Bubble's Brain搜索⌘K
EN
登录

AlphaFold 为什么没有真正解决蛋白质折叠

  1. Brandon Anderson主持

    很高兴来到这里。今天上午真让人兴奋，发布了好多很酷的东西，我觉得生命科学的未来就在这里被宣布。这一场是建模专场，我们都是做模型的，所以聊数据再自然不过。

    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.

  2. Brandon Anderson主持

    说到数据，我很喜欢想的一件事是怎么把它正确地扩展上去。这让我想到「苦涩的教训」，不过是放在数据的框架里重新问。给不做 AI 的朋友解释一下：苦涩的教训说的是，能扩展的方法最终会赢，只要规模够大就能赢。所以我的问题，先从 Sal 开始：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?

  3. Sal CandidoBiohub

    当然有。显然你得有对的数据，它才能起作用。关于扩展定律有一个误解，以为扩展定律到处都有、一直存在。其实很大一部分工作就是去找到那条扩展定律。我们做的很多事，是想弄清楚：在什么样的设定下，投入更多算力、喂进更多数据，就能得到更好的结果？

    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?

  4. Sal CandidoBiohub

    那是个很好的处境，因为一旦找到了，你就可以开始摇曲柄，它变成了一个工程问题，而这正是我喜欢的。这和架构有关，但也和数据关系很大。如果你的数据里没有解决问题所需的信息和统计规律，你就得不到拥有你想要的能力和理解的模型。你只能从手头的数据里提取信息，再用它去泛化到数据之外。

    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.

  5. Brandon Anderson主持

    说到收集数据，你怎么考虑哪些模态最好？还有一个相关的问题：做模型的人是不是更倾向于去做有现成数据的问题，而不是最符合目标、或者对转化医学影响最大的问题？

    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?

  6. Sal CandidoBiohub

    确实是这样。我不代表所有做模型的人，但我很懒，所以我会用现成的东西。这件事有好的一面，也有坏的一面。

    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.

  7. Sal CandidoBiohub

    好的一面是：做常规机器学习的时候，你往往会去找最干净、质量最高的数据样本。但如果你像我一样，就会钻到后屋里，翻翻别人堆着的旧东西，看看有什么。一个具体的例子是用宏基因组序列训练蛋白质语言模型，那可不是质量最高的数据。说实话，我敢保证其中很多数据根本不是一个真实、完整的蛋白质。可它还是让模型的表现提升了，无论是设计真正能用的蛋白质，还是理解我们已知存在的蛋白质。

    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.

  8. Sal CandidoBiohub

    这是好的一面。但它也可能把你带到不好的地方，让你说「那就把容易生成的数据扩展上去吧」。我不认为那一定是对的做法。这也是今天这里讨论的 BBI 让我兴奋的地方之一：它是主动去问，「解决这个问题，我们需要什么数据？」

    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?"

  9. Sal CandidoBiohub

    这关乎对的资源，也关乎对的社区。我们在 Biohub 努力做的一件事，是开放地工作，和社区一起，推动整个社区往前走。这一点至关重要，因为如果只有人在建模型，我们就会偏向已经存在的数据；如果只有人在生成数据，他们就会偏向那些容易生成的东西。但如果能作为一个社区一起工作，每一步都开放，你就能弄清楚真正需要的是什么数据，然后才能找到那条扩展定律。

    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.

  10. Brandon Anderson主持

    Pushmeet，你怎么看数据的苦涩的教训？

    Pushmeet, what do you think about the bitter lesson for data?

  11. Pushmeet KohliGoogle DeepMind

    Rich 在 DeepMind 的时候我也在，那时我们就在琢磨苦涩的教训这个想法。我从 Rich 当年在 DeepMind 的那场讲座里得到的，并不是数据有没有用、或者该怎么看待机器学习这种具体的观点。我的理解是，他讲的是一件更概念性的事。

    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.

  12. Pushmeet KohliGoogle DeepMind

    有时候我们看问题，会用一种近乎信仰的方式去想解法：「我是做模型的」，或者「我是做数据生成的」。我觉得这才是苦涩的教训。如果你带着这种心态去做一个问题，可能就不会成功。问题先行，在解法上要保持灵活，两种都要试。你要理解自己想解决的问题：需要在建模上下功夫，就去建模；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.

