从人工智能扩展的“苦涩教训”，到蛋白质折叠尚未破解的奥秘，Google DeepMind 的 Pushmeet Kohli 和 Biohub 的 Sal Candido 正在重新思考：要打造真正理解生物学的人工智能，究竟需要什么。在 Brandon Anderson 主持的这场特别圆桌讨论中，他们探讨了 AlphaFold 的突破为何只是起点、为什么仅靠扩大算力和数据规模无法解决生物学问题，以及新一代人工智能模型将如何改变我们对蛋白质、细胞和人类疾病的认识。From the Bitter Lesson of AI scaling to the unsolved mysteries of protein folding, Google DeepMind’s Pushmeet Kohli and Biohub’s Sal Candido are rethinking what it takes to build AI that truly understands biology. In this special panel moderated by Brandon Anderson, they explore why AlphaFold’s breakthrough was only the beginning, why scaling compute and data alone won’t solve biology, and how the next generation of AI models could transform our understanding of proteins, cells, and human disease.

我们深入探讨 AI 驱动的生物学的未来：如何在生物数据中发现尺度定律、科学直觉与通用架构之间的权衡、为什么蛋白质结构预测远未解决，以及要构建生命系统的预测模型需要具备什么条件。Pushmeet 回顾了 AlphaFold 背后的经验教训、人类可解释性的局限，以及未来的前沿模型能否比我们更好地理解其他 AI 系统。Sal 解释了为什么蛋白质语言模型可能已经蕴含着我们尚未发掘的科学知识、生物建模为何必须超越单个蛋白质，以及要实现 Biohub 治愈所有疾病的使命，为何必须着眼于 10 倍级突破，而非渐进式改进。We go deep on the future of AI-driven biology: finding scaling laws in biological data, the tradeoffs between scientific intuition and general-purpose architectures, why protein structure prediction is far from solved, and what it would take to build predictive models of living systems. Pushmeet reflects on the lessons behind AlphaFold, the limits of human interpretability, and whether future frontier models could understand other AI systems better than we can. Sal explains why protein language models may already contain scientific knowledge we haven’t unlocked, how biological modeling must move beyond individual proteins, and why achieving Biohub’s mission to cure all disease requires thinking in terms of 10x breakthroughs rather than incremental improvements.

我们讨论：We discuss:

• 生物学领域的苦涩教训：为什么仅仅扩大算力和数据规模还不够

  The Bitter Lesson for biology: why scaling compute and data isn’t enough

• 为什么找到合适的缩放定律比盲目增大模型规模更重要

  Why finding the right scaling law matters more than blindly increasing model size

• 低质量宏基因组数据如何提升蛋白质语言模型的性能

  How low-quality metagenomic data can improve protein language models

• 为什么人工智能研究人员优化的是现有数据，而不是最重要的科学问题

  Why AI researchers optimize for available data instead of the most important scientific problems

• DeepMind 关于如何平衡建模、数据生成与科学专业知识的经验教训

  Lessons from DeepMind on balancing modeling, data generation, and scientific expertise

• 为什么构建虚拟细胞需要从根本上不同的数据集

  Why building a virtual cell requires fundamentally different datasets

• AlphaFold 的人工设计架构与科学直觉的作用

  AlphaFold’s handcrafted architecture and the role of scientific intuition

• 为什么优质数据比单纯增加数据量更重要

  Why good data matters more than simply having more data

• 归纳偏置、缩放定律与专用 AI 架构的未来

  Inductive biases, scaling laws, and the future of specialized AI architectures

• 为什么我们尚未进入后 Transformer 时代，但架构正在不断演进

  Why we aren’t in a post-Transformer world, but architectures are evolving

• 为什么 AlphaFold 并没有真正解决所有蛋白质折叠问题

  Why AlphaFold didn’t actually solve all of protein folding

• 蛋白质动力学、无序性与静态结构预测的局限性

  Protein dynamics, disorder, and the limitations of static structure prediction

• 冷冻电镜显微照片如何解锁更丰富的生物学表征

  How cryo-EM micrographs could unlock richer biological representations

• 从单个蛋白质模型转向整个生物系统模型

  Moving from models of individual proteins to whole biological systems

• 费曼著名的原则，以及为什么 AI 如今能够创造出我们无法理解的事物

  Feynman’s famous principle and why AI can now create things we don’t understand

• 蛋白质语言模型中隐藏的生物学知识

  The hidden biological knowledge inside protein language models

• 为什么可信度和不确定性校准比完全可解释性更重要

  Why trustworthiness and uncertainty calibration matter more than full interpretability

• 前沿 AI 模型是否比人类更能理解其他 AI 系统

  Whether frontier AI models could interpret other AI systems better than humans

• 当 AI 能将药物发现效率提升 10 至 100 倍时

  When AI could deliver 10x–100x acceleration in drug discovery

• 为什么治愈所有疾病需要我们追求十倍级突破，而不是百分之十的改进

  Why curing all disease requires thinking about 10x breakthroughs instead of 10% improvements

• X: https://x.com/pushmeet

• LinkedIn: https://www.linkedin.com/in/pushmeet-kohli-4838994/

萨尔·坎迪多——BiohubSal Candido — Biohub

• Biohub: https://biohub.org/team/salvatore-candido/

• X: https://x.com/salcandido

• LinkedIn: https://www.linkedin.com/in/salcandido/

布兰登·安德森——主持人Brandon Anderson — Moderator

• LinkedIn: https://www.linkedin.com/in/brandon--anderson

Timestamps

00:00:00 引言：生物数据的苦涩教训00:00:00 Introduction: The Bitter Lesson for Biological Data

