法将其收敛降低，无疑将具有极其深远的科学价值。因此，我们针对二氧化碳浓度的精准建模与估算，正是朝着攻克该难题迈出的坚实一步。
Original English
**Guest**: I think I talked about this, right? The CO2 thing. That was pretty fun because it's still quite a... The reason why estimating CO2 in the atmosphere is an interesting problem is we actually don't know what the carbon flux is in and out of the biosphere. I mean, we do—we know that the biosphere captures... right, we emit a bunch of CO2 out into the atmosphere, and some of it gets absorbed into the ocean with sort of mostly inorganic chemistry, some phytoplankton, and a lot of it gets absorbed on land. But the error bars about what happens are moderately large, and the error bars 50 years from now are very large, like the models in 2100. We don't know how the biosphere will react to the ever-increasing temperatures and CO2. So we don't actually know how much the CO2 will be absorbed, and the error bars are 300 ppm of CO2 just from the uncertainty of what gets absorbed. And just to point out, you know, right now there's about, what, 440, 450 ppm. So it's huge. I mean, it could be seriously amazingly awful or not great, but you know, the 300 ppm is like enormous uncertainty. So it would be really nice to figure out, you know, can we reduce that? So the CO2 concentration is like one step towards solving that.
### 天气预报与气候模拟的本质分野
**Host**: 据我所知，谷歌在气候建模和天气预测领域同样取得了多项突破性的重大进展，对吧？我今年参加了上一届 NeurIPS 顶会，期间专门去听了气候科学专题的分享。虽然当时因为行程原因只听了两场报告，但带给我的震撼是极其强烈的——在过去大概十年左右的时间里，气候与气象建模领域所发生的跨越式剧变简直令人难以置信。我知道这其中有相当大一部分突破都诞生自谷歌。你能否详细聊一聊，在谷歌以及其他科研机构内部，究竟发生了怎样的技术演进，才促成了天气与气候建模领域如此巨大的范式转变？
Original English
**Host**: And I know that Google has made some really big improvements in climate modeling and weather prediction as well, right? I was at NeurIPS this year, this last NeurIPS, and I stopped by the climate track. And you know, I maybe only had a chance to listen to two talks or something, but it really blew my mind, the sort of step change I think that's happened in the past, I don't know what it is, maybe 10 years or whatever, in terms of climate modeling. And I know a lot of that happened at Google. Can you talk a little bit about what has happened in Google and other places that has allowed that really big transition in climate and weather modeling?
**Guest**: 好的，让我来梳理一下。公众在日常讨论中往往习惯将“气候”与“天气”混为一谈，这在直觉上很自然，因为它们在底层遵循的大气物理学定律是一致的。
Original English
**Guest**: Okay, so let me... People often sort of collapse climate and weather together. Well, because they're fundamentally the same physics.
**Host**: 确实是这样。
Original English
**Host**: Yeah.
**Guest**: 但必须说明的是，两者的物理一致性主要局限于纯粹的大气动力学范畴。一旦你的研究视野拓展到冰川冻土和陆地演变，两者的差别就显现出来了——简而言之，“气候”本质上是长周期维度下的统计天气表现。正因如此，一个完整的地球系统模型（Earth System Model，即真正的全要素气候模型）的内部复杂度，要远远超出单纯的大气环流模型。在气候模型中，你必须极其细致地计算水汽与二氧化碳在陆地与大气之间的双向通量交换，必须精准模拟极地冰盖与海冰的消长动态。
而在过去几年中真正迎来跨越式质变的，首先是天气预测模型。这里我想要把概念界定清晰：所谓天气预报，针对的通常是未来大约 15 天以内的短期大气演变。之所以界限在两周左右，是因为我们的大气系统本身是一个高度典型的混沌系统。
Original English
**Guest**: Although at least for the atmospheric physics, they're... Obviously, when you start having ice and land, you know, climate is long-term weather. And so the complexity of a full Earth system model, which is a climate model, is much, much bigger than an atmospheric model. Like you have to actually measure what's the water flux and the CO2 flux in and out of the land, or what will happen with ice. And so there has been a step change with weather models. Right... Sorry, I want to make this... So weather is up to 15 days, okay, approximately, because, you know, weather itself or the atmosphere appears to be chaotic.
### 混沌效应、机器学习与热带气旋路径预测
**Guest**: 呃，我不知道是否需要展开解释混沌的概念。简而言之，它就是著名的“蝴蝶效应”：微小如一只蝴蝶振翅所带来的微弱扰动，就足以导致两到三周之后全球的大气状态发生彻底翻天覆地的改变。因此，传统天气预报的核心任务，就是在约两周的时间窗口内，尽可能准确地预测大气状态的具体演变轨迹。
而这正是近年来发生翻天覆地重大变革的领域。这种飞跃在很大程度上甚至并不是依赖当下最新的大语言模型（LLM）技术，而是建立在 2018 年前后成熟起来的高级机器学习方法基础之上，再结合极其庞大的历史观测数据储备以及前所未有的海量算力支撑。在这期间，学界与业界开展了大量极富智慧的算法创新，其中谷歌贡献了极为突出的核心成果，接连推出了新一代机器学习天气预报模型，实际效果非常惊艳。
事实上，我们在该领域取得了一项非常亮眼的实际突破：如今的新模型能够以远超传统动力学模型的精度，提前多天高精度预测热带气旋（台风与飓风）的移动轨迹。例如不久前牙买加遭遇了一场极其罕见的毁灭性飓风袭击，当时许多经典的数值气象预报模型甚至根本未能成功提前预警。这在很大程度上是因为传统模型很难捕捉飓风极其强烈的突发增强过程，而这种强度突变几乎完全由海洋表面的水温决定——很多人可能并不清楚，飓风在热力学本质上就是一台庞大无比的热机，它持续不断地将海洋表面积聚的热能直接转化为剧烈狂暴的大气机械运动。
所以在短期天气预报领域，数据驱动模型已经取得了巨大的成功。然而，长期的气候建模就要困难棘手得多。因为在气候研究中，你关心的根本不是 2070 年某一天西雅图会不会下雨这种具体的轨迹问题，而是关注整个气候系统在宏观长周期下的平均统计特征。而让气候预测变得极端困难的核心症结在于：整个气候演变系统具有高度的“非平稳性”（Non-stationary）。这意味着系统赖以运行的物理环境参数与底层机制本身正在发生动态偏移，你面对的物理现实与过往经验……
Original English
**Guest**: Uh, I don't know if I should explain chaos—essentially it's the butterfly effect, right? That small perturbations—like a butterfly flaps its wings and the weather will be completely different in, you know, two or three weeks. So weather is trying to predict the actual trajectory of the atmosphere over, say, two weeks. And that has been a huge step change, and that's because that's been a lot of—not even the new LLM stuff—that was based on, you know, the 2018-era machine learning stuff, and just a large amount of data and a large amount of compute. So there's been a lot of sort of very clever work, and a lot of it from Google, making new weather models, and it's been great.
In fact, we had a really neat breakthrough cuz now we can apparently predict tracks of cyclones, tropical cyclones, much more accurately many days in advance. And so places like Jamaica had got hammered by a terrible hurricane. And a lot of the classic models didn't actually predict it, partially because it's often that—especially the intensification—it's all being driven by what the surface temperature is, cuz hurricanes—people might not realize—are heat engines essentially. They convert sort of heat in the ocean to big atmospheric motions.
So weather has been great. Climate is much more difficult because you don't actually care about—you're not trying to predict whether it's going to rain in Seattle in 2070. You're trying to get kind of like averages. And what makes it difficult is that it's what they call non-stationary. So that the... in fact, literally the... it's like the underlying physics or the underlying... like, you know...
### 气候建模的低数据困境与蝴蝶效应
**Speaker A**: 植物的生理响应正在发生改变，冰川的行为模式也在改变，因此想要将传统的经典机器学习直接应用于真正的气候模型是极其困难的。这也正是你们之前讨论的核心所在——关于描述性模型以及多重假设检验的那些问题，在气候科学领域体现得尤为严峻。因为我们根本没有任何未来30年后的观测数据，同时我们也不可能干等上30年或50年，才去验证当初的预测究竟是正确的，还是仅仅发生了严重的过拟合。
所以无论我们采取何种研究手段，都必须保持高度的审慎，尝试将整体系统拆解、剥离出一个个相对独立的子问题。而其中最核心的挑战恰恰在于：你究竟该如何将已知的规律注入进去？如何构建一个既能对遥远未来做出长程预测，同时又能严格受限于人类已知物理科学法则约束的大型模型？这是一个极其引人入胜的科学难题。我认为这个问题至今仍未被完全攻克，但它无疑是一个极具价值的绝佳课题，因为这里面充满了太多的不确定性。我们确实非常迫切地想要确切知道，60年之后全球气候究竟会发生怎样的演化。所以这依然是一个悬而未决的现实课题。这是一个非常非常值得深入钻研的方向，但坦白讲，迄今为止人工智能尚未在这个领域带来颠覆性的变革，因为气候系统对纯数据驱动的方法表现出了极强的抵触性——归根结底，这本质上是一个“低数据”场景的问题。
Original English
**Speaker A**: plants are behaving differently and ice behaves differently, and so it's very very difficult to use sort of classical ML on sort of true climate models. And so that's why sort of the whole discussion you guys had about, you know, what we're talking about about descriptive models and multiple hypothesis testing that is incredibly severe in climate, because we have no data from 30 years out, and we don't want to wait 30 or 50 years to find out whether we were right or that we overfit.
So whatever things we do, you have to be kind of careful and try to peel off sub-problems. And the problem of exactly how do you inject, how do you build a big model that can predict into the future, but is still constrained by what we know—it's a fascinating problem. I think it's still unsolved, but it's a great problem to have because of, again, these uncertainties. We really would like to know what will happen in 60 years to the climate. So it's still a thing. It's a very very interesting problem to work on, but so far AI has not revolutionized it because it's very very resistant, again, because of this data problem. It's a low-data problem.
**Speaker B**: 那么所谓的气象“蝴蝶效应”——天气系统那本质上的混沌特性——是否也会直接冲击到长期的气候预测？抑或是因为气候的时间尺度足够宏大，从而可以被视作一个自封闭的动力系统，即使它可能在两极之间来回振荡，但在宏观尺度上整体表现得更为平稳呢？
Original English
**Speaker B**: Does the butterfly effect—the chaotic nature of weather—does that also impact climate, or is the time scale so large that you have a closed system for which, you know, maybe it's oscillating between poles or whatever, but when you look at it on that time scale, it's more stationary?
**Speaker A**: 不幸的是，它恰恰是以一种截然不同的方式呈现出“非平稳性”（non-stationary）。如果要追溯整个混沌理论的源头——虽然其他学科或许也有相关提出，但在气象学界，它最早可以追溯到一位名叫洛伦茨（Lorenz）的学者，他提出了那个极其经典的简易常微分方程动力系统模型。天气与气候的核心区别就在于：天气关注的是你当前正处于吸引子（attractor）的哪一个具体位置，
Original English
**Speaker A**: It's unfortunately non-stationary in a different way. But the original sort of whole chaos thing was—well, maybe people came up with it, but in meteorology it was back to a person named Lorenz, who had this very simple model ODE. So the difference between climate and weather is: weather is where are you on the attractor,
**Speaker B**: 而气候关注的则是吸引子本身的宏观统计特性。
Original English
**Speaker B**: and climate is about the statistics itself of the attractor.
**Speaker A**: 然而气候面临的真正难题在于：人类正在主动改变它。这意味着吸引子本身正在发生形变与位移。同时还可能存在着大家常说的“临界点”（tipping points），这意味着吸引子的几何结构可能会遭遇突变。而最棘手的地方在于，这种相变是极难甚至几乎无法提前精确预报的。
Original English
**Speaker A**: The problem with climate is that we're altering it. So the attractor itself is changing, is moving, and there could be—everyone talks about tipping points—that means that the attractor suddenly changes. And the trouble is that's very, very, very difficult to predict.
**Speaker B**: 所以甚至连吸引子的几何形态本身也会迅速发生变异。
Original English
**Speaker B**: So even the attractor's shape changes quickly,
**Speaker A**: 确实可能如此。而且麻烦还在于，当你运行数值模拟器时——
Original English
**Speaker A**: or could. And the trouble is when you run a simulator,
**Speaker B**: 你根本无法分清：当前出现的失稳崩溃，究竟是因为自己构建的数学模型存在缺陷，还是物理真实世界中客观存在的物理动力学失稳？
Original English
**Speaker B**: you don't know, like, did it go unstable because my model's not great, or is it an actual physical instability,
**Speaker A**: 这两者的边界极难分辨，想要将它们区分开来可以说是难如登天。
Original English
**Speaker A**: and it's extremely difficult to tell the difference.
### 过程模型的困局与微物理学的还原难题
**Speaker B**: 那研究人员究竟该怎么办？尤其是在我看来，不仅我们完全没有关于未来的真实数据，即便回顾过去，我们掌握的高精度历史数据其实也寥寥无几。固然人们可以通过古气候学手段，利用冰芯钻探等各种复杂技术进行间接重建，但要明白，一百年前的地表根本没有先进科学仪器密布的网络。
Original English
**Speaker B**: What do you do? Especially when—I mean, to me it strikes me that not only do you not have future data, you really don't have much past data. You can do some measurements and ice cores and lots of stuff to try to do that. But there was nobody with an instrument 100 years ago.
**Speaker A**: 没错，哪怕你手头拥有按年统计的监测数据，如果你运气足够好，在任何一个特定的观测点位上，满打满算可能也只有50个有效数据点而已。正因如此，我们无论采取什么手段，都必须受到现有科学认知边界的极其严格的物理约束。但这依然艰难万分——我只是在向你剖析当今科学家们所面临的两难绝境。
科研人员不得不转而构建所谓的“过程模型”（process models）——事实上整个应用科学领域都在使用这种范式，而我认为气候学则是这种思路走得最极致的典型。这类模型的具体运作方式我亲自审查过代码：大家普遍采取一种还原主义（reductionist）的路径，试图将极其错综复杂的气候巨系统拆解细分为上千个细小的独立环节；接着，你去找专门发表过针对第763号细分环节学术论文的专家，而那位专家当年正是通过对某种局部试验数据拟合了一条三次曲线来建立参数化方案的。
举个非常典型的例子：关于飞机尾迹（contrails），有一个至今依然充满未知的物理现象，那就是云层中的冰晶微观行为究竟是怎样的？正如我们之前所言，任何科学问题一旦你往深层挖掘，就会发现其底层机制异常复杂（笑）。事实表明，当飞机在巡航高度产生凝结尾迹时，这段尾迹到底能在大气中持续留存多久？这取决于诸多非线性物理过程：因为凝结尾迹要消散蒸发，正如我提到的，冰晶会首先凝结聚集并逐渐长大，随后由于自重开始向下沉降。然而，这些冰晶向下沉降的速度完全取决于晶体的微观形态与几何构型，而这在目前是根本无法被精确测定的；与此同时，凝结尾迹内部潮湿空气向外与周围干空气相互湍流掺混的扩散程度又是多少？科学家们只能依赖经验近似模型，但无人真正清楚其精确物理机制。
因此，多重环节的认知不确定性会层层叠加、严重混杂在一起。这绝非单纯一句“哎呀，只有约翰才会在意飞机尾迹”那么轻描淡写，而是因为事实证明，冰晶的实际微观物理过程（microphysics），对宏观全球气候模型的整体演化轨迹有着极其重大的决定性影响。而对此，我们目前在科学上确实知之甚少。
所以我想表达的是，这是一个极其盘根错节、荆棘密布的复杂难题，它绝非一个已被解决的陈旧问题。我个人真正寄予厚望的，是未来新一代的智能工具——也许并非今天的这代大模型，而是明天的AI范式。因为正如我刚才所说，大模型不仅能够拟合数值数据，它更具备通读和理解海量学术文献的能力。关键在于，它阅读和消化的论文量级远远超出任何人类个体的极限；也许，未来我们能够让它把人类历经严苛同行评议所积累的全部科学知识与观测数据全面融会贯通，这要远远胜过单个人类程序员徒手编写一串代码去拟合局部数据的做法。如果真能做到那一步，那将是无与伦比的科学壮举。今天的AI确实还做不到这一点，但这正是我对未来更具革命性的科学工具所抱有的殷切期许：一种能够真正以极具物理严谨性、甚至更加理性稳健的方式编写科学代码的系统，因为它是在人类迄今为止所积淀的全部科学知识框架的强约束下运转的。那将是无比令人惊叹的未来，尽管我们今天尚不具备这样的能力。
Original English
**Speaker A**: So if you have annual data or whatever, maybe you have, if you're lucky, 50 data points in any one location, right? So whatever we do has to be very constrained by what we know. But it's just very diff—I'm just telling you sort of the horns of the dilemma people are on. People make these—in fact, people in applied science in general, I would say climate is the most extreme—make these things called process models where what you do, and I've seen the code: "Oh well, you know, I'm going to be reductionist and I'm going to sort of take the horrible complicated climate thing and sort of boil it down to a thousand pieces. And then I'm going to, you know, find the expert who wrote a paper about, you know, piece number 763 and he fit a cubic to some data."
Like, for example, one thing that's very mysterious which is related to contrails is how does ice behave in clouds? It turns out—you might again, everything is complicated once you dig into it. [laughter] But it turns out that, like when you make a contrail, how long does it last? Well, it depends on—because the way contrails can evaporate is ice starts to accumulate, as I said, and then the ice crystals get big and then they fall. But of course, how quickly they fall depends on their shape, which is not known. And how much does the contrail mix from the moist inside the contrail out? Again, people have approximations, but they don't know. And so the uncertainty is very much compounded. And it's not just, oh, John who cares about contrails. It turns out that the actual physics of—microphysics of ice has very strong implications about what climate models do. And it's sort of we just don't know.
So I'm trying to say it's very gnarly and it's not a solved problem. My hope is that with tools—maybe not like today's era, but maybe tomorrow's era—because remember, as I was saying before, it not only can fit data, it can read papers, right? And the question is, it can read a lot more papers than we can. So maybe, maybe we can integrate all the data or all the knowledge that people have carefully evaluated, much more than any one person writing a piece of code and fit data. I mean, that would be—that would be utterly glorious. We don't have that today, but that's sort of one of the hopes that I have even for a more amazing tool in the future is something that really can write code in a sane way, even much more sane because it'll be constrained by all the scientific knowledge that we've accumulated so far. That would be amazing. We don't have that today.
**Speaker B**: 这给我的感觉，非常类似于生物学面临的复杂局面。
Original English
**Speaker B**: I mean, that strikes me as being very similar to biology.
**Speaker A**: 噢，太对了！天哪（笑）。确实如此。只要你接触过生物学，哪怕只是稍微翻看了解过生物学文献，你就会发现生命系统中存在着无穷无尽的特例，以及演化过程中东拼西凑出来的权宜补丁机制。是的，千真万确。因此，如果我们真能拥有一个能够将人类已知的全部科学知识与观测数据深度统一融合、并借此合成归纳出全新物理模型与科学洞见的智能系统，那绝对是革命性的突破。
Original English
**Speaker A**: Oh, yes. Oh, boy. [laughter] Right. If you've ever played with biology or even looked at biology, there's so many exceptions and so many hacks in the biological systems. Yes. Oh boy. So, yeah. It would be amazing if we could have a thing that could really integrate all known scientific knowledge with data and try to synthesize sort of new models and new things.