  13. Pushmeet KohliGoogle DeepMind

    当时机器学习圈里的情况是，大家会说：「我们是机器学习研究员。数据集就在那儿，这是训练数据，这是测试数据，我们只管优化模型。」如果你的最终目标是解决问题，这种做法是有缺陷的，因为你必须同时看到进入这个过程的两个方面。

    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.

  14. Pushmeet KohliGoogle DeepMind

    而进入这个过程的不只是数据或建模，还有专业知识。比如做 AlphaFold 的时候，我们看的是用现有数据集能做到什么，因为我们既没有核心的专业能力，也没有资源去说「我们把 PDB 扩充一个数量级吧」。全世界的机构和科学家为构建那批数据投入的心血是无价的。所以在那里，你得把钱花在建模上，争取最大的回报。

    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.

  15. Pushmeet KohliGoogle DeepMind

    但在别的领域，比如细胞基因组学，我们用同样的方法问：「用细胞-基因数据能做到什么？」做了大量工作之后，结论很清楚：数据还没到位，还撑不起构建虚拟细胞这个宏大目标。这能把大家聚到一起，专注于推动科学的真正挑战，而不是信仰式地追随数据生成或建模上的进展。

    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.

  16. Brandon Anderson主持

    我很喜欢这个回答。那能落地的结论是什么？永远先定义问题，再决定在哪个解法空间里找。但更宽泛地说，在我们迈向下一代转化医学的时候，整个社区应该怎么看这件事？

    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?

  17. Pushmeet KohliGoogle DeepMind

    我给任何刚进入这个领域的人的建议是：把自己当成一个跨学科的人。先理解问题：你为什么做它？你想达成什么？然后再想需要什么，是建模、扩展算力，还是生成数据。理解问题极其重要，当然你也需要积累自己的专业能力。

    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.

  18. Pushmeet KohliGoogle DeepMind

    还有约束。也许你能生成的数据只有那么多，或者你负担得起的模型只有那么大。理解这些约束，尽量快速失败，再看哪些方法从长期看是可行的，能把你带到想要的影响力水平。

    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.

  19. Brandon Anderson主持

    这正好引出我的下一个问题。回看建模的演进，AlphaFold 2 基本上是一件艺术品，有大量精心手工设计的特征，解法的每个部分都经过非常刻意的思考。Google 或者说 Alphabet 的一些工作仍然走这条路，但普遍的共识似乎在转向更可扩展、更通用的策略。

    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.

  20. Brandon Anderson主持

    某些问题还需要手工打磨的精巧解法吗？还是资源更应该先花在规模上？资源和钱是固定的，你可以投给算力、人才或者数据。这个取舍该怎么看？

    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?

  21. Pushmeet KohliGoogle DeepMind

    我们从第一性原理来看。构建一个更好的模型，这门手艺并不是偶然得来的。我们做了很多实验，但背后有一个愿景。来自生物物理和生物化学的科学直觉告诉我们，氨基酸残基并不是各干各的，它们会受到其他残基的影响。那就把这一点直接写进模型。如果你从科学界学到了什么，就把这些信息用上，给模型那个「不公平的优势」。

    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.

  22. Pushmeet KohliGoogle DeepMind

    这让模型的数据效率高得多，因为它不必从头学一切。我还得说，整理好的数据本身就是一门艺术。不是说一句「多放点数据进去」就行了。如果你只是复制同样的数据，你哪儿也去不了。关键不只是大数据，而是好数据，以及理解要取得进展需要数据覆盖到哪些范围。这本身就是一个有意思得多、也难得多的问题。

    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.

  23. Brandon Anderson主持

    看起来你有想法，Sal。你怎么看？

    It looks like you have a thought, Sal. What do you think?

  24. Sal CandidoBiohub

    我非常同意，这取决于要解决的问题。如果数据量比较小，在模型里放更多归纳偏置会帮到你，在小数据规模下你确实需要它才能出结果。随着数据越来越多，有时模型会发现一些你原本并不知道的东西，而那些归纳偏置如果不完全正确，有时反而会拖你后腿。随着情况变好，会有一个临界点。

    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.