00:01:00 寻找生物学中的尺度定律与合适的数据00:01:00 Finding Scaling Laws and the Right Data for Biology

00:04:54 DeepMind 的苦涩教训：解决问题与扩大模型规模</final>00:04:54 DeepMind’s Bitter Lesson: Solving Problems vs. Scaling Models

00:06:50 AlphaFold、数据局限性与构建虚拟细胞00:06:50 AlphaFold, Data Limitations, and Building the Virtual Cell

00:09:52 手工设计的架构 vs. 扩展算力00:09:52 Handcrafted Architectures vs. Scaling Compute

00:11:53 优质数据、归纳偏置与模型设计00:11:53 Good Data, Inductive Bias, and Model Design

00:13:39 超越 Transformer：人工智能架构的未来00:13:39 Beyond Transformers: The Future of AI Architectures

00:14:38 为什么 AlphaFold 尚未解决蛋白质折叠问题00:14:38 Why AlphaFold Hasn’t Solved Protein Folding

00:17:39 蛋白质动力学、设计与冷冻电镜00:17:39 Protein Dynamics, Design, and Cryo-EM

00:19:09 从单个蛋白质到整个生物系统00:19:09 From Individual Proteins to Whole Biological Systems

00:20:43 费曼原理：在不理解的情况下创造00:20:43 Feynman’s Principle: Creating Without Understanding

00:22:18 蛋白质语言模型内部隐藏的知识00:22:18 The Hidden Knowledge Inside Protein Language Models

00:23:48 AlphaFold、可信度与可解释性00:23:48 AlphaFold, Trustworthiness, and Interpretability

00:25:56 人工智能能否比人类更好地理解其他人工智能模型？00:25:56 Could AI Understand Other AI Models Better Than Humans?

00:27:43 人工智能何时将彻底革新药物发现？00:27:43 When Will AI Revolutionize Drug Discovery?

00:29:47 为什么治愈所有疾病需要 10 倍级的突破00:29:47 Why Curing All Disease Requires 10x Breakthroughs

Transcript

引言：数据是否也存在一条苦涩教训？Introduction: Is There a Bitter Lesson for Data?

Brandon Anderson [00:00:04]：很高兴来到这里。今天早上真是令人振奋。发布了这么多很酷的消息。我觉得生物科学的未来就在这里揭晓。这是建模专题。我们都是建模人员，所以谈论数据当然再自然不过了。Brandon Anderson [00:00:04]: Great to be here. What an exciting morning. So many cool announcements. I think the future of bioscience is being announced right here. This is the modeling session. We’re all modelers, so of course it’s natural for us to talk about data.

谈到数据时，我喜欢思考的一个问题是如何恰当地扩展数据规模。这让我想到了“苦涩的教训”，不过是从数据的角度重新审视它。对于不了解人工智能的人来说，“苦涩的教训”指的是：能够扩展规模的方法最终会胜出。只要规模足够大，它就会胜出。所以，我想先问问萨尔：数据也有“苦涩的教训”吗？One of the things I like to think about when it comes to data is how to scale it properly. This has brought me to the question of the bitter lesson, but recast in the frame of data. The bitter lesson, for those of you who are not AI people, is the statement that methods that scale win eventually. If you can scale enough, it wins. So my question, starting with Sal, is: Is there a bitter lesson for data?

Sal Candido [00:01:00]：当然。显然，要让它奏效，你需要合适的数据。关于缩放定律，一个常见的误解是它们无处不在，而且总是存在。实际上，很多工作都在于找到这种缩放定律。我们做的很多事情，都是在试图弄清楚：在什么情况下，投入更多算力、使用更多数据，就能得到更好的结果？Sal Candido [00:01:00]: For sure. You obviously need the right data in order for it to work. One misconception of scaling laws is that scaling laws are everywhere and they always exist. A lot of the work is actually finding that scaling law. A lot of what we do is trying to figure out: What’s a situation where, if you put more compute into it, if you put more data into it, you get a better result out?

这是个很好的局面，因为一旦到了这一步，你就可以按流程推进了。问题就变成了一个工程问题，而这正是我喜欢的。它既关乎架构，也在很大程度上关乎数据。如果你没有包含正确信息和统计特征的数据来解决你想解决的问题，就无法得到具备你所期望能力和理解力的模型。你实际上只能从现有数据中提取信息，并利用这些信息进行泛化，超越数据本身。That’s a great situation because, once that happens, you can turn the crank. It becomes an engineering problem, which is something I like. That has to do with architecture, but it also very much has to do with data. If you don’t have data with the right information and statistics to solve the problem you want, you’re not going to get a model that has the capabilities and understanding you want. You can only really pull information from the data you have and use that to generalize beyond it.

选择合适的数据：可用性与科学影响力Choosing the Right Data: Availability vs. Scientific Impact

Brandon Anderson [00:02:09]：说到数据收集，你认为哪些模态最合适？还有一个相关的问题：模型研究人员是否倾向于研究那些有现成数据的问题，而不是最符合自身目标或对转化医学影响最大的问题？Brandon Anderson [00:02:09]: When it comes to data collection, how do you think about which modalities are best? A related question: Do modelers tend to work on problems where the data is available, rather than the problems that best serve their goals or have the greatest impact on translational medicine?

Sal Candido [00:02:32]：当然。我不代表所有建模师发言，但我比较懒，所以会直接用现成的东西。这件事有好的一面，也有不好的一面。Sal Candido [00:02:32]: For sure. I won’t speak for all modelers, but I’m lazy, so I’m going to work with what’s available. There’s a good side and a bad side to this.

