Claude-shaped science“Claude 风格”的科学

Summary: In this guest post, Prof. Matthew Schwartz returns to describe a new approach to AI-accelerated science. In Vibe Physics, Schwartz discussed similarities in capability between Claude and a physics graduate student. Here, he describes what happened when he stopped fighting Claude and allowed Claude to find “Claude-shaped” problems: ones best suited to the capabilities of the current generation of LLM tools. This led him to build BootLoops, a toolkit for exact calculations in quantitative science. Because similar calculations often turn up across very disparate areas of science, Claude found connections to ecology, population genetics, and a dozen other fields. These connections were often technically correct but scientifically unremarkable at first, so Schwartz worked with domain experts to steer BootLoops toward questions those fields care about. Below, we share more about these projects and how BootLoops came about.摘要：在本篇客座文章中，Matthew Schwartz 教授回归，介绍了一种 AI 加速科学研究的新方法。在《Vibe Physics》一文中，Schwartz 探讨了 Claude 与物理学研究生在能力上的相似之处。本文中，他描述了当他不再试图“对抗”Claude，而是允许 Claude 去寻找“Claude 风格”的问题——即最适合当前一代大语言模型（LLM）工具能力的问题——时所发生的情况。这促使他构建了 BootLoops，这是一个用于定量科学精确计算的工具包。由于相似的计算往往出现在截然不同的科学领域，Claude 发现了其与生态学、群体遗传学及其他十几个领域之间的联系。这些联系起初在技术上是正确的，但在科学上并不引人注目，因此 Schwartz 与领域专家合作，引导 BootLoops 转向这些领域所关注的问题。以下我们将分享更多关于这些项目的内容，以及 BootLoops 的诞生过程。

Agentic AI is improving rapidly. Everyone notices the models seem smarter: they know more, make fewer mistakes, and have better ideas. If you follow the trend lines, it is easy to speak confidently about the potential of AI to revolutionize science. However, academic scientists trying to use the models today in their own work often feel a disconnect. The models may be solving challenging and longstanding problems, but so far these have mostly been well-scoped applications of existing techniques. Many of the headlines seem to be in mathematics, the one part of science where a problem can be stated completely and an answer checked absolutely. But most of science is not like that. And for many of us researchers and students, the distance between those headlines and what happens when we use these models can leave us frustrated and anxious.代理式 AI 正在迅速进步。每个人都注意到模型似乎变得更聪明了：它们了解更多知识，犯错更少，并能提出更好的想法。如果你跟随趋势线，很容易自信地谈论 AI 彻底改变科学的潜力。然而，今天尝试在自己的工作中应用这些模型的学术科学家们，往往会感到一种脱节。这些模型可能正在解决具有挑战性的长期问题，但到目前为止，这些大多只是现有技术的良好应用。许多头条新闻似乎都集中在数学领域，这是科学中唯一一个可以完整陈述问题并能绝对核对答案的分支。但大部分科学并非如此。对于我们许多研究人员和学生来说，这些头条新闻与我们实际使用这些模型时的体验之间的距离，可能会让我们感到沮丧和焦虑。

The core conflict, as I see it, is that although these models are brilliant, working like a human scientist is not what current LLMs do best. Claude and GPT are good at science, but they are not scientists: yes, they are smart, but it can take a lot of hand-holding to get them to produce anything of scientific value. Physicists would call this an “impedance mismatch”: two systems that each work fine but are poorly matched, so most of what one puts in never gets through to the other. Here, the mismatch is between what scientists want and what AI does well. 在我看来，核心矛盾在于：尽管这些模型非常出色，但像人类科学家那样工作并不是当前大语言模型最擅长的事情。Claude 和 GPT 虽然擅长科学，但它们并不是科学家：是的，它们很聪明，但要让它们产出任何具有科学价值的成果，往往需要大量的引导。物理学家会将此称为“阻抗失配”（impedance mismatch）：两个系统各自运行良好，但匹配度很差，导致一个系统输入的大部分内容无法有效地传递给另一个系统。这里，失配存在于科学家想要的东西与 AI 擅长做的事情之间。

So how can we fix it?那么，我们该如何解决这个问题？

I started looking for examples where the impedance mismatches are less acute. I began by having Claude build an accessible suite of tools for mathematical physics. Before long, the tools found uses for other problems. This iterative process generated a set of software and scientific protocols, which I call BootLoops. BootLoops functions as a kind of harness for the LLM, much like Claude Code or Claude Science is a harness for Claude, or Codex is a harness for GPT. I’ve found BootLoops especially well-suited for a class of quantitative problems in science. It is also open-source, so it can be used with whatever model you like.我开始寻找那些阻抗失配不那么严重的例子。我首先让 Claude 构建一套易于使用的数学物理工具。不久之后，这些工具就在其他问题中找到了用途。这个迭代过程产生了一套软件和科学协议，我将其称为 BootLoops。BootLoops 的功能类似于大语言模型的“套件”，就像 Claude Code 或 Claude Science 是 Claude 的套件，或者 Codex 是 GPT 的套件一样。我发现 BootLoops 特别适合科学中的一类定量问题。它也是开源的，因此可以与你喜欢的任何模型配合使用。

Once Claude had BootLoops, it kept noticing the same pattern: many fields have problems that a technique from mathematics, physics, or computer science would solve outright if anyone knew it existed. I started calling these “Claude-shaped” problems, and followed them outside high-energy theoretical physics—my home turf—into geology, biology, economics, and linguistics. In these other areas, I could not rely on my own expertise to know whether what Claude found was interesting. So I found some experts and asked. With their guidance, BootLoops was able to make substantive advances in many research areas. 一旦 Claude 有了 BootLoops，它就不断注意到同一个模式：许多领域都存在一些问题，如果有人知道数学、物理或计算机科学中的某种技术存在，这些问题就能迎刃而解。我开始称这些为“Claude 风格”的问题，并跟随它们走出高能理论物理学（我的主场），进入地质学、生物学、经济学和语言学领域。在这些其他领域，我无法依靠自己的专业知识来判断 Claude 发现的东西是否有趣。所以我找到了一些专家并向他们请教。在他们的指导下，BootLoops 能够在许多研究领域取得实质性进展。