  25. Sal CandidoBiohub

    不过我也要对你的问题小小地提个异议，因为扩展这一侧同样有很多手艺。把模型拿来、用更多数据训练、投入更多算力、把它做大，这里面有大量算法上的工作。而且这不只是基础设施层面的事，也不只是让模型推理更快。我们正看到架构从标准 Transformer 走向更定制化的设计。我一点也不认为我们已经进入了「后 Transformer 时代」，但我们在修改这些架构，让它们更贴合用途、效果更好，哪怕是面对互联网规模的数据。

    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.

  26. Sal CandidoBiohub

    随着更多数据进来，挑战不只是整理数据、或者挑出模型需要的下一批信息，还在于每一步、每个规模上都要问：什么样的架构才能从这些信息里榨出最多东西？

    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.

  27. Brandon Anderson主持

    想想蛋白质结构预测现在的状况。新闻会说蛋白质结构预测已经被解决了。但如果你去问我的朋友们，他们会说：「这里还有太多事要做。」功能、动态和设计都还是完全开放的问题。AlphaFold 2 发布已经大约五年了，也确实取得了进展。

    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.

  28. Brandon Anderson主持

    Pushmeet，你觉得还会有下一次大飞跃吗？解决这些问题有什么拦路虎？是我们没有对的数据、算法或想法？还是只是时间问题？

    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?

  29. Pushmeet KohliGoogle DeepMind

    科学的运作方式是先把某个东西隔离出来，再一步一步取得进展。当人们说蛋白质折叠问题已经被解决，在概念层面上确实有进展。但想想「蛋白质是构建生命的积木」这种说法。蛋白质不是积木，也不像积木那样运作。我自己也一直说蛋白质是构建生命的积木，但我其实并不这么认为。

    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.

  30. Pushmeet KohliGoogle DeepMind

    蛋白质极其复杂，它们是无序的，形状可能随所处环境而变。John Jumper 和我以前就讨论过：我们到底在解决什么？我们并不知道蛋白质真正的基态，也不知道它们实际呈现的结构分布。我们当时想做的，是复现某个人测出来、存进 PDB 的那一个结构。我们做到的就是这个。而它恰好很有用。

    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.

  31. Pushmeet KohliGoogle DeepMind

    但这不代表我们已经搞懂了蛋白质动力学的全部。在最高层面，说我们取得了进展更容易传播，但在座的科学家都知道还有多少事要做。值得庆祝的东西很多，但别停止资助蛋白质结构预测和蛋白质动力学的研究，因为我们才刚刚开始。

    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.

  32. Brandon Anderson主持

    最大的拦路虎是什么？如果你能挥一挥魔杖，说「我们有更多这个了」，从而加速功能、动态或设计的研究，你会变出什么？

    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?

  33. Pushmeet KohliGoogle DeepMind

    我的背景挺杂的。我最早做安全研究，后来做计算机视觉、贝叶斯理论、判别式机器学习、深度学习、AI 写代码，最后才是科学。所以你问我这个问题，我心里那个计算机视觉研究员就会对冷冻电镜的显微照片特别兴奋。

    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.

  34. Pushmeet KohliGoogle DeepMind

    我当时想：「PDB 这些数据算什么？我应该到源头去，去看冷冻电镜的显微照片。我不要那些结构，它们肯定漏掉了大量数据。我应该直接在冷冻电镜显微照片上做。」

    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."

  35. Pushmeet KohliGoogle DeepMind

    如果能有模型在那个层面上扩展，配上合适的数据量，把其中记录下来的动态和分布信息提取出来，那会非常了不起。我试过，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.

  36. Brandon Anderson主持

    还有工作要做，但你相信这条路能给你那些信息？

    There's still work to do, but you believe that's a route that could give you that information?

  37. Pushmeet KohliGoogle DeepMind

    是的。我觉得到某个时候，也许会有比我更厉害的人去试一试，然后我们会有所突破。

    Yeah. I think at some point maybe some people better than me will take a stab at it, and we'll get somewhere.

  38. Brandon Anderson主持

    Sal，你怎么看？

    What do you think, Sal?

  39. Sal CandidoBiohub

    这挺有意思的，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解决的问题。它们服务于一个非常具体的目的，你也可以拿它们去做别的事。比如你可以用它们来设计新的蛋白质，而这未必是你一开始会想到的用途。

    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.