好的一面是，在做传统机器学习时，你通常会寻找能找到的最纯净、质量最高的数据样本。但如果你跟我一样，就会跑到后屋里翻找，看看人们的废弃物里有什么。一个具体例子是用宏基因组序列训练蛋白质语言模型，而这些序列并不是质量最高的数据。事实上，我可以向你保证，其中很多数据甚至都不是真正完整的蛋白质。然而，这些数据却能提升模型在设计真正有效的蛋白质，以及理解已知存在的蛋白质方面的表现。The good side is that, when you’re doing conventional machine learning, you’re often looking for the most pristine, high-quality data examples you can find. But if you’re like me, you’re rooting around in the back room, looking through people’s junk to see what’s there. A concrete example is training a protein language model on metagenomic sequences, which are not the highest-quality data. In fact, much of that data, I can guarantee you, isn’t even a real, whole protein. And yet it makes the performance of the model go up for designing real proteins that work and understanding proteins that we know exist.

这是积极的一面。但它也可能把你引向消极的一面，让你说：“我们就把那些容易生成的数据扩大规模吧。”我认为这未必是正确的做法。今天我们在这里谈论的 BBI 有一点让我很兴奋，那就是它会主动去问：“我们需要什么数据才能解决这个问题？”That’s the positive side. But it can lead you to a negative place where you say, “Let’s just scale up the data that we can generate easily.” I don’t think that’s necessarily the way to do it. That’s one of the things that’s exciting to me about what we’re talking about here today with the BBI. It’s going out and asking, “What is the data that we need to solve the problem?”

这不仅关乎合适的资源，也关乎合适的社区。在 Biohub，我们努力以开放的方式开展工作，与社区合作，并推动整个社区向前发展。这一点至关重要，因为如果只有人们在构建模型，我们就会倾向于使用现有的数据；如果只有人们在生成数据，他们就会倾向于生成能够生成的数据。但如果整个社区能够一路携手合作，并以开放的方式开展工作，就能弄清楚我们真正需要什么数据。然后，就能找到那条缩放定律。It’s about the right resources, but also the right community. One thing we try to do at Biohub is work in the open, work with the community, and move the whole community forward. That’s critical because, if you just have people building models, we’re going to lean toward the data that exists. If you just have people generating data, they’re going to lean toward the things that can be generated. But if you can work together as a community, every step of the way, in an open fashion, you can figure out what data you actually need. Then you can find that scaling law.

问题优先：建模、数据与专业知识The Problem Comes First: Modeling, Data, and Expertise

Brandon Anderson [00:04:50]: Pushmeet, what do you think about the bitter lesson for data?

Pushmeet Kohli [00:04:54]：Rich 和我们共事时，我也在 DeepMind，当时正在思考“苦涩的教训”这个理念。我从 Rich 在 DeepMind 最初那场演讲中领会到的，并不是数据是否有用这一具体问题，也不是我们应该如何看待机器学习。我的理解是，他谈的是更具概念性的东西。Pushmeet Kohli [00:04:54]: I was at DeepMind when Rich was with us and thinking about this idea of the bitter lesson. What I took from Rich’s original lecture at DeepMind wasn’t about the specific notion of whether data is useful or how we should think about machine learning. My take was that he was talking about something more conceptual.

有时我们看待问题时，会以一种非常教条的方式思考解决方案：“我是做建模的”，或者“我是做数据生成的”。我认为这就是那个苦涩的教训。如果你带着这种心态去解决问题，可能不会成功。问题才是第一位的，你应该在解决方案上保持灵活。两种方法都应该尝试。你应该弄清楚自己要解决的问题。如果需要建模，就做建模；如果需要收集更多数据，那就去收集数据。Sometimes when we’re looking at problems, we think about solutions in a very religious way: “I’m a modeler,” or “I’m a data generation person.” I think that is the bitter lesson. If you approach a problem with that mindset, you might not succeed. The problem comes first, and you should be flexible in your solution space. You should try both things. You should understand the problem you’re trying to solve. If it requires modeling effort, do the modeling effort. If it requires collecting more data, then collect data.

当时机器学习领域的情况是，人们会说：“我们是机器学习研究人员。数据集就在那儿。这里有一些训练数据，这里有一些测试数据，我们只要优化模型就行了。”这种做法存在问题，因为如果你的最终目标是解决这个问题，就必须同时关注整个流程中的两个方面。What happened in machine learning at the time was people saying, “We’re machine learning researchers. The dataset is there. Here’s some training data, here’s some test data, and we’ll just optimize the model.” That is broken in the sense that, if your eventual goal is to solve the problem, you have to look at both aspects of what goes into the process.

而投入这一过程的，不只是数据或建模，还有专业知识。以 AlphaFold 为例，我们研究了利用现有数据集能做到什么，因为我们既缺乏核心专业知识，也没有足够资源去说：“我们来把 PDB 的规模大幅提升一个数量级。”世界各地的机构和科学家为构建这些数据所投入的心血，价值无可估量。因此，在这种情况下，你必须专注于投资建模，以获得最大的投入产出比。And what goes into the process is not just data or modeling. It’s also expertise. With AlphaFold, for example, we looked at what was possible with existing datasets because we didn’t have the core expertise, or even the resources, to say, “Let’s augment the PDB by a significant order of magnitude.” The investment that organizations and scientists across the world had put into constructing that data was invaluable. So there, you had to focus on getting the biggest bang for your buck by investing in modeling.