Below, I share how BootLoops came about, describe some early findings in areas where I have been applying it, and share some of my thinking on how to resolve the impedance mismatch between human and AI scientists today.下面，我将分享 BootLoops 是如何产生的，描述我在应用它的一些领域中获得的早期发现，并分享我对如何解决当今人类科学家与 AI 科学家之间阻抗失配问题的一些思考。

Claude, take the wheel!Claude，接手吧！

Last December, I tried using Claude as a research assistant, and found that Claude Opus 4.5 performed like a strong graduate student at 20 times the speed. Despite Claude producing a high-quality paper at the end of the experiment, it was a slog to get there. I had to correct every sentence it wrote, steer it away from irrelevant threads, and pull it back from dead ends.去年 12 月，我尝试让 Claude 担任研究助理，发现 Claude Opus 4.5 的表现相当于一名强力的研究生，但速度快了 20 倍。尽管在实验结束时 Claude 产出了一篇高质量的论文，但过程非常艰辛。我必须纠正它写的每一句话，引导它远离无关的思路，并将其从死胡同中拉回来。

This summer, I tried doing something different: instead of treating Claude like the collaborator I wanted it to be, I started to treat it like the collaborator it actually is. This required looking for problems suited to its strengths. Right now, Claude is just not able to help me with deep conceptual questions—but it does have a virtually unlimited breadth of knowledge across all domains, incredible coding skills, leading-edge knowledge of mathematics and statistics, and the ability to parse papers, appendices, and data at machine speed.今年夏天，我尝试了一些不同的做法：我不再试图把 Claude 当作我希望它成为的那种合作者，而是开始把它当作它实际上的那种合作者来对待。这需要寻找适合它优势的问题。目前，Claude 无法在深层的概念性问题上为我提供帮助——但它在所有领域都拥有几乎无限的知识广度、令人难以置信的编码技能、对数学和统计学的尖端了解，以及以机器速度解析论文、附录和数据的能力。

A natural place to start looking for Claude-shaped problems was in areas where coding could help. When Anthropic released Claude Fable 5 in Summer 2026, I wanted to see whether its cyber capabilities would translate to scientific computing. So I sought to test it by having it port, code up, and improve various methods from a handful of my papers and the adjacent literature on scattering amplitudes.寻找“Claude 风格”问题的自然切入点是编码能有所帮助的领域。当 Anthropic 在 2026 年夏季发布 Claude Fable 5 时，我想看看它的网络能力是否能转化为科学计算。因此，我试图通过让它移植、编码并改进我的一些论文以及散射振幅相关文献中的各种方法来进行测试。

Scattering amplitudes are how we interpret data from the Large Hadron Collider: smash two protons at 13 trillion electron volts, and the amplitude is the theoretical bridge between the debris and whatever new particle, a Higgs boson or something unknown, the collision produced. At their core are Feynman diagrams, multidimensional integrals of a particular form. The ones we are struggling with now can each be a PhD thesis, or occupy a group for years.散射振幅是我们解释大型强子对撞机数据的方式：以 13 万亿电子伏特的能量撞击两个质子，振幅就是碎片与碰撞产生的任何新粒子（如希格斯玻色子或某种未知粒子）之间的理论桥梁。其核心是费曼图，即特定形式的多维积分。我们现在正在攻克的每一个积分都可能是一篇博士论文，或者让一个研究小组忙碌多年。

Over the past 20 years an alternative has grown up: the S-matrix bootstrap. In the bootstrap approach, instead of grinding out the integral, you impose physical constraints until only one answer is possible. Knowing where the amplitude is infinite (its “singularities”) might narrow it to 20,000 options; a symmetry cuts that to 500; and so on down to one. The traditional bootstrap is purely analytic and has gone furthest in the most symmetric theories, where the constraints reach all the way to a single option (the nine-loop amplitude in N=4 super-Yang–Mills is an example). Closer to the real world you often run out of constraints before the end. A newer pivot, the semi-numerical bootstrap, closes the gap when you can also compute the amplitude at a handful of points to absurd precision (sometimes 1,000 digits): if few enough options remain, those numbers pin down the remaining coefficients exactly.在过去的 20 年里，一种替代方法兴起了：S 矩阵自举（S-matrix bootstrap）。在自举方法中，你不是死记硬背地计算积分，而是施加物理约束，直到只剩下一个可能的答案。知道振幅在何处无穷大（其“奇点”）可能会将其范围缩小到 20,000 种选项；对称性将其削减到 500 种；依此类推，最终只剩一种。传统的自举方法纯粹是解析性的，在对称性最强的理论中走得最远，约束条件可以直接推导至单一选项（N=4 超对称杨-米尔斯理论中的九圈振幅就是一个例子）。在更接近现实世界的情况下，你往往会在得出结果前就耗尽了约束条件。一种较新的转向，即半数值自举（semi-numerical bootstrap），可以在你也能以极高精度（有时达到 1,000 位数字）计算出振幅的几个点时弥补这一差距：如果剩下的选项足够少，这些数字就能精确确定剩余的系数。