  40. Sal CandidoBiohub

    我觉得我们现在建的模型，就像一个想搞懂自行车怎么运作的人，却在给车上的一根辐条建模。这些模型会越做越好，但你真正需要的是从辐条的模型走到车轮，再走到整辆自行车，因为那才是人们想理解的。把这个比喻再往下说，感觉就像人们想用一辆自行车的模型，去设计一辆皮卡上的零件。

    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.

  41. Sal CandidoBiohub

    当我们把这些模型放进它们实际运作的具体生物学环境里，就能在更大的尺度上了解这些相互作用。我认为事情正朝这个方向走。但我同意，我们应该继续做折叠模型，因为它们还会越来越好。

    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.

  42. Brandon Anderson主持

    说到设计，有一句著名的费曼名言：「我不能创造的东西，我就不理解。」现在我们身处的世界，很容易在完全不理解的情况下创造出东西。有能帮助人类理解事物的模型，和有那种能一次就做出皮摩尔级结合剂之类东西的神奇黑盒，相比之下前者有多重要？

    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?

  43. Sal CandidoBiohub

    早在 AI 出现之前，我们就一直在用神奇的黑盒设计东西了，从某种意义上说那仍然有用。但理解真的很重要，这也是我思考 AI 时想得最多的事情之一。

    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.

  44. Sal CandidoBiohub

    这些模型能学的东西太多了。随着智能越来越便宜、越来越容易获得，你可以部署它去越来越多地了解世界上正在发生的事。但怎么把那些知识从机器里拿出来，让我也能理解？也许这只是我小众的好奇心吧。我觉得有太多东西可以学了。

    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.

  45. Brandon Anderson主持

    很多科学家真的很想把事情搞懂，终点对他们来说可能没那么重要。但我们在这里，是要把转化医学当作一个问题来解决的，对吧？

    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?

  46. Sal CandidoBiohub

    我觉得你越往深里挖，就越能找到继续推进的正确方向。有一点让我印象特别深：这些模型里包含的信息，比我们知道的要多得多。

    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.

  47. Sal CandidoBiohub

    比如我们在模型的可解释性上下了很多功夫，在里面能找到很多信息。大家都知道蛋白质语言模型会在表征里学到某种结构的概念，但我们发现里面也有关于功能和运动的信息。即使是现有的模型，也还有很多东西有待解锁。在我们提升它们能力的同时，理解这一点很重要。

    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.

  48. Sal CandidoBiohub

    说到底，把这么多信息压缩进一个模型，你会期待一个世界模型从中浮现。它是怎么完成任务的？它是怎么设计出一个蛋白质的？它把来自进化的信息压缩进了模型。除了能给我们产出有用的东西，光是看看它里面有什么，肯定也能学到东西。

    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.

  49. Brandon Anderson主持

    你怎么看，Pushmeet？设计还是理解？

    What do you think, Pushmeet? Design versus understanding?

  50. Pushmeet KohliGoogle DeepMind

    我的看法不太一样，我认为某种程度的理解是必要的。让我解释一下我指的是哪个程度。

    I have a different take in the sense that some level of understanding is necessary. Let me explain what level I mean.

  51. Pushmeet KohliGoogle DeepMind

    AlphaFold 并不完美。AlphaFold 2 当时在那个测试集上的 GDT 分数大约是 90。但即使它的 GDT 分数是 95，如果它的 pLDDT 置信分完全没有校准，谁会相信它？想象一下，它神奇地给出了好答案，却在错的时候也告诉你它非常有把握，然后你在上面干了一整年，才发现它完全错了。不确定性指标的校准极其重要。

    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.

  52. Pushmeet KohliGoogle DeepMind

    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上取得了很大进展。这和理解它内部是怎么找到答案的不是一回事。在那一点上，我同意 Sal 的说法：可解释性问的是，它为什么有效？它为什么给出这个答案？

    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?

  53. Pushmeet KohliGoogle DeepMind

    在最高层面，AlphaFold 2 在行为和泛化能力上是可解释的，而且我们在发布之前并没有把这些全部摸清。我们发布 AlphaFold 2 并公开权重之后，大家发现它是一个很棒的无序蛋白预测器，能判断出蛋白质里哪些部分是无序的。这说明它能泛化。

    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.

