推荐微信扫码登录以获得更好的体验!另:🎉建议更新到最新版,以体验新功能!🎉 版本更新记录→

人类医学史最大拐点?扎克伯格放出狠话:在AI面前,百年前的目

字幕摘录

时间英文中文
0:00We just want to give tools to the whole scientific community.我们只是想给整个科学界提供工具。
0:02We want to understand how biology works.我们想知道生物学是如何运作的。
0:04I want to understand the genetics of this person.我想了解这个人的遗传学
0:06I want to understand the risks they have to different illnesses.我想了解他们感染不同疾病的风险
0:09My goal is to be able to treat the individual as an individual,我的目标是能够把个人当作个人对待,
0:13understand the mechanisms, and be able to intervene.了解各种机制,能够进行干预。
0:17We'll have a bigger impact by getting this in more scientists' hands quicker通过让更多的科学家 更快地掌握这个,我们就能产生更大的影响
0:20by doing it as open-source projects instead.作为开源项目处理。
0:23It's not just like there's some factory somewhere that you can pay to produce the data.这不仅仅是一些工厂 你可以支付生产数据。
0:27You actually need to invent new novel scientific approaches.你实际上需要发明新的科学方法。
展开字幕全文(978 条)
序号英文中文
1We just want to give tools to the whole scientific community.我们只是想给整个科学界提供工具。
2We want to understand how biology works.我们想知道生物学是如何运作的。
3I want to understand the genetics of this person.我想了解这个人的遗传学
4I want to understand the risks they have to different illnesses.我想了解他们感染不同疾病的风险
5My goal is to be able to treat the individual as an individual,我的目标是能够把个人当作个人对待,
6understand the mechanisms, and be able to intervene.了解各种机制,能够进行干预。
7We'll have a bigger impact by getting this in more scientists' hands quicker通过让更多的科学家 更快地掌握这个,我们就能产生更大的影响
8by doing it as open-source projects instead.作为开源项目处理。
9It's not just like there's some factory somewhere that you can pay to produce the data.这不仅仅是一些工厂 你可以支付生产数据。
10You actually need to invent new novel scientific approaches.你实际上需要发明新的科学方法。
11The theory isn't that we're going to cure the diseases.理论不是说我们要治好这些疾病.
12我们没有 We're not.
13It's that we want to help accelerate the pace of progress for the whole scientific field.我们想帮助加快整个科学领域的进步步伐.
14We folded over 1.1 billion proteins and predicted their structures.我们折叠了超过11亿个蛋白质,并预测了它们的结构。
15And we didn't design a model for antibodies.我们没有设计抗体模型。
16We didn't design a model to be able to find one particular target.我们没有设计一个模型来找到一个特定的目标。
17We just designed a model that could understand proteins.我们刚刚设计了一个能理解蛋白质的模型。
18If we could design a protein to actually change the physiology,如果我们能设计一种蛋白质来改变生理
19then we can actually cure someone.这样我们就能治愈某人了
20Today on No Priors, we're joined by Mark Zuckerberg, Priscilla Chan, and Alex Reeves.今天在"No Presidents"节目中,我们与马克·扎克伯格,普里西拉·陈(英语:Priscilla Chan)和亚历克斯·里夫斯(英语:Alex Reeves)一起演出.
21We'll be talking about Biohub and all their various efforts to now start applying AI at scale我们将讨论Biohub 和他们所有的努力 现在开始大规模应用AI
22to do world models of cells and different levels of interactions across biology.做世界细胞模型 和不同水平的相互作用 跨生物学。
23Mark, Priscilla, thank you for doing this.马克,普瑞西拉,谢谢你的帮助
24Yeah, thanks for having us.是啊,谢谢你邀请我们。
25Great to be here.很高兴在这里。
26Alex, congratulations on new missions.亚历克斯,恭喜你的新任务
27Thank you.谢谢
28You guys made Biohub your primary philanthropic effort你们把Biohub作为你们的主要慈善事业
29and then committed $500 million to this virtual biology initiative.然后为这个虚拟生物学倡议承诺5亿美元。
30Can you tell us a little bit about why do that and how did you go from你能跟我们说说为什么吗?
31we should fund this to this is like who we are?我们应该为此提供资金 就像我们是谁?
32So Biohub in its current form, we're super excited about.所以现在的Biohub 我们非常兴奋
33We feel like it's a really good fit for who we are and what we bring to the table我们觉得这很适合我们自己和我们带来的东西
34and what we can achieve together.以及我们能一起实现的
35But this work started 10 years ago when we were thinking about how can we give back.但是,这项工作始于10年前,当时我们正想着如何回报。
36And Mark wanted to build an organization that could cure, prevent,马克想建立一个组织 可以治愈,预防,
37and manage all disease by the end of the century.并管理所有疾病 到本世纪末。
38And we had a series of hilarious meetings with scientists that我们和科学家们举行了一系列有趣的会议
39famous Nobel Prize-winning scientists were just laughing at us.著名的诺贝尔奖得主科学家只是嘲笑我们。
40Was that your starting line?那是你的起点线吗?
41We're just going to cure all disease?我们只是要治好所有疾病?
42No, no.没有,没有。
43And to be clear, we don't think that we're going to be the ones curing the diseases.且明说,我们不认为我们是治疗疾病的人。
44Our goal was always to build tools that could accelerate the whole scientific fields.我们的目标总是建立能够加速整个科学领域的工具。
45That way, the scientific field collectively could cure all the diseases.这样,科学领域就能共同治愈所有疾病。
46But still, people thought that by the end of the century was a stretch.然而,人们仍然认为,到本世纪末,这是一个延伸.
47Now, I think it's too conservative.我觉得太保守了
48And so we kept being like, okay, well, we had these series of funny, awkward,所以我们一直想,好吧,好吧,我们有 这些有趣的,尴尬的系列,
49educational conversations where we were like, okay, but why?教育对话 我们像,好吗, 但为什么?
50Why do you think it's impossible?何以以为不可.
51And just being the person in the room was just like, oh, I don't know why.而只是作为人 在房间里只是一样, 哦,我不知道为什么。
52You tell me.你告诉我。
53Finally, we got people to like, they're like, fine, if you really must know.最后,我们得到了人们喜欢, 他们喜欢,他们一样,很好, 如果你真的必须知道。
54And we're like, you know, we do.我们就像,你知道,我们做到了。
55It seems important.这似乎很重要。
56They were like, well, we work in silos.他们就像,嗯,我们工作 在仓。
57And when you publish, information doesn't get shared.而当你发布时,信息不会被分享.
58It gets locked up for long periods of time.它被锁了很长时间。
59And we don't have tooling.我们没有工具。
60They gave the example of like, we build a great tool by one postdoc in the lab,他们举例说 我们用实验室的博士后设计了一个伟大的工具
61and it lives on their computer.它生活在他们的电脑上。
62And when they graduate, the tool is gone.当他们毕业,工具就不见了。
63And they just, it was, what we heard was very hard to build shared tools to move他们只是,这是,我们听到的 很难建立共享工具 移动
64science faster, build the shared knowledge base to quickly move science faster.科学更快,共享知识库建设迅速推进科学建设.