但在其他领域，比如细胞基因组学，我们也采用了同样的方法，提出了这样的问题：“利用细胞-基因数据，你能做些什么？”经过大量工作，我们很清楚地认识到，现有数据尚不足以实现构建虚拟细胞这一宏大目标。这让人们能够凝聚起来，专注于推动科学发展的实际挑战，而不是盲目追随数据生成或建模方面的进展。But in other areas, say cell genomics, we took the same approach and asked, “What can you do with cell-by-gene data?” After a lot of work, it was very clear that the data was not there yet to pursue that grand ambition of building the virtual cell. This brings people together to focus on the actual challenges of advancing science rather than religiously following advances in data generation or modeling.

Brandon Anderson [00:08:23]：我非常喜欢这个回答。我们能从中得到什么可操作的启示？始终先定义问题，再确定要在哪个解空间中寻找答案。但更广泛地说，随着我们迈向下一代转化医学，社区应该如何看待这件事？Brandon Anderson [00:08:23]: I really like that answer. What’s the actionable takeaway? Always define the problem first and then figure out which solution space to search over. But more broadly, how should the community think about this as we move toward the next generation of translational medicine?

Pushmeet Kohli [00:08:47]：我给任何刚开始涉足这个领域的人的建议是，把自己看作一个多学科人才。首先要理解问题：你为什么要研究这个问题？你想实现什么目标？然后再考虑需要什么，无论是建模、算力扩展，还是数据生成。理解问题至关重要，当然，你也需要积累自己的专业知识。Pushmeet Kohli [00:08:47]: The advice I give to anyone starting in this area is to think of yourself as a multidisciplinary person. Understand the problem first. Why are you working on it? What are you trying to achieve? Then think about what’s needed, whether that’s modeling, compute scaling, or data generation. Understanding the problem is extremely important, and of course you need to build your expertise.

当然也存在一些限制。你可能只能生成一定数量的数据，或者只能负担得起训练特定规模的模型。了解这些限制，尽早验证方案是否可行，并思考哪些方法从长远来看切实可行，能够帮助你达到预期的影响力。There are constraints, too. Maybe there’s only a certain amount of data you can generate, or a certain model size you can afford to train. Understand those constraints, try to fail fast, and look at which approaches will be feasible in the long term to get you to the level of impact you’re aiming for.

科学归纳偏置与规模扩展：构建模型的技艺Scientific Inductive Bias vs. Scaling: The Craft of Building Models

Brandon Anderson [00:09:52]：这正好引出了我的下一个问题。回顾建模的发展历程，AlphaFold 2 几乎是一件艺术品，其中包含许多经过精心手工设计的特征。解决方案的每个部分都经过了深思熟虑。Google 或 Alphabet 的一些工作仍然处于这一领域，但普遍的看法似乎是，大家正转向更具可扩展性、更通用的策略。Brandon Anderson [00:09:52]: This leads right into my next question. When I look at the evolution of modeling, AlphaFold 2 was essentially a work of art, with a lot of carefully handcrafted features. There was very intentional thought put into every part of the solution. Some of Google’s or Alphabet’s work still stays in that space, but the general consensus seems to be moving toward more scalable, general strategies.

对于某些问题，我们是否仍然需要匠心打造的解决方案？还是说，最好先把资源投入到规模化上？在资源和资金有限的情况下，你可以投资算力、人才或数据。我们该如何看待这种取舍？Do we still need artisanal, craft solutions for certain problems? Or are resources better spent focusing on scale first? With fixed resources and money, you can invest in compute, talent, or data. How should we think about that trade-off?

Pushmeet Kohli [00:10:48]：我们从第一性原理出发来思考这个问题。构建更优秀模型的艺术与技艺并非偶然所得。我们进行了大量实验，但背后始终有一个愿景。生物物理学和生物化学的科学直觉告诉我们，氨基酸残基并不是各自为政，它们会受到其他残基的影响。所以，我们要把这一点融入模型。如果你从科学界学到了什么，就把这些知识用起来，赋予模型这种不公平的优势。Pushmeet Kohli [00:10:48]: Let’s approach it from first principles. The art and craft of constructing a better model wasn’t accidental. We did a lot of experimentation, but there was a vision behind it. Scientific intuition from biophysics and biochemistry told us that amino acid residues are not just doing their own thing. They’re influenced by other residues. So let’s bake that in. If you’ve learned something from the scientific community, use that information. Give the model that unfair advantage.

这使得模型的数据效率高得多，因为它不必重新学习所有内容。我还必须提到，筛选优质数据本身就是一门艺术。并不是说你只要说“多放些数据进去”就行了。如果你只是重复相同的数据，就不会有任何进展。这不仅仅关乎大数据，更关乎优质数据，以及理解取得进展所需的数据覆盖范围。这本身就是一个更有趣、更具挑战性的问题。That makes the model much more data-efficient because it doesn’t have to relearn everything. I also have to mention that curating good data is an art in itself. It’s not as if you can just say, “Put in more data.” If you replicate the same data, you’re not going anywhere. It’s not just about big data; it’s about good data and understanding the coverage of the data necessary to make progress. That’s a much more interesting and challenging problem in itself.

Brandon Anderson [00:12:20]: It looks like you have a thought, Sal. What do you think?

Sal Candido [00:12:24]：我非常赞同。这取决于要解决的问题。如果数据量较少，模型具有更多归纳偏置会有所帮助。要在较小的数据规模下取得成果，你实际上需要这一点。随着数据越来越多，有时模型会发现一些你未必知道的东西；而如果归纳偏置并不完全正确，有时反而会限制模型。随着模型不断改进，这其中会出现一个转折点。Sal Candido [00:12:24]: I very much agree. It depends on the problem to be solved. If you have a smaller amount of data, having more inductive bias in the model is going to help you. You actually need that to get results at smaller data scales. As you get more and more data, sometimes the model finds things you didn’t necessarily know about, and sometimes that inductive bias, if it wasn’t exactly correct, can hold you back. There’s a tipping point as things get better.