  54. Pushmeet KohliGoogle DeepMind

    但可解释性还会引出另一个问题：对谁可解释？如果你说的是对人类这种理性系统可解释，受限于人脑的认知和计算能力，那么不，AlphaFold 2 是不可解释的。但如果你问的是，AlphaFold 2 的工作方式对一个大得多、复杂得多的模型来说是否可解释，也许是的，只是我们看不懂。

    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.

  55. Pushmeet KohliGoogle DeepMind

    作为 AlphaFold 2 的使用者，我们确实需要理解它能做什么、不能做什么。理解它的行为特征、长处和局限极其重要。没有这种刻画，我们不能就这么用这些模型，否则它们非但帮不上忙，还可能伤害我们。

    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.

  56. Brandon Anderson主持

    所以你的看法是，严格意义上的可解释性并不是必需的，大家应该关注的是确保模型值得信任，让人可以据此做出能落地的决定。

    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.

  57. Pushmeet KohliGoogle DeepMind

    没错。可解释性也是因人而异的：谁在解释它？如果是一位人类科学家想解释模型怎么做出预测，那和给另一个大得多的 LLM 开放激活层，问它「你能预测 AlphaFold 会怎么做吗？」是不同的问题。也许未来那些前沿模型能预测出来，并提出一套理论，说明 AlphaFold 2 是怎么理解和产出结果的。

    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.

  58. Brandon Anderson主持

    还剩两分钟，最后问两位一个问题。AI 很有可能在不远的将来大幅改善人类健康。我喜欢量化的预测。请给出你们的最佳猜测：还要多久我们才会开始在临床上看到 AI 的成果？Pushmeet，你先来？

    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?

  59. Pushmeet KohliGoogle DeepMind

    我觉得「AI 在临床上的成果」是个提法有问题的问题。今天 AI 已经被用在药物发现流程的每个环节里了。从这个角度说，它已经到了。

    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.

  60. Pushmeet KohliGoogle DeepMind

    但如果你问的是什么时候能看到时间线加速 10 倍或 100 倍，那下一个问题就是：我们在加速什么？是先导化合物优化？靶点发现？还是临床前研究或毒理学？

    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?

  61. Pushmeet KohliGoogle DeepMind

    在接下来几年里，我们需要攻克生物学里的一些硬骨头，这也是今天宣布的这项工作极其重要的原因。只有那样，我们才能真正解锁那种大幅改变药物发现的加速。AI 会一直被用在走进临床的东西里，但更大的加速，只有在更好地理解这项工作想要构建的生物学模型之后才会到来。

    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.

  62. Brandon Anderson主持

    好的，谢谢。Sal，如果可以的话快速回答一下，我们快没时间了。

    All right, thanks. Sal, a quick answer, if you can. We're almost out of time.

  63. Sal CandidoBiohub

    我真想戴上我的生物黑客帽子来回答这个问题。重要的是想清楚，推动这个领域往前走意味着什么。我们真正想看到的是结果被改变。什么时候会有一种药完全由 AI 做出来？我不知道，这很难预测。但我确实认为我们会很快看到快速的进展，因为这些工具都已经在用了。

    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.

  64. Sal CandidoBiohub

    回到你说的基础研究和基础理解，我很久以前在 Google 学到一件事：有时候，问「怎么才能提升 10 倍」比问「怎么提升 10%」更容易找到出路。不是因为 10 倍的路一定更容易走，而是它让你把视野放宽，看到你一直没去尝试的解法。你会回到第一性原理去问：「我们要怎么做到？我们要怎么真正把它推上去？」

    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?"

  65. Sal CandidoBiohub

    10% 和 10 倍这两种思路都有价值，我们两种都应该做。我很高兴在 Biohub 我们有一个美好而远大的使命：治愈所有疾病。如果想做到这一点，你真的需要想清楚那条 10 倍的路是什么。能去做这件事，是一个很好的机会。

    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.

  66. Brandon Anderson主持

    太棒了。谢谢两位，谢谢你们坐在这名副其实的「热座」上。非常有意思。

    Awesome. Thank you both. Thank you for being in the literal hot seat. Very interesting.

本页英文取自节目官方发布的文字稿，时间点按节目音频对齐，中文由 AI 翻译，可能存在对齐和翻译误差；以节目原文为准。发现错漏欢迎告诉我们。

⌘K搜索知识库

最近收录

正在载入知识索引…