The semi-numerical bootstrap seemed ideal for agentic AI. It draws on mathematics, physics, and computer science that no one person has mastered; it needs a great deal of coding and algorithm development; and it is checkable, since the same numerics let anyone, expert or not, verify the final answer against the integral to as many digits as they like by running two scripts. The community's expertise is also unevenly distributed: good ideas sit in Wolfram Language, C++, Python, or Julia, and many more sit in papers with no code at all. So my first assignment for Fable 5 was to port all of it to a common framework, and to write the code the papers never provided.半数值自举似乎非常适合代理式 AI。它借鉴了没有任何一个人能完全掌握的数学、物理和计算机科学知识；它需要大量的编码和算法开发；而且它是可核验的，因为相同的数值计算让任何人（无论是否是专家）都可以通过运行两个脚本，将最终答案与积分进行任意位数的比对。该领域的专业知识分布也不均匀：好的想法存在于 Wolfram 语言、C++、Python 或 Julia 中，而更多的想法则存在于根本没有代码的论文中。因此，我给 Fable 5 的第一个任务是将所有这些内容移植到一个通用的框架中，并编写论文中从未提供的代码。

Claude did this effortlessly. I was surprised when it reproduced the results from my paper in around 20 minutes, while the code I wrote to do it took me weeks. However, I was not surprised when it informed me that I was doing something very inefficiently and that there was a better algorithm I was unaware of. Claude 轻而易举地完成了这项工作。令我惊讶的是，它在大约 20 分钟内就重现了我论文中的结果，而我编写同样的程序则花了数周时间。然而，当它告诉我我的做法效率极低，并且存在一种我不知道的更好算法时，我并不感到惊讶。

Then, I asked Claude to search for unsolved amplitudes it could compute. It turns out that problems that are simple enough for the S-matrix bootstrap are also simple enough for humans to do—and, indeed, most have been done. It nevertheless found a few unsolved problems. After some discussion with the model, it became clear that Claude was limiting itself to amplitudes with the simplest family of functions: logarithms. So I asked, could it do the same thing for the next simplest family, elliptic functions?接着，我让 Claude 搜索它能计算的未解振幅问题。事实证明，对于 S 矩阵自举来说足够简单的问题，对于人类来说也足够简单——实际上，大多数已经完成了。尽管如此，它还是发现了一些未解问题。在与模型进行了一些讨论后，很明显 Claude 将自己限制在具有最简单函数族（对数函数）的振幅上。所以我问，它能否对下一个最简单的函数族——椭圆函数——做同样的事情？

Elliptic integrals are really hard, even for people like me who spend a lot of their time computing integrals. Only a handful of elliptic Feynman integrals have ever been computed, and none completely by the bootstrap, at least to my or Claude’s knowledge. The issue is not that the methods wouldn’t work, but rather that nobody had tried, since the expertise needed to do so is distributed among many humans. Claude, by contrast, easily generalized all of the machinery it had ported and built for the logarithmic case to these other integral classes. This time it wrote most of the software itself, or borrowed it from mathematics rather than physics. As the toolkit grew, it started to land one integral after another. Soon we had 30 integrals BootLooped from end to end, comprising 15 reproductions of known results by this new method and 15 that had never before been computed.椭圆积分确实很难，即使对于像我这样大部分时间都在计算积分的人来说也是如此。据我或 Claude 所知，只有少数几个椭圆费曼积分被计算出来过，而且至少没有一个是完全通过自举法计算出来的。问题不在于这些方法行不通，而是没有人尝试过，因为完成这项工作所需的专业知识分散在许多人手中。相比之下，Claude 轻松地将它为对数情况移植和构建的所有机制推广到了这些其他积分类别。这一次，它自己编写了大部分软件，或者从数学而非物理学中借用了代码。随着工具包的增长，它开始一个接一个地攻克积分。很快，我们就通过 BootLoops 完成了 30 个积分的端到端计算，其中包括用这种新方法重现的 15 个已知结果，以及 15 个从未被计算过的结果。

That was all after only a few weeks. Initially, I thought I would be satisfied just to do a write-up on that, but I was too tempted to see what else we (that is, me and Claude with BootLoops) could do.这一切仅在几周后就完成了。起初，我以为只要写一份报告就够了，但我太想看看我们（即我和使用 BootLoops 的 Claude）还能做些什么。

“I know Kung Fu”“我会功夫了”

A serendipitous feature of science is that the same equations often appear over and over again in different contexts. In physics, for example, the diffusion equation, Fokker-Planck equation, and Schrödinger equation all have the same mathematical form, so if you develop a method to solve one problem, you can often apply it to many others. I knew that computations BootLoops was good at were relevant elsewhere: in cosmology and string theory, for example. What I didn’t know, but Claude was happy to tell me, was that these integrals could also map onto Bayesian evidence integrals in population genetics, or that the finite-field methods used for Feynman integral reduction could also apply to problems in evolutionary biology.科学的一个偶然特征是，相同的方程经常在不同的背景下反复出现。例如，在物理学中，扩散方程、福克-普朗克方程和薛定谔方程都具有相同的数学形式，因此如果你开发出解决其中一个问题的方法，通常可以将它应用于许多其他问题。我知道 BootLoops 擅长的计算在其他领域也很重要：例如在宇宙学和弦理论中。我所不知道、但 Claude 很乐意告诉我的是，这些积分也可以映射到群体遗传学中的贝叶斯证据积分，或者用于费曼积分简化的有限域方法也可以应用于进化生物学中的问题。