不过，我对你的问题也有点异议，因为在扩展规模这方面同样有很多技术门道。要把模型用更多数据训练、投入更多算力，并把模型做得更大，这背后需要大量算法工作。这不仅仅是从基础设施角度出发，或是为了让模型更快地运行推理。我们看到，发展已经超越了标准的 Transformer，开始采用更多定制化的架构。我不认为我们以任何方式、任何形式进入了后 Transformer 时代，但我们正在改进这些架构，让它们更符合特定用途、表现更好，即便面对互联网规模的数据也是如此。But I also object to your question a little because there’s a lot of craft in the scaling part of things as well. There’s a lot of algorithmic work that goes into taking models, training them on more data, putting more compute into them, and making them bigger. And that isn’t only from an infrastructure perspective or about making the models run inference faster. We’re seeing things go beyond standard transformers to more bespoke architectures. I don’t think we’re in a post-transformer world in any way, shape, or form, but we’re modifying those architectures to make them more fit for purpose and work better, even with internet-scale data.

随着数据不断涌入，挑战不只是整理这些数据，或挑选模型下一批需要的信息，还包括在每一步、每个规模上都要思考：怎样的架构才能最大限度地利用这些信息。As more data comes in, the challenge isn’t only curating that data or selecting the next batch of information the models need. It’s also asking, at every step and at every scale, what the right architecture is to get the most out of that information.

为什么 AlphaFold 并未解决蛋白质生物学的所有问题Why AlphaFold Didn’t Solve All of Protein Biology

Brandon Anderson [00:14:36]：如果你看看当前蛋白质结构预测领域的情况，新闻会说蛋白质结构预测问题已经解决了。但如果你和我的朋友们聊聊，他们会说：“我们在这方面还有太多工作要做。”功能、动态变化和设计仍是尚未解决的开放问题。AlphaFold 2 发布至今已有大约五年，期间也取得了进展。Brandon Anderson [00:14:36]: If you think about the state of protein structure prediction right now, the news will say that protein structure prediction has been solved. But if you talk to my friends, they’ll say, “We have so much left to do here.” Function, dynamics, and design are still wide-open problems. It’s been about five years since AlphaFold 2 was announced, and progress has been made.

Pushmeet，你觉得还会有另一次重大突破吗？解决这些问题是否存在障碍？是因为我们缺少合适的数据、算法或想法吗？还是说这只是时间问题？Pushmeet, do you think there’s another big leap coming? Are there blockers to solving these problems? Do we not have the right data, algorithms, or ideas? Or is it just a matter of time?

Pushmeet Kohli [00:15:28]：科学的运作方式是先把某个问题孤立出来，然后一步一步地取得进展。人们说蛋白质折叠问题已经解决了，从概念层面来说，确实取得了一些进展。但想想“蛋白质是构成生命的基本积木”这种说法。蛋白质不是积木，也不会像积木那样发挥作用。我总是说蛋白质是构成生命的基本积木，但我其实并不相信这一说法。Pushmeet Kohli [00:15:28]: Science operates by isolating something and then making progress step by step. When people say the protein-folding problem has been solved, at a conceptual level there have been advances. But think about the narrative of proteins being the building blocks. Proteins aren’t blocks, and they don’t act as blocks. I say proteins are the building blocks all the time, but I don’t actually believe it.

蛋白质极其复杂。它们是无序的。它们的形状可能会因所处情境而改变。我以前常和 John Jumper 讨论这个问题：我们究竟想解决什么？我们并不知道蛋白质实际所处的真实基态，也不知道它们实际呈现的结构分布。我们想做的是复现某人获得并存入 PDB 的结构。我们做的就是这件事。碰巧的是，这样做很有用。Proteins are extremely complex. They’re disordered. Their shape might change depending on the context. John Jumper and I used to discuss this: What are we trying to solve? We don’t know the actual true ground state that proteins take, or the actual distribution of structures that proteins take. What we were trying to do was replicate a structure that somebody had obtained and deposited in the PDB. That’s what we did. And it just so happens that it’s useful.

但这并不意味着我们已经完全理解了蛋白质动力学。从宏观层面来看，我们很容易传达出我们已经取得进展这一点，但在座的科学家们都知道，还有多少工作有待完成。我们有很多值得庆祝的成果，但不要因此停止资助蛋白质结构预测和蛋白质动力学研究，因为我们才刚刚起步。But that doesn’t mean we’ve understood all of protein dynamics. At the top level, it’s easier to communicate that we’ve made progress, but the scientists in this crowd know how much remains to be done. There’s a lot to celebrate, but let’s not stop funding protein structure prediction and protein dynamics, because we’re just getting started.

冷冻电镜、分子动力学与缺失的信息Cryo-EM, Molecular Dynamics, and the Missing Information

布兰登·安德森 [00:17:39]：最大的阻碍是什么？如果你能挥动魔法棒，说一句“我们需要更多这样的东西”，就能加速功能开发、动态效果或设计，你会让什么变成现实？Brandon Anderson [00:17:39]: What’s the biggest blocker? If you could wave a magic wand and say, “We have more of this,” and that would accelerate function, dynamics, or design, what would you bring into existence?