### 从零碎拼图到全局综合：AI与搜索的智能进路
**Speaker B**: 我想你所表达的核心在于，人工智能在这一领域确实能够成为关键的突破钥匙——在很大程度上是因为现有的各类科学模型实在是太过零碎拼凑了。在人类认知局限下，它们不得不被拆解得支离破碎。因此，为了把这幅错综复杂的拼图严丝合缝地拼接起来——恕我使用这个拼图的比喻——单凭模型本身庞大的参数规模与惊人的处理吞吐能力，实际上就能带来决定性的助力。
Original English
**Speaker B**: I think what you're saying is that AI can be an unlock here to some extent because the models are so piecemeal. They're necessarily piecemeal. And so being able to assemble the jigsaw puzzle—not to mix metaphors, but to assemble really this jigsaw puzzle—having just scale and capacity actually helps a lot.
**Speaker A**: 完全正确。这类前沿AI所独具的核心优势在于——你看，人类学者就算再博览群书，像我自认为读过的文献已经算非常广泛了，但若让我把毕生读过的几千篇论文彼此之间进行跨维度的关联与综合分析，对我个人的大脑认知带宽而言也是不可承受之重；而大模型数以千亿计的内部参数中压缩了如此海量的常识与规律，加之你还可以直接赋予它解析阅读数百万篇PDF学术论文的外部工具接口，它便能够真正开始捕捉并串联起人类专家根本无从联想到的跨学科规律。我们在ERA系统的实验中已经开始隐约窥见这类能力的雏形与曙光。我并不是吹嘘今天的ERA已经彻底达到了那种神境，但毫无疑问，这正是我坚信并期待未来技术演进的必由之路。
Original English
**Speaker A**: That's right. The one thing that these AIs have is somehow, you know, humans—even I'm pretty well-read, I think, but it's just difficult for me to kind of integrate across the n-squared different... my 'n' is the number of papers I've read in my life, it's pretty big, and it's just difficult for me to even do that n-squared thing. But somehow there's just so much data in those billions and billions of parameters, and you can also give it access to read PDFs, that it can somehow start to pull things together that people wouldn't do. So that's again, I'm starting to see little indications of that inside of ERA. I'm not claiming that's what ERA does today. But yeah, that's sort of my hope of where this is going to go.
**Speaker B**: 我注意到业界很多人都在提出类似的论断：通往通用智能的有效路径，本质上正是将大语言模型与某种形式的高效搜索机制深度结合。所以ERA当前的实现方式其实与其有异曲同工之妙，对吧？它将超大规模数据库的高精度检索能力与强大的搜索算法协同配合，这显然是一条极其可行的路径。当然，大家现在普遍热衷的检索增强生成（RAG）也属于这个演进脉络。
而且如果仔细推敲，Google本身最早起家的那套展示“10条蓝色链接”的搜索引擎，即便早在大型语言模型横空出世之前的蛮荒年代，本质上也是一种极其强大的人工智能形态，不是吗？因为即便把你扔回2010年或2015年的互联网环境，你几乎可以向Google搜索框键入关于世间万物的任何疑难杂症，而它竟然总能神奇地给你呈现出切中要害的答案，对吧？
Original English
**Speaker B**: I think I've heard a lot of people suggest something like the route to intelligence is to combine LLMs with some form of search. So it's actually amazingly like ERA doing that, right? Something which you know is maybe a very strong database lookup with a good search algorithm is one way. And of course, I mean, people still do, I guess, the whole RAG thing, of course. Um, and if you think about it, Google itself, you know, the 10 blue links thing, it was or is a form of AI even before we had LLMs, right? Because you could cast yourself back to whatever, 2010 or 2015, you could ask Google about literally anything and it will tell you stuff, right? And so—
**Speaker A**: 事实上，它的检索效果往往好得令人难以置信。
Original English
**Speaker A**: Surprisingly well, actually.
**Speaker B**: 确实，效果好得出奇，因为互联网的某个隐秘角落里大概率早已有人探讨过这个问题了。所以只要搜索引擎能做好相关度的精准匹配——
Original English
**Speaker B**: Yeah, surprisingly well, because somebody on the internet has written about it probably. So if you can match that—
**Speaker A**: 说老实话，当年我之所以渴望加入Google工作，核心动力之一正是因为我觉得这套搜索系统实在太惊艳、太不可思议了。
Original English
**Speaker A**: In fact, that was one of the reasons why I wanted to come to Google, is just that was such an amazing thing, right?
**Speaker B**: 我很好奇我们当下的听众群体里，究竟有多少人曾经历过Google诞生之前的互联网早期搜索年代，切身体会过当年的搜索体验到底有多么糟糕（笑）。我至今清晰地记得大约是在1998年，当时我还在主力使用AltaVista搜索引擎，紧接着Google刚刚问世。我第一次试用了Google，随后我立刻把旧引擎——不好意思了，Digital Equipment Corporation——就像丢掉烫手山芋一样彻底弃用了它，从此毫不犹豫地全面转向了Google。
Original English
**Speaker B**: I wonder how much of our audience did a search pre-Google and just know how bad that experience was. [laughter] Yeah, I remember 1998 I think, when Google—like I was using AltaVista, and Google had just gotten released, and I used it and I just—sorry, Digital—I just dropped it like a hot potato or something and started immediately using Google.
### 还原论与黑箱模型：科学发现的可解释性
**Speaker A**: 不，所以从某种程度上说这也是一种人工智能的形式。因此，是的，这或许只是因为能够访问所有这些信息，并且能够同时在脑海中对它们进行统筹考虑。
Original English
**Speaker A**: No, it's so that is sort of a form of AI and so yes it might be uh yeah it could be that just having access to all of that and sort of keeping it in mind at the same time.
**Speaker B**: 这种科学发现的模式，只要它最终行得通，其实也是挺让人安心的，因为它是还原论的，你可以去审视每一个独立的组成部分并理解它们。也就是说，它找到了完全正确的组装方式，但这些构件在本质上可能原本就是人类已经发明出来的东西，或者是它在人类成果的基础上进行了迭代。因此，所有这些微小的碎片在个体层面上都是可以被人类理解的，然后你也可以将它们拼凑成一个连贯的全局图景。我认为对于生物学或气候科学等许多领域的问题而言，除非最终给出的答案呈现出这种形式，否则我们恐怕不会完全信任它。
Original English
**Speaker B**: That model of sort of scientific discovery in as much as it pans out is kind of comforting too because it is reductionist so that you can look at the individual parts and understand them. So it found the exact things to assemble, but they're all actually maybe fundamentally things that people have invented or it's done iterations on, and so all those little pieces are individually understandable and then you can also put them together into a coherent picture. I think for a lot of problems like biology or climate science, I don't think we would trust the answer unless it was in that shape.
**Speaker A**: 是的。
Original English
**Speaker A**: Yeah.
**Speaker B**: 因为如果有一个庞大的黑箱模型跳出来说：“瞧，细胞就是这样运作的。”大家就会想：我该相信它吗？我的意思是，我都不知道自己能不能信它，因为我根本无法对其进行内部机制的审查。所以……
Original English
**Speaker B**: Because if there was some giant blackbox model that said, "Oh, this is how a cell works." It's like, do I believe it? I mean, I don't know if I believe it because I can't examine it. So,
**Speaker A**: 不过要是反驳这种观点的话，如果这个模型的效果确实非常非常好……
Original English
**Speaker A**: But I mean to argue against it though, like if it works really well,
**Speaker B**: 但你必须收集足够的数据，你显然必须进行极其严格的测试。然而对于统计模型而言，它始终面临一个根本问题：它必须具备外推能力。
Original English
**Speaker B**: But you'd have to gather—you'd have to, I mean, you have to test it obviously, but a statistical model, again it has to extrapolate.
**Speaker A**: 是的。
Original English
**Speaker A**: Yeah.
**Speaker B**: 是的，它必须能够外推到极端情况，或者说应对那些黑天鹅事件。
Original English
**Speaker B**: Yeah. And it has to extrapolate to the extreme or the sort of the black swan events.
**Speaker A**: 确实如此。所以这非常困难。这也是为什么像自动驾驶汽车这样的技术会是一个极其棘手的难题，因为它全都是边缘极端情况（Corner Cases）。
Original English
**Speaker A**: That's right. And so it's very hard. This is why things like self-driving cars are a very difficult problem. It's all corner cases.
**Speaker B**: 确实。
Original English
**Speaker B**: Yeah,
**Speaker A**: 不过客观来说，它们目前能做到这种程度已经相当惊人了。
Original English
**Speaker A**: It's kind of amazing how well they've done.
### 从物理第一性原理到数据驱动：科学范式的转变
**Speaker B**: 思考这一点非常有意思，特别是当你来自物理学界时——在物理学的传统中，一个模型通常就是一个唯一定义系统的方程或者少数几个方程，它涵盖了系统的方方面面，你只需要解出这个方程组，你就掌握了一切所需的信息。我认为像 AlphaFold 这样的突破对很多人来说是一次认知范式的转变。在这之前，人们普遍认为蛋白质折叠是一个“只要我们找到合适的力场并拥有足够强大的计算引擎，就一定能彻底解决”的物理问题。
Original English
**Speaker B**: That's an interesting point thinking about like when you know coming from the world of physics where a model was usually a single equation or a small number of equations which uniquely define a system and everything about it and you just crank you just find a solution to the system and you now know everything you need to know. And I think something like AlphaFold was kind of a shift for a lot of people where before they thought oh protein folding is you know problem where you just if we find the right force field and we have the right computational engine we will solve protein folding.
**Speaker B**: 而真正用数据驱动的方式来解决这个问题的想法，实际上在 AlphaFold 1 问世之前的几年才刚刚萌芽。我觉得非常耐人寻味的是，这种全新的人工智能建模方式在某种程度上迫使人们重新去审视并评估“究竟什么是科学”。因为 AlphaFold 以及类似的深度学习模型固然极其强大，它们作为工具开创了无尽可能，但在其核心层面上，它们往往无法像历史上物理学家所习惯和追求的那样，为人类提供清晰透彻的物理直觉。所以我想这正好印证了那句名言：“所有模型都是错的，但有些是有用的。”
Original English
**Speaker B**: And the thought of even really solving it in a datadriven way was only appeared a few years before AlphaFold 1 came out. And it's sort of forced the new AI modeling has I think forced people to re-evaluate almost what is science because AlphaFold is and similar models are incredibly powerful. There's a lot of things that they've opened up as tools, but at their core they often times don't give intuition in nearly the same way that most physicists historically would have wanted. And so I guess it's sort of there's this old saying: all models are wrong, some are useful. Yes,
**Speaker A**: 乔治·博克斯（George Box）说的。
Original English
**Speaker A**: Box said that.
**Speaker B**: 是的，没错。那你认为在什么情况下数据驱动的模型就完全足够了，而在什么情况下我们又必须追求人类能够真正理解的可解释模型？
Original English
**Speaker B**: Yeah. Yeah. When do you find the datadriven models to be sufficient and when do you want sort of like something which is interpretable that humans can actually understand?
### 数据丰富度与封闭系统：天气、蛋白质与气候的本质区别
**Speaker A**: 我觉得这归根结底类似于天气与气候之间的区别。如果你处于一个数据极其丰富的机制体系中……
Original English
**Speaker A**: I think it boils down to almost like the difference between weather and climate. If you're in a data-rich regime...
**Speaker B**: [清了清嗓子]
Original English
**Speaker B**: [clears throat and snorts]
**Speaker A**: 比如天气预报，甚至包括蛋白质结构预测——因为有蛋白质数据库（PDB）的存在，你会感到“是的，我掌握了足够的数据来覆盖整个状态空间”。因此像 AlphaFold 这样的统计模型就能发挥绝佳的效果。事实上很多人都在非常愉快地使用它。虽然我不是生物化学家，但据我了解它确实彻底革新了这个领域，大家似乎都对它赞不绝口。而且 DeepMind 做出的一项令人惊叹的壮举，就是详尽穷尽地在整个 PDB 数据集上运行了该模型并将结果公开发布，这实在太酷了。
Original English
**Speaker A**: Um like weather or even proteins because of PDB you can feel oh yes in other words I've got enough data to kind of cover and so a statistical model like AlphaFold does. And so in fact a lot of people happily use it. I think it's really revolutionized—my understanding, I'm not a biochemist, but people seem to love it. And one amazing thing they did is they exhaustively just ran it on all PDB and published it, which is just really really cool.
**Speaker B**: 包含了大约六十亿个蛋白质预测结构，其中绝大多数的准确度都相当高。
Original English
**Speaker B**: It's like six billion protein producers. So the vast majority are actually quite accurate. Yeah.
**Speaker A**: 是的，所以那真的非常惊人。但从某种意义上说，它感觉是一个“封闭”的问题，你明白我的意思吧。
Original English
**Speaker A**: Yeah. So that's just amazing. So but it feels closed if you know what I mean.
**Speaker A**: 但是当面对气候这种系统时，它是一个开放的、非平稳的（Non-stationary）系统，或者你需要做极其大跨度的外推，这时你就必须更加审慎。在生物学中也是如此，生物学里某些细分领域可能也是这样。比如之前 Ark Institute 举办的那个“虚拟细胞挑战赛”（Virtual Cell Challenge）……
Original English
**Speaker A**: But when it's like climate and it's open and it's non-stationary or you have to make these big extrapolations, you have to be much more cautious. Or maybe biology again, and there's probably there may be parts of biology like oh there was this virtual cell challenge from the Ark Institute...
**Speaker B**: 那个比赛得出了一个很有意思、甚至有点讽刺的结果。
Original English
**Speaker B**: That had a funny result.
**Speaker A**: 据我所知，人们发现复杂的深度学习模型可能出现了过拟合……
Original English
**Speaker A**: I know that people were—there may have been some overfitting for at least that would say more, sorry.
**Speaker B**: 或者从高层次的角度来看，那些简单的基准模型（Baselines）其实……
Original English
**Speaker B**: Or um I guess maybe at a high level I think the simple baselines for...
**Speaker A**: 表现得非常非常出色，确实如此。
Original English
**Speaker A**: Work very very well things, yeah yeah just like...
**Speaker B**: 在生物学研究中，一条经典的守则就是永远先从最简单的基准模型开始。这大概也是通用机器学习实践中的好习惯——先充分理解你的最简单基准情况。而在生物学里，有许多问题即便你拥有海量数据，它们也极难被超越简单基准的模型攻破。
Original English
**Speaker B**: One of the classic things whenever you do biology is just always start with a simple baseline. Maybe this is probably just good ML in general—understand your simplest case. And in biology there are many problems where they're extremely resistant to anything beyond the simple baseline even if you have a lot of data.
**Speaker A**: 一点也没错。事实上我也一直对别人这么说：永远先拟合一个线性回归模型，直接先上它……
Original English
**Speaker A**: That's right and in fact I tell people the same thing. I said always just fit linear regression, just fit...
**Speaker B**: 直接做。
Original English
**Speaker B**: Just do it.
**Speaker A**: 直接做，先跑线性模型。
Original English
**Speaker A**: Just do it, just do linear...
**Speaker B**: 或者是支持向量机（SVM）。我的意思是，SVM 本质上也只是一种变体……
Original English
**Speaker B**: Or SVMs. I mean SVMs are just a different uh...
**Speaker A**: 另一种形式的……
Original English
**Speaker A**: Different form of...
**Speaker B**: 线性回归的扩展形式，是的。那么到底何时才真正需要采用那种更注重物理与化学底层机制的过程模型（Process Models）呢？我认为关键在于……
Original English
**Speaker B**: Linear regression, yes. So when do you need the more process modely thing? I think it's just when you...
**Speaker A**: 在于你面对的问题处于什么位置。气候问题可能处在光谱的这一端，而天气预报大概在另一端——尽管这种对比可能稍显绝对。但本质上取决于你在“数据丰富度”这个光谱上的具体位置：你什么时候能有把握地认为“这是一个封闭的问题，而且我的数据已经充分覆盖了所有的可能边界”？
Original English
**Speaker A**: When you have—I mean sort of climate is on one end and I don't know weather maybe on the other end, maybe that may be too extreme, but I think just where are you on the data richness thing? When can you feel like, "Oh no, I really have a closed problem and I think I can actually cover it"?
**Speaker B**: 没错，一个数据能够全面覆盖的封闭问题。以我的经验来看，这个判断逻辑非常合理。
Original English
**Speaker B**: Yeah, closed problem that your data fully covers. I think that that makes a lot of sense in what I've seen as well.
### 气候干预与尾迹云消除：局部收益与不对称物理效应
**Speaker B**: 我还有一个非常好奇的问题：当你在做气候建模时，你刚才提到了飞机尾迹云（Contrails），也提到了二氧化碳（CO2）预测。那么从宏观角度来看，你们到底试图实现什么目标？我猜其中一个目标是制定干预措施，另一个可能是为保险公司等机构提供预测支持，或者帮助人类适应某种气候变化。对你个人或对整个气候科学研究界而言，最核心的目标到底是什么？
Original English
**Speaker B**: Yeah I think one thing I'm kind of curious about is when you're working on climate modeling, what are the—you talked about contrails, you've talked about I guess CO2 predictions, what are the broad things you're trying to accomplish? So, one of them is I guess making interventions and the other one might be making predictions for things like insurance or like how do you help adjust for some sort of climate change. What are the principal goals I guess for you specifically or the community at large?
**Speaker A**: 我想和任何研究社区一样，不同的人有许多不同的目标。对我以及我的团队而言，我们对“气候干预”（Interventions）极其感兴趣，尤其是评估哪些干预手段在相对成本上是实际可行的。飞机尾迹云治理就是一个非常不可思议的例子，因为事实证明它的干预成本非常低。而且尾迹云还有一个非常奇妙的特点：与二氧化碳这类在全球均匀混合的气体不同，尾迹云的影响具有高度的局部性。因此，如果一个国家决定消除其上空的尾迹云，虽然它具有全球层面的外溢效应，但它主要改善的是该国本土上空的气候，所以各国非常乐意推动这项举措。
Original English
**Speaker A**: I think just like any community there's probably many different goals. For me or and my team we're very very interested in interventions. So like which ones are possible at relative cost. I mean contrails was kind of amazing because it turns out that the intervention is quite low cost and also one amazing thing about contrails is they are local unlike things like CO2. So if a country decides to fix contrails over itself it actually improves its—I mean it has global effects but it mostly improves the climate a little bit over themselves. So they like that.
**Speaker B**: 不过我猜想，如果你处在一个寒冷的气候区，而你正好希望让气候变暖，这难道不会变成一种可以自行操纵的手段吗……
Original English
**Speaker B**: I guess though if you are in a cold climate and you want to warm it up this is now your own—you could own...