This kicked off a particularly fertile period of searching for and solving Claude-shaped (and, more narrowly, BootLoops-shaped) problems. Some ideas were immediately obvious as great applications. For example, the science of phylogenetics studies how to organize species into family trees using DNA; the relevant computation is a Bayesian evidence integral, whose output is a single number saying how well a candidate tree accounts for the observed DNA, an integral of the same kind BootLoops was originally built to do. As the ideas came, I insisted that Claude both use old tools and build new ones, so that the harness would grow. Each tool, even from a failed project, opened up new doors. Claude was getting more and more capable. It was like Neo in The Matrix after waking up from the Kung Fu download.这开启了一个寻找并解决“Claude 风格”（更窄一点说是“BootLoops 风格”）问题的特别肥沃的时期。有些想法立刻显现为绝佳的应用。例如，系统发生学研究如何利用 DNA 将物种组织成家族树；相关的计算是一个贝叶斯证据积分，其输出是一个单一数字，说明候选树解释观察到的 DNA 的程度如何——这正是 BootLoops 最初构建时所做的那种积分。随着想法的涌现，我坚持要求 Claude 既使用旧工具又构建新工具，以便让这个套件不断成长。每一个工具，即使是来自失败的项目，也打开了新的大门。Claude 变得越来越有能力。这就像《黑客帝国》中的尼奥从功夫下载中醒来一样。

As the projects drifted from my professional comfort zone, however, I started to worry. When Claude claims something it did in my field is fantastic, I can judge whether that’s true or not (it often isn’t). But when it claims something it did in another field is fantastic, I find myself agreeing. My suspicion heightened, I knew I needed to bring in some experts to be sure. Indeed, I found that in almost all cases, Claude was technically correct, but the result was not all that interesting until the expert helped steer us.然而，随着项目偏离我的专业舒适区，我开始担心起来。当 Claude 声称它在我的领域所做的事情非常出色时，我可以判断这是否属实（通常不是）。但当它声称它在另一个领域所做的事情非常出色时，我发现自己竟然表示赞同。我的怀疑加深了，我知道我需要引入一些专家来确定。事实上，我发现几乎在所有情况下，Claude 在技术上都是正确的，但在专家帮助我们引导之前，结果并不那么令人感兴趣。

One such project was on neutral biodiversity theory in ecology. In any ecosystem, some species thrive while others die off. Neutral theory asks how much of that life history is due to random chance. Building on earlier work, in 2001, the ecologist Stephen Hubbell suggested, provocatively, that perhaps it was all random. In 2005, the ecologist Rampal Etienne put an equation to Hubbell’s theory allowing it (at least in principle) to be tested in a precise way. Unfortunately, for 20 years nobody could solve the equation at scale. Claude recognized Etienne’s equation as BootLoops-shaped, and solved it. Applying the computation to data, we established that in ecology’s most studied forest on earth, Barro Colorado Island in the Panama Canal, the mix of tree species changes 4.5 times faster than neutral theory allows.其中一个项目是关于生态学中的中性生物多样性理论。在任何生态系统中，一些物种繁荣而另一些物种消亡。中性理论探讨了这种生命史中有多少归因于随机偶然。在早期工作的基础上，生态学家 Stephen Hubbell 在 2001 年提出了一个挑衅性的观点，认为也许这一切都是随机的。2005 年，生态学家 Rampal Etienne 为 Hubbell 的理论提供了一个方程，使其（至少在原则上）能够以精确的方式进行测试。不幸的是，20 年来没有人能够大规模地求解这个方程。Claude 识别出 Etienne 的方程是“BootLoops 风格”的，并解决了它。将该计算应用于数据，我们确定在地球上研究最深入的森林——巴拿马运河的巴罗科罗拉多岛，树种组合的变化速度比中性理论允许的速度快 4.5 倍。

Excited by this result, I brought it to James O’Dwyer, a professor in plant biology and an expert in neutral theory. James was patient with me. Though he was impressed by the technical feat, he said the results would likely “be met with a shrug by many ecologists.” Ecologists had already observed, more qualitatively, that neutral theory can’t keep up with real forests. But James had a better idea: subtract the neutral prediction and study the remainder. This would give us a better sense of what changes were attributable to natural selection, competition, and the differences between species.受此结果的鼓舞，我将其带给了植物生物学教授、中性理论专家 James O’Dwyer。James 对我很有耐心。虽然他对这一技术成就印象深刻，但他表示，这些结果很可能会“被许多生态学家耸肩以对”。生态学家们已经更定性地观察到，中性理论无法跟上真实森林的变化。但 James 有一个更好的主意：减去中性预测并研究剩余部分。这将使我们更好地了解哪些变化归因于自然选择、竞争以及物种之间的差异。

James and I proceeded to work intensively with Claude to craft a successor model. James has a physics background, so the disciplinary language barrier was more easily surmountable, but, like many scientists, he had not yet really appreciated the power of agentic AI. As the collaboration progressed, I became a sort of Claude handler, translating Claude-speak to James and keeping the model on track, while James pushed the model to produce something ecologists might value. The final result is something all three of us are proud of: a minimal predictive model of life histories in excellent agreement with data. We are currently extending the model from Panama to other global forest plots, using datasets that Claude has helped curate.James 和我继续与 Claude 深入合作，构建了一个后续模型。James 有物理学背景，因此学科语言障碍更容易克服，但像许多科学家一样，他之前并没有真正体会到代理式 AI 的力量。随着合作的进展，我成了某种 Claude 处理员，将“Claude 语言”翻译给 James，并让模型保持在正轨上，而 James 则推动模型产出生态学家可能重视的成果。最终的结果是我们三个人都引以为豪的：一个与数据高度吻合的生命史最小预测模型。我们目前正在利用 Claude 协助整理的数据集，将该模型从巴拿马扩展到全球其他森林样地。

Another project involved looking at exact calculations in population genetics, a field that studies how genes vary within a population and why. Claude first used methods imported from mathematical physics to solve a 30-year-old integral expression for how natural selection shapes rare mutations. We applied it to gnomAD, the largest public catalog of human genetic variation. Claude was very excited about this result, but I wasn’t so sure. I had to write to three different biologists for validation before one responded, but at last I was able to recruit my colleague Michael Desai, who works in the field, though not on this exact problem. As with James, Michael was impressed by the technical result but not compelled by the science.另一个项目涉及研究群体遗传学中的精确计算，这是一个研究基因如何在群体内变异及其原因的领域。Claude 首先利用从数学物理学中引入的方法，解决了一个关于自然选择如何塑造稀有突变的 30 年前的积分表达式。我们将其应用于最大的公共人类遗传变异目录 gnomAD。Claude 对这个结果非常兴奋，但我不太确定。我不得不写信给三位不同的生物学家进行验证，才有一位回复，但最终我能够招募到我的同事 Michael Desai，他在该领域工作，尽管不是针对这个确切的问题。和 James 一样，Michael 对技术结果印象深刻，但并未被其科学意义所折服。