65And that's sort of where we began and thinking about, okay, like, if those are the problems,这就是我们开始思考的地方, OK,喜欢,如果这些都是问题,
66like, what can we contribute?像,我们可以贡献什么?
67Yeah, I mean, so the original Biohub model was basically focus on long-term tool是啊,我的意思是,所以最初的Biohub模型 基本上专注于长期的工具
68development by bringing together engineers and scientists across multiple universities.将多所大学的工程师和科学家聚集在一起,从而实现发展。
69To focus on long-term tool development.注重长期工具开发.
70And it basically, it like worked.基本上,它就像工作。
71And, you know, we started off with CZI doing a number of different things.我们从CZI开始做一些不同的事情。
72And I think over time, we just felt like, okay, the science piece is really working.我想随着时间的推移,我们只是觉得,好吧, 科学作品真的起作用了。
73And we just kept on investing more and more and more in it until now it is basically the而我们只是继续投入 越来越多的投资到现在
74primary and main thing that we're doing.我们的首要任务
75And we've expanded the original San Francisco Biohub to a handful now at this point.我们把原来的旧金山比奥胡布 扩大到了一小撮
76There's New York, there's Chicago.纽约,芝加哥。
77The real focus and the unifying theme at this point is the virtual biology initiative around目前真正的重点和统一主题是虚拟生物学倡议
78taking the unique data sets that are able to be generated in order to model effectively使用能够生成的独特数据集以有效建模
79starting with the smallest pieces of proteins, but then eventually cells and whole biological首先是最小的蛋白质, 但最终细胞和整个生物
80systems.系统。
81But that's kind of how we've evolved is, you know, this idea that we talk about around但这就是我们进化的方式, 你知道,这个想法,我们谈论
82that some of this is an AI problem and you want to build a frontier AI lab,有些是人工智能问题 你想建立一个前沿人工智能实验室
83but you need to couple that with a frontier biology effort that can do the work of basically但是,你需要 与边疆的生物学努力 基本上可以做的工作
84being able to understand and get the data that you need to actually be able to build these能够理解和获得你需要的数据 真正能够构建这些
85models.模特儿们
86Because unlike language models, where there's just like a lot of data out there on the internet,因为和语言模型不同 互联网上有很多数据
87that's not really the case with biology.生物学上的情况并非如此
88There are obviously a bunch of different data sets that exist that academia and scientists很明显, 学术界和科学家们有一堆不同的数据集
89have generated over the decades, but a lot of the stuff that I think we want to put into this,已经产生了几十年, 但是很多的东西 我想我们想放进这个,
90it doesn't exist.它不存在。
91It's like you want to be able to visualize things that people haven't been able to see就像你想能够想象人们所看不见的东西
92before, which is why we're doing the imaging work.之前, 这就是为什么我们做成像工作。
93You want to be able to record things that are going on inside the body, which is why你希望能够记录 身体内部发生的事情, 这就是为什么
94we're doing the kind of cellular engineering work.我们正在做细胞工程
95You want to be able to measure things like inflammation in ways that haven't been possible,你想用不可能的方法 来测量炎症
96which is why the Chicago Biohub is focused on building those kind of devices and being这就是为什么芝加哥的Biohub 专注于建造这些设备 并成为
97able to do that.能够做到这一点。
98And that will fundamentally create new types of data sets that will allow new types of这将从根本上创造新的数据集类型,允许新的类型。
99models.模特儿们
100And I think it's just a very exciting thing that I'm going back to what you're saying.我认为这只是一件非常激动人心的事,我要回到你所说的。
101If the scientific field, it primarily needs kind of tool development that now is going如果科学领域,它首先需要一种 工具开发,现在正在进行
102to empower scientists across the field to be able to do their work faster.增强整个领域科学家的能力,使其能够更快地开展工作。
103That's what we think we can provide through this kind of long-term focus on tool development.我们认为,通过这种长期关注工具开发,我们可以提供这些.
104I think there's a fun through line on where we started and bringing us to our work that我觉得我们从何而来 带我们参加我们的工作
105Alex is driving now, is that our very first request for application RFA here was aroundAlex正在开车 是因为我们第一次申请RFA 是在附近
106single cell sequencing.单细胞测序.
107And we wanted to look at sort of like the RNA that is transcribed in individual cells.我们想看看在单个细胞中转录的RNA。
108And that was possible, but it was still pretty early on in understanding how different cells有可能,但还很早才明白 如何不同的细胞
109were expressing their DNA to the point where at the beginning we were just funding methods,他们的DNA表达到 我们一开始只是资助方法
110like getting people to describe how to do it so that others could share that methodology.比如让人们描述如何做到这一点,以便其他人可以分享这种方法。
111And then that became us funding the Human Cell Atlas, which is now one of the largest之后我们资助了人类细胞图集,它现在是最大的
112databases of single cell transcriptomes.单细胞转录器的数据库。
113It was getting hard for scientists to annotate the data.科学家很难对数据进行注释。
114So we built Cellbygene, which was like a very simple annotation tool that scientists could于是我们建立了Cellbygene, 这就像一个非常简单的注释工具 科学家可以
115use to make use of that data.利用这些数据。
116然后一个社区来到了Cellbygene周围, 建造在Cellbygene周围,
117more and more data that we had nothing to do with sort of creating or funding or making越来越多的数据,我们与创造、资助或制作无关
118happen in the world.发生在世界上。
119And now Cellbygene is a corpus of knowledge that a lot of the transcriptomic-based models而现在的Cellbygene 是一个知识库 很多基于记录的模型
120are based off of and is used regularly by the scientific community.以科学界为基础并定期使用。
121But still there are always critiques, like this is just stamp collecting.但还是有批评, 就像这只是邮票收集。
122You're just gathering bits of data, and we're not going to be able to pull scientific你只是收集一些数据 我们无法利用科学
123knowledge and wisdom and insights out of.知识、智慧和洞察力。
124And we're like, well, we didn't have an answer for a while.我们好像有段时间没有答案了
125And then imagine our delight when large language models became a huge topic of conversation然后想象一下当大语言模型成为大对话话题时 我们的快乐
126that could make sense of large amounts of data.这可以解释大量的数据。
127And for me it's like, what if we could actually understand how biology worked?对我来说,它就像, 如果我们能真正理解 生物学是如何工作的?
128Move it from a discovery-based science to an engineering-based science, where we could把它从基于发现的科学转移到基于工程的科学, 在那里我们可以
129systematically understand how living beings, living cells, worked and be able to understand系统地理解生命、细胞、工作方式和理解
130why things go wrong.为什么事情会出问题。
131And so when we saw that moment, we're like, this is it.所以,当我们看到那一刻, 我们喜欢,这就是它。
132Something really big could happen here.这里可能会有大事发生
133Alex 你从MetaFair开始 但你正准备...
134a team at evolutionary scale and you raised venture and you were making progress in your一个进化规模的团队 你提升了风险 并在你的
135models.模特儿们
136What was the pitch from Mark and Priscilla where you said, like, that's actually the马克和普里西拉的音响是什么?
137right way to go after the mission?任务完成后走的路吗?
138Well, I think for me it was really kind of the moment when I understood that, you know,我想对我来说,这真的是 那种时刻,当我明白,你知道,
139they really saw this as an integration of frontier AI and frontier biology.他们真的把这看作是前沿AI和前沿生物学的结合.