**Speaker A**: 事实证明这种效应在物理上是轻微不对称的。尾迹云带来的温室变暖效应在本质上是持续且全球普遍的，而冷却效应只有在太阳处于合适入射角度（白昼反射阳光）时才会发生。因此，很少有哪种尾迹云的净效应会在不确定性置信区间之外明确呈现为冷却。在绝大多数情况下，尤其是在夜间产生的绝大部分尾迹云，其增温效应极其显著，我们在两个标准差（2-sigma）的统计显著性水平上对此非常笃定。基本上极少有尾迹云能让你确凿无疑地断定“它百分之百能产生降温效果，因此我们希望产生更多这类尾迹云”。大概只有在极地地区的极昼夏季，你才能确信尾迹云起到的是净冷却作用，进而如果在那里消除尾迹云反而会导致变暖。然而南极洲上空基本没有商业航班，夏季高纬度极区上空的航班也非常稀少，所以……
Original English
**Speaker A**: Yeah. It turns out that it's a little bit asymmetric. The warming is constant essentially and global. The cooling only happens when the sun is at a good angle over you. So it's very rare that the uncertainty—there are contrails where our uncertainty bounds in terms of the warming... There are many many contrails, mostly at night of course, where it's largely warming and we're very sure in terms of like two sigma. There's not very many contrails where you say, "Oh I know for sure that it's cooling and I want more of them." Only over sort of the poles in polar summer do you know that the contrails are cooling and therefore if you got rid of them they would warm up, but there are essentially no flights over Antarctica and not that many over the poles in the summer. So...
**Speaker B**: 在夏季的极地上空。所以……
Original English
**Speaker B**: Poles um in the summer. So...
**Speaker B**: 所以生活在寒冷气候区的人，并不会真正为了恶意取暖而利用这一点。
Original English
**Speaker B**: So no one who lives in a cold climate is going to use this maliciously.
**Speaker A**: 可以这么说。因为他们无法确切判断具体某条尾迹云究竟是在增温还是在降温，如果贸然操作可能会适得其反。因此对于那些高度不确定的航线，我们通常直接忽略，不建议航空公司去调整航路。
Original English
**Speaker A**: Well yes, in that well they wouldn't know for sure whether it was warming or cooling and so they would do stuff. So mostly we just sort of ignore, we don't recommend that people fly those.
**Speaker B**: 你刚才还提到了一个关于经济学的非常关键的论点。我的意思是，在历史上，很多人对于某些特定的应对气候变化的干预手段曾抱有极大的抵触情绪，但在某种程度上，市场力量现在已经完全接管了这一进程，例如……
Original English
**Speaker B**: You also brought up an interesting point about the economics. I think there was a lot of resistance historically about certain climate change interventions which have in some sense the market has just taken over like...
### 可再生能源的局限与核电池设想
**Speaker A**: 在这一点上，毫无疑问，像可再生能源和电池这样的技术，几乎在所有方面都普遍且明确地优于传统替代方案。
Original English
**Speaker A**: this point unambiguously like renewables and batteries are just almost universally unamiguously just better than alternatives
**Speaker B**: 对于非移动场景确实如此。但对于移动场景……我是说，你说得非常对。比如飞机，我们目前根本就没有任何现成的解决方案。
Original English
**Speaker B**: for for for non-mobile I mean you mobile I mean that's a really good point yeah yeah like planes we we do not have a solution to
**Speaker A**: 没错。我的意思是，现在确实有一些电池驱动的飞机，但它们的体积非常小，而且飞行航程极其有限。
Original English
**Speaker A**: correct I mean there are some battery powered planes, but they're very small and have to go very limited range.
**Speaker B**: 它们大概永远也无法真正达到……
Original English
**Speaker B**: They probably will never actually be uh
**Speaker A**: 很难想象那样的物理条件能行得通。除非我们能研发出像核电池之类的东西，那会非常惊人，但我不知道……我们目前根本不知道该怎么造出这种东西。
Original English
**Speaker A**: it's hard to imagine the physics would be very very unless we came up with something like nuclear batteries would be kind of amazing but I don't we don't know how to do that
**Speaker B**: 哪怕我们真造出来了，我认为公众也会非常担忧，大家会极度害怕核电池发生故障或者出事故之类的风险。
Original English
**Speaker B**: or even if we did I think the the risk of like people would be too afraid of a nuclear battery going wrong or something.
**Speaker A**: 确实是这样。既然我们现在连核电池具体是什么形态都不知道，自然也就无从确切评估其风险。我想正因为还没有这项技术，我们目前倒也不必承担这个风险。
Original English
**Speaker A**: Oh yeah. Yeah. Since we don't know what they are, we don't know what the risk. I guess we don't have the risk. Yeah.
### 气候变化的“悲伤饼图”与可控核聚变的前景
**Speaker B**: 没错，所以我们对此还一无所知。我之前做过关于气候变化的演讲，在演讲中我经常会展示一张所谓的“糟糕饼图”或者说“悲伤饼图”。
Original English
**Speaker B**: Yeah. So we don't know. So yeah, that's that's the problem with I I talk about I have given talks about climate change and I talk about the p the pie chart of badness, pie chart of sadness,
**Speaker A**: 这张饼图说明应对气候变化根本没有单一的“银弹”，对吧？因为在整个现代经济体系中，有太多不同的领域和行业都在排放温室气体。因此，我们几乎必须对所有这些领域进行脱碳改造，或者说至少要彻底解决其中的绝大部分问题。所以不存在某一种能搞定一切的单一灵丹妙药。我之前研究过受控核聚变，聚变技术非常酷，如果它的发电成本足够便宜，实际上有望一下子消灭饼图中的很大一部分温室气体排放。但我们现在还不确定，因为我们还不知道它到底能不能最终成功运转。
Original English
**Speaker A**: which is there's no one silver bullet for climate change, right? There's so many different things that contribute greenhouse gases just from across our economy. So they sort of all have to be fixed or many many of them have to be fixed. So there's no one single thing. I mean I've worked on fusion. Fusion is cool and it might actually knock a lot of them out if it's cheap enough, which we don't know cuz we don't know if it'll work yet.
**Speaker B**: 受控核聚变确实是个很有意思的领域，过去大家常开玩笑说“可控核聚变永远还需要30年”。但我觉得，现在的实际距离可能已经不到30年了。
Original English
**Speaker B**: I mean, fusion is one of those interesting things where the joke was always fusion is 30 years away, but I think it's actually now less than 30 years away, maybe.
**Speaker A**: 没错。不仅如此，我认为甚至有很大的概率，在这个十年结束之前（甚至未来三年左右），就会有人制造出具有商业相关价值的聚变堆。所以我觉得它距离实现大概只有三年了，绝不再是遥不可及的三十年。
Original English
**Speaker A**: Yeah. No, I think there's a there's a def definite probability that that someone will make uh commercially relevant fusion even by the end of this decade. So, I think it's like three years away, not not 30 years away.
**Speaker B**: 这是非常切实的进展。有趣的是，这里我想顺便宣传推广一下我们其他的几期节目，但聚变领域的很多突破归根结底其实取决于材料科学的发展。这在很大程度上是非常关键的……
Original English
**Speaker B**: This is very real. Interestingly enough, I think a lot of that um I'm going to now just stump or uh advertise some of our other episodes, but a lot of it actually comes down to material science. Um interesting enough in that
**Speaker A**: 确实是这样。当然，如果你愿意的话，我们随时可以深入探讨聚变。
Original English
**Speaker A**: Well, I'm Oh, super. Sure. Sure. Sure. Well, yes. Sorry, we could talk about fusion if you
**Speaker B**: 好的。其实关键主要在于两点：其一是聚变技术本身，其二则是更先进的控制系统，我认为这一点其实也非常关键。
Original English
**Speaker B**: Yeah. Yeah. There was actually two two things. One is fusion. The other is um better control systems, which I think is actually
### 托卡马克的不稳定性与等离子体破裂风险
**Speaker A**: 确实如此。事实上，Google DeepMind 一直在致力于研发用于托卡马克（Tokamak）装置的等离子体智能控制系统，以确保等离子体不会陷入失稳状态并发生破裂。完全没错。
Original English
**Speaker A**: Yes. And in fact uh uh Google DeepMine has been working on uh uh control systems for Tokamax uh uh to make sure they don't uh uh essentially go unstable and go disrupt. Yeah, that's right.
**Speaker B**: 等离子体破裂（Disruption）本身就是一种非常特殊的物理现象。
Original English
**Speaker B**: Disruptions are are quite interesting to themselves.
**Speaker A**: 是的。
Original English
**Speaker A**: Yes.
**Speaker B**: 基本上就是托卡马克中所有的等离子体能量瞬间汇聚成束，准直成一道极细的能量束，然后……
Original English
**Speaker B**: Yeah. It's basically the entire the all energy in the the tokamac columnates into one little beam and then
**Speaker A**: 然后它直接猛烈轰击在真空室的内壁上，那一瞬间你会感到无比绝望和悲伤。
Original English
**Speaker A**: and it hits your vacuum chamber and you're very very sad.
**Speaker B**: 那确实非常惨烈。
Original English
**Speaker B**: Very sad.
**Speaker A**: 确实。大家都认为像 ITER（国际热核聚变实验反应堆）这样耗资高达300亿美元的庞然大物，如果刚启动就发生一次严重的破裂，基本上就相当于直接砸出了一个价值300亿美元的大废铁块。
Original English
**Speaker A**: Yeah. Yeah. I think people believe that eater could be could turned on after $30 billion disrupt and then basically have a 30 billion $30 billion brick or something.
**Speaker B**: 是啊，不过我想人们大概还可以尝试去修补打补丁吧。我记得……
Original English
**Speaker B**: Oh yeah, I guess I guess you could try to patch it. I remember um
**Speaker A**: 在大语言模型出现之前，我们就曾与一家名为 TAE 的核聚变公司合作过。当时我就在他们的主控制室里，现场确实让人捏一把汗。你必须极其谨慎小心。当时我们正在构建一套用于推荐新实验参数的算法系统，而他们的实验人员态度极其怀疑和审慎。这完全在情理之中，因为即便是由经验丰富的人类专家完全控制，我也亲眼见过实验过程中突然传来一声震耳欲聋的巨响，大家心里瞬间咯噔一下“大事不好了”。随后，整套实验装置就不得不彻底停机两周，好让他们去维修修补受损的部件。
Original English
**Speaker A**: uh working uh we we uh again uh before LM we worked uh with uh a fusion company called TAE and I was in their control room and yes it was kind of sad. You have to be very careful. we were making systems to recommend new experiments and uh they were very very uh skeptical and jaundice which way they should because because I've even been there even under human control it's like uh they were doing some experiment then you hear this big bang and it was like oh no and then it's like you know then the apparatus is down for two weeks as they patch some
**Speaker B**: 你当时竟然亲历了一次破裂事故？
Original English
**Speaker B**: you were there during a disruption
### 场反转构型（FRC）与等离子体磁约束机制
**Speaker A**: 噢不，不是托卡马克的破裂。抱歉，他们采用的是场反转构型（Field-Reversed Configuration，简称 FRC）。
Original English
**Speaker A**: oh no this is this is sorry they have filled reverse configuration
**Speaker B**: 原来如此。场反转构型自身也有其特性，比如会产生放电电弧。不过那具体是什么原理？抱歉我对这方面不太熟悉。
Original English
**Speaker B**: oh okay which has its own I mean things you know there's some arc So, so what is that? Sorry, I'm not familiar.
**Speaker A**: 什么是场反转构型？托卡马克虽然是目前被研究得最深入的等离子体约束形式，但它绝不是唯一的路径。实际上存在许多不同种类的约束架构，本质上都是在探索如何去稳定和压缩高能等离子体。场反转构型在形态上基本上是一个自包含的橄榄球状等离子体，其内部与外部的磁场方向恰好相反。它们之间被一个被称为“分界线”（separatrix）的边界隔开。从理论计算来看，这种构型本该是不稳定的，但在物理实验中它其实能够稳定存在。例如，当你运行磁流体力学（MHD）模拟代码时，在那些假设条件下它会显示为不稳定，但这只是一种理论假设，现实世界的物理规律并非如此运作。正因如此，FRC 曾经被冷落了许多年，但像 TAE 以及 Helion 等公司都在采用 FRC 路径，因为它们在实际运行中具有相当高的鲁棒性。你甚至可以把等离子体碰撞到腔壁上，它们依然能保持形态稳定。当然，放电现象依然存在，有时还是会在真空室壁上打出孔洞，这确实挺让人头疼的。
Original English
**Speaker A**: Oh, oh, what's the field reverse configuration? Well, it turns out toamax are not although they're perhaps the most studied form of plasma. There's many different kinds of architectures, essentially ways to try to stabilize and and compress plasmas. Uh there was a a shape essentially, it's essentially a self-contained football plasma called a field reversive configuration where essentially the magnetic field inside and outside are opposite. So, they're separated by something called a separatrix. And uh that is sort of in theory unstable but in practice stable. Uh like for example when you run magnet hydrodnamics MHD code uh it it's unstable under that assumption but that's that's an assumption that's not the way the real world works. And so yeah it was kind of disfavored for many years but um uh TA and other people I think Helion have have uh F FRC's uh because they are actually relatively robust. uh you can actually knock them against walls and they'll they'll still they stay stable and yes, but you can still get discharges and things that punch holes in your vacuum chamber which is kind of unfortunate.
**Speaker B**: 为了理清脉络我想确认一下：我们建造的这些聚变装置（或者说致力于成为实用反应堆的实验装置），会在内部产生等离子体，而这些等离子体带磁荷……
Original English
**Speaker B**: For clarification, so you have these fusion reactors. They are or trying to be reactors maybe and apparatuses. Appares and you create a plasma. The plasma is magnetically charged
**Speaker A**: 或者说是被磁场约束。是的。
Original English
**Speaker A**: or confined. Yes.
**Speaker B**: 或者说是磁约束。等离子体被磁场约束在特定空间内。所以装置配备了某种可以通过计算机调节的磁场控制系统，然后由计算机系统竭尽全力去维持这一约束状态。
Original English
**Speaker B**: Or confined. So it's confined by a magnetic field. So you have some sort of magnetic system that is tunable by a computer and then the computer tries to kind of maintain the confinement.
### 劳森判据与聚变技术的三大核心瓶颈
**Speaker A**: 在场反转构型（FRC）中，一旦成功形成等离子体，它在一定程度上是能够自我维持的，当然有各种不同的方法来确保持续约束。归根结底，核聚变的所有物理挑战都可以归结为所谓的“劳森判据”（Lawson criterion）。这从本质上解释了为什么核聚变如此困难。你可以非常轻松地在信封背面粗略推导出来：等离子体的密度、温度，以及表征能量损耗速率的“能量约束时间”（也就是等离子体中能量衰减至原本 1/e 所需的时间倒数），这三个物理量的乘积必须大于某个特定的常数，只有达到这个门槛，聚变反应才能持续发生并输出净能量；如果达不到，聚变就无法实现。正因为它是三个关键参数的乘积，才导致核聚变极其艰难，因为每一种聚变路径都有其致命的阿喀琉斯之踵——总有其中一个参数非常不理想，然后研发人员不得不拼尽全力去提升那一个短板参数。
Original English
**Speaker A**: Well F FRC's kind of once you make them they're sort of sustained. There's different ways of of trying to make sure you Okay, so all of fusion boils down to something called the loss in criteria. There's essentially uh and it's it explains why fusion is hard. Essentially, you can just very easily in on the back of an envelope just show that the density, the temperature, and essentially the energy loss, it's called the confinement time. It's one over the amount of time it takes for the energy to decay away uh 1 over e in a plasma. So the the product of those three numbers has to be bigger than some constant and then you can get fusion and if you don't then you don't. And that's explains the the fact that it's a product of three numbers explains why fusion is so hard because every approach has an Achilles heel where one of those numbers is not very big and then they try to desperately make that be higher.
**Speaker A**: 而且每种聚变路径的技术特性都截然不同。对于核聚变，公众必须对媒体上那些耸人听闻、大呼小叫的夸张新闻保持警惕。因为很多新闻报道会大肆宣扬“约束时间刷新纪录，稳定运行了 X 分钟”之类的消息，但这其实仅仅强调了三个参数中的某一个数字，而实现真正的受控核聚变必须同时满足全部三个参数的要求。我认为整个聚变领域目前确实在取得巨大的实质性进展，前景非常令人振奋，但对于那些只报喜不报忧、只挑单一参数大做文章的夸张报道，大家确实需要保持一份理性和审慎。
Original English
**Speaker A**: Uh and every approach is different. Every approach to fusion is kind of different and a lot of you have to be a bit skeptical when there there's all these sort of breathless news things about fusion because it'll say you know now confinement time is start like oh stable for x minutes or whatever and it's talking about like one of the three numbers but you have to have all three numbers before it you can get fusion. I I think it the whole field is is making a lot of progress and it's very exciting but you do have to you have to be a little bit cautious about the breathless news articles that only talk about one number.
### 控制算法实现与气候转型的经济现实
**Speaker B**: 那么，计算机和算法在这其中具体扮演了怎样的角色？
Original English
**Speaker B**: So what is the computational part of that?
**Speaker A**: 无论好坏，这完全取决于所采用的聚变路径。正如布伦丹（Brendan）之前所提到的，对于托卡马克来说，等离子体在大部分时间里是基本稳定的，但会偶尔爆发这种破坏性极强的失稳现象，它会瞬间汇集所有能量并猛烈砸向内壁某处，因此你必须依靠控制系统时刻将一切维持在可控范围内。所以这是一个复杂的反馈控制系统。而对于 FRC 构型来说，它本身的失稳模式则非常简单。例如它们具有所谓的 Z 轴不稳定性。整体上它是稳定可控的，只是会来回晃动，在物理空间上像橄榄球一样前后摆动。你只需要设计一个常规的 PID 控制器，通过动态调节将那个“橄榄球”牢牢稳定在反应堆的正中心，一切就能安然无恙地运行。
Original English
**Speaker A**: Oh, it unfortunately for better for worse it depends on the approach. So for tokamax as Brendan said um it's uh that there's this it's mostly stable except that there's occasionally this instability that takes all the energy and max smacks it into one place and so you have to sort of keep everything sort of under control. So it's a control system. Uh F FRC's themselves have very simple instabilities. So, for example, they have what they call a Z instability. So, it's fine. It's stable. It'll just wobble. It'll literally wobble back and forth, but you just make what they call a PID controller that just keeps the football in the center of the reactor and and things are fine.
**Speaker B**: 它是通过动态调节磁场来做到这一点的吗？
Original English
**Speaker B**: And it does that by adjusting the magnetic field.
**Speaker A**: 对。其实更准确地说，我认为它主要是通过调节电场，以此来前后推挽校正位置。人们所面临的具体控制挑战，完全取决于他们选择采用哪种等离子体物理架构。气候问题之所以演变成政治议题，主要根源在于经济成本，可能大部分是因为经济考量，或许也夹杂着其他因素。但在经济层面上，你要么必须说服大众愿意为清洁能源支付更高的成本，要么你就必须创造出一种极佳的技术方案，达成一种双赢的巧合——它不仅在经济成本上更具优势，同时又对气候环境更加友好。但要做到这一点，难度极大。
Original English
**Speaker A**: Yeah. It sort of it actually adjusts, I think, the electric field. It sort of by sort of knocks it back and forth. The issue that people have is really depends on which which sort of plasma architecture they're deciding to use. Climate is, you know, sort of political because of economics basically probably mostly maybe other stuff, but the the the economics of it, you know, you have to persuade people to to somehow spend more or you have to have a solution that it has like this happy coincidence where it's both economically better and and better for the climate. That that's hard.