Michael noted, however, that more impactful results might come from studying correlations between pairs of mutations on a single chromosome, using the same or similar methodology. Within this scope, Claude found and built new tools, like the inequality certificates used for computer-assisted proofs in mathematics, and added them to the BootLoops kit. We then analyzed 5.7 billion pairs of nearby mutations in genomes from the 1000 Genomes Project, and found evidence for a mechanism called gene conversion. This is an important finding because nearly every analysis that uses linked genetic variation, from population history to disease mapping, ignores gene conversion. The bigger lesson is that AI can help find biology hidden in enormous genomic datasets, of which there are thousands sitting in public archives.然而，Michael 指出，如果利用相同或类似的方法研究单条染色体上成对突变之间的相关性，可能会产生更有影响力的结果。在此范围内，Claude 发现并构建了新工具，例如数学中用于计算机辅助证明的不等式证书，并将它们添加到了 BootLoops 工具包中。然后，我们分析了来自“千人基因组计划”基因组中的 57 亿对邻近突变，并发现了名为“基因转换”（gene conversion）机制的证据。这是一个重要的发现，因为几乎所有使用连锁遗传变异的分析（从群体历史到疾病映射）都忽略了基因转换。更大的教训是，AI 可以帮助在巨大的基因组数据集中找到隐藏的生物学信息，而这些数据集中有数千个正躺在公共档案中。

Other projects had a similar arc: Claude comes up with an initial finding I find compelling. I bring it to an expert, who is unmoved but sees the potential, and together we sculpt it into meaningful science. Interestingly, some of these drifted from BootLoops-shaped quantitative calculations to more general Claude-shaped work. For example:其他项目也有类似的弧线：Claude 提出了一个我认为很有说服力的初步发现。我把它带给专家，专家虽然不为所动，但看到了潜力，于是我们一起将其塑造成有意义的科学。有趣的是，其中一些项目从 BootLoops 风格的定量计算转向了更通用的 Claude 风格工作。例如：

• Economics. Economics journals now ask authors for a replication package, and an editor is often tasked with checking it. That is careful, manual work. Working with two economists, we built an AI data editor: it ported the packages of 4,452 papers in five leading journals from MATLAB, Stata, and other commercial tools to open-source code (some 30,000 routines) and checked essentially every validatable number against the published tables. The results appear in an NBER working paper.经济学。经济学期刊现在要求作者提供复制包，编辑通常负责检查。这是细致的手工工作。与两位经济学家合作，我们构建了一个 AI 数据编辑器：它将五大顶刊中 4,452 篇论文的包从 MATLAB、Stata 和其他商业工具移植到了开源代码（约 30,000 个例程），并对照已发表的表格检查了几乎每一个可验证的数字。结果发表在 NBER 的一份工作论文中。
• Linguistics. Word-stress catalogs are typically assembled by hand, with a few hundred languages, selection effects, and human preferences. Claude scoured every openly accessible source and, working with three linguists, we produced AccStack: a database of word stress covering 6,072 languages, quoting the deciding passage for nearly every entry, plus a bibliography of 160,000 phonology works.语言学。词重音目录通常是手工汇编的，涉及数百种语言、选择效应和人类偏好。Claude 搜罗了每一个公开可访问的来源，并与三位语言学家合作，我们制作了 AccStack：一个涵盖 6,072 种语言的词重音数据库，引用了几乎每一条目的决定性段落，外加一份 160,000 部音韵学著作的参考文献。

Some additional highlights, each of which was done in collaboration with experts, and each of which is undergoing further exploration and verification, include:一些额外的亮点，每一项都是与专家合作完成的，每一项都在进行进一步的探索和验证，包括：

• Phylogenetics. We made the Bayesian evidence for evolutionary trees fast to compute and exactly checkable, and found that single genes are often tied between competing trees by less than the error of standard sampling programs.系统发生学。我们使进化树的贝叶斯证据计算变得快速且可精确核验，并发现单基因在竞争树之间的关联度往往小于标准采样程序的误差。
• Earth science. We combined equations and data from atmospheric science, geochemistry, and climate modeling to build a quantitative, predictive model of how the Great Oxidation Event unfolded through four glaciations.地球科学。我们结合了大气科学、地球化学和气候模拟的方程与数据，构建了一个定量的预测模型，展示了“大氧化事件”是如何在四次冰期中展开的。
• Genomics. We studied the statistics of single-cell RNA counts at scale to characterize bursting rates and the departure from the textbook telegraph model of gene expression.基因组学。我们大规模研究了单细胞 RNA 计数的统计数据，以表征爆发率以及与教科书中基因表达电报模型之间的偏差。
• Sunspots. We deduced the lifecycle of sunspots using the methods of human population demographics, then extended the methodology to characterize starspots, a confounding factor in many exoplanet searches.太阳黑子。我们利用人类人口统计学的方法推导出了太阳黑子的生命周期，然后将该方法扩展到表征星斑，这是许多系外行星搜索中的一个混淆因素。
• Mathematical physics. We solved Watson’s “final problem”: the exact return probability of a 3D random walk with three unequal hopping rates, the last of the lattice integrals George Watson began in 1939.数学物理学。我们解决了沃森的“最终问题”：具有三个不等跳跃率的 3D 随机游走的精确返回概率，这是乔治·沃森自 1939 年开始研究的晶格积分中的最后一个。
• Cosmology. We built a complete theoretical toolkit to fit cosmological parameters to data on the universe’s large-scale structure, including the full two-loop power spectrum and the one-loop trispectrum.宇宙学。我们构建了一套完整的理论工具包，用于将宇宙学参数拟合到宇宙大尺度结构的数据中，包括完整的双圈功率谱和一圈三谱。
• Statistics. Mixture models are notoriously ill-behaved; we derived precise approximations to their Bayesian evidence that make model selection practical, with applications from biostatistics to textual analysis.统计学。混合模型以难以处理而闻名；我们推导出了其贝叶斯证据的精确近似值，使模型选择变得切实可行，并应用于从生物统计学到文本分析的各个领域。