140And I think I had developed conviction that, you know, this is really a new era of science我想我已经产生了信念,你知道, 这是一个科学的新时代
141that's just beginning kind of what's going to be possible with artificial intelligence.这才刚刚开始 人工智能是可能的
142And, you know, we're in the age of information theory at scale and we have these systems that而且,你知道,我们处于信息理论的时代 在规模,我们有这些系统
143can basically kind of predict the next token and they can, you know, learn world models基本上可以预测下一个信号 他们可以,你知道,学习世界模型
144from that.从之.
145They can learn biology from the data.他们可以从数据中学习生物学.
146And so, you know, I think that it just, it was really clear that, you know, to build所以,你知道,我认为它只是, 它真的很清楚,你知道,
147kind of that next kind of institution for the next era, you would really need to have那种下个时代的下一种机构, 你真的需要拥有
148frontier artificial intelligence.前沿人工智能.
149You would have to have frontier biology.你必须有前沿生物学。
150You would need to start to put those things in feedback and really have models that are你得开始把这些东西放进反馈中 并且真的有模型
151learning from the biology.学习生物学。
152And I think, you know, it's just, and you need the right scale and the right people.我认为,你知道,这只是, 你需要正确的尺度和正确的人。
153And so this just really felt, I think, like the way to do that.所以,这真的感觉,我认为, 像这样做的方式。
154There's a variety of different models that you all have been working on.有很多不同的模型,你们都在研究。
155And I think it's kind of interesting because some of the earliest breakthroughs in biology我认为这很有趣 因为生物学上一些最早的突破
156were things like Alpha Fold where, you know, it was a Google model that showed that you就像Alpha Fold一样 谷歌模型显示你
157could do protein folding at scale in a really interesting way that people didn't realize可以做蛋白质 规模叠叠 真正有趣的方式 人们没有意识到
158was very tractable.非常可拉动。
159And this was pretty sort of the really big transformer waves that came later.这是相当的 真正的大变压器波 之后。
160And then you're working on a variety of different things at different scale, right?然后,你的工作 各种各样的不同的东西 在不同的尺度,对不对?
161You're doing incremental molecular modeling and protein folding.你在做增量分子模型和蛋白质折叠。
162You're doing cell-based stuff.你在做基于细胞的东西。
163You're thinking about interrogating larger scale systems in biology.你在考虑在生物学中 审讯更大的规模系统
164How well do you think that extends from sort of the micro to the macro?你认为从某种微观到宏观能有多好?
165You mentioned almost starting with building blocks and building up.你提到几乎从构件和积木开始。
166But modeling cellular behavior is very different from modeling protein folding.但建模细胞行为与建模蛋白折叠有很大不同.
167The data is very different.数据差异很大。
168The modeling is different.模特儿不一样
169I'm just curious, like, do you think it's all similar in terms of this data?我只是好奇,喜欢, 你觉得这 数据都是相似的?
170And you train stuff or do you think it's actually there's some differences in terms你训练的东西 或者你认为它 实际上是有一些不同的条件
171of how you actually have to deal with these systems?你如何应对这些系统?
172I mean, there are probably some differences.我是说 可能有些不同
173I mean, you can probably talk more to the specifics around this.我的意思是,你也许可以多谈谈 围绕这个的细节。
174But like, I mean, I think each layer is going to end up being但是,就像,我的意思是,我觉得每一层 都会是
175somewhat qualitatively different, right?质量上有点不同,对吧?
176I mean, but you need to be able to understand the protein interactions in order to be able我的意思是,但你需要能够 理解蛋白质的相互作用
177to understand how cells work.了解细胞是如何工作的
178So you can't just go straight to cells in a way without understanding the protein modeling.所以,你不能只是直接去细胞的方式 不理解蛋白质模型。
179And then if you're trying to understand something like the, you know, the way the如果你是想了解 类似的东西,你知道的,方式
180system works or a bunch of cells interact together, then, you know, it's tough to do系统工作或一群细胞相互作用, 那么,你知道,这很难做
181that without first understanding cells.没有第一理解细胞。
182I mean, you might be able to add like a very high level of abstraction, simulate a system.我的意思是,你也许可以像一个高度的抽象, 模拟一个系统。
183But if you really want to like understand how it's going to work, you kind of want to但是,如果你真的想 了解它是如何运作的, 你有点想
184build the simulations at each level hierarchically.分层次构建模拟。
185So that's basically the approach that we're going through, starting with the building所以基本上这就是我们要走的方法, 从大楼开始
186blocks and the protein.块和蛋白质。
187But yeah, I mean, I think that there's going to be different types of data that you want但是,是的,我的意思是, 我认为会有 不同类型的数据 你想
188to collect for each.用来收集每个。
189The modeling techniques, I think we'll see.建模技术,我想我们会看到。
190That'll all keep on advancing across the board.这一切都会继续 全面推进。
191But I do think that a big part of the strategy is this view that you need to build it up但我确实认为这个策略的一大部分 是这个观点 你需要建立起来
192hierarchically.分等级。
193And, you know, one of the things that's unique about us in the space is we were very在太空中我们独有的一件事 就是我们非常
194intentional that the AI efforts and the wet lab efforts were a single effort.故意认为人工智能和湿实验室的努力是单一的努力。
195And we've done a lot of work to bring them together.并且我们做了很多工作来使他们团结起来.
196And the really neat thing that we can do is really try to pull and gather data that helps我们能做的非常整洁的事情是 真正尝试拉动和收集有助于
197{\fn黑体\fs20\shad2\2aH82\3aH20\4aH33\fscx95\3cH592001\be1}我们跨越等级
198You know, you can look at transcriptomics with space within a cell and look at where你知道,你可以看 记录片与空间 在细胞里,看看哪里
199it's localizing.它正在本地化。
200We can look at translucent zebrafish and look at the development across different cells我们可以观察半透明斑马鱼 观察不同细胞的发展
201and when the brain develops.而当大脑发展。
202We have sensors that allow us to look at cell-cell communication in different molecules.我们有感应器 能够让我们观察不同分子的细胞间通信
203And so we can be strategic about the types of experiments and data we want to collect这样我们就可以对想要收集的实验和数据类型 采取策略
204that helps us bridge across these, that makes it so that there's some connective tissue帮助我们跨越这些, 这使得有一些连接组织
205that helps drive the modeling magic that happens.帮助驱动 模拟魔法发生。
206Yeah, the reason I ask the question, by the way, is I used to be a biologist.是的,我问这个问题的原因,顺便说一句, 我曾经是一个生物学家。
207So I have a PhD in biology and I worked in wet labs for almost a decade and everything所以我有生物学博士学位 我在湿实验室工作了将近十年
208else.别的
209Are you looking for a job?你在找工作吗?
210We can talk about that later.我们可以稍后再谈
211Not a no.没有
212I'm like Danny Glover, you know, and I'm almost at retirement.我就像丹尼・格洛弗,你知道, 我几乎在退休。
213But I think, you know, one of the things that was always lacking was this integrative nature但我认为,你知道, 总是缺少的一件事 就是这种融合性
214across the different layers of biology and the developmental biologists would work on不同层次的生物学和发展生物学家将致力于
215their own, the molecular biologists would be doing different experiments.他们自己的,分子生物学家 会做不同的实验。
216And so that's what I was curious about.这就是我想知道的。
217Typically there's a reductionist view of biology and there's a systems view.典型的生物学观点有递减论,系统观点也有.