**Speaker B**: 确实。在很多情况下，这种双赢并不存在，或者说这种经济优势目前根本还没有被证明……
Original English
**Speaker B**: Yeah. In many cases, it's not. I mean many cases it has hasn't been earned but
**Speaker A**: 嗯，你……
Original English
**Speaker A**: well you
### 能源干预与经济性：核聚变的多重副产物与炼金术设想
**Interviewer**: 试想一下，比如对天气的预测，如果你能提前预测天气，就可以提前做好准备，这显然具有巨大的经济效益。那么，你们在干预措施方面开展了哪些工作？这类干预又是如何与经济因素相互作用的？听起来就像飞机尾迹云（contrails）的那个案例一样，我做过分析后觉得太棒了，因为它的经济成本极低，却能带来极高的价值。
Original English
**Interviewer**: Think about like, okay, predicting even weather, right? You can prep and you could see how that could be economically beneficial. So what kind of work are you doing with interventions, and how does that kind of interact with economics? Like it sounds like the contrails one, I did it in an analysis and said actually this is great because it's very low economic impact but high value.
**Google Researcher**: 确实如此。如果要推进这类干预，我认为首先存在能源层面的干预。你必须去和现有的能源形式展开竞争。这绝非易事，除非它本身能带来协同效益（co-benefit），或者存在某种非常巧妙的协同效益。比如——虽然这再次具有很强的投机和设想性质，并不是我们的研究工作——有一家初创公司，不知你是否看过新闻，大概是去年的报道，有人发现如果向核聚变反应堆中注入汞（水银），聚变产生的中子流实际上可以将汞嬗变成黄金，然后你就可以把黄金卖掉变现。[笑] 我觉得这个想法非常巧妙。它最终可能行不通，但确实很聪明。
Original English
**Google Researcher**: That's right. So if you try, I think there's sort of energy intervention. So you have to sort of compete with existing forms of energy. And that's not trivial unless there's a co-benefit or there's some sort of clever just co-benefit. Like, this again is highly speculative, it wasn't our work. There was a startup that was—I don't know if you saw the news, it was last year, I think, where someone figured out if you inject mercury into a fusion reactor that the neutron flux can actually transmute the mercury into gold and then you can sell the gold. [laughter] Which I thought was very clever. It might not work, but...
**Interviewer**: 你知道，作为一名物理学家，我从聚变反应堆中最希望得到的东西其实是氦气，不过那是另一个话题了，抱歉扯远了。
Original English
**Interviewer**: You know, as a physicist, the one thing I want out of a fusion reactor is helium, but that's a different story. Sorry.
**Google Researcher**: 哦，你指的是氦-3（helium-3）吧。其实普通的氦-4挺平淡无奇的，不过由于战略储备库被关闭，现在的供应量确实在不断减少，是的。
Original English
**Google Researcher**: Oh, helium-3. Well, I mean, helium-4 is kind of boring, although it is getting—because the strategic reserve has been shut down, there's less of it. Yes.
**Interviewer**: 没错，我当然想要氦-3。倒不仅是用来做核聚变，哪怕仅仅是用来给稀释制冷机做制冷剂……
Original English
**Interviewer**: Uh and of course I want helium-3. Well, not even just a fuse, just to make dilution refrigerators for...
**Google Researcher**: 用于量子计算之类的设备，对吧。如果我们耗尽了氦气，我们所设想的许多现代科技实际上就彻底玩完了。
Original English
**Google Researcher**: Quantis or so. Yeah. So much technology we think about actually just goes out the window if we run out of helium.
**Interviewer**: 确实如此。
Original English
**Interviewer**: That's true.
**Google Researcher**: 根本没人在关注这个。抱歉，这完全扯偏了。
Original English
**Google Researcher**: No one's thinking about—sorry, that's like a complete aside.
**Interviewer**: 确实，美国设立氦气战略储备原本是为了满足我们那支极为重要的飞艇舰队的需求。
Original English
**Interviewer**: Uh yes. The fact that the US had a strategic helium reserve was for our very important blimp fleet.
**Google Researcher**: 是的。[嗤笑][笑]
Original English
**Google Researcher**: Yeah. [snorts] [laughter]
**Interviewer**: 但不管怎样，他们还是将它保留了数十年，这算是一件好事。但后来我们停止了储备，彻底处理掉了它，所有的氦气都升空飘散了。
Original English
**Interviewer**: But they kept it for decades anyway. So that was nice. But then we stopped. We got rid of it all. It all went up in the air.
**Google Researcher**: 没错，都充进气球之类的东西里飞走了。
Original English
**Google Researcher**: Yes. In balloons and stuff.
**Interviewer**: 或者是直接从天然气井中泄漏逸散了。
Original English
**Interviewer**: Or out of natural gas wells.
**Google Researcher**: 确实。
Original English
**Google Researcher**: Yeah.
**Interviewer**: 抱歉，我们刚才扯到氦气去了。
Original English
**Interviewer**: Um sorry, now we're talking about helium.
### 全面电气化的瓶颈与清洁基荷电力的挑战
**Google Researcher**: 没事，我们回到干预措施上。[笑] 那么对于干预措施，有哪些最令人兴奋、最有趣的方向呢？
Original English
**Google Researcher**: Yeah. No. So, [laughter] so interventions. What are some of the most exciting interesting ones?
**Interviewer**: 嗯，我对核聚变感到非常兴奋。我不知道这算不算是一种干预手段，它更像是一种根本性的能源来源。因为如果我们能够让核聚变成功运转，并且能将其初始资本建设成本降得足够低，那将会带来极其巨大的帮助。因为根据目前的模型，可再生能源固然非常好，理想情况下你希望将所有领域都实现电气化，对吧？但全面电气化本身面临重重困难：例如民航飞行就无法实现纯电化，不过你可以尝试将许多其他领域电气化，比如电动汽车（EV）；但你还必须设法解决水泥制造或钢铁冶炼等高能耗工业的电气化难题。这些都极其困难，尤其是炼钢工艺本身就需要还原动力，本质上必须添加碳来还原铁矿石。因此，在“实现一切电气化”的道路上存在许多非常艰巨的阻碍。
退一步说，就算你能够将一切都实现电气化，那么全社会所需的电力总量也将暴增整整五倍。你或许可以尝试通过大幅扩建可再生能源加上储能电池来满足需求，但若想榨干最后所有的化石能源缺口，成本就会呈指数级攀升。因为要覆盖最后那几个百分点，甚至最后的10%到20%的不稳定用电缺口，你将需要数量极其庞大惊人的储能电池。因此，我们确实需要某种能够填补最后20%空缺的电源，也就是某种可靠的基荷电力（base load）。而受控核聚变可能正是胜任这一角色的关键技术。因此，这非常令人振奋。不过话说回来，并不存在能够包治百病的“万灵丹（silver bullet）”。我很乐意深入探讨任何具体的场景，但现实世界的情况极其纷繁复杂。
Original English
**Interviewer**: Well, I'm very excited by fusion. I mean, I don't know if that's intervention. That's sort of a source of energy cuz if we can make it work and we can make it be sort of low enough capital cost that that will actually help a lot. Because at least the current models are renewables are great, ideally you'd like to electrify everything, right? Which has problems because like you can't electrify flights, but you can try to electrify a lot of stuff, you know they're EVs, you'd have to figure out how to electrify things like cement or steel. Those are hard, especially things like making steel want reduction power anyway to essentially you're adding carbon and you're reducing iron ore. So there's a lot of sort of things that are difficult about electrifying everything. But if you could electrify everything then the amount of electricity required would grow by a factor of five and you could try to grow renewables plus battery. And trying to squeeze all of it out, it starts getting ever more expensive because you just need ever more—you need like a huge number of batteries to sort of cover the last, you know, few percent or even, you know, 10 or 20%. So, we do need some sort of power that can cover the last 20%, something that's base load. So, fusion might be a thing for that. So, that's super exciting. Again, there's no one sort of silver bullet that can sort of cover all the cases. So I'm happy to sort of talk about any specific case, but it's sort of like...
**Google Researcher**: 整个世界是一个极其复杂的系统，全球经济也是一个错综复杂的庞然大物。因此，要泛泛而谈各类干预措施是非常困难的。
Original English
**Google Researcher**: The world is a very complicated place and the global economy is a very complicated place. So it's super hard to sort of talk about sort of interventions in general.
**Interviewer**: 确实是这样。
Original English
**Interviewer**: Yeah.
### 从减缓到适应：野火灾害与社会韧性建设
**Interviewer**: 也许我们可以抛开抽象的干预概念，我非常好奇的一点是：这些研究与预测是如何影响具体决策的？例如，我们究竟该建造什么设施？应该如何进行工程建设？我记得你以前常住洛杉矶，对吧？或者至少你在那里生活过……
Original English
**Interviewer**: Maybe instead of interventions, one thing I'm curious about is how does this affect decisions and to, for example, like what do we build? How do we build? I think you're from LA, right? Or at least you...
**Google Researcher**: 我在洛杉矶生活了11年。
**Interviewer**: 好的，所以你在洛杉矶度过了人生中很长的一段时光。我的意思是，洛杉矶的大片区域基本上刚刚被野火烧毁了。而且那些受灾区域很可能就靠近你以前生活过的地方。很多人其实早就预料到了这类灾害的发生，部分原因可能是监管层面的疏漏，部分原因则是其他问题。然而灾难来临时，我们完全处于猝不及防的状态，而且现在看来，在如何应对后续局面或如何调整适应方面，依然严重缺乏准备。我的意思是，你们是否从事过针对新风险评估的预测工作，或者提出过类似“我们究竟需要做出哪些实际改变来强化整个社会的韧性”的建议？哪怕只是为了让我们能承受住即将到来的灾难，无论我们最终是否真能采取行动去从根本上解决底层的气候问题。
Original English
**Google Researcher**: Well, I spent 11 years.
**Interviewer**: Okay. So yeah, you spent a lot of your life in LA. I mean, LA just basically large parts of it just burned down. And maybe probably close to where you used to live. So this is something that I think a lot of people kind of saw coming. Maybe partially due to regulatory issues, but partially due to other issues, you know, and we were completely unprepared and it seems like there is a lack of preparation about what to do next or to sort of adjust for this. And I mean have you worked on basically predicting like new risk assessments or you know suggestions of like what do we actually change to maybe harden society even that for what's coming regardless of whether or not we're actually do something to make solve the underlying problem.
### 卫星红外星座与 AI 代理模型：Google 的野火早期预警与火情传播预测
**Google Researcher**: 完全没错。事实上，Google 在一个名为“危机韧性（Crisis Resilience）”的领域投入了巨大精力。我们开展过一个非常有趣的项目，叫做野火监测项目（FireSat）。我不知道你是否了解这个项目。事实证明，对于森林野火而言，如果能够在火灾刚起步的极早期及时发现，扑灭一处只有这间屋子大小的山火是极其容易的；然而一旦火势蔓延到一个英亩（约4046平方米）的规模，扑救难度就会呈数量级剧增。当然，在某些气象和干燥条件下，火势可能会以指数级迅猛扩散，从一间屋子大小瞬间席卷至一英亩。要捕捉这种突发扩散固然困难，但在大多数情况下，它们起初往往火势较小，并且会在小范围内潜燃酝酿相当长的一段时间。
因此我们意识到，如果能在低地球轨道（LEO）部署一个全球卫星星座，搭载能够探测中波红外（midwave IR）波段的传感器——这又回到了黑体辐射原理，本质上中波红外恰好对应了火焰燃烧的温度特征——那么森林火灾在中波红外波段下就会显得异常清晰夺目。
基于此，我们设计了一种高灵敏度传感器。根据具体的轨道参数规划，只需发射大约50到80颗卫星，就能够从太空中精准发现大约相当于这间屋子大小——大概每边长5米见方，或者比这间屋子稍大一点——的初起微小火点。无论火灾发生在地球上的任何角落，取决于卫星星座的组网数量，如果部署80颗左右的卫星，我们就能在火情爆发后短短15到20分钟之内成功捕获目标。
一旦在十几分钟内探测到火情，人类就能够真正采取针对性的干预行动。应急部门也可以自主研判：如果希望让野火受控燃烧以清理地面可燃物载荷、且评估当前环境可控安全，就可以选择暂不干预；但如果监测到火势极可能失控爆发为灾难性大火，就能果断出击。
为了推进这个项目，我们与一家名为“地球火灾联盟（Earth Fire Alliance）”的非营利组织展开合作，Google 也是其重要发起成员之一。目前他们已经开始行动，我们已经成功发射了一颗原型试验卫星，并与 Muon Space 公司紧密合作来具体制造和交付卫星硬件。这非常振奋人心。
此外，我们还构建了野火火线边界检测技术，并将这些实时火情信息通过 Google 的生态网络广泛分发。我们可以根据现有的公开卫星遥感数据与各类地面实时数据流，精准计算并标定火势蔓延的前沿边界，随后通过安卓手机推送通知或在 Google 搜索结果中直接向受灾地区的普通民众发出预警提醒。
不仅如此，我们还与美国国家森林局（US Forest Service）深入合作，共同研发了火情动态传播的新一代预测模型。因为现有的传统火蔓延模型大多基于20世纪70年代由一位名叫罗瑟梅尔（Rothermel）的学者提出的基于物理过程的模型，计算开销很大。我们利用全新的神经网络技术构建了一个轻量高效的 AI 代理模型（neural network proxy model），能够在极短时间内实现火势扩散的高速推演与模拟计算。我们正是在这一前沿方向上与林务局携手攻关。
所以是的，我们对最大程度减少野火危害怀有极高的热情与投入。很多人可能没有意识到，根据世界卫生组织（WHO）的权威估算，全球每年因野火烟雾导致的超额死亡人数高达到惊人的30万人。
Original English
**Google Researcher**: That's right. So in fact there's a big effort at Google into something called crisis resilience. And so we had a very fun project called FireSat. I don't know if you know about this. So it turns out that for wildfires, a lot of these wildfires you could—if you only caught them early enough, it's very easy to put out a wildfire the size of this room, but even if it's like an acre it gets much, much harder. And so, and of course under certain circumstances they can grow exponentially from the size of this room up to an acre. So that might be hard to catch, but they often sort of start small and spend a while. So we figured out that, oh, if you had a global constellation of low earth orbit satellites that could detect in the midwave IR, that which goes back to the black body essentially that's the temperature of fire. The fire stands out in the midwave IR. And so, we designed a sensor that if you built—it depends on exactly what their orbits, but, you know, roughly 50 to 80 of them, you could actually find fires about the size of this room, about 5 meters on a mount, maybe it's a bit larger than this room, 5 meters on a side, anywhere on the planet. And again depending on how many satellites you had, within like 15 to 20 minutes. You'd have to put a fair number up, like 80 to get them within 15 minutes, and then you could actually intervene. You could decide not to if you wanted to have the fire burn fuel and you thought it was safe, but if it was going to blow up to something unsafe. So you worked with a now a nonprofit called Earth Fire Alliance that we're part of, and so they're starting to—we've launched one satellite which is a prototype. We've worked with a company named Muon Space to actually sort of make the satellite. So that's cool. We have wildfire boundary detection and we propagate that information out through Google. So we can actually sort of figure out from existing satellites and existing data feeds where the boundaries of fires are, and then we sort of tell people through their Android phones or through search about fires. We've worked with the US Forest Service on making new models for how fires propagate, because again that goes back to these process-based models from the 70s by a person named Rothermel. So we've actually made a little neural network proxy model based on a new essentially to sort of be able to run it very, very quickly. So we worked with the forest service on that. So yeah. Yeah, we're very, very interested in trying to minimize because it turns out people might not realize the World Health Organization estimates that there are 300,000 excess deaths a year across the world from wildfire smoke.
### 野火烟雾的致命威胁与健康影响评估
**Interviewer**: 是的。我清晰地记得——大概是几年前吧，如果没记错的话，那是前几年一个特别严重的火灾季，大概四五年前，有一大团遮天蔽日的浓烟横扫了整个美国北部和加拿大全境，当时引发了极其严重的呼吸系统疾病问题。
Original English
**Interviewer**: Yeah. I mean I remember—was it if it's been a few years since we had a really bad fire season. Maybe what, four or five years ago there was this cloud of smoke which like crossed all of northern US and Canada and caused a lot of respiratory issues I think.
**Google Researcher**: 确实。
Original English
**Google Researcher**: Yeah.
**Interviewer**: 是的，而且这种影响极其难以直接追踪。我的意思是，你必须通过严谨的流行病学统计学方法才能估算出这些超额死亡人数。
Original English
**Interviewer**: Yeah. And it's very hard to track. I mean, you have to get these—you have to get the estimate these excess deaths from statistical means.
**Google Researcher**: 但不管怎样，这确实是一个……
Original English
**Google Researcher**: But yeah, it's a...
### 气候变化应对：从根本治理到适应韧性
**Speaker A**: ……非常严重的公共卫生问题，而且令人深感恐惧，它会烧毁人们的房屋，后果是极其灾难性的。
Original English
**Speaker A**: ...very serious public health problem and also just very scary and burns people's houses down and it's terrible.
**Speaker B**: 在我看来，气候变化在某种程度上就像一种严重的疾病。面对这种严重疾患，你究竟是去治疗症状——也就是说，去被动适应；还是努力去根除背后的病因？答案显而易见：如果病情足够严重，这两件事你必须双管齐下。
Original English
**Speaker B**: My view is that climate change is sort of like a serious disease. Do you treat the symptoms, i.e. do you adapt, or do you try to attack the underlying thing? And the answer is, well, if it's serious enough, both,
**Speaker A**: 没错吧？
Original English
**Speaker A**: right?
**Speaker B**: 确实如此。正因如此，我们在 Google 非常重视这种适应性举措，尤其是在气候韧性方面。我们致力于为人们提供各类信息工具来帮助应对。这也正是我们积极投入研发天气预报系统、气旋与台风预测等技术的部分原因。所有这些工作实际上都是紧密交织、相辅相成的。所以这绝不仅仅是单维度的干预手段，你说的很对，它关乎全面的气候韧性建设。
Original English
**Speaker B**: And so yes, so we take sort of adaptation especially around climate resilience very seriously at Google and we try to give people informational tools to sort of help. That's part of the reason why we're working on weather and then sort of cyclone prediction and things. So it all actually hangs together. So it's more than just—you're right—it's more than just interventions. It's climate resilience too.
**Speaker A**: 是的。我自己在美国西海岸亲身经历过四五个野火山火季，那种场面真的极其恶劣糟糕。以前那里可不是这个样子的。我在内华达山脉（Sierra Nevada）有一座度假小木屋。以前到了夏季，那里气候宜人、非常舒适美好；可现在呢，虽然并非年年如此，但也差不多成了常态——一年四季几乎变成了冬季、春季、夏季，以及……“浓烟季”。
Original English
**Speaker A**: Yeah. Having lived through four or five fire seasons on the West Coast, they can be quite, quite nasty. And it used to not be—I mean, I have a cabin up in the Sierra Nevada mountains, and yeah, it used to be, "Oh, you know, summertime, it's nice." And then now it's like, well, not every year, but yeah, there's like, you know, winter, spring, summer, and smoke.