Pushmeet Kohli [00:17:54]：我的背景相当多元。我最初是一名安全研究员，之后涉足计算机视觉、贝叶斯理论、判别式机器学习、深度学习、用于编程的 AI，最后转向科学研究。所以你问我这个问题时，我内心的计算机视觉研究员会对冷冻电镜显微图像感到非常兴奋。Pushmeet Kohli [00:17:54]: My background is quite eclectic. I started as a security researcher, then went into computer vision, Bayesian theory, discriminative machine learning, deep learning, AI for coding, and finally science. So when you ask me that question, the computer vision researcher in me gets very excited about cryo-EM micrographs.

我心想：“这些 PDB 数据是什么？我应该从源头入手。我应该看冷冻电镜显微图像。我不想要那些结构数据。它们一定遗漏了所有数据。我应该直接处理冷冻电镜显微图像。”I thought, “What is this PDB data? I should be working at the source. I should be looking at cryo-EM micrographs. I don’t want those structures. They must be missing out on all the data. I should just operate directly on cryo-EM micrographs.”

如果能训练出可扩展到那个规模的模型，并提供恰当数量的数据，从中提取所蕴含的动态特征和分布信息，那就太棒了。我试过了，但还需要更多工作。Getting models that can scale at that level, with the right amount of data, and extract the dynamics and distributional information captured there would be amazing. I tried it, but it requires more work.

布兰登·安德森 [00:18:57]：还有工作要做，但你认为这条途径能够为你提供那些信息吗？Brandon Anderson [00:18:57]: There’s still work to do, but you believe that’s a route that could give you that information?

Pushmeet Kohli [00:19:00]：是的。我想，也许有一天会有比我更厉害的人试着解决这个问题，我们就能有所进展。Pushmeet Kohli [00:19:00]: Yeah. I think at some point maybe some people better than me will take a stab at it, and we’ll get somewhere.

从蛋白质结构到更大型的生物系统From Protein Structures to Larger Biological Systems

你怎么看，Sal？Brandon Anderson [00:19:07]: What do you think, Sal?

Sal Candido [00:19:09]: 这很有意思，因为这些模型相当有用，但它们解决的并不完全是大多数人想要解决的问题。它们服务于一个非常具体的目的，你也可以用它们来做其他事情。比如，你可以用它们设计新的蛋白质，而这未必是你最初会想到的用途。Sal Candido [00:19:09]: It’s interesting because these models are quite useful, but they’re not exactly the problem that most people want to solve. They solve a very specific purpose, and you can also use them to do other things. For example, you can use them to design new proteins, which isn’t necessarily what you would start with.

我把我们现在构建的模型想象成这样一种东西：有人试图理解自行车是如何运作的，却只在给自行车上的一根辐条建模。这些模型可以随着时间推移变得越来越好，但真正需要做的是从辐条模型转向车轮模型，再转向整辆自行车的模型，因为人们想理解的正是这些。继续这个类比，似乎人们想用自行车模型来设计一辆皮卡的零件。I think of the models we’re building now as someone who’s trying to understand how a bicycle works, but is modeling a spoke on it. Those models can get better and better over time, but what you really need to do is move from models of spokes to wheels to whole bicycles, because that’s what people want to understand. To continue the analogy, it seems like people want to use a model of the bicycle to design a part for a pickup truck.

随着我们将这些模型置于其所处的具体生物学背景中，就能从更宏观的层面进一步了解这些相互作用。我认为这正是未来的发展方向。不过我也同意，我们应该继续研究折叠模型，因为它们会不断改进。As we put these models into the particular biological context in which they’re operating, we’ll be able to learn more about these interactions on a broader scale. That’s where I think things are going. But I agree that we should keep working on folding models, because they’re going to keep getting better.

蛋白质设计与科学理解之比较Protein Design vs. Scientific Understanding

Brandon Anderson [00:20:43]：谈到设计，有一句费曼的名言：“我无法创造的东西，我就不理解。”如今，我们身处这样一个时代：人们很容易创造出某些东西，却完全不理解它们。相比那些能有效地一次性设计出皮摩尔级结合物之类的“神奇黑箱”，那些能帮助人类理解事物的模型有多重要？Brandon Anderson [00:20:43]: With regard to design, there’s a famous Feynman quote: “That which I cannot create, I do not understand.” Now we’re in a world where it’s really easy to create things without understanding them at all. How important is it to have models that help humans understand things, versus magical black boxes that can effectively one-shot a picomolar binder or something like that?

Sal Candido [00:21:18]：早在 AI 出现之前，我们就已经在用神奇的黑箱设计各种东西了。从某种意义上说，这仍然很有用。但理解非常重要，也是我认为 AI 领域的一大关键问题。Sal Candido [00:21:18]: We were designing things with magical black boxes long before AI came around. In some sense that’s still useful. But understanding is really important, and it’s one of the big things I think about with AI.

这些模型有太多东西要学了。随着智能变得越来越廉价、越来越容易获得，你可以部署它，让它去了解世界上正在发生的事情，而且了解得越来越多。但你要怎样才能从机器中提取这些知识，让我也能理解呢？也许这只是我个人的一点冷僻好奇心。我觉得还有太多东西值得去学习。There’s so much for these models to learn. As intelligence gets cheaper and more available, you can deploy it to learn more and more about what’s going on in the world. But how do you pull that knowledge out of the machine so that I can understand it? Maybe that’s just my esoteric curiosity. I think there are so many things to learn.

Brandon Anderson [00:22:05]：很多科学家确实很想弄明白事物的原理，最终成果对他们来说可能没那么重要。但我们在这里是要解决转化医学中的问题，对吧？Brandon Anderson [00:22:05]: A lot of scientists really want to understand things, and the endpoints may not be as important to them. But we’re here to solve translational medicine as a problem, right?