218And those people didn't really work together deeply.这些人并没有真正合作。
219And so one of the exciting things about what you're doing actually is how you're bridging所以一个令人兴奋的事情 关于你在做什么 实际上,你是如何连接
220that.那个
221And so that was kind of the basis for the question as well.因此,这也是问题的基础。
222Yeah.对
223And if I could add something there, you know, I think that, you know, we're in如果我能增加一些东西, 你知道,我认为,你知道,我们
224the age of this kind of information theory in biology.这种信息理论在生物学中的年代。
225And so, you know, there are levels of complexity and hierarchy and biology and kind of each所以,你知道,有 复杂程度 层次和生物学 和类型
226level is made up of and, you know, constituted by the lower levels.级别由较低级别组成,你知道的。
227And so as you want to have that kind of more complete description, you want to have systems所以,既然你想有那种 更完整的描述, 你想有系统
228that can really generalize and begin to actually answer, you know, experimental questions它可以真正概括 并开始实际回答, 你知道,实验性的问题
229digitally that you could ask in the lab.数字化的,你可以问 在实验室。
230You know, you need to have kind of the right你知道,你需要 有某种权利
231basis for modeling at every level.各级建模的基础。
232And so I think what's really unique about所以我认为什么是真正独特的
233what we can do is to, as Priscilla and Mark were saying, you know, really build我们可以做的是,正如Priscilla和Mark说, 你知道,真的建设
234information at each of these different layers, collect them, collect kind of those connection收集信息 收集信息 收集这些连接
235points, but then also really kind of do it at the scale that will reveal that underlying点数,但同时也是真正的类型 做到这一点的规模 将揭示基础
236information architecture.信息架构。
237And that's going to be really critical to actually be able to build这对真正能够建造
238digital representations that can answer new experimental questions.可以回答新的实验性问题的数字化表示.
239One of the things that inspires me most about this effort is really what Priscilla said,我对这项努力最有启发性的一件事 就是普里西拉说的
240which is like, well, there's so much we actually don't understand about biology and what if这就像,嗯,有这么多 我们实际上不明白 关于生物学,如果
241we could, which I think is actually very different from lots of other incredibly我们可以,我认为 这实际上 与很多其他不可思议的
242interesting and useful AI problems we attack.有趣的和有用的AI问题 我们攻击。
243We're like trying to replicate human behavior.我们就像试图复制人类的行为。
244I'm like a lot of that data is, you know, on the internet or captured without pretending to我就像很多数据是,你知道, 在互联网上或者捕捉 不假装
245understand all human behavior.了解所有人类的行为
246You can predict a lot of it.你可以预测很多
247I thought one of the most interesting我觉得最有趣的一个
248things in your release was actually, you know, the like mechanistic interpretability stuff you你释放中的东西 其实是,你知道, 像机械解释的东西你
249alluded to, which is, can we actually extract new knowledge from, you know, what the model believes意思是,我们能从模型所相信的东西中 吸取新的知识吗?
250is happening?发生什么了?
251Right.对
252Can you talk a little bit about that?你能谈谈吗?
253Yeah, I'm really excited about that.是的,我真的很兴奋。
254I think, you know, in mechanistic interpretability, kind of traditionally it's been applied to large我认为,你知道,在机械解释性, 那种传统上 它被应用到大
255language models with the goal of understanding, you know, kind of what is the representation语言模型的目标是理解, 你知道,什么是代表
256space of a large language model?大型语言模型的空间 ?
257How does it compute things and does that really connect to,它是如何计算事物 和真正连接,
258you know, what we understand about our intuitive understanding of the world?你知道,我们对世界的直觉了解多少?
259And so there's,所以说,
260I think, this really rich toolkit that has been developed to start to be able to ask those我认为,这个非常丰富的工具箱 已经开发出来 开始问那些
261questions.问题。
262So kind of what does that mean for biology?这对生物学意味着什么?
263One of the classes of models that we我们的模型之一
264train are these protein language models.火车就是这些蛋白质语言模型。
265So they're really, you know, just trained on the所以,他们真的,你知道, 只是训练
266codes of proteins.蛋白质编码.
267And so anything they learn about biology is kind of emergent.所以他们学到的生物学知识 都有点突然。
268And we've seen that我们看到了
269they can learn things like biological structure and biological function.它们可以学习生物结构 和生物功能
270And that's just kind of而这只是一种
271emergent from this, you know, token prediction training task.从这个,你知道的, 象征性的预测训练任务。
272So, you know, as we think about所以,你知道,我们想
273mechanistic interpretability in those models, you know, we're really seeing the unknown because the这些模型的机械可解释性, 你知道,我们真的看到了未知的,因为
274models have been trained on billions of protein sequences.模型接受了数十亿个蛋白质序列的培训.
275They've been trained on, you know,他们受过训练
276both known and unknown biology.已知和未知的生物学
277And yet they're developing these representations that start to然而,他们正在开发 这些陈述开始
278kind of capture things that we can really see correspond to that reductive picture of biology那种捕捉的东西,我们真的可以看到 对应的还原 生物图片
279that's been built up over the centuries.这是几个世纪以来建立的
280So kind of, you can start to connect the dots between所以,你可以开始连接点之间
281proteins where we kind of really don't know anything about them with proteins where we do蛋白质,在那里我们真的不知道 与蛋白质,我们做什么
282know something because there's that kind of underlying structure grammar that's linking知道一些事情,因为有 那种基础结构语法 是连接
283them in the representation space of the model.在模型的表示空间。
284And at the extreme, it could be, you know,在极端,它可能是, 你知道,
285we're going to understand systems in the body that we didn't before, or the mechanism of action for我们要了解体内我们以前没有的系统 或行动机制
286a new treatment, because we can ask the model, right, interrogate that representation.新的治疗,因为我们可以问模型, 对,询问这个代表。
287That's这是
288right.对
289The hope is that you kind of really learn the underlying basis for how it's making the希望你们能真正了解 如何创造
290predictions.预言。
291And so you open up the black box and you can actually understand kind of the biology所以你打开黑匣子 你就能真正理解生物学
292that the model is representing.模型是代表着。
293So asking for a friend, you know, you guys all believe in你们都相信
294venture-backed companies as a way to have impact on the world.风险支持的公司是影响世界的一种方式。
295What was it like collecting data on它是什么样子的数据收集 关于
296zebrafish or the span of the data or the wet lab work or just the scale?斑马鱼或数据长度 或湿实验室工作或只是尺度?
297Like what makes this a就像是什么使这个
298fit for this big nonprofit ecosystem effort versus a venture-backed company?适合这个大型的非营利生态系统 相对于一个风险支持的公司?