**Speaker B**: 是啊。一到了野火山火季，整个人就只能……
Original English
**Speaker B**: Yeah. Fire season you're just—
**Speaker A**: 哈哈，没错。[笑声]
Original English
**Speaker A**: Yeah. [laughter]
**Speaker B**: 一心只想确保自己待在火线蔓延方向的西侧上风向。
Original English
**Speaker B**: I want to stay to the west of the fire line and—
### 微型卫星与红外遥感：AI 赋能的超分辨率野火监测
**Speaker A**: 是的，千真万确。所以这也引出了另一个让我非常感兴趣、同时 Google 也有着极高研究热情的领域，那就是应对气候变化所必需的技术韧性。比如，现在的红外（IR）传感器体积是否已经做得足够小，以至于能够搭载到类似微型低轨卫星星座网络中进行组网？或者换个角度说，如果是去直接向那些已经在运营和监控现成遥感卫星星座的厂商付费采购数据，会不会成本更低？
Original English
**Speaker A**: Yes. And yes. So that is another thing that I'm interested in and that Google's also very interested in is climate resilience. Are these IR sensors small enough that they could hitch a ride in like a micro satellite grid? Like would it make sense to—would it be almost cheaper just to pay someone who's watching a constellation?
**Speaker B**: 噢，它们其实并没有小到那种微型程度。核心难点在于，这套系统需要专门的制冷机制……
Original English
**Speaker B**: Oh, they're not that small. The thing is you need refrigeration because—
**Speaker A**: 噢，我明白了，原来是这样。
Original English
**Speaker A**: Oh, okay. Okay.
**Speaker B**: 因为它工作在中波红外（midwave IR）波段。这就意味着你必须配备冷却系统，让传感器始终保持在极低温度下工作。
Original English
**Speaker B**: because it's midwave IR. So, you have to keep it cool.
**Speaker A**: 这么说来，它们必须得是专门定制的专用卫星了。
Original English
**Speaker A**: So, these would have to be their own special—
**Speaker B**: 必须是专用卫星。不过它们本身的物理尺寸倒也没有庞大到夸张的程度。
Original English
**Speaker B**: satellites. They're not super large.
**Speaker A**: 明白。
Original English
**Speaker A**: Okay.
**Speaker B**: 它们绝不像地球同步轨道上那些因为塞满了庞大光学镜头组件等复杂设备而体积极其庞大的庞然大物。不过，是的……
Original English
**Speaker B**: They're not like the satellites in geosynchronous orbit that are these giant monsters because of all the optics and who knows what. But yeah,
**Speaker A**: 而且从本质上讲，它们主要搭载的就是红外光学传感设备，原生空间分辨率大概是 5 米乘 5 米……？
Original English
**Speaker A**: and they're basically just IR sensors with a resolution of 5x5—
**Speaker B**: 不，不，并不是。这正是这套方案极为精巧的地方：传感器原始的物理地面分辨率其实只有大约 50 米乘 50 米，但你可以引入超分辨率（Super-Resolution）算法进行重构。由于它本质上是多光谱成像，而且你在先验上能够大致掌握火灾常发区域的环境特征与物理规律，因此我们在其底层数据之上融入了相当深度的 AI 算法，最终成功实现了 5 米乘 5 米级别的高精度空间分辨率反演。
Original English
**Speaker B**: Uh, no, no, that's the other cute thing is the resolution is about 50 by 50 m, but you can use super resolution because it's essentially multispectral and you sort of know where fires are and you have—so yeah, there's a fair sprinkling of AI on top of them to reach that 5x5 meter.
**Speaker A**: 原来如此，非常精妙。
Original English
**Speaker A**: Yeah. Yeah.
**Speaker A**: 与此同时，由于传感器读出扫描时存在相对位移，这里面其实还存在着时间与空间维度的卷积效应，对吧？
Original English
**Speaker A**: When you also have a convolution over the what you're reading out, right? As well—
**Speaker B**: 我记不太清它确切的帧率参数了。卫星本身是在高速轨道运动中的，我一时也记不得它的点扩散函数（Point Spread Function, PSF）具体是多少了，实在抱歉。但你说的完全准确，卫星确实处于高速运动中，虽然我记不清具体的运动角速度数值。此外，这套设备还采用了一种所谓的推扫式扫描传感器（broom sensor）。因此这就带来了一个非常有意思且棘手的问题：你需要在一个维度上将光谱信号展宽色散，同时还要兼顾空间推扫，整个光学成像与信号采集系统相当复杂，它绝不是拿拍立得相机拍张快照那么直接，而是一个构造极其复杂的综合传感系统。
Original English
**Speaker B**: I forget the frame rate. The satellite is moving. I don't remember what the point spread function is. I'm sorry. I don't—but you're right. They do move, but I don't remember how fast they—this also has something called a broom sensor. So, there's this funny thing of trying to—you kind of spread out the spectrum one way and it's a somewhat complicated thing. It isn't just like a Polaroid. It's a complicated sensor.
### AI for Science 的范式巨变：从专用模型到通用基础大模型
**Speaker A**: 确实是这样。接下来让我们稍微转换一下话题。我们知道，你在“人工智能赋能科学研究”（AI for Science）这一交叉前沿领域深耕探索了相当长的时间，可以说一路亲历并见证了这个交叉领域的起伏演进与来回迭代。
在你的观察中，这个领域究竟经历了怎样的演变历程？因为我个人明显感觉到，当下它的演进迭代节奏变得异乎寻常的迅猛。在这一路探索的过程中，你个人总结出了哪些深刻的经验教训？你认为整个科研学术界共同沉淀出了什么认知？如果面对的是年轻一代的科学家或年轻的技术从业者，这种时代的剧变应该如何重塑他们规划与拥抱未来科研生涯的方式？
Original English
**Speaker A**: Yeah. Sort of switching gears a little bit. You've been at the intersection of AI and science for quite some time. I think you've sort of wound your way into and out of it back and forth.
How do you see the field has evolved? Because I feel like it's evolving very quickly now. And like what are the sort of lessons that you've learned that you think the community has learned, and how do you think this should change if you are a young scientist or young practitioner? How should this change how you should approach the future?
**Speaker B**: 在我看来，在过去的 12 到 18 个月时间里，整个领域经历了一场真正意义上的“相变”（phase change）。回顾过去，正如我之前谈到的，我们长久以来的典型工作模式是针对某一个具体的独立学科难题，去专门构建高度定制化的特化 AI 模型。如果你把这种“发现一个问题、攻克一个问题；接着再找下一个问题、再去攻克它”的过程当作自己的本职工作，那确实充满了探索的乐趣与成就感。
Original English
**Speaker B**: Well, I think there has been a phase change in the last 12 to 18 months. A lot of what we used to do, as I said, was build these specialized models to solve individual problems. And if you think that's your job, it's kind of fun. You find a problem, you solve it. You find another problem, you solve it.
**Speaker B**: 但现如今，我们手中拥有了能力通用性远比以往强大得多的通用 AI 系统与前沿模型。我认为眼下整个 AI for Science 共同体其实仍处于一种集体摸索、试探边界的阶段。这些通用大模型在各个科学维度上展现出的惊人有效性是如此前所未有且新颖，以至于我们在集体层面上尚未完全确立共识：究竟怎样做才是最优的技术解法？抑或根本就不存在所谓的单一最优解，而是需要一套全新的复合工具链（tool chain）。我想所有人目前都在极力推演与弄清楚：我们接下来究竟应该怎样开展科学研究？
Original English
**Speaker B**: But now we have these much more general AI things. And I think the whole AI for science community is kind of still feeling around. The fact that they're working is so new that collectively we're not sure what's the best thing to do, or maybe there's no one best, maybe there's a tool chain, and I think we're all trying to figure out like what should we do.
### 给年轻科学家的建议：深耕领域专长、锤炼品味与掌握新一代工具
**Speaker B**: 紧接着就会引发那个根本性的问题：年轻一代的科学家现在究竟应该怎么做？我想，这不由得让我联想到我的儿子。他刚满 21 岁，目前对人工智能算法、软件工程编码以及生物化学都抱有极其浓厚的探索热情。每当我审视他的学习路径时，我都由衷觉得他正走在完全正确的轨道上：因为他一方面在疯狂吸收专业知识，努力让自己在 RNA 等核心前沿生物领域建立扎实深厚的领域专业认知（domain expertise）；而另一方面，他积极拥抱各种最前沿的技术工具，充分践行“氛围编码/直觉式编程”（vibe coding），熟练调用并驾驭市面上的所有新型研发辅助工具。我认为这正是当下的标准答案与最优解。
Original English
**Speaker B**: And so there's a question of what should young scientists do? I have a son who just turned 21 and he's really into both sort of AI and coding and chemistry, and I look at him, I think he's doing the right thing because he's both learning a lot and trying to be a domain expert about RNA, but he's also sort of using vibe coding and using all the tools. I think that's the right answer.
**Speaker B**: 为什么必须坚持走这条路？正因为如今所有人都还在重新摸索前行的技术逻辑，深厚的专业领域底蕴依然无可替代。我坚信领域专业深度不仅绝不会被取代或消亡，反而会回归到许多前人一再强调的本质上来——它最终考验的是一个研究者的“科学审美品味”（taste）。可接踵而至的深层挑战是：如果在如今这个 AI 时代，年轻人在成长过程中完全跳过了那些繁琐底层的苦活、累活、基础脏活（grunt work），他们究竟该如何培养出卓越犀利的科学品味？这是一个极具深度且悬而未决的开放式议题。但无论如何，一方面务必全力深耕垂直领域的专业深度，另一方面也要大胆尝试并熟练把玩目前涌现出的所有新颖工具与前沿平台。因为当前的行业态势绝非什么“未来的技术图景早已被完全看透、前方的资深专家全知全能地洞悉一切”；完全不是这样，我们这些所谓资深同行同样每天都在面对未知展开高频的实验摸索。
因此我的核心建议就是：务必扎扎实实地建立起不可替代的领域深度知识，积极主动地利用好这些前沿 AI 工具，竭尽所能地去挑战并攻克那些真正宏大且棘手的硬核科学难题。
此外，当前依然横亘着一个极其巨大的核心瓶颈与开放课题：面对现实世界中无可回避的实体实验室物理实验（physical lab work），我们究竟该如何破局？实体实验是绝对不可能凭空消失的，因为归根结底，真实物理世界中的实验验证才是检验科学真理的唯一标准（ground truth），而且……
Original English
**Speaker B**: Because everyone's figuring it out, still be a deep domain—I don't think domain expertise is going away because it goes back to a lot of people who said it goes back to taste, and trying to figure out how people get taste without doing all the grunt work. That's an interesting open question. But develop domain expertise, but also try and play, I would say, with all the different tools that are available, because it's not like, "Oh yes, we know what's going to happen and the smart old people know what's..." Like no, we're experimenting too. And so yeah, so I would say definitely develop domain expertise and try to use these tools, and try to solve big, hard scientific problems as best you can. There's still the huge open issue about what do you actually do about physical lab work that is not going away, because, you know, experiments are the ground truth, and—
**Speaker A**: 而且物理实验本身就是一个极其庞大的通量瓶颈。
Original English
**Speaker A**: And a bottleneck—
**Speaker B**: 并且是一个绝对的核心瓶颈。虽然业内大家如今都在热议所谓的“人在回路/自动化闭环实验室”（lab-in-the-loop），但这个方向目前依然处于极其早期、高度开放的探索状态。因为试问谁能够打造出一套真正通用的、无所不能的通用自动化实验室呢？据我所知目前根本无人做到。市面上现存的绝大多数方案，都只是针对极其狭窄、特定化学或生物反应流程的高度专用可控实验室。
所以总的来说，我们正在共同经历这样一场深刻的范式相变，这种变革前景令人感到无比振奋。再次强调，我真诚地建议年轻研究人员：放开手脚去玩转目前能够触及的所有工具，竭尽所能地去打磨专属于你自己的深厚领域洞见与科学审美鉴赏力。与此同时，千万不要心存畏惧，要勇于动手做各种大胆激进的尝试。在 Google，我们每年都会迎来许多杰出的学生研究员，他们来到这里后总是会去尝试做各种充满想象力、看似天马行空甚至近乎疯狂的探索，而这往往总能碰撞出令人惊喜赞叹的成果。所以，年轻人就应该敢于去尝试各种看似离经叛道、疯狂大胆的新想法，去看看最终究竟能孕育出怎样颠覆性的突破。
Original English
**Speaker B**: and a bottleneck. I mean, people are talking about a lab in the loop, but that's still very, very, very open because no one has, I think, as far as I know, a general lab that does everything. There's a lot of very specific, you know, labs that are controllable. So I think it's just we've gone through this phase change. It seems super exciting. Again, I would advise people to play with whatever tools are available, and to develop sort of deep domain expertise and taste to the extent you can. And I would advise people also not to be scared, and like try stuff. We have student researchers at Google and they come and they do sort of wild and crazy things and that's always just delightful. So yeah, people should be trying sort of wild and crazy things and see what happens.
### 从底层造轮子到开箱即用：科学基础能力的训练如同比赛体能训练
**Speaker A**: 确实如此。接下来这个问题或许未必有一个确切的定论，但当我回想我自己过去究竟是如何逐步建立起自身技术专长、以及许多顶尖同仁是如何培养出扎实专长时，我们会发现，大家最初往往都是从一个界定清晰的简单具体问题切入，然后反复硬碰硬地去死磕、攻关磨砺它。而在这种深度探索与反复试错的过程中，你的认知边界不断拓宽，有时思维走向更加宽广的全局视野，有时则在垂直维度扎向更深的本质机理。但归根结底，正是那种“不断把头撞向坚硬南墙、咬牙死磕具体技术细节”的磨砺过程，塑造并赋予了你后续解决更复杂、更困难问题所必备的核心工程与科研硬实力。
可如今，当年那些需要我们耗费数月死磕的问题，在现代 AI 工具面前几乎瞬间就能迎刃而解。面对这一现实，你又会给你的儿子提供怎样的求学与成长建议呢？
Original English
**Speaker A**: Yeah, this may be a question without an answer, but when I think about how I developed expertise and how a lot of people developed expertise, it was by starting with a simple defined problem and then hammering it, and then in that process of exploration, you learn more and some ways you go broader, some ways you go deeper, but the process of just banging your head against the problem, which now would be instantly solvable, teaches you the skills you need to solve harder problems. What advice would you give to your son for that?
**Speaker B**: 实话说，我也很难给出标准答案。但或许这在某种程度上就像高山徒步登山一样。很显然，你不可能什么地方都指望靠开车一路开过去；虽然在很多情况下，你确实完全有条件直接开车一路开到山顶……
Original English
**Speaker B**: You know, I don't know. Maybe it's a bit like hiking, which is you—yes, I mean, you obviously can't drive everywhere. Or you could drive up the mountain.
**Speaker A**: 是的。
Original English
**Speaker A**: Yeah.
**Speaker B**: 但你也可以选择穿上登山鞋，一步一个脚印地亲自徒步攀爬登顶。即便你明知自己随时能够开车轻松上山，但偶尔特意选择依靠双腿徒步登顶，不仅是完全合理且有益的，甚至往往是一件极具乐趣、锻炼心志的事情。
Original English
**Speaker B**: Or you could hike up the mountain. And maybe it's okay, even fun to occasionally hike up the mountain even if you can drive up the mountain.
**Speaker A**: 确实是这样。
Original English
**Speaker A**: Uh yeah. I mean—
**Speaker B**: 回想起以往的老日子——我这么说听起来可能真像个上了年纪的老古董了，在当年那些旧时光里……[笑声]
Original English
**Speaker B**: in the old days—I'm going to sound like a real old man. In [laughter] the old days,
**Speaker A**: 哈哈，你是指……六个月前的“旧时光”吗？
Original English
**Speaker A**: in the old days of six months ago?
**Speaker B**: 哈哈，不，不，我指的是更久远的 80 年代和 90 年代的软件工程岁月。如今很多人理所当然地享受着极其丰富的生态红利：“瞧，这里有海量成熟的开源软件包，有 Scikit-learn，有现成齐备的科学计算框架……”而在我们那个年代，这些基础底座统统不存在。我必须全部亲手一行行编写自己的基础数值计算库（numerical library），必须亲手实现自己的机器学习算法架构。我自己用四种截然不同的编程语言，将 Boosting 算法前前后后完整重写了不下四次！但正是经历了这种甚至可以说是严苛重复的造轮子历程，如今我对 Boosting 算法底层的每一处数学精髓与工程细节都有着深入骨髓的认知。
因此，或许明智的做法就是不要总是贪图极致的安逸与捷径。显然，这里面永远存在着一种权衡取舍：你一方面本能地渴望尽可能提升产出效率与生产力；这是无可厚非的。但与此同时，你必须清醒地意识到，自己的基础工程底子与科研“核心肌肉群”必须得到针对性的锻造。这在很大程度上，其实就类似于一名职业运动员的日常体能训练。
Original English
**Speaker B**: Well, no, no, I was even thinking of in the old days of the 80s and 90s, you know, a lot of people take it like, "Oh, there's open source packages, there's scikit-learn, there's all sorts of things." We didn't have that! I had to write my own numerical library, I had to write my own machine learning. I've written Boosting, probably rewritten it four times in four different languages. And so now I know Boosting, you know, it's like—and so maybe not taking the totally easy—I mean obviously, I mean there's this trade-off like, "Oh, but I want to be as efficient and productive as possible." Yes. But you also have to develop the muscles. So it's a little bit maybe like being an athlete. Like—
### 探索与榨取的平衡：为什么我们需要保护“20%时间”
**Speaker A**: 在很多时候，你确实需要全力“榨取”（exploit）现有成果，拼命往前跑，跑得越快越好。但同时，你也必须留出专门的训练和沉淀时间，大家终究是需要不断学习与训练的。
Original English
**Speaker A**: There are places, there are times when you're actually doing exploit when you're trying to run as fast as you can. And then there's also training time, and so maybe people just have to train.
**Speaker B**: 这一点完全说得通。即便你花时间踏踏实实地啃硬骨头、做最苦最累的基础工作，当下看起来进展缓慢，但长远来看，它为你未来的生产力带来的复利回报是极其可观的。哪怕在某个局部节点上，你因为没有直接调用大语言模型（LLM）走捷径而显得不是那么“高效”，这种投入最终也是会沉淀下来的。
Original English
**Speaker B**: And it's entirely plausible that if you spend time actually hammering away and doing the hard work, even if it goes slower there, that pays dividends into your larger, you know, productivity long term. Like even if locally that one moment you are not being maximally productive by not exploiting an LLM that feeds into something.