Overall, this approach of iterative improvement of BootLoops, searching for Claude-shaped applications, and pushing to the scientific frontier with experts has been astonishingly productive. More information about these projects and others can be found on bootloops.ai.总的来说，这种迭代改进 BootLoops、寻找 Claude 风格应用并与专家共同推向科学前沿的方法，已经产生了惊人的成效。关于这些项目及其他项目的更多信息，可以在 bootloops.ai 上找到。

The technical details技术细节

Running many projects at once (36 manuscripts in 18 fields with 19 coauthors over three months, out of some 400 candidate problems) takes a lot of coordination, both for me and for Claude. The setup that makes this manageable is fairly simple. The Claude Code sessions all run in terminals on Google Cloud virtual machines, linked to various GitHub and Overleaf repos depending on my collaborators’ preferences. I have separate sessions for each project, plus a master session that coordinates the others, allocates compute, and validates results. Each session involves agents that run the computations in the background, with intermediate results stored in markdown files in their respective folders. These subagents are helpful because I regularly run into Fable 5’s classifiers; instead of these blocks corrupting my whole session, they just shut down a single agent. I also have separate sessions for writing results; creating the GitHub repos, tool manuals, and the BootLoops website; writing and validating all the separate tool codes; and checking and rechecking results as an adversarial referee.同时运行许多项目（三个月内在 18 个领域有 19 位合著者完成了 36 份手稿，从约 400 个候选问题中选出）需要大量的协调，对我个人和 Claude 来说都是如此。实现这一点的设置相当简单。所有的 Claude Code 会话都在 Google Cloud 虚拟机的终端中运行，并根据合作者的偏好链接到各种 GitHub 和 Overleaf 仓库。我为每个项目设置了单独的会话，外加一个协调其他会话、分配计算资源并验证结果的主会话。每个会话都涉及在后台运行计算的代理，中间结果存储在各自文件夹的 Markdown 文件中。这些子代理很有用，因为我经常遇到 Fable 5 的分类器；这些阻塞不会破坏我的整个会话，只会关闭单个代理。我还设有单独的会话用于撰写结果；创建 GitHub 仓库、工具手册和 BootLoops 网站；编写和验证所有单独的工具代码；以及作为对抗性裁判检查和复核结果。

Still, the sessions require a lot of guidance. For example, Claude has no sense of time, and loves to be dramatic. At the end of one project, it told me “Matt — the formula is solved. Two years of campaign, four deep marches, …”. It had only been working for three days. When I ask for ETAs, the estimates are always either way too long or way too short. I tried telling Claude not to commit to an estimate until it ran a test to confirm, but that didn’t work well either. Over time, I started to get a feel for how long things would take myself, since I couldn’t rely on Claude to tell me.尽管如此，这些会话仍需要大量的指导。例如，Claude 没有时间感，而且喜欢戏剧化。在一个项目结束时，它告诉我：“Matt，公式解开了。两年的征程，四次深入的探索……”。而它实际上只工作了三天。当我询问预计完成时间（ETA）时，估算总是要么太长要么太短。我试过告诉 Claude 在运行测试确认之前不要做出估算，但效果也不好。随着时间的推移，我开始自己摸索出事情需要多长时间，因为我无法指望 Claude 告诉我。

A bigger problem was that Claude’s default approach seemed to be to try to grind through a long, multiday calculation rather than build a new tool that would make that same calculation take mere minutes. Time and again, I had to tell it to think smarter, not harder. Over the course of these long projects, compaction would often kick in, causing Claude to lose important context. To guard against this, I had Claude periodically organize its files and consolidate them, so it always had access to the latest version of the plan. I’ve written protocol skills and other tools into the BootLoops harness to address this. Some of these problems have also been solved in good commercial harnesses, like Claude Science. But many are just the sorts of irritations inherent in the current generation of agentic AI that I expect the models will eventually outgrow.一个更大的问题是，Claude 的默认方法似乎是试图通过漫长的、持续数天的计算来解决问题，而不是构建一个新工具来让同样的计算只需几分钟。我一次又一次地告诉它，要思考得更聪明，而不是更努力。在这些长期项目的过程中，压缩（compaction）往往会发生，导致 Claude 丢失重要的上下文。为了防止这种情况，我让 Claude 定期整理其文件并进行整合，这样它总能访问到计划的最新版本。我已将协议技能和其他工具写入 BootLoops 套件以解决此问题。其中一些问题也在一些优秀的商业套件（如 Claude Science）中得到了解决。但许多问题只是当前一代代理式 AI 固有的烦恼，我预计模型最终会克服这些问题。

Here are a few more failure modes I encountered, and tips for dealing with them:以下是我遇到的一些其他失效模式，以及处理它们的技巧：