299Well, I think we just want to give tools to the whole scientific community.我想我们只是想给整个科学界提供工具。
300And I think in order我按顺序思考
301to have the biggest impact, part of it is just we're, I mean, it's not actually clear that we影响最大, 一部分只是我们,我的意思是, 它实际上不清楚,我们
302couldn't run it as a business if we wanted to.如果我们想的话 就不能当生意经营
303I just think that we'll have a bigger impact by我只是觉得,我们会有一个更大的影响 通过
304getting this in more scientists' hands quicker by doing it as open source projects instead.把它作为开源项目 更快地放到更多的科学家手中
305So yeah, I mean, I think that that's kind of the approach, but I don't know.所以是的,我的意思是,我认为那是一种方法, 但我不知道。
306It's an interesting这很有趣
307question.问题。
308I'm not sure that, I mean, obviously you were doing it as a for-profit company,我不确定,我的意思是,很明显, 你这样做作为一个盈利公司,
309a bunch of the modeling before, then you run into certain issues.之前很多模特儿,然后你碰到一些问题
310I mean, you have to raise我的意思是,你必须提高
311a large amount of money in order to build the compute clusters.为了建立计算组,需要大量资金。
312I think in a lot of ways,我觉得很多方面,
313the data is actually even more of a constraint.数据实际上更是一个制约因素。
314And because if you look at like the scale of狦癬ㄓ钩
315models compared to language models, they're smaller, but they're smaller because the amount与语言模型相比,模型较小,但因为数量少
316of data is less.数据较少。
317In order to get the data, it's not just like there's some factory somewhere that为了得到数据 不只是在某处有工厂
318you can pay to produce the data.您可以付费生成数据。
319Like you actually need to invent new novel scientific approaches to就像你实际上需要发明新的科学方法
320be able to do the, for example, the type of cellular engineering we're doing in New York能够做,比如说, 我们在纽约做的蜂窝工程的类型
321or the types of devices in Chicago, which is why when we're talking about this concept of或芝加哥的装置类型, 这就是为什么当我们谈论这个概念时,
322frontier biology and frontier AI, the frontier biology is you need to do real science to advance边疆生物学和边疆AI,边疆生物学是需要做真正的科学才能进步的.
323different biological methods in order to be able to observe the things that create the data that不同的生物方法 以便能观测 产生数据的东西
324go into the model.进入模型。
325So it's not just like an off the shelf thing that you can create.所以,这不只是一个离架子的东西 你可以创造。
326Now that's现在
327a pretty big effort.相当大的努力。
328I don't know that there are like that many things like that that are done as我不知道有很多事情是这样的
329biotechs.生物科技.
330I think it's just the scale of the ambition of what we're doing, the horizon我想这只是我们所做的事情的雄心壮志的尺度 地平线
331over which we're committed to doing it.我们承诺要做到这一点。
332I think part of the theory is like,我认为部分理论是,
333if you're building tools that are this complicated, you kind of want to have a如果你正在建造这些复杂的工具, 你希望有一个
33410 to 15 year time horizon on building out these efforts.在10至15年的时间内,将开展这些努力。
335And then the scale of capital required.然后,资本规模需要。
336I mean, I guess there's no rule that said that you couldn't do it as like an incredibly well我的意思是,我猜没有规则说 你不能做到这一点 像一个令人难以置信的井
337funded startup, but I think that this just made more sense.资助启动, 但我认为这只是更合理。
338And then it also is simplifying然后也简化了
339strategically to not have to think about how you're going to make money with the different things.从战略上讲,不用想着 如何用不同的东西赚钱
340We want to get the models in people's hands.我们要让模特们掌握在人们手中
341We release them as open source.我们释放他们作为开源。
342I think that that's like a very valuable thing to do.我认为这就像一个非常有价值的事情做。
343And again, I mean, the theory isn't that再说一遍 理论不是
344we're going to cure the diseases.我们要治好疾病
345我们没有 We're not.
346It's that we want to help accelerate the pace我们想加快速度
347of progress for the whole scientific field.整个科学领域的进展。
348As the person least experienced with making money作为赚钱经验最差的人
349here, I would say that the sort of neutral nonprofit nature of our work actually helps在这里,我要说,那种中立的非营利性质 我们的工作实际上帮助
350harness more people to enter this effort and to actually achieve the mission of like understanding利用更多的人参与这一努力 并真正实现类似理解的使命
351the totality of human biology and to cure, prevent, manage all these, you actually do need为了治疗、预防、管理这一切 你确实需要
352the entire academic biotech industry to come together and to work on this in a sort of整个学术生物技术产业 聚集在一起 并努力在某种程度上
353unified way in part because there's a lot of talent out there and it's not helpful to leave统一的方式 部分因为那里有很多人才 离开也无济于事
354any talent, exclude any talent from the effort.任何人才,排除任何人才的努力。
355And there's a super long tail of diseases.并有超长尾疾.
356There are the common ones and even the common ones, I think if you unbundle有共同的,甚至共同的,我想如果你解开
357heart disease, cancer, neurodegenerative diseases, even if you unbundle like dementia or depression,心脏病 癌症 神经退化疾病 即使你像痴呆症或抑郁症一样解剖
358there are many, many, many subcategories that become more and more niche.有很多,很多,许多亚类 变得越来越适合。
359And that's not even这还不算
360looking at the long, long tail of rare diseases.看着罕见疾病的长尾巴
361Those often get orphaned and don't get brought那些经常成为孤儿,不带孩子
362along when we're sort of looking at the most efficient way to impact the lives of many.当我们在寻找最有效的方法 影响很多人的生活时
363But if you sort of decentralize the effort and put the tools in many people's hands,但如果你把精力分散到很多人手中
364you start getting people who are like, you know what, I am super interested in spinal我对脊椎很感兴趣
365mass muscular atrophy.大规模肌肉萎缩.
366And that's something I care deeply about.我非常关心这个
367And if you put the tools如果你把工具
368in that person's hands, they're going to be able to make progress in a way if you had to在那个人的手中,他们会 能够以某种方式取得进步 如果你不得不
369focus your efforts and make big bets.集中努力,下大赌注
370You probably wouldn't because it's just a niche你可能不会,因为它只是一个小地方
371individual small group disease that actually will in turn, if we can understand that disease如果我们能理解这种疾病,那么这种疾病将反过来发生
372process, helps us unlock knowledge about a lot more about how the human body works.帮助我们解开更多关于人体工作的知识。
373Do you have any thoughts or predictions in terms of what disease areas this work will您是否对这项工作将涉及哪些疾病领域有任何想法或预测?
374impact first?先撞击?
375I know it's very hard to be predictive about these things, but just given我知道很难预测这些事情,但是...
376the nature of the work and the nature of the models, are there areas you're most optimistic工作性质和模型性质 你最乐观的领域
377about in the short to medium term?中短期?
378That's actually not how I think about it,其实我不是这么想的
379at least.起码
380The way I think about it is like, we want to understand how biology works.我的想法是,我们想知道生物学是如何运作的。
381The ideal world is you would say, I understand the genetics of this person.理想的世界是,你会说, 我理解这个人的遗传。
382So I want to think所以我想考虑
383about people at the individual level.关于个人层面的人。
384I want to understand the genetics of this person.我想了解这个人的遗传学
385I wantи稱
386to understand the risks they have to different illnesses.来了解他们面临的不同疾病的风险。
387I want to understand the mechanistic我想了解机械学
388connection between, say, a gene variant, a protein, and a disease process.基因变体、蛋白质和疾病过程之间的联系。
389Because if因为如果
390you understand that through chain, then you can design a protein, design a drug, bespoke你明白,通过链, 然后你可以设计蛋白质, 设计药物,说话
391to them, and actually make an intervention.对他们来说,并进行干预。
392And right now, I'm sure we've all had experiences现在,我相信我们都有经验
393being sick.已经生病了。
394And if you have something that's even remotely non-standard, you go into PubMed,如果你有超标准的东西, 你进入PubMed,
395you look up a paper, you look up the supplement, and then you start going through the methods,你找报纸,你找副刊, 然后开始研究方法,
396and you're like, am I represented in this paper?而你喜欢, 我代表在这篇论文?