**Speaker A**: 我真希望如此。但我拿不准的是，现在的职场环境对从业者来说实在太难了，因为整个世界似乎都在狂热地追求全盘优化，恨不得榨干每一丝价值。这甚至像成了一种错——不，这不是你的错，是我的错。但现实就是这种弥漫的氛围，让人身不由己。所以很多时候，你必须刻意去划出一块不受打扰的时间。比如在 Google，尤其是在我带领的团队里，我们一直保留着“20%自由时间”（20% time）的传统，而我也一直在尽全力在自己组里捍卫这一制度。也就是说，你想学什么、想尝试什么完全由你自己决定，你甚至都不用跟我汇报。真的，你完全没必要告诉我，甚至我可能根本就不该过问。大家尽管去为了学习本身折腾事情，去探索未知，因为那才是真正孕育创造力火花的源泉。我不想把大家的时间填得密不透风，以至于大家完全没有空间去玩耍、去钻研、去尝试各种异想天开的新点子。我知道在当今环境下保留 20% 时间显得有些特立独行，因为周遭似乎总有一股强大的推力在催促你像我刚才说的那样去极限优化、榨干所有产出。但一旦你过度优化，某种程度上就相当于“过拟合”了。
Original English
**Speaker A**: I hope so. I hope the thing I don't know is I hope people in their careers it's hard because right the the the whole world seems to want to optimize everything and your fault. No, I don't. It's No, I don't. It's my fault, but it's just sort of the the the the you know, uh and sometimes you have to set aside time. Like at Google, especially in my group, we have this concept of 20% time, which I still I very very strongly in my own group try to protect. It's like you can do whatever you if you want to learn stuff, if you want to try stuff, you don't even have to tell me. You don't even have to tell me. In fact, I probably shouldn't tell. Uh you know, just do stuff for for Exactly. for learning and also because that's where the sort of creative juices are. I don't want to so occupy people's time where they have nothing like where they they can't feel like they can play or learn or try new crazy things. So I know 20% time is unusual and there just seems to be this strong impetus in the world to just like like I said optimize and squeeze everything out but you do lose something when you hyperop you sort of overfit. [laughter]
**Speaker B**: 你直接过拟合到了眼前的狭隘生产力指标上。
Original English
**Speaker B**: You overfit to productivity as you're so
**Speaker A**: 没错，完全就是这样。所以我知道，我给出的这个建议很可能是在逆着当下主流的文化惯性前行。
Original English
**Speaker A**: that's right. So I I know that might be my advice might be swimming upstream against uh perhaps cultural norms.
---
### LLM 时代的新技能：从执行者走向中层管理者
**Speaker B**: 在这个问题上，我脑海中始终在盘旋的一点是：技术环境和问题本身并不是静止的（not stationary）。未来势必会出现一套全新的、真正契合未来需求的技能树。我一直在反复琢磨的是：究竟哪些技能才是真正具有持久生命力的“恒久技能”（enduring skill）？有些能力可能昨天还算不上持久核心，但到了今天却变得不可或缺。举个显而易见的例子：当你深度使用大语言模型时，你的角色本质上变成了一个管理者。那么，传统的技能迁移到底能不能行得通？
Original English
**Speaker B**: The thing that always comes up for me here is that I there's a the problem is not stationary. There's a new skill set that will be the the right skill set for the future, right? And that the question in my mind is always just tangling. Okay. Is this a skill that is an enduring skill that Yes. that or or or like maybe it it wasn't enduring yesterday, but today it will be enduring because the like I've seen that um you know like for just as an obvious one you become sort of like a manager when you're when you're using LL skills transfer. Well,
**Speaker A**: 有些技能确实无法迁移，但有相当一部分是可以迁移的。作为一名管理者，你不可避免地会从底层的执行细节中抽离出来，你必须信任你的下属、AI Agent 或者系统能把具体细节处理好，以便他们能向上汇报，回答那些宏观层面的大问题，并对琐碎的具体事项做出准确的判断。难道我们最终的命运就是变成一个个中层管理者吗？虽然我自己做过一点管理工作，但我必须说：你绝对不能沦为一个空架子（empty suit）。
Original English
**Speaker A**: some of them don't, but a lot of them do. And so that as a manager, you lose track of of the details of what's going on and you trust your people or agents or whatever to have that managed so that they can report up to you and answer uh you know sort of the highle questions and and get the judgment about the little things correctly and and so that is that what we've come to is that we're just like middle managers now. No, I mean I don't know. Again, I I I have uh a little management work, but you can't I don't you don't want to be an empty suit.
**Speaker B**: 换句话说，正因为这些工具往往会犯错——特别是大语言模型这类行为模式十分诡异的系统，它们犯的错误往往跟人类完全不一样。
Original English
**Speaker B**: In other [laughter] words, because the things might get it wrong, especially LM that are sort of really weird. They don't make the same kind of mistakes that humans make.
**Speaker A**: 没错，所以你绝对不能完全盲信它们。你必须保持严谨，反复挑刺、不断质询和验证。不过话又说回来，对你自己亲手写的代码，你也本就应该这样去反复挑刺。换句话说，连你自己都不能轻信。这是我多年来学到的一大教训。费曼（Richard Feynman）是怎么说的来着？你绝对不能自欺欺人，而……
Original English
**Speaker A**: And so you you can't, you know, fully trust them. You have to be rigorous and like, you know, poke at it and make sure. Although, you should be poking at software that you write yourself, too. I mean, you shouldn't trust yourself. That's that's one thing I've learned. What did Fman say? You know, you absolutely can't fool yourself and you're
**Speaker B**: 你自己就是最容易被自己愚弄的那个人。
Original English
**Speaker B**: you're the easiest person to fool.
**Speaker A**: 正是如此。所以我很难用量化指标去界定究竟什么才是恒久的，但那种触及世界底层规律的“根本性认知”一定是长存的。比如真正掌握一个关于现实世界的学科领域——这也是为什么我非常推崇生物学和物理科学，我认为这类根本性的东西是永不过时的；数学同样极其恒久。但除此以外，甚至像批判性严谨思维、持续验证这类素质也是如此。这又绕回了刚才说的管理维度：你必须切实确保大模型输出的内容是正确的、没有在某个环节糊弄作弊。然而在本质上，你也同样需要用这种审视标准来检验你自己。
Original English
**Speaker A**: So, um, so that might be I don't know if I I can quantify what's enduring, but somehow fundamentalness I mean really learning a domain that is about the world for example. So this is why I like well biology or physical sciences I think those will those are enduring sort of fundamental things. math is very enduring, but even things like rigor and checking and that sort of thing, which goes back to maybe management that that you want to really make sure that the LMS are producing the right things or they haven't cheated in some way, but again, you should be doing that to yourself, too.
**Speaker B**: 是的。
Original English
**Speaker B**: Yeah.
**Speaker A**: 所以我认为这里面存在某些历久弥新的东西，此外创造力和打破常规的创意思维（thinking out of the box）同样具有恒久的价值。这些都是具有持久生命力的品质。在这一切的深处，都存在着某种非常根本的底层逻辑。
Original English
**Speaker A**: So, I think there's some enduring and also just the enduring value of sort of creativity and thinking out of the box. And so, I think those are enduring. I don't know. There's some there's something very fundamental about all of those.
---
### 加州理工往事：费曼的计算物理课与“费曼效应”
**Speaker B**: 既然你刚才提到了费曼，如果你不介意的话，我们不妨换个话题聊聊他。
Original English
**Speaker B**: So, you mentioned Fineman. Um, if you don't mind me uh changing gears.
**Speaker A**: 我当年曾经上过费曼的课。而且那可绝不是一门普通的课。
Original English
**Speaker A**: I took a class from Fineman. Yeah. Not just any class.
**Speaker A**: 没错，就是他开的那门“计算物理学”（Physics of Computation）课程。那是他和霍普菲尔德（John Hopfield）以及卡弗·米德（Carver Mead）一起联合讲授的。那段经历非常有意思。但在当时，恐怕连费曼自己都未必完全清楚整个方向到底在哪，我感觉当时我们所有人甚至连真正的问题是什么都没搞明白。费曼当时大概是试图提出“用量子系统来做量子模拟”，后来的事实证明这确实是击中要害的正确答案。不过在那个时候，首先这门课有 DARPA 的经费资助，所以原则上学生每周可以蹭到一顿上好的上等牛肋排大餐（prime rib dinner）。当然即便没有这顿饭，我每周也照样会风雨无阻地去听课。
Original English
**Speaker A**: Yes. The the the his sort of uh physics of computation class. He did it with in fact Hopfield and and and uh uh Carver me. Uh that was fun. At the time I don't think any maybe Fman knew. I felt like none of us knew what even the problem was. I mean I guess Fman was trying to say oh let's do quantum simulation which I guess is uh it turned out to be the right answer but yes at the time it was well I guess uh it was cool first of all it was DARPA funded it so you were supposed to go once a week to get a prime rib dinner uh but I just went every week [laughter] uh anyway for the students.
**Speaker B**: 是的，再交代一点时代背景的话：这门课基本上就是费曼以及其他人提出量子计算机雏形概念之后不久开的。当时大家甚至都还没搞清楚量子计算机到底长什么样，只知道在底层物理上蕴藏着某种全新的可能性……
Original English
**Speaker B**: Yeah. So ju just for a little bit more context, this class was basically the class right after Feman and I forget who else proposed the concept of a quantum computer without really knowing what it was, but knowing that there was some sort of
**Speaker A**: 费曼当时写过一篇著名的演讲文章叫《底层的空间还大得很》（*There's Plenty of Room at the Bottom*），不过那是更早期的事了。我上这门课是在 1982 年。
Original English
**Speaker A**: well there was plenty of there was plenty of room at the bottom essay which I think was in the very early this I took it in '82. Okay.
**Speaker B**: 原来那是他亲自授课的时间。
Original English
**Speaker B**: Oh, that was when he taught it.
**Speaker A**: 他在课堂上也确实强调过“底层的空间还大得很”。但那门课当时的授课形式非常奇特：通常是周二邀请一位外部学术嘉宾来做客座演讲，然后到了周四，费曼就会大摇大摆地站上讲台，当场给大家拆解为什么周二那位嘉宾讲的全都是错的！这简直太搞笑了。而且在那门课上，我们亲身体验到了后来许多人都提到过的“费曼效应”（Feynman effect）：他的个人魅力实在太强了，当他站在台上推导解释时，你会频频点头觉得“全懂了，太透彻了！”；然而当你走出教室门冷静下来一琢磨，才猛然惊呼：“等等，我刚才根本就没听懂！”
Original English
**Speaker A**: And he did say there was plenty of room at the bottom. But the way the class was structured, it was um uh it was like a guest lecture on Tuesday and then Fman would stand up on Thursday and explain why that was all wrong. [laughter] Uh which was pretty fun. And then we encountered something which other people have also encountered something which called the Fman effect. Maybe he was so charismatic or something. He would explain things and he would say, "Yes, yes, I understand. Yeah. And then you walk out to think, "No, no, I I [laughter] did not."
---
### 可逆计算、兰道尔极限与 1982 年的算力荒原
**Speaker A**: 是的，那门课确实极富乐趣。不过从那些请来的客座嘉宾身上，也恰恰折射出了当时学界探索初期的混沌与迷茫。当时请来了很多大咖，比如丹尼·希利斯（Danny Hillis）等一系列重量级人物。如果把他们探讨的内容汇总起来看，能极其清晰地看出当时整个领域的大混乱：大家当时都在热烈争论“计算机到底能不能做成可逆计算（reversible computing）？”因为大家都在思考，每次逻辑操作产生的耗热极限难道真的可以是零吗？还是说存在着某种不可逾越的热力学极限？这在当时是极其核心的大课题，虽然放到今天大家已经不太把它当成什么不可逾越的大阻碍了。
Original English
**Speaker A**: Yeah. So, it was kind of fun. But a lot of the guest people, I mean, maybe the sort of showed the chaos. I mean, a lot of people, I think it was um Danny Hillis came, uh there was all these interesting guests. So, I would say in the union of all of them, I think showed the sort of mass confusion of what was going on because there was a lot of like, oh, should we make computers reversible because we have to make sure, you know, can they even be reversible? Can can the sort of the bottom limit of of the heat per operation be zero or is there some sort of thermodynamic limit that was like a big deal. I don't think that's I don't think that's a big deal now.
**Speaker B**: 对，但是……
Original English
**Speaker B**: Yeah. But
**Speaker B**: 等等，你这里说的其实是指兰道尔极限（Landauer's limit），对吧？还是指别的热力学理论？
Original English
**Speaker B**: wait. So so you're talking about the land hour limit, right? or not what's on them.
**Speaker A**: 当时围绕这个领域的所有争论都在于：我们能否构建出像“台球计算机”（billiard-ball computer）那样完全理想的可逆物理计算模型？这种物理系统到底有没有可能做到无能耗可逆？此外大家也在探索需要什么样的新型计算架构。这也正是为什么丹尼·希利斯会来演讲的原因，当时正值“连接机”（Connection Machine）以及早期的思考机器公司（Thinking Machines Corporation）创立的时代，也就是 80 年代最初的那批架构雏形。
Original English
**Speaker A**: Uh well there at the time it was all this this question about like can you have rever like billyard ball computers and can they be reversible and things so and also sort of like what kind of computing that's why Danny Hillis came like you know I think that was in the era of the connection machine and what the original thinking machines not mad but the original one in the 80s
**Speaker B**: 所以他当时亲自去了现场。
Original English
**Speaker B**: uh so he came
**Speaker B**: 那在 1982 年的时候，通用的经典计算发展水平到底处于什么状态？我是指在那个时间节点上……
Original English
**Speaker B**: what was the state of like general computation in 1982 like I mean I mean at this point
**Speaker A**: 按照四舍五入的舍入误差来算，当时的通用算力几乎约等于零！我记得我刚到加州理工的时候，担任卡弗·米德实验室的系统管理员（sysadmin）。我们全组只有一台 DEC VAX-11/750 小型机，整台机器的处理性能大概只有 1 MIPS（每秒一百万次指令运算），而且整个研究组共用一个容量仅为 80MB 的硬盘驱动器，那个硬盘的体积简直跟一台洗碗机一样大！在当时拥有这套配置已经让人无比兴奋了。那么当时的计算复杂性理论（complexity theory）又发展到了什么地步呢？因为我知道当下大家对量子计算复杂性理论以及它跟量子引力之间的内在关联有极大的研究兴趣，所以我很好奇，当年对于复杂性理论……
Original English
**Speaker A**: to rounding error we had zero [laughter] I mean we had I remember when I sorry I got there uh I was Carver me's um uh CIS admin and we had a vax 11 750 that maybe did a myip 1 million 1 million operations and the whole uh research group shared an 80 megabyte disc drive that was the size of a dishwasher. It was it was very exciting. Uh so what about complexity theory? What what was it? Because I know that there's a lot of interest in quantum complexity and how it relates to gravity right now. And I wondered I don't have a mind in my mind about when complexity
**Speaker B**: 我们可以回头具体查一下文献。不过你说的对，现在的确建立起了完整的量子复杂性类谱系（hierarchy of quantum complexity classes）。
Original English
**Speaker B**: we we could try to look it up. I don't think there's of course the whole there's a whole hierarchy of you're right quantum complexity classes.
**Speaker A**: 我觉得那一整套完备的理论框架应该都是在 1982 年之后才逐步发展建立起来的。因为在 1982 年那堂课上，事实上我甚至都怀疑当时是否已经有了明确的“量子门”（quantum gate）这个概念，因为在当时的课程讨论里，我们根本就从来没有提及过量子门……
Original English
**Speaker A**: I think that was developed after that because this was in 1982. In fact I I'm not even sure there was a gate. We never talked about
### 量子计算早期与经典教材
**Speaker A**: ……门模型。当时根本没有量子计算的门模型——
Original English
**Speaker A**: ... the gate model. There was no gate model of quantum--
**Speaker B**: 你是指量子门吧，没错，确实如此。
Original English
**Speaker B**: I mean quantum gate. Yeah. Yeah. Yeah.
**Speaker A**: 是的，我一时忘了，抱歉。我觉得很多这些概念……是的，我认为那些直到90年代左右才真正提出来。对，我记得读过尼尔森和庄（Nielsen & Chuang）写的那本经典量子信息——
Original English
**Speaker A**: Yeah. Yeah. I forget. I'm sorry. I think a lot of those--yeah, I don't think those came out until like the '90s or something. Yeah, I mean, I remember reading Nielsen and Chuang, the classic quantum information--
**Speaker B**: 对，那本经典的量子信息教科书。现在回想起来，那本书确实提到了很多这类观点。那本教材大概是写于90年代末或2000年代初，没错吧。
Original English
**Speaker B**: Yeah, quantum information textbook which brings up a lot of those points now I think about it. That textbook was written in what, the late '90s or early 2000s, that's right.
**Speaker A**: 我觉得那是第一本将该领域的通用知识系统化整理出来的教科书。不过我也可能记错了。我想，当年我还年轻的时候，大概只是想当然地认为这就是所有人都在读的那本书……
Original English
**Speaker A**: And I think that was the first textbook which like put down kind of the general knowledge of the field, but I could be wrong. I guess I think now that when I was young, I don't think I took it just for granted that this is the book that everyone--
### 80年代神经网络热潮与Hopfield网络
**Speaker B**: 确实。不过你还记得在80年代初的时候，大家对神经网络有一股巨大的热情，这非常耐人寻味，因为当时我们真的不知道自己在做什么，根本没有人知道自己在做什么。
Original English
**Speaker B**: Yeah, but this is a--well, you remember in the early '80s, I mean there was a whole bunch of excitement around neural networks and it's really interesting cuz again we really did not know what we were doing. No one knew what they were doing.
**Speaker A**: 那时经历过一段有趣的演进：先是神经网络，然后是支持向量机（SVM），接着又是神经网络。
Original English
**Speaker A**: There was like an interesting progression: like neural networks, then SVMs, then neural networks.
**Speaker B**: 不不不，这比那个阶段还要早得多！因为在80年代，所有人都在说，这完全有能力彻底改变计算科学。
Original English
**Speaker B**: Oh no no no, this is before that because in the '80s everyone said, everyone said this has the capability of revolutionizing computing.
**Speaker A**: 但那是基于什么呢……？
Original English
**Speaker A**: But what does--
**Speaker B**: 当时围绕霍普菲尔德网络（Hopfield networks）有着极大的轰动。事实上，NeurIPS（当时的NIPS）之所以诞生，就是因为当时在雪鸟（Snowbird）举办了一个名义上是小范围闭门的研讨会，结果所有人都争相跑去参会，于是他们便由此创办了NIPS。
Original English
**Speaker B**: Well, I mean there was a tremendous excitement around Hopfield networks. In fact NeurIPS came out of--because there was a workshop at Snowbird that was nominally private, but everyone tried to crash, and so they spun up NIPS. And so that was--
**Speaker A**: 雪鸟研讨会就像是一场附带了计算学术会议的滑雪之旅。
Original English
**Speaker A**: Snowbird is a skiing trip with a computation conference attached.
**Speaker B**: 没错！我就是在那儿学会滑雪的，因为我本来不会滑雪，学会后就一直滑下去了。话说回来，那个研讨会源自1985年的圣巴巴拉研讨会，而后者又是从加州理工学院本地称为“Hopfests”（霍普菲尔德节）的研讨活动中衍生出来的。
Original English
**Speaker B**: That's right! That's how I learned to ski, because I didn't know how to ski and then I kept going. Anyway, that workshop came out of the Santa Barbara workshop in 1985, which came out of some local things at Caltech which is called Hopfests.