• Claude loves to declare victory. “Done, with one asterisk” is often “not done at all.” “Exactly that, with one refinement” usually means “no.” Giving Claude clear and rigid standards for what success looks like can help with this. On one project, it was proud of its proof, up to “one unproven lemma.” That lemma was the whole proof! I said no unproven lemmas. Then “done” again, but now with a new axiom.Claude 喜欢宣布胜利。“完成了，带一个星号”通常意味着“根本没完成”。“完全如此，带一个修正”通常意味着“不”。给 Claude 提供清晰且严格的成功标准可以对此有所帮助。在一个项目中，它对自己的证明感到自豪，直到“一个未被证明的引理”。那个引理就是整个证明！我说不要未被证明的引理。然后它又说“完成了”，但这次又带了一个新的公理。
• Look at everything yourself. I always ask to see plots. Even with all the monitors I set up, I’ve found the automated checks still can’t be trusted. Beware of qualitative claims like “good agreement.”凡事亲自查看。我总是要求查看图表。即使设置了所有的监控，我发现自动检查仍然不可信。要警惕诸如“良好的一致性”之类的定性说法。
• Question the conclusions. Claude is good at performing calculations, but the conclusions it draws can be wrong. Have it explain what it found until you believe it.质疑结论。Claude 擅长执行计算，但它得出的结论可能是错误的。让它解释它发现了什么，直到你相信它为止。
• Supply the taste. Claude can find Claude-shaped problems, but there are thousands, perhaps millions, of them. Claude seems to favor old debates, highly cited but long forgotten. Asking it to focus on what is newly possible, not what is merely faster or involves more digits, helps, but you still can’t trust its judgment of whether something is interesting.提供品味。Claude 可以找到 Claude 风格的问题，但这样的问题有成千上万，甚至数百万个。Claude 似乎偏爱古老的争论，那些被高度引用但早已被遗忘的话题。让它专注于什么是新可能，而不是仅仅更快或涉及更多数字的内容是有帮助的，但你仍然不能相信它对某事是否有趣的判断。
• Follow up. Once Claude has a great result, ask for an even greater one. It will always come up with something, and occasionally it’s brilliant.跟进。一旦 Claude 取得了伟大的成果，就要求它取得更伟大的成果。它总能想出点什么，偶尔还会非常出色。
• Watch out for the grind. Honestly, I never succeeded in getting Claude to estimate time well. Instead, I developed a sense of how long something interesting should take, and I learned to sense whether Claude was making what I considered progress on a sensible time scale. The model will grind forever if you let it.小心陷入死循环。老实说，我从未成功让 Claude 很好地估算时间。相反，我培养了一种对有趣的事情应该花费多长时间的感觉，并学会了感知 Claude 是否在我认为的合理时间范围内取得了进展。如果你任由它去，模型会永远死磕下去。

It also bears mentioning that these projects were compute- and token-intensive. Part of the expense, for me, came from trying to build tools that would be useful beyond any one project: connecting fields, and assembling a package that others could pick up. I hope the BootLoops harness can be used broadly across quantitative science without each user having to rebuild everything from scratch. Agentic AI makes it easier to connect ideas and share codebases, and if we learn to use that strength, the community can move faster than any individual could alone.值得一提的是，这些项目是计算和 Token 密集型的。对我来说，部分开销来自于试图构建在任何单个项目之外都有用的工具：连接各个领域，并组装一个其他人可以拿来就用的包。我希望 BootLoops 套件能够在定量科学领域得到广泛使用，而无需每个用户都从头开始重新构建一切。代理式 AI 使得连接思想和共享代码库变得更容易，如果我们学会利用这种优势，整个社区的进步速度将远超任何个人单打独斗。

Outlook展望

No matter how smart the models become, most scientific progress comes from real-world data that has to be acquired, understood, and checked, with each round informing the next question. AI can sit inside that loop, and that is genuinely valuable, but it does not collapse the loop to a point. Understanding is built up in increments, each one resting on the last; a cure for cancer or a fusion reactor will arrive that way too, not from a single prompt.无论模型变得多么聪明，大多数科学进步都来自于必须获取、理解和核验的现实世界数据，每一轮数据都会为下一个问题提供信息。AI 可以置身于这个循环之中，这确实很有价值，但它并不能将循环压缩成一个点。理解是逐步建立起来的，每一步都建立在前一步的基础上；癌症的治愈或核聚变反应堆也将以这种方式实现，而不是来自单一的提示词。

The focus on Millennium Prize Problems and Big Science may be, as Claude likes to say, a “footgun”: a feature that makes it easy to shoot yourself in the foot. Here, the risk is that unrealistic expectations may discourage productive uses that are already possible. Real progress will probably come the way it always has: by strengthening the foundations of scientific research and constructing increasingly sophisticated applications upon them. It will be sped up dramatically—as illustrated by the work I’ve described—but I don’t see any evidence or need to revisit the scientific method.对千禧年大奖难题和大科学（Big Science）的关注，正如 Claude 所言，可能是一个“footgun”（容易导致自残的功能）：一种让你很容易搬起石头砸自己脚的特性。这里的风险在于，不切实际的期望可能会阻碍已经可能的生产性应用。真正的进步很可能会像过去一样到来：通过加强科学研究的基础，并在其上构建日益复杂的应用。正如我所描述的工作所展示的那样，这一过程将被戏剧性地加速，但我没有看到任何证据或必要去重新审视科学方法本身。

A term I like that is sometimes used to discuss AI science is the convex hull. Take a shape and connect every pair of its points with a straight line; the new shape you get is the smallest convex shape that contains the original: its convex hull. Science is now very disjointed: there are jagged frontiers everywhere. Biology pokes out one way, math another. One lab might spend 20 years learning everything about a single set of genes with a single method they have mastered, while neighboring genes and alternative methods lie fallow. A harness like BootLoops can connect these points, filling in the hull.我喜欢的一个术语有时被用来讨论 AI 科学，即“凸包”（convex hull）。取一个形状并用直线连接其每对点；你得到的新形状就是包含原始形状的最小凸形：即它的凸包。科学现在非常脱节：到处都是参差不齐的前沿。生物学向一端突出，数学向另一端突出。一个实验室可能花 20 年时间利用他们掌握的一种方法了解关于一组基因的一切，而相邻的基因和替代方法却处于休眠状态。像 BootLoops 这样的套件可以连接这些点，填补凸包的空白。