397And we're just making guesses.我们只是在猜测
398We really have no我们真的没有
399mechanistic understanding.机械理解.
400We're saying, like, okay, you're kind of like these people that we我们说,像,好吧, 你有点像这些人,我们
401studied.学习。
402And this drug kind of impacts the pathway that we think is implicated.而这种药物会影响我们认为涉及的路径。
403Let's try and see if让我们试试看
404anything happens.任何事都会发生
405And time passes, and sometimes it works, and sometimes it doesn't.时间流逝,有时有用,有时则不行。
406So my goal所以我的目标
407is to be able to treat the individual as an individual, understand the mechanisms, and be能够将个人作为个人对待,理解各种机制,以及
408able to intervene.能够干预。
409And there are different diseases that are at different stages of filling out that有不同的疾病 处于不同的填充阶段
410whole through line.整个线。
411And so for some diseases, you just want to understand which gene variants所以对于一些疾病,你只是想了解 哪些基因变异
412actually cause disease and which don't.实际造成疾病,但不会。
413And that in itself can be super empowering to patients.这本身就能够对病人产生超强的能量。
414And beyond that, there are some diseases where we understand the chain.除此之外,还有一些疾病,我们理解了链条。
415We just can't intervene我们不能干涉
416and change a specific protein function.并改变特定的蛋白质功能。
417That's super exciting, too.这也太令人兴奋了
418If we could design a protein如果我们能设计出蛋白质
419to actually change the physiology, then we can actually cure someone.来改变生理, 然后我们就可以真正治愈某人。
420But to me,但对我来说,
421that is just as exciting as contruing to our understanding of how someone gets sick in the这同样令人兴奋,就像我们理解 一个人是如何生病的
422first place.初总标.
423That's a very exciting vision because you're basically saying you can bring这是一个非常令人兴奋的愿景 因为你基本上说 你可以带来
424generalizable tools to provide very personalized things for each individual person.为每个人提供个性化物品的通用工具。
425And that's the这就是
426power of the approach is you have these big models that you build that can then apply anywhere.方法的力量是,你拥有这些大模型, 这些模型可以应用到任何地方。
427I know that you mentioned earlier that you were going to try and cure prevent all diseases我知道你刚才说过 你要尝试治疗所有疾病
428within 100 years.在100年内。
429And you mentioned, hey, it could actually be sooner now, given all the advances in你提到,嘿,现在可能更早了,因为所有的进步
430AI.大赦国际。
431Do you have some thought of when we think we'll be closer to that goal?你有没有想过我们什么时候能更接近这个目标?
432I'm optimistic it'll be sooner.我乐观地认为会更早
433I think the thing that's complicated is that it's a dynamic system.我认为复杂之处在于它是一个动态系统.
434If you fix something, there will obviously be future things that you need to work on.如果你修补了东西,显然会有你需要研究的未来事物.
435So这么说
436I don't think that the current set of things that we're aware of are going to be the only我不认为我们所知的 现今的一组事情 将是唯一
437things that need to get worked out.需要解决的事情。
438But I think that the progress with AI is obviously very但我认为人工智能的进展显然非常明显
439exciting on this.令人兴奋的这个。
440The other thing that I'd say, just adding to what you were saying a second ago,还有一件事我想说, 只是补充你刚才说的话,
441is we really look at more systems than specific diseases.我们的系统比特定的疾病要多
442So for example, one area that seems比如说,一个似乎
443really important to understand is inflammation.真正重要的是了解炎症。
444We talked about this a bunch.我们谈过了
445This is a big focus这是重点
446of the Chicago Biohub.芝加哥比奥胡布。
447There's a lot of data on that.相关数据很多.
448It seems quite clear that it's connected很明显,它连接在一起
449to a bunch of different diseases.给一堆不同的疾病。
450But rather than studying the specific diseases, we think that by但是,我们不研究特定的疾病,而是认为
451trying to understand inflammation more broadly, that will make it so that other companies that试图更广义地理解炎症 这将使其他公司
452can then use these tools can work on specific therapies.然后可以使用这些工具对特定疗法起作用。
453Another example is I think that the另一个例子是我认为
454some of the work that we're doing in cellular engineering and when we kind of ladder up from当我们在细胞工程中做一些工作, 当我们从某种阶梯上升起
455proteins to cells to like whole dynamic systems within the body, I think that that one makes sense蛋白质对细胞来说就像体内的整个动态系统, 我认为这有道理
456and it's sort of privileged.它是一种特权。
457The cells can travel around through the body, all that.细胞可以穿梭在身体中,所有这一切。
458So obviously,很明显
459that has a big part in addressing different diseases.这在治疗不同疾病方面有着重大作用。
460How do you make the immune system如何使免疫系统
461function better?功能更好吗?
462But exactly how do you connect that last mile, I think is going to be more但确切的说,你如何连接 最后一英里,我认为将是更多的
463something that biotech or other academics individually studying things will be better生物科技或其他学者单独研究的东西会更好
464suited to do.适合这样做。
465So this is kind of how we think about building up the tool set that just helps因此,这就是我们如何考虑 建立工具集,只是帮助
466accelerate all these other folks.加速这些其他人。
467Whether the timeline is 10 years, hopefully less than 100时间是否为10年,希望不到100年
468now, I think it's useful for maybe your average doctor or patient, human being, everybody's a现在,我认为它有用 也许你的普通医生或病人, 人,每个人都是一个
469patient, to think about what's externally visible in the progress here.耐心想想在进步中能从外部看到什么
470You worked with patients你和病人共事过
471for a long time at UCSF.长期在UCSF工作.
472What should doctors look out for?医生应该注意什么?
473What should people look out人应该注意什么
474for if you're actually accelerating progress?如果你真的在加速进展?
475I'm super excited about the progress, especially我对进步非常兴奋,尤其是
476with this launch that Alex and his team have put forth.亚历克斯和他的团队 推出的这次发射
477And I think it's very clear that science我觉得科学是很明显的
478is going to start moving pretty quickly.将开始快速移动。
479And I think the thing that's less clear to me is我觉得我不太清楚的是
480exactly how we translate to the clinic and what that looks like.确切地说,我们如何翻译 到诊所,什么样子。
481And I think what has to change我觉得应该改变什么
482is actually the way we do clinical research.其实就是我们临床研究的方式
483And my hope is that we're really shortening the我希望我们真的在缩短
484distance between bench research and patient impact.长椅研究和病人撞击之间的距离。
485But there's a lot of steps there但那里有很多台阶
486that we need people who actually take care of patients to think creatively and我们需要真正照顾病人的人 创造性地思考
487think about how to deploy safely.想想怎么安全部署
488And that's a gap that we have some work in.这是一个差距,我们有一些工作。
489We partner with我们合作
490Jennifer Doudna和我们的CRISPR在UCSF的治疗程序.