**Speaker B**: 当时人们觉得，“哇，这里面肯定蕴含着某些深刻的东西”。事实上，讽刺的是，这与联想记忆（associative memory）密切相关；如果你真正深究Transformer的本质，它们其实就是联想记忆。所以后来甚至有一篇论文就叫《霍普菲尔德网络就是你所需要的一切》（Hopfield Networks is All You Need）。
Original English
**Speaker B**: So, so yeah, people thought oh wow, something involved. In fact, ironically, there was like yeah, something about associative memory, and if you actually dig down into what transformers are, they are associative memory. So in fact there was even a paper called, you know, "Hopfield Networks is All You Need." So--
**Speaker A**: 对，我都忘了那个标题其实是在向当年的那篇论文致敬。
Original English
**Speaker A**: Yeah, I forgot that title was a reference to that paper.
**Speaker B**: 没错，向那个时代致敬。所以整件事情就像是兜兜转转又回到了原点。其实我认为，作为一个领域，我们集体确实彻底改变了计算机科学，只是当年的希望与憧憬完全超出了当时的实际能力，因为回过头来看，我们当时几乎没有任何实际算力。
Original English
**Speaker B**: To the era, to the era. So it's actually like the whole thing has sort of come full circle. And in fact I think we did--we have collectively as a field revolutionized computer science, but the hopes and dreams completely outstripped the capabilities, because again we effectively had zero compute.
### 算力演进、硬件协同与英伟达的转型
**Speaker A**: 是的，我认为机器学习的历史就是不同范式的演进，随着算力与内存、模型扩展与数据在不同层面的可用性逐步建立起来。
Original English
**Speaker A**: Yeah, I think the history of machine learning is different paradigms as like compute versus memory scaling versus data become available in different levels.
**Speaker B**: 确实如此。而且我觉得很多人并没有意识到，神经网络之所以胜出，是因为它们是受算力约束（compute-limited）的——尽管现在有了Transformer之后，情况又开始变得受内存带宽约束（memory-limited）了。但最关键的是，神经网络是构建在基础线性代数子程序（BLAS）之上的。由于GPU等硬件在持续对BLAS进行深度优化……虽然我并不确切知道人脑是不是靠矩阵乘法运作的——事实上我非常确定大脑并非通过矩阵乘法工作（笑）——但这正是算法与硬件共同演化的结果。
Original English
**Speaker B**: That's right. And the fact that I think people don't realize that the reason why neural networks won is they're the one compute-limited thing--although again now that we have transformers, things are getting memory-limited again--but and the fact that they ride on top of BLAS. And so the fact that BLAS was being optimized by things like GPUs. I mean, I don't actually know if--in fact I'm pretty sure brains don't work by matrix multiply [laughter], but it was just that the algorithms co-evolved with the hardware.
**Speaker A**: 这也是我们走到今天这一步的原因。谁知道呢？如果我们当初走了另一条路，人们更关注其他类型的计算方式，天晓得我们最终会得到怎样的架构。我也不清楚。话说，这一切最初很大程度上是不是始于有人破解PlayStation游戏机来训练神经网络？或者甚至是更早之前用在超算上？
Original English
**Speaker A**: And that's why we're here. Who knows? I mean, if we'd gone down some other path where people really cared about some other compute making, who knows what architecture we have ended up with. I don't know. I mean, didn't a lot of this start with like I think people were hacking PlayStations to train neural networks or something, or to do--I guess maybe it was even before that for like super--
**Speaker B**: 这里有个有趣的故事，是我读研时的朋友布莱恩·卡坦扎罗（Bryan Catanzaro，现任NVIDIA应用深度学习研究副总裁）讲的。布莱恩，如果我记错了细节请见谅——他刚加入英伟达时，很难推行自己的想法。当时英伟达在游戏业务上押下了双倍、三倍的重注。后来有一天，他跟黄仁勋（Jensen Huang）开了一次会，成功说服了他，布莱恩说：“你看，现在有那么多人正在把CUDA用于BLAS计算，尤其是用于深度学习。”他打动了黄仁勋。据他说，那只是一次大约15分钟的交谈，第二天整个公司就完成了全面转向（笑）。
好吧，我自己从来没在英伟达工作过。不过我有朋友在那里——第一任首席科学家戴夫·柯克（Dave Kirk）和比尔·达利（Bill Dally）都是我读研时的同窗好友。所以情况大体如此，但具体细节我并不全了解。
不过你要记得，像辛顿（Geoffrey Hinton）团队在2010年左右就已经在用GPU做深度学习了，但在2007、2008年左右，GPU相比CPU其实并没有快那么多。它们真正开始大幅超越CPU——我认为这绝非巧合——正是在初代ImageNet突破以及部分语音识别技术突破的那个时期。因此，我认为深度学习在GPU超越CPU之时强势复兴，这两者绝不是巧合。
Original English
**Speaker B**: There's a funny story by a friend of mine from grad school, Bryan Catanzaro at NVIDIA, where he--Bryan, sorry if I get this wrong--but he came to NVIDIA and was having a lot of trouble getting traction. And basically they were doubled and tripled down on gaming. And he basically one day had a meeting with Jensen and convinced him, "Let's, you know, we have all these people using CUDA for BLAS basically, and for deep learning in particular." And convinced Jensen, and he said it was like a 15-minute conversation and they pivoted the whole company the next day or whatever [laughter].
Okay, this--I've never worked at NVIDIA. I mean I have friends--Dave Kirk, who's the first chief scientist, and Bill Dally were both friends of mine from grad school. So yes, but I don't know the details of way. Now remember that people like in Hinton's group, even around 2010 they were using GPUs to do deep learning, but even in I would say 2007, 2008 they weren't that much faster than CPUs. Again, they only started really exceeding--in fact I think this is not a coincidence--in the era of, you know, the original ImageNet and some of the speech recognition stuff. So I think that's not a coincidence that they really resurged when GPUs passed CPUs. So there's that--that's not a coincidence either.
### 在加州理工发现小行星与小行星命名
**Speaker A**: 是的。我真的很想知道，一个人究竟怎样才能获得给小行星命名的机会呢？
Original English
**Speaker A**: Yeah. I really want to know how does one get the opportunity to name an asteroid.
**Speaker B**: 哦，那还是在加州理工学院的时候。当时有两位非常棒的人——尤金·舒梅克（Gene Shoemaker）和卡罗琳·舒梅克（Carolyn Shoemaker），他们在开设一门行星科学课程。我当时很喜欢行星科学，所以在课程的一部分中，他们带我们体验了小行星搜索发现的整套流程。
那是在80年代，现在的探测系统已经极其先进惊人了。后来有一段时间出现了一个叫LINEAR的项目实现了自动化搜索，但在当年，任何事情都还没有自动化。
所以，这只是课程作业的一部分。当时的做法是——虽然我们在课上实际操作的顺序有所调整，但40年前的标准步骤是这样的：你要去帕洛马山天文台（Palomar Observatory），那里有一台快速巡天望远镜。那台望远镜现在已经成了陈列在他们访客中心的历史博物馆展品了，但在当时可是一台真正的科研仪器。你在望远镜里装上一张底片，拍下一张照片；然后等待几分钟，再对着天空完全相同的方位拍摄另一张照片。
回到加州理工后，把两张底片放进闪烁立体比较仪（stereoscope）里。你四处观察，看是否有任何星点在“悬浮漂浮”。因为如果是小行星，它在两张底片之间会发生极其微小的位移。由于两只眼睛看到的图像位置微小不同，它就会产生视差，直接凸显悬浮出来，让你能明显看到它被投影在不同的空间位置上。
Original English
**Speaker B**: Oh well again at Caltech there were some wonderful people, Gene and Carolyn Shoemaker, and they were teaching a class in planetary science. I liked planetary science and so yes, as part of that class they took us through the sort of asteroid discovery thing. Now this is in the '80s so now there's all sorts of amazing systems. Well there was for a while there something called LINEAR which automated the thing, but this was before anything was automated.
So yeah, it was just part of a class. And what you would do--the steps, even though we did it out of order in the class, but the steps we used to do, this is 40 years ago, is you would go to Palomar, there would be a fast telescope which literally now a museum piece that's in their visitor center, but at the time it was a real thing. You would put a piece of film in it, you would take a picture, and then you'd wait for a few more minutes and take a same picture of the same point of sky. You'd put in a stereoscope like back at Caltech, you would see if anything--you would look around, you would see if anything floated because it would move by a tiny bit. So it would pop up and you can see it was because it's different in different eyes. Then you would be able to see it as something that was projected in a different place.
**Speaker B**: 没错，它在视觉上会直接跳出来呈现在你眼前。接着你会走到一台测量显微镜前，以已知参考恒星为基准，精确测量这个漂移天体的位置坐标。
如果你测量的数据足够精确，就会把它寄给一个人——我想现在应该已经不是他了，他叫布莱恩·马斯登（Brian Marsden），在亚利桑那州的小行星中心（Minor Planet Center）工作。他当时有一套庞大的软件系统，用来将这些所谓的“冲日观测记录”（apparitions）拼接关联起来。如果你找到了某个小行星，而你的观测恰好是最后一次确认记录、使得他的软件能够将其连成一条完整的轨道，那么你就会获得该小行星的发现权，并且可以为这颗小行星命名。
不过现在的情况已经令人叹为观止了。智利现在建起了一座薇拉·鲁宾天文台（Vera C. Rubin Observatory），配备了一台极其出色的西蒙尼巡天望远镜（Simonyi Survey Telescope），基本上把整个流程完全自动化了。它基本上可以对天空进行高频连续拍摄，是一台令人震撼的仪器。他们仅在6周时间内就发现了11000颗小行星。
Original English
**Speaker B**: That's right. So it would pop out at you literally. And then you would go to a measuring microscope and you would take measurements of known star references and where this floater is. And so if you got accurate enough measurement, then you would send it off to a person, I don't think it's him anymore, his name Brian Marsden at the Minor Planet Center in Arizona. And then he had a big software system to sort of piece together--these are called apparitions. And if you found, and if you happen to have done an observation, which was the final apparition which allowed his software to connect it into one big orbit, then you would get discovery rights and you could name the asteroid.
But now it's like amazing. There's this observatory now in Chile called the Vera Rubin Observatory and there's this amazing telescope called the Simonyi Survey Telescope. Essentially automates this process. Essentially can just take many frames of the sky. It's an utterly stunning instrument. And so they discovered 11,000 asteroids in 6 weeks.
**Speaker A**: 哇，太厉害了。
Original English
**Speaker A**: Oh wow.
**Speaker B**: 是啊，所以现在——
Original English
**Speaker B**: Yeah. So it's now--
**Speaker A**: 现在已知的小行星一共有多少颗？至少……
Original English
**Speaker A**: How many of them are there? I mean at least--
**Speaker B**: 这个得看按什么尺寸截断标准来算了，不过是的，可能存在数以百万计的小行星。
Original English
**Speaker B**: Well, it depends I guess on the cut off but yeah. So, but there probably millions of asteroids.
**Speaker A**: 这么说来，命名的机会其实还很多。
Original English
**Speaker A**: So, there's still a lot of opportunity.
**Speaker B**: 确实如此。但我甚至不知道他们现在还愿不愿费力气去逐个命名了（笑）。
Original English
**Speaker B**: It's true. But I don't even know if they bother name. [laughter]
**Speaker B**: 所以我也不清楚，也许根本不需要逐个命名了吧。不过现在人们在做恒星掩星（occultation）观测——抱歉，聊起这个我能没完没了地说下去。有种现象叫做掩星观测。比如我发现的其中一颗小行星，我之前正在跟一位业余天文爱好者发邮件交流。如果一颗小行星恰好从……前方经过……
Original English
**Speaker B**: So, I don't know. Maybe you don't need--I don't know. But people are doing occult--sorry, I'll talk about this forever. There's stuff called occultation. Like one of my asteroids, I was talking, I was sending email to an amateur. If an asteroid happens to pass in--
### 小行星卫星的意外发现
**John Platt**：当小行星从恒星前方掠过时，恒星的亮度会发生下倾变暗，就像天文学家发现系外行星时那样。如果你运气极佳，它会先变暗一次，紧接着由于存在卫星，亮度会再次下降。就在去年，一位业余天文爱好者在我发现的一颗小行星周围发现了一颗小卫星。
Original English
**John Platt**: ...front of a star, it dips. Like when they discover exoplanets. But if you're super lucky, it'll dip and then it will dip again because there's a moon. And so one of my asteroids, this amateur found a little moon around it. That was, I think, last year.
**主持人**：在那颗小行星周围？
Original English
**Host**: Around the asteroid?
**John Platt**：没错。
Original English
**John Platt**: Yeah.
**主持人**：真有意思。
Original English
**Host**: Interesting.
**John Platt**：所以这确实挺酷的，这意味着业余爱好者依然有探索空间……
Original English
**John Platt**: So that's pretty cool. So there's still room for...
**主持人**：你的小行星竟然还有一颗属于它自己的卫星。
Original English
**Host**: Your asteroid has a moon to it.
**John Platt**：是的。事实上，这背后其实还有一段小故事［笑］。这严格来说并不完全算是我发现的小行星——我一共找到了两颗。我本来想把其中一颗以我父亲的名字命名，但我当时犹豫不决。于是 Carolyn 就以华盛顿大学一位教授的名字给它命名了。我说：“哎呀，但我本来想给它命名的。”由于我有些懊恼，她人很好，就把她发现的一颗小行星赠送给了我。后来他们就用我母亲的名字命名了那一颗，而恰恰就是那一颗小行星被发现拥有卫星。
Original English
**John Platt**: Yes. In fact, apparently [laughter] there's a little bit of a story there because it's not really my asteroid that I found two of them. I named one of them after my dad. I waffled. So Carolyn named it after a professor at University of Washington, and I said, "Oh, but I wanted to name it." And so I whined, and she was nice, and so she gave me one of hers. And so then they named that one after my mom, and that's the one that has the moon.
**主持人**：原来是这样。
Original English
**Host**: Oh, okay.
**John Platt**：所以，以我母亲名字命名的小行星竟然带有一颗卫星。
Original English
**John Platt**: So the asteroid named after my mom has a moon.
**主持人**：它有多大？
Original English
**Host**: How big is it?
**John Platt**：主天体的直径他们估计大约是四公里宽——我尽量把单位说准确。而那颗卫星大约是一公里。所以实际上这是一个相当大的双星系统（binary system），虽然算不上孪生双星，但尺寸已经很可观了。
Original English
**John Platt**: It's like the main body they think is about four kilometers across. I'm trying to get the units right. And the moon is about 1 km. So it's actually a pretty big binary. They're not quite twins, but yeah.
**主持人**：那期课程里还有其他人发现小行星吗？
Original English
**Host**: Did other people in that class discover?
**John Platt**：我想大概还有另外一个人找到了吧。这事多少需要碰运气。确实如此，而我一个人能找到两颗，算是非常不同寻常的了。
Original English
**John Platt**: I think there was one other person. It's a little bit of a crapshoot. Yeah. And the fact that I found two was quite unusual.
**主持人**：当时能找到是有什么特殊技巧，还是纯凭运气，抑或是整天泡在那里搜寻的结果？
Original English
**Host**: Was there a reason why you... Was it just pure luck or just staying there all day?
**John Platt**：这算是努力的结果吗？我也不好说。我当时确实尽量观察得很仔细，不过……
Original English
**John Platt**: Do I try? I know. I tried to be careful, but...
### 从实习生研究到奥斯卡科学技术奖
**主持人**：对了，一个人怎样才能拿奥斯卡奖呢？
Original English
**Host**: Yeah. How does one get an Oscar?
**主持人**：也就是奥斯卡金像奖（学院奖，Academy Award）。
Original English
**Host**: Yeah. Academy Award.
**John Platt**：嗯，这或许也是因为我在对的时间出现在了对的地方。我的博士论文导师名叫 Al Barr。1986年时，我还是个学生实习生，当时在一家名叫斯伦贝谢（Schlumberger）的公司实习。那本来是一家石油勘探技术公司，但他们内部有一个人工智能实验室，大家都在那里从事计算机图形学相关的研究。
当时我们所有人都在思考——在那个年代，这种想法可谓极具颠覆性，尽管我知道它在今天看来已经平常得甚至有些乏味——那就是其实可以利用物理模拟器来制作计算机动画电影。而在当时，我觉得这简直太酷了。于是我说，完全可以利用弹性理论（theory of elasticity）来模拟软体形态（floppy things）。我基于弹性力学理论，编写了一套弹性物理模拟器，成功模拟出了织物布料以及富有弹性的物体等等。
斯伦贝谢的人看到后惊叹：“哇，这太棒了！”后来，这套技术的衍生成果演化成了许多物理模拟系统，被皮克斯（Pixar）等公司广泛应用于他们的多部经典动画电影中。所以，我现在常常半开玩笑地对实习生们说：如果你在实习期间干得足够出色，你甚至可能拿到一座奥斯卡奖［笑］。
Original English
**John Platt**: Well, again, maybe I was at the right place. My thesis advisor is named Al Barr, and I was a student intern actually in 1986 at a place called Schlumberger, which is an oil discovery place, but they had an AI lab and so we were all doing sort of computer graphics research. And we were all thinking, at the time this was revolutionary—I realize it's now considered incredibly boring—like you could actually use physics simulators to make computer graphics movies. And at the time I was like, wow, that's really cool. So I said, you could use the theory of elasticity to make floppy things. And so I said, here's the theory of elasticity, and I wrote an elastic simulator, and I made fabric and stretchy things and stuff.
So they said, oh wow, that's cool, and so the descendants of that became a lot of the physics simulators that people used in Pixar and their various movies. So yes, I tell interns sort of half-jokingly, well, if you do a really good job as an intern, you can get awarded. [laughter]
**主持人**：从你完成那项研究工作，到最终获奖，中间隔了多久？二十年？
Original English
**Host**: So how long was that between the time you did that work and then... 20 years?
**John Platt**：整整过了二十年。
Original English
**John Platt**: It was 20 years.
**主持人**：这其实挺合乎常理的，对吧？因为当学院给你颁发奥斯卡科学技术奖时，他们必须确认这项技术确实经过了充分验证，并且已经在整个行业中被大家广泛使用。在这漫长的二十年里，这套技术显然已经普及成了行业标准，所有人都习以为常了。
Original English
**Host**: Which is actually not atypical, right? Cuz when they give you an Academy Award, they want to make sure like, oh yeah, it's sort of well used and everyone uses it and stuff. So, yeah, but of course obviously like in that 20 years like, well, everyone does it. It's obvious, but you know...