Sal Candido [00:22:18]：我觉得，你对事物挖掘得越深入，就越能找到推动它们不断向前发展的正确方法。有一点让我印象特别深刻：这些模型所包含的信息，比我们所知道的要多得多。Sal Candido [00:22:18]: I think the more you dig into things, the more you find the right way to keep pushing them forward. One thing that’s really salient to me is that these models have a lot more information in them than we know.

例如，我们在模型可解释性方面做了很多工作，你会从中发现大量信息。人们知道，蛋白质语言模型会在其表征中学习到某种结构概念，但我们还发现了有关功能和运动的信息。即使是从我们现有的模型中，仍有很多有待发掘的内容。随着我们不断提升模型的能力，理解这一点对我们很重要。We’ve worked a lot on interpretability for our models, for example, and you find a lot of information there. People know that protein language models learn some notion of structure within their representations, but we find information about functions and motions as well. There’s a lot still to be unlocked, even from the models we have now. That’s important for us to understand as we raise their capabilities.

说到底，你期待通过将所有这些信息压缩进一个模型，最终形成一个世界模型。它是如何发挥作用的？它是如何设计蛋白质的？它将进化中的信息压缩进了这个模型。除了能够产出对我们有用的东西之外，光是看看模型内部有什么，肯定也能让我们学到一些东西。At the end of the day, you expect a world model to emerge from compressing all this information into a model. How does it do its job? How does it design a protein? It has compressed information from evolution into that model. In addition to being able to produce something useful to us, there’s certainly something to learn just by looking at what’s inside.

AlphaFold 置信度、校准与可解释性AlphaFold Confidence, Calibration, and Interpretability

布兰登·安德森 [00:23:48]：你怎么看，Pushmeet？设计与理解之间的关系呢？Brandon Anderson [00:23:48]: What do you think, Pushmeet? Design versus understanding?

Pushmeet Kohli [00:23:53]：我的看法有所不同，我认为还是需要一定程度的理解。让我解释一下我所说的程度。Pushmeet Kohli [00:23:53]: I have a different take in the sense that some level of understanding is necessary. Let me explain what level I mean.

AlphaFold 并非完美。当时，AlphaFold 2 在该数据集上的 GDT 分数约为 90。但即使它的 GDT 分数达到 95，如果它的 pLDDT 分数完全未经校准，又有谁会相信它呢？想象一下，它似乎能给出不错的答案，还告诉你它非常有把握；结果你接下来一年都在研究这些答案，最后才发现它们完全错了。因此，对不确定性度量进行校准至关重要。AlphaFold isn’t perfect. AlphaFold 2 had a GDT score of about 90 on that set at the time. But even if it had a GDT score of 95, if its pLDDT score were completely uncalibrated, who would trust it? Imagine that it magically gave good answers but told you it was very confident, and then you worked on it for the next year only to find out it was completely wrong. Calibration of the uncertainty measure was extremely important.

从这个意义上说，我们确实理解 AlphaFold 2，也在理解它的行为方式方面取得了很大进展。但这不同于理解它内部是如何运作并找到解决方案的。在这一点上，我同意 Sal 的看法：可解释性要问的是，它为什么能奏效？它为什么会给出这个答案？In that sense, we do understand AlphaFold 2, and we made a lot of progress in understanding how it behaves. That’s different from understanding how it worked internally to find the solution. There I agree with Sal that interpretability asks: Why did it work? Why did it give this answer?

从最高层面来看，我们可以理解 AlphaFold 2 的行为及其泛化能力，但在发布之前，我们并没有发现它的所有特性。发布 AlphaFold 2 并开放其权重后，人们发现它是一个很出色的无序蛋白质预测器。它能够判断蛋白质中的哪些部分是无序的。这说明它具有泛化能力。At the highest level, AlphaFold 2 was interpretable in terms of its behavior and ability to generalize, and we didn’t discover everything about that before launching. When we launched AlphaFold 2 and made the weights available, people found that it was a great disordered-protein predictor. It could figure out which elements of the protein are disordered. That shows it generalizes.

但可解释性提出了另一个问题：对谁而言可解释？如果你说的是对具有人类大脑认知和计算局限的人类理性系统而言可解释，那么答案是否定的，AlphaFold 2 并不可解释。但如果你问的是，对于一个规模更大、能力更强的模型来说，AlphaFold 2 的工作原理是否可解释，也许是可以的。只是我们还没弄明白。But interpretability asks another question: Interpretable by whom? If you’re saying interpretable by a human rational system, with the cognitive and computational limitations of the human brain, then no, AlphaFold 2 is not interpretable. But if you’re asking whether AlphaFold 2 is interpretable to a much larger, more sophisticated model in terms of how it works, maybe it is. We just don’t get it.

作为 AlphaFold 2 的使用者，我们确实需要了解它能做什么、不能做什么。深入了解它的行为特征、优势和局限性极其重要。我们不能在不了解这些特性的情况下就使用这些模型，否则它们非但不能帮助我们，反而可能造成伤害。As users of AlphaFold 2, we do need to understand what it can and cannot do. Understanding its behavioral characteristics, strengths, and limitations is extremely important. We can’t just use these models without that characterization, because otherwise, rather than being helpful, they can harm us.

Brandon Anderson [00:26:53]：所以你的看法是，严格来说，可解释性并非必要，但人们应该关注的是确保模型值得信赖，以便人类能够据此做出可执行的决策。Brandon Anderson [00:26:53]: So your take is that interpretability, strictly speaking, isn’t necessary, but ensuring that a model is trustworthy so humans can make actionable decisions is what people should focus on.