491So we're dipping our toe in understanding所以我们在努力理解
492how the deployment of research needs to change, given how quickly research will be progressing.研究的部署需要如何改变,因为研究将如何迅速取得进展。
493But that one is still, I think, is still shaping up.但是,我认为,这个仍在形成。
494Maybe I could say something about our most也许我可以说一些关于我们最
495recent launch, because I think it also kind of illustrates.最近推出,因为我认为它也说明了这一点。
496Oh yeah, please.哦,是的,请。
497We should ask you我们应该问你
498about now.现在。
499So we announced the new ESM fold.所以我们宣布了新的无害环境管理折叠。
500And so this is basically所以这基本上是
501an open system for scientific discovery in protein biology.一个开放的蛋白质生物学科学发现系统。
502It's a world model of protein biology这是蛋白质生物学的世界模式
503that's been trained.已经训练过了
504It's a language model base.语言模型基础.
505So it's been trained on billions of protein因此,它接受了数十亿蛋白质的训练
506sequences, kind of learns these emergent representations of protein biology.学习蛋白质生物学的 新兴表现
507And then礛
508we can use it to make predictions of atomic resolution protein structure.我们可以用它来预测 原子分辨率蛋白质的结构。
509And it's really这是真的
510fast.快点
511So it's blazing fast.故火速炽盛.
512So it's kind of illustrating this Pareto optimal frontier of所以,它有点说明 这个帕雷托最佳前沿
513speed and accuracy in structure prediction.结构预测的速度和准确性.
514And so this allows us to kind of characterize因此,这让我们可以 一种特征
515really vast kind of stretches of the protein universe.蛋白质宇宙的伸展很广
516So we folded over 1.1 billion proteins所以我们折叠了超过11亿个蛋白质
517and predicted their structures and identified kind of features connecting all of them through并预测了它们的结构 并确定了将它们连接起来的特征
518mechanistic interpretability.机械解释.
519But I think the thing that I thought was most exciting about this但我认为这件事最令人兴奋
520model is it's this really general model of kind of protein biology.模型是这种蛋白质生物学的普通模型。
521And so you can use it as a这样你就可以把它当作一个
522world model.世界模式。
523You can actually really start to search the space of the world model to design你可以真正开始搜索世界模型的空间来设计
524new proteins.新的蛋白质。
525And it's really hitting state of the art across pretty much every structure并且它真的打击 最新艺术 几乎所有的结构
526prediction benchmark, and especially on protein-protein interactions and protein-antibody预测基准,特别是关于蛋白质-蛋白质相互作用和蛋白质-抗体的基准
527interactions, which is really critical for therapeutic design.互动,这对治疗设计来说非常关键。
528And so what we found is you所以我们找到了你
529can actually now use the model to design proteins and to design actually single-chain antibodies.现在可以使用模型来设计蛋白质 并设计出单链抗体
530And so you can do all of this digitally and then really in a small number of experimental trials,所以,你可以做所有这些 数字化,然后真的 在少数实验试验,
531basically like a 96-well plate, select from hundreds of thousands of trajectories digitally,基本上像一个96井的盘子, 从数十万的轨道数字,
532actually synthesize 96 proteins tested in the lab in a really kind of short, easy experimental在实验室中测试的96个蛋白质 是一种很短很简单的实验
533cycle.循环。
534And we found nanomolar binders there.我们在那里发现了纳米软膜粘合剂。
535And so that's really the level for therapeutic activity.这就是治疗活动的水平。
536So it's really, I think, showing that you can have these kind of general purpose models.所以,我认为,它真的,显示 你可以有这种通用模型。
537We didn't design a model for antibodies.我们没有设计抗体模型。
538We didn't design a model to be able to bind我们没有设计一个能装订的模型
539one particular target.一个特定的目标。
540We just designed a model that could understand proteins.我们刚刚设计了一个能理解蛋白质的模型。
541And you kind of你有点
542get protein design as an emergent property.获得蛋白质设计作为新兴属性.
543And then I also think it illustrates the power of之后我还认为这说明了
544science and open source because we released this as basically an open discovery engine.科学和开放源代码 因为我们发布这个基本上是 一个开放的发现引擎。
545And so这样
546really anyone can build on it.任何人都可以继续发展
547And so it takes what are these really intensive laboratory因此,它需要什么 这些真正密集的实验室
548experiments where you have to screen through hundreds of thousands or millions of antibodies实验,你必须筛选 数十万或数百万抗体
549and high throughput screens in the lab.还有实验室的高吞吐量屏幕
550And you can really just kind of spin up an instance and而你真的可以只是 那种旋转一个实例,
551compute and now be able to generate antibodies.计算,现在可以产生抗体。
552You should say more about sort of like we took你应该多说说我们拿的
553that data when we did an antibody screen and then we validated it.当我们做一个抗体屏幕 然后我们验证了数据
554We looked at PDL in cells.我们在牢房里看了PDL
555And then we looked at it under the cryo-EM and sort of how all that complement validated what然后我们在低温电磁层下 研究了它 并研究了如何补充
556you were seeing in the models.你在模特儿身上看到过
557That's right.没错
558Yeah.对
559So I mean, I think it's really critical所以我觉得这很重要
560to actually go and characterize these molecules in the lab.来描述实验室中的这些分子
561And we have a structural biology我们有一个结构生物学
562centered here.以这里为中心
563We have incredibly powerful cryo-EM microscopes.我们有惊人强大的低温EM显微镜。
564And so we're really able所以我们真的可以
565to kind of look at these proteins biophysically and functionally.从生物物理和功能上看这些蛋白质
566And so, you know, we designed所以,你知道,我们设计
567proteins for several therapeutically relevant targets and we're able to confirm their function.蛋白质对几个治疗相关目标来说 我们能够确认它们的功能
568It's delightful when it works the way it's supposed to.以如是法行时令人欢喜.
569Yeah, that's very amazing.是啊,这是非常惊人的。
570We're able to look at the structure also.我们也可以看结构
571So you can see atomic resolution kind of at the binding of our faces.因此,你可以看到原子分辨率 那种在我们脸上的束缚。
572I know a lot of your work is really focused on basic research and kind of building other我知道你的很多工作都集中在 基础研究上 和建设其他
573fundamentals.基础
574If I look at actual translation into drugs or drug development, often a clinical如果我看 实际翻译成药物 或药物开发, 经常是一个临床
575trial will be 15 years.审判时间为15年。
576It'll cost $1.5 billion.耗资15亿元.
577About $50 million of that often is the molecule大约5000万美元 经常是分子
578in preclinical work and it's a few years of work.在临床前的工作, 这是几年的工作。
579And then the other $1.45 billion in decade plus然后另外14.5亿美元在十年之外
580is actually the drug development side of it.实际上是药物开发的一面。
581A lot of that seems to be gated on some regulatory其中许多似乎 门在一些监管
582issues.问题。
583Some of it's recruitment.有的是在招募.
584It's a variety of things, but a lot of it also has to do with有很多事情,但很多事情都和
585failure of drugs and trials around things like absorption or toxicity or things like that.药物的失败和实验 围绕诸如吸收或毒性之类的东西。
586Have you considered at all tackling that other chain of sort of molecular design and thinking,你有没有考虑过 解决其他的分子设计和思考链,
587or is the primary focus more on the basic biology and sort of the initial sort of molecules?或者说,首要的焦点是 基本的生物学和某种 最初的分子类型?