**主持人**：那是专门为某一部特定电影所开发的工作吗？
Original English
**Host**: This was specific work done for a specific movie or something?
**John Platt**：不，那其实是一篇发表在 SIGGRAPH 上的学术论文。
Original English
**Host**: No, it was like a paper in SIGGRAPH.
**主持人**：那是一篇论文，后来就逐渐演变成了皮克斯等动画制作流程的核心关键技术。
Original English
**Host**: It was a paper and then Pixar just became, like I think somewhat important core to like a lot of their...
**John Platt**：没错。事实上，我的很多朋友都参与了大量这类物理模拟器的后续研发。确实如此。
Original English
**John Platt**: Yes. And in fact some of my friends, in fact a lot of my friends did a lot of these simulators. So yeah.
### 量子计算的发展轨迹与时代定位
**主持人**：我对你过去的背景很好奇，你曾接触过量子计算领域，并且之前在 Google 应用科学部门直接参与过量子计算工作。虽然我知道你目前已经不在那个团队了，但我很好奇你想怎么看……
Original English
**Host**: I'm curious about, since you have been quantum computing adjacent and worked on quantum computing directly at Google Applied Sciences for a while. I think you're not currently on that, but I'm curious to see what...
**John Platt**：我现在偶尔也还会涉猎一些。
Original English
**John Platt**: I'm still dabbling in...
**主持人**：你依然在保持关注和涉猎，很好。那么纵观这些年，你如何看待量子计算未来的发展轨迹？因为这似乎有点像核聚变技术：最初出现时让人无比振奋，随后陷入长期的停滞或进展缓慢，让人感觉似乎毫无突破；而到了现在，我们似乎又重新看到了许多激动人心的曙光。或者这只是我作为一个外部旁观者的片面感受？
Original English
**Host**: You're still dabbling. Yeah. So where do you see the trajectory of quantum computing over the years going? Because this is one of those things which, I guess kind of like fusion, had a sense at first like being very exciting, and then seeming like it wasn't going anywhere for a long time, and then maybe now we're seeing hints again of it being exciting. Or maybe that's like my sort of quantum adjacent...
**John Platt**：我认为很多人都经历过这种心态起伏。而我倾向于将视角拉得更长远一些。大概是因为我年纪已经足够大了，习惯把事物放在数十年的宏观跨度里去平均看待，其本质一直都是稳步的技术进步。
Original English
**John Platt**: I think a lot of people have gone through that. I tend to average things out over the... I guess I'm old enough now where I sort of average things out over the decades, and it's just progress.
**主持人**：但纵观历史进程，从理查德·费曼最初提出构想、试图厘清这一概念究竟意味着什么，一直发展到如今，比如近期谷歌 Willow 芯片在量子纠错方面取得的突破性进展，这似乎切实表明凭借正确的标度律（scaling laws）等支撑，量子计算是完全行得通的。
依你看，量子计算距离实现某种简单但真正成功运行的实用算法，究竟还需要二十年吗？你觉得这个进程正在加速，还是说它依然需要相当漫长的攻坚过程？因为我们在很多技术领域都见证过令人意想不到的突飞猛进，你认为量子计算也会迅速突破，还是依旧难以一蹴而就？
Original English
**Host**: But I mean, see going from, you know, starting from Feynman just trying to figure out even what this means as a concept, all the way up to now where, there's this recent Willow result of quantum error correction, which actually seems genuinely achievable with the right scaling laws and stuff. Would you say that quantum is 20 years away till some simple but practical algorithm which actually succeeds? Do you see this accelerating, or do you think this is still going to be something... Because there's a lot of areas where we saw this advanced very quickly and it was very unexpected, and I'm wondering if this is a thing that you think will be quick or will not be?
**John Platt**：我认为它的发展速度介于两者之间。量子计算绝非一个纯粹的软件问题，因为我们必须制造出规模足够大、稳定性足够高的物理系统，使其成为一台真正的通用量子计算机，而不仅仅是一套停留在实验室里的“量子仪器装置”。
我们目前正处于所谓的 NISQ（含噪声中等规模量子，Noisy Intermediate-Scale Quantum）时代。
Original English
**John Platt**: I think sort of in between. It's not a purely software thing because we need to build systems that are large enough and stable enough to be able to be a quantum computer instead of a quantum apparatus. We're in the what they call the NISK era...
**主持人**：中等规模量子。
Original English
**Host**: Intermediate scale quantum...
**John Platt**：这个词我想是由 John Preskill 提出的。虽然这个命名并不尽如人意，但行业内已经普遍沿用了。
在现阶段，Google 量子团队发表了一篇非常精彩的论文，我也是众多合著者之一。论文提出了一种名为“量子回波算法”（Quantum Echoes Algorithm）的方法，它有望在现阶段就派上用场。从本质上讲，这种思路继承了费曼最初的设想——其专业术语是指：利用量子系统拟合核磁共振（NMR）等实验观测数据中的哈密顿量（Hamiltonian）。
具体来说，就是构建一个参数化的物理模型，利用量子计算机在闭环迭代中调整参数，反推出精确的系统状态，从而实现诸如解析 NMR 复杂参数的功能。因此，这是一项非常具象且具备实用价值的应用。
Original English
**John Platt**: ...which was I think coined by John Preskill. It's not a very good term, sorry, but I guess that's the term we have. And the quantum team made this wonderful paper, which I think I'm co-author on, one of many, for something called the Quantum Echoes Algorithm, which could be applicable now. Fundamentally it's in the style of Feynman's proposal, which essentially—the technical term is—you could try to fit a Hamiltonian to observed data like an NMR. What that means is you have a physical model that's parameterized, and you use the quantum computer to kind of adjust the parameters and try to figure out inside of a loop what the right... so you could, for example, decode NMR parameters. So that is a practical thing that's used.
### 实用量子计算的现实挑战与时间表
**John Platt**：现在的关键问题在于：这些系统的规模是否足以支撑颠覆性的科学突破？这仍有待进一步验证。
谷歌的量子计算团队一直表现优异，他们极其出色地推进着 Hartmut Neven 几年前制定的技术路线图。团队正坚定不移地沿着既定方向向前迈进，逐步扩大系统规模。当他们攻克路线图上的最后一个里程碑时，就应该能够交付一台能够完成惊人任务的量子计算机。
目前他们仍在持续推进这一进程。所以在我看来，这个时间跨度大概在几年到若干年之间。我并没有时刻跟踪他们对外公布的具体时间节点，所以具体何时实现，建议去深入了解团队的最新表态。但无论如何，他们正在稳扎稳打地推进。
现在真正引人关注的问题在于：超导量子计算最终会成为胜出者，还是其他某种替代性的技术路径会后来居上？这依然充满悬念。不过我个人仍然认为超导路线极具前景，因为它具备极高的工程可扩展性。
所以，我认为实现实用化绝不需要等上三十年，但也绝非一朝一夕之功。我不认为短期内会出现断崖式的突跃，哪怕硬件底层突然出现重大突破也是如此。因为量子计算系统极其脆弱精细，即使有人提出了极其惊艳的构想，要将其工程化落地仍需假以时日。毕竟在当前的 NISQ 时代及未来相当长的一段时间里，这些机器本质上都是模拟计算机（analog computers），这决定了它们极其敏感脆弱。因此，我不认为行业会突然发生翻天覆地的跃迁。
目前还没有人彻底解决如何在任意规模下让量子比特保持良好相干互联的难题，现有系统大多仍停留在 100 到 200 个物理量子比特的量级……
Original English
**John Platt**: Now the question is, will it be big enough to make breakthroughs? That's still TBD. The quantum team at Google has been executing amazingly well against a roadmap that Hartmut Neven laid out a few years ago, and they're just continuing to march down this thing where they're scaling up. And when they hit their last milestone, they should be able to have a quantum computer that does amazing things, and they're continuing to march it along.
So yeah, I think on the scale of a few to several years. I haven't sort of kept up on exactly what date they're saying, so you should ask hard exactly when that's going to happen. But yeah, they're marching along.
So the main, the interesting question is will the superconducting computing be the winner, or will one of the other sort of alternate technologies? And that's still TBD. I still think superconducting is a very promising thing because it is very scalable. So yeah, no, I don't think it's 30 years away. I don't think it's tomorrow. I don't think unless—well, even if there's a sudden hardware breakthrough, these are very finicky things. So someone might have a brilliant idea, but it will still be a while before... Cuz fundamentally, at least in the NISK era or for a while, these are fundamentally analog computers and so they tend to be very, very, very finicky. So I wouldn't expect all of a sudden some phase change happens. [clears throat] No one has figured out how to scale up qubits in a way that interact in just to kind of arbitrary size, they're still on the order of 100, 200 qubits I think, or...
**主持人**：你是从具体哪个角度来衡量这一点的？
Original English
**Host**: You mean in terms of... Okay, so... In terms of...
**John Platt**：对，这有点像是……没错，确实如此。所以对于超导量子比特而言，大家虽然……
Original English
**John Platt**: Well, it's a little like... Yeah, yeah, exactly. So for superconducting qubits, people—I mean, although...
### 量子纠错与超导量子比特的工程挑战
**John**: 人们正在努力攻关这个问题。就达到目标所需的错误率而言，物理量子比特与逻辑量子比特之间的比例目前仍然相对偏高。因此，也许在这个方向上会出现重大突破，我也不好说；或者也可能存在其他技术路径，并不需要那么高的比特比例。因为在很大程度上，这一比例与芯片的二维连通性直接相关——你是在二维芯片上排布这些量子比特的。而像中性原子这类的体系，理论上它们可以在任意两点之间实现全互联。但当然，在工程实践中我们目前还不能完全确定，也并不真正清楚中性原子路线的技术极限在哪里。至少我个人还不清楚它的极限所在，也许顶尖的领域专家们心里有数。
Original English
**John**: People are working on it. The ratio of the number of physical qubits to logical qubits is still relatively large for the error rates that you need. And so maybe there'll be a breakthrough there, I don't know, or there are other things that might not need that ratio to be so high, because a lot of it has to do with the 2D connectivity of the chips, that you lay out your qubits on 2D chips. And for things like neutral atoms, they in theory can connect anything to anything else, but of course in practice we don't really know, and we don't actually know what the limitations of neutral atoms are. At least I don't know the limitations of what neutral atoms are, maybe the super experts know.
**Host**: 特别是超导量子比特，它内部似乎天然存在着某种张力：一方面，你希望拥有非常干净整洁的谐振器，而要获得干净的谐振器，你就必须将其与外界环境解耦；但另一方面，为了获得优良的相互作用，你又必须让谐振器之间发生耦合，而这种耦合本身就不可避免地会牵涉到环境噪声。所以这看起来像是一个……
Original English
**Host**: Superconducting qubits in particular have this sort of tension where you want to have, you know, nice clean resonators, which you get a clean resonator by decoupling from the environment, and then you get good interactions by coupling resonators, which involves coupling to the environment. So it seems like...
**John**: 确实如此。但至少在现实中，虽然存在整个外部环境的干扰，但你其实可以通过开启一些极其微小的通道，只精准地连接到相邻的比特上。
Original English
**John**: True, but at least, yes, but there's like the whole environment, and then there's like little tiny holes that you want to go through to your neighbors.
**Host**: 所以这并不是某种像测不准原理那样根本性的物理定律限制，而更多是工程技术层面的限制与挑战。
Original English
**Host**: This is not like a fundamental uncertainty relationship or something. It's like this is a technological limitation or difficulty or something.
**John**: 没错。谷歌量子团队的硬件团队技术非常非常精湛，水平极其高超。因此，他们在提出这些精妙的设计，并切实将它们变为现实方面真的非常擅长。我对他们的成果感到由衷的敬佩。
Original English
**John**: Yeah, the hardware team in Google Quantum is very, very skilled. They're very, very skilled. So, yeah, they're really good at making these designs and making these things actually work. So, I find them impressive.
**Host**: 确实，看到这项技术的持续演进令人十分激动。
Original English
**Host**: Yeah. It'd be exciting to see that advance.
**John**: 是的，非常令人期待。
Original English
**John**: Yeah. Yeah. Yeah.
**Host**: 看来我们目前也只能屏住呼吸，拭目以待了。
Original English
**Host**: Yeah. Um, I guess we'll just have to hold our breath and wait.
### 提前二十年布局与卷积神经网络的命名
**John**: 耐心等待就好。我想我是一个非常有耐心的人，所以往往很早就开始着手研究了。通常是提前二十年……
Original English
**John**: Just wait. Yeah. I guess I'm just very patient, so I just start working, so 20 years ahead of...
**John**: 负责谷歌量子软件团队的戴夫·培根（Dave Bacon）经常拿这事调侃我。大概十年前，他就打趣我对我说：“约翰，天哪，我现在可做不了量子计算，因为你总是超前时代整整二十年，所以我还得等上二十年呢。”结果你看，现在十年已经过去了，正如大家开玩笑说的那样……
Original English
**John**: That's what Dave Bacon, who runs the software team in Google Quantum, always teases me like: "John, like, oh no, I can't work in quantum computing," was like 10 years ago, "because you're always 20 years ahead of time, and so I have to wait for 20 years." Like, okay, well, you know, 10 years have gone by, like you... [laughter]
**Host**: 已经走完一半路程了！
Original English
**Host**: Halfway there.
**John**: 走完一半了！当然，我并没有把二十年当作什么绝对的时间定律。不过话说回来，我也是在十年前就开始投身受控核聚变研究了，我们拭目以待吧。[笑声]
Original English
**John**: Halfway there. I don't take that as an absolute, yeah. But I also started working on fusion 10 years ago. We'll... Yeah. Or... [laughter]
**Host**: 也许你本身就是原因所在——就像是你只要开始涉足某个前沿方向，客观现实随后就会顺理成章地追赶上来。
Original English
**Host**: Maybe you're actually causal, like you start working on something and reality just, you know, catches up, catches up.
**John**: 也许多多少少有一点吧，谁知道呢？这我也说不准。其实在90年代初，我就非常喜欢并开始研究卷积网络了。
Original English
**John**: I guess so. Maybe. Who knows? I don't know. I started working—I loved convolutional nets in the early '90s.
**Host**: 扬·勒昆（Yann LeCun）声称的那些事实证明确实立得住脚。
Original English
**Host**: Yann LeCun claimed... checks out.
**John**: 据我所知，“卷积网络”（convolutional net）这个术语其实是我创造的。据目前能考证的情况看，事实应该确实如此。因为当时所有人都在管它叫“LeNet”，那是一个非常具体、特定的网络架构，而且大家当时都为扬工作。但我当时对他们说：“我又不是扬的手下，我可不想叫它 LeNet。”于是我就称它为“卷积网络”。不管怎么说，这是一个更加通用的学术名词，也是个很贴切的好名字。
Original English
**John**: I coined the term "convolutional net." As far as I can tell, that may be true. Because everyone called it LeNet because it was a very specific thing. They all worked for Yann. I said, "Well, I don't work for Yann. I don't want to call it LeNet." So I called it a convolutional net. So I don't know, that was a more generic term. It's a good term.
### AI赋能科学研究的未来愿景
**Host**: 确实是一个经典术语。在节目结束之前，您有什么特别希望广大听众和观众带走的思考吗？或者有什么想要传达的核心信息？
Original English
**Host**: Yeah. Before you go, is there anything you want the audience to take away? Any messages you want to deliver?
**John**: 我觉得我们刚才讨论的科学发现中的误差与不确定性就是很好的例子。但我内心深处真正感到无比兴奋的，是人工智能在科学领域的巨大潜力。我认为它对科研工作者而言，将成为一件不可思议的强大杠杆与利器。我坚信科学家绝不会被AI取代；事实上我更希望看到的是，科学家们能够把所有的精力和时间重新倾注在真正具有创造性的工作上、严谨求证的研究上，以及科学哲学的深层次思辨上。所以我坚信，未来的科学发展将会变得极其酷炫。
Original English
**John**: I think error is an example of it. But I'm really amazingly excited about the potential of AI for science. I think it's going to be an amazing power tool for scientists. And I think scientists won't be replaced. In fact, I'm hoping that they'll spend all their time on again the creative stuff, on the rigorous stuff, on the philosophy stuff. And so I think it's going to be way cool.
**Host**: 我也由衷地希望如此。我的直觉告诉我，虽然我也听过不少人持有这种乐观论调，但但愿他们都是对的，但愿这不仅仅是人类面对令人不安的现实时的一种心理应对机制。[笑声] 另外还有一个我们差点忘了问的核心问题：如果您能凭借一道绝对的意志、像施展魔法一样直接消除本行业中的某一个瓶颈——无论您如何定义这个行业——那会是什么？
Original English
**Host**: I hope so, too. Yeah. No, yeah. My intuition is that—I've heard a lot of people say that. I hope they're right. I hope it's not just a sort of coping with [laughter] a reality that's uncomfortable. The other question we almost forgot to ask: If you could remove a bottleneck in your industry, however you want to define that, by fiat, like magic...
**Host**: 没错，如果真有魔法，那会是怎样的一个愿望？
Original English
**Host**: Yeah. By magic, what would that be?
### 终极愿望：“全能实验室”与机器人难题
**John**: 如果真能给我一个魔法许愿的机会，我会许愿说：请某位高人造出一个“全能实验室”（the everything lab）吧！你只需要向它发送一段 JSON 数据包，它就能自动完成世间任何你想要的实验。
Original English
**John**: If I could get a magic wish, I would say: someone please make the "everything lab" that you could like send a JSON blob to, and it will do any experiment at all.
**Host**: 好的，一个完全自动化的实验室……
Original English
**Host**: Okay. Automated...
**John**: 但关键在于它必须能做“任何实验”。所以从本质上讲，这就意味着我们必须彻底攻克某种具有“AI完全性”（AI-complete）的通用机器人学难题。但如果我们真的做到了这一点，那将产生无与伦比的颠覆性影响。因为在现阶段，像气候模型 ERA 这类的研究全都是纯计算层面的；现实中必须得有人辛辛苦苦去现实世界采集和汇总数据。
Original English
**John**: But it'd have to be anything. So essentially, I guess we have to solve the sort of AI-complete robotics problem, I guess. But if we did, then that would be stunning, because right now things like ERA, it's all computational. So someone has to gather the data.
**Host**: 没错，如果我们能够打破这个数据采集的瓶颈，那简直太不可思议了，绝对是令人叹为观止的飞跃。
Original English
**Host**: So yeah, if we could just break that. Oh, that would be so amazing. That would be so utterly amazing.
### 结语与道别
**Host**: 太棒了。非常感谢您今天特意抽出宝贵的时间来与我们交流。据我所知，您似乎是在飞行的途中专门调整了自己的行程……
Original English
**Host**: Yeah. Um, great. So I really appreciate you taking the time to see us. Um, and I think you kind of flew in and adjusted your schedule a little bit to...
**John**: 没错，我原本正要在回家途中飞经旧金山上空，于是我便临时决定在旧金山降落。
Original English
**John**: Yeah, I was sort of flying over San Francisco to get home, and so I said I landed in San Francisco.
**Host**: 所以您真的是付出了极大的努力来到现场。我们由衷地感激，今天能和您交谈真的非常愉快。
Original English
**Host**: So you made a big effort to be here. We really appreciate that. It was really fun to talk to you.
**John**: 嗯，确实如此。
Original English
**John**: Mhm.
**Host**: 这真是一场酣畅淋漓的精彩对话。
Original English
**Host**: Yeah. Listen, it's been a blast. Yeah.
**John**: 好的，太棒了，非常感谢你们邀请我。
Original English
**John**: Okay, cool. Thank you for having me.
**Host**: 不客气，这是我们的荣幸。
Original English
**Host**: Yeah, you're welcome.