Things are changing fast, and it’s practically impossible to plan ahead. Why apply for a grant for three years of funding to compute a problem an AI model might end up solving overnight? I still don’t know how to train graduate students. Since I published my piece on vibe physics, that feeling has only become more acute. In some fields, such as computer science, the disruption is downright scary. Two years ago, I would have recommended a course on “Python for engineers” as essential. Today, it’s unnecessary. I have spent a lot of time on machine learning, building models to study various physical phenomena. Now, all of those models can be built by AI, so even learning the ins and outs of neural networks is of limited utility. Just tell Claude to find the state-of-the-art ML approach and it will do it for you.事情变化得很快，提前规划几乎是不可能的。为什么要申请三年的资助来计算一个 AI 模型可能一夜之间就能解决的问题呢？我仍然不知道如何培养研究生。自从我发表了关于“Vibe Physics”的文章以来，这种感觉变得更加强烈。在某些领域，如计算机科学，这种颠覆简直令人恐惧。两年前，我会推荐一门“工程师 Python”课程作为必修课。今天，它已经没必要了。我花了很多时间研究机器学习，构建模型来研究各种物理现象。现在，所有这些模型都可以由 AI 构建，因此即使是学习神经网络的细节也效用有限。只需告诉 Claude 寻找最先进的 ML 方法，它就会为你完成。

I think BootLoops shows that if you find the right problem, much of the technical work of solving it can now be automated. That is liberating, in a way. But humans are still needed for the conceptual part. Because of that, I hope we can figure out a way to give humans the credit they deserve even for Claude-shaped science. Curiously, AI runs the risk of inverting the usual business model, where the person at the top gets the credit and the people working on the problem hands-on get none. I don’t know how this resolves, but it won’t be resolved by pretending the human contribution was the typing.我认为 BootLoops 表明，如果你找到了正确的问题，解决它的大部分技术工作现在都可以自动化。这在某种程度上是一种解放。但概念部分仍然需要人类。正因为如此，我希望我们能找到一种方法，让人类即使在“Claude 风格”的科学中也能获得他们应得的认可。奇怪的是，AI 有可能颠覆通常的商业模式，即高层的人获得认可，而亲手解决问题的人却一无所获。我不知道这会如何解决，但绝不是通过假装人类的贡献只是打字来解决的。

So where does this leave Joe Scientist? If you ask me, in a great place, actually. We now have tools at our fingertips that can make progress on seemingly intractable problems with lightning speed. I can foresee huge growth on the horizon in data-rich-but-theory-starved fields like systems biology. I can also easily imagine human minds being freed from the tedium of calculation and analysis to do the deeper intellectual work of direction and guidance. And most of all, I anticipate scientists branching out from their home turf to find synergy and collaboration, as I have with BootLoops. The most refreshing outcome of focusing on Claude-shaped science is that it helps us appreciate what human-shaped science looks like.那么，这让“Joe Scientist”（普通科学家）处于什么境地呢？如果问我，实际上处于一个很好的位置。我们现在触手可及的工具，可以以闪电般的速度在看似棘手的问题上取得进展。我可以预见，在数据丰富但理论匮乏的领域（如系统生物学）将出现巨大的增长。我也很容易想象人类思维从繁琐的计算和分析中解放出来，去从事更深层次的指导和引领的智力工作。最重要的是，我期待科学家们走出自己的舒适区，去寻找协同效应和合作，就像我通过 BootLoops 所做的那样。专注于“Claude 风格”科学最令人耳目一新的结果是，它帮助我们欣赏了“人类风格”的科学是什么样子。

Details of the BootLoops harness and the science it has enabled can be found at www.bootloops.ai. The BootLoops harness is on GitHub for anyone to use and contribute to.关于 BootLoops 套件及其所实现的科学成果的详细信息，可以在 www.bootloops.ai 上找到。BootLoops 套件已在 GitHub 上发布，供任何人使用和贡献。

Acknowledgements致谢

The science I produced with Claude and BootLoops would not have been anywhere near as interesting without the guidance and collaboration from my fellow humans: Isaiah Andrews, Nima Arkani-Hamed, Michael Desai, Scott Edwards, Noam Elkies, Cecilia Garraffo, Matthew Gordon, Thomas Grimm, Martin Hemberg, Mikhail Ivanov, David Johnston, Gary King, Paul Lewis, Brendan Meade, Amara McCune, Joe Pater, Subhabrata Sen, Siddharth Mishra-Sharma, James O’Dwyer, Kevin Ryan, Jesse Shapiro, and Xiaoyuan Zhang.如果没有我的同类人类的指导和合作，我与 Claude 和 BootLoops 共同产出的科学成果绝不会如此有趣。他们是：Isaiah Andrews, Nima Arkani-Hamed, Michael Desai, Scott Edwards, Noam Elkies, Cecilia Garraffo, Matthew Gordon, Thomas Grimm, Martin Hemberg, Mikhail Ivanov, David Johnston, Gary King, Paul Lewis, Brendan Meade, Amara McCune, Joe Pater, Subhabrata Sen, Siddharth Mishra-Sharma, James O’Dwyer, Kevin Ryan, Jesse Shapiro 和 Xiaoyuan Zhang。

Disclosure披露

During this project, Schwartz has been working as a visiting researcher at Anthropic. BootLoops is not an Anthropic project; it is owned and maintained by Matthew Schwartz.在本项目期间，Schwartz 一直以访问研究员的身份在 Anthropic 工作。BootLoops 并非 Anthropic 的项目；它由 Matthew Schwartz 拥有并维护。