588I mean, at least my hope in building this comprehensive model of how cells work is我的意思是,至少我的希望 建立这个全面模式 细胞是如何工作的
589actually also being able to predict off-target effects.也能够预测非目标效应
590I think you can do some of that actually我想你可以做一些实际上
591with biological models because right now some of the off-target effects are we just didn't know和生物模型 因为现在一些非目标效应 我们只是不知道
592your kidney cell also expressed this receptor.你的肾细胞也表达了这个受体。
593And then when we test it in human, we see it然后当我们测试它 在人,我们看到它
594happening and we see renal toxicity.并且我们看到肾脏的毒性
595And if you have a single cell atlas that looks at all the如果你有一个单细胞图集 来查看所有的
596different cell types, some of which actually were not predicted before we modeled them,不同的细胞类型, 有些实际上没有预测 在我们模拟之前,
597you can start looking at which cells actually do have receptors for the target you thought you were你可以开始观察哪个细胞真的有 受体的目标,你以为你是。
598exclusively targeting and be able to predict some of these downstream effects before we get完全针对目标,并且能够在我们获得
599into the human trials.进入人类的审判。
600And I think that that's actually one of the more exciting applications我觉得这其实是更刺激的应用之一
601of like a transcriptomic model to understand actually how the different cells will react就像一个记录器模型 来理解不同的细胞会怎样反应
602when you intervene and do something.当你介入并做一些事情。
603But I think when you think about delivery mechanisms但我想当你考虑 交付机制
604and patient care, that's where you start having to be creative about when you ask,和病人的护理, 这就是你必须开始 创造性 当你问,
605what disease do you want to care first?你想先照顾什么病?
606There are certain diseases that will be easier to有些疾病比较容易感染
607deliver a therapeutic to or the risk reward makes more sense.提供治疗或风险奖励更合理。
608And I think we were all inspired by我觉得我们都是从
609Baby KJ I think last year now when the team at CHOP was able to deliver a CRISPR therapeutic to宝宝 KJ 我想去年当CHOP的团队 能够提供CRISPR治疗
610a mutation that he had that would have inevitably led him to significant neurotoxicity他的突变会不可避免地导致他的神经毒性
611and altered his life.并改变了他的生活。
612But we were able to, that disease was very carefully chosen because we但是,我们做到了, 这种疾病是精心选择的,因为我们
613needed to target his liver cells.需要瞄准他的肝细胞
614And if we could easily deliver a product that would work in his如果我们能轻松地交付一个产品 能够帮助他
615liver.肝脏
616And I think that's when the creativity, the wherewithal to choose the right applications我认为正是那时 创造力 以及选择正确申请的手段
617can help us unlock the first applications.可以帮助我们解锁第一批应用程序。
618Maybe something just to add to that also,也许还有别的
619because I mean, kind of you described the conventional drug development process, right?我的意思是,你有点描述 传统的药物开发过程,对不对?
620And I think these kind of tools have the potential to have a lot of impact on that process.我认为这些工具有可能对这一进程产生很大影响。
621But what's interesting is to really start to think about kind of the new paradigms that can但有趣的是真正开始思考 那种新的范式可以
622open up and what does it mean if you can, the barrier to develop a drug, to design a molecule,打开它是什么意思 如果你可以, 障碍 开发药物,设计分子,
623to kind of get through all of those stages is so much lower.进入所有这些阶段都低得多
624And so you have programmable biology所以你有可编程的生物学
625and you can really start to create a medicine for every individual patient.你可以开始为每一个病人 创造一种药物。
626I think that我觉得
627has enormous implications for how we do drug development and what the future of medicine对我们如何开发毒品 以及医学的未来有着巨大的影响
628looks like.看起来像。
629It'll be an exciting day when the FDA accepts like a virtual clinical trial for the当FDA接受虚拟临床试验
630day's one or something, or, you know, that's based on some person's view of that person.一天或什么的, 或者,你知道,这是根据 某人的观点的人。
631Yeah.对
632Or even short of that, like thinking about the specific like mechanisms where you see this或甚至没有,像思考 具体的机制,你看到这一点
633acceleration, like I imagine if people feel like they can predict impact in kidney cells or have加速,就像我想象的 如果人们觉得他们可以预测 影响肾细胞或有
634a stronger perspective on tox because they have this broader understanding, they'll be willing to对毒剂有更强烈的视角 因为他们有更广泛的理解 他们愿意
635try many more programs, right?尝试更多的程序,对不对?
636Yeah.对
637The recruitment could also change.征聘工作也可能改变。
638I mean, we have this program我的意思是,我们有这个程序
639as one and the basic idea is that a lot of people focus on the most common diseases,一个基本的想法是 很多人都关注最常见的疾病
640but there's this long tail and the economics don't quite work out for companies to focus on但有这么长的尾巴 经济学并不太适合公司专注于
641those diseases.这些疾病。
642But if you can make it so that the groups of patients can kind of come together and但是,如果你能做到这一点, 这样,病人群体 可以聚集在一起,
643organize and say, Hey, we would take an experimental drug on this.组织起来说,嘿,我们会服用一个实验药物 关于这个。
644Then it actually, because of the其实是因为
645costs that you're talking about and how that's a huge amount of the overall cost, if you can flip如果你能翻转的话,你所说的成本 以及这如何是一个巨大的总成本
646that, then it actually makes it so that the economics make a lot more sense to then if you那么,实际上它使它 使经济学 更有意义,然后如果你
647can generate something more easily and you can pair it with a group of people.可以更容易地产生一些东西 你可以把它和一群人配对
648I think one of我觉得其中之一
649the interesting things from science and engineering is that often you can hit your head against the科学和工程学的有趣之处在于 你常常可以撞到头
650{\fn黑体\fs22\bord1\shad0\3aHBE\4aH00\fscx67\fscy66\2cHFFFFFF\3cH808080}但很多时候你喜欢学很多
651more about a system from finding some kind of rare or like weird side thing that's happening.更多关于一个系统 从发现某种稀有 或者像怪异的侧面 正在发生的事情。
652Yeah.对
653So I don't know.故不识.
654I think that that's like always been kind of an interesting part of this我觉得这就像总是 有点有趣的部分
655that actually connects pretty well to this because now you're going to be able to enable a long tail这实际上连接得很好 因为现在你可以 启用长尾巴
656of new kind of ideas to get tried and enable them to potentially get tested more easily.新的想法 尝试, 并使它们有可能 更容易接受测试。
657Yeah.对
658That's a really good point on rare.难得的道理
659In our rare disease cohorts, first of all,在我们罕见的疾病群中,首先,
660they're incredibly inspiring and powerful, but patient groups are self-organizing patient他们令人难以置信的鼓舞 和强大的, 但患者群体是自我组织病人
661registries, natural history registries, biobanks.登记册、自然历史登记册、生物库。
662They're organizing their own clinical trials.他们正在组织自己的临床试验.
663There's gene therapy that one disease group has moved forward over the course of like,有基因疗法 一个疾病组 已经向前发展 类似,
664I want to say like three to five years rather than decades and the speed is so fast because我想