Pushmeet Kohli [00:27:04]：没错。可解释性也取决于观察者。是谁在解释？如果是人类科学家试图理解模型如何做出预测，那和让另一个规模大得多的 LLM 访问激活层，然后问：“你能预测 AlphaFold 会做什么吗？”是两个不同的问题。也许未来的前沿模型能够预测这一点，并提出一种理论，说明 AlphaFold 2 是如何解读并生成其结果的。Pushmeet Kohli [00:27:04]: Exactly. Interpretability is also in the eye of the beholder. Who is interpreting it? If it’s a human scientist trying to interpret how the model makes a prediction, that’s a different question from giving another, much larger LLM access to the activation layers and saying, “Can you predict what AlphaFold will do?” Maybe those frontier models of the future will be able to predict that and come up with a theory of how AlphaFold 2 was interpreting and producing its results.

人工智能何时会改变药物发现和临床医学？When Will AI Transform Drug Discovery and Clinical Medicine?

Brandon Anderson [00:27:43]：我们还剩两分钟，所以最后问你们两位一个问题。AI 很有可能在不久的将来显著改善人类健康。我喜欢定量预测。你们的最佳估计是：还要多久，我们才能在临床上看到 AI 带来的成果？Pushmeet，你想先回答吗？Brandon Anderson [00:27:43]: We have a two-minute warning, so one last question for both of you. There’s a real chance that AI will make dramatic improvements in human health in the immediate future. I like quantitative predictions. Best guess: How long until we start seeing AI results in the clinic? Pushmeet, do you want to start?

Pushmeet Kohli [00:28:08]：我认为，“AI 在临床中的成果”这个问题本身就定义得不够清晰。如今，AI 已经应用于药物发现过程的各个环节。从这个角度来看，它早已在其中发挥作用。Pushmeet Kohli [00:28:08]: I think “AI results in the clinic” is an ill-posed question. AI is already being used today in every part of the drug discovery process. From that perspective, it’s already there.

但如果你问的是，何时才能看到研发时间线加快10倍或100倍，那么接下来的问题就是：我们要加速的是什么？是先导化合物优化？靶点发现？临床前研究还是毒理学研究？But if you’re asking when we’ll see a 10x or 100x acceleration in timelines, then the next question is: What are we accelerating? Is it lead optimization? Target discovery? Preclinical work or toxicology?

在未来几年里，这也是今天宣布这项努力极其重要的原因，我们需要攻克生物学领域的一些艰巨挑战。只有这样，我们才能实现真正的加速突破，彻底改变药物发现。人工智能将继续被用于各种持续进入临床应用的领域，但只有更深入地理解这项努力旨在构建的生物学模型，才能释放更大幅度的加速效应。Over the next few years, and this is why the effort announced today is extremely important, we need to tackle some of the hard challenges of biology. Only then will we get the true unlocks of acceleration that dramatically transform drug discovery. AI will continue to be used in things that go into the clinic all the time, but larger acceleration will only be unlocked with a better understanding of the biological models this effort is trying to create.

布兰登·安德森 [00:29:43]：好，谢谢。萨尔，如果可以的话，请简短回答。我们快没时间了。Brandon Anderson [00:29:43]: All right, thanks. Sal, a quick answer, if you can. We’re almost out of time.

Sal Candido [00:29:47]：我很想真的戴上我的生物黑客帽子来回答这个问题。重要的是要想清楚，推动这个领域向前发展意味着什么。我们真正想看到的是，这些工具能对结果产生影响。什么时候才能有一种药物完全由 AI 研发出来？我不知道，这很难预测。但我确实认为，我们会很快看到迅猛的进展，因为所有这些工具已经在投入使用了。Sal Candido [00:29:47]: I’m tempted to literally put on my biohacker hat to answer this question. It’s important to think about what it means to push the field forward. What we really want to see is outcomes being affected. When will a drug be made entirely by AI? I don’t know. That’s hard to predict. But I do think we’re going to see rapid progress very quickly, because all these tools are already being used.

回到你关于基础研究和基本理解的观点，我在职业生涯早期、还在 Google 工作时学到了一件事：有时候，与其思考如何改进 10%，不如问问要怎样才能实现 10 倍提升，这样更容易着手解决问题。这并不是因为实现 10 倍提升的路径一定更容易，而是因为这能让你从更广阔的视角看问题，发现自己之前没有考虑过的解决方案。你回到第一性原理，问：“我们要怎么做到这一点？要怎样才能真正推动这件事？”Going back to your point about basic research and basic understanding, one thing I learned a long time ago in my career, back at Google, is that sometimes it’s easier to approach a problem by asking what it would take to make a 10x improvement rather than a 10% improvement. That’s not because the 10x path is necessarily easier. It’s because it allows you to take a broader view and see solutions you haven’t been approaching. You go back to first principles and ask, “How are we going to do this? How are we going to really push this?”

这两种方法——10% 和 10 倍——都很有价值，我们两者都应该尝试。我很高兴 Biohub 拥有一个宏大而美好的使命宣言：治愈所有疾病。如果你想做到这一点，就真的需要弄清楚什么是 10 倍方法。这是一个绝佳的机会，让我们能够去实现这一目标。Both approaches, the 10% and the 10x, are valuable, and we should do both. I’m very happy that at Biohub we have a beautiful and lofty mission statement: to cure all disease. If you want to do that, you really need to figure out what the 10x approach is. It’s a great opportunity to be able to go do that.

Brandon Anderson [00:31:46]：太棒了。谢谢你们两位。感谢你们坐在字面意义上的“热座”上。非常有意思。Brandon Anderson [00:31:46]: Awesome. Thank you both. Thank you for being in the literal hot seat. Very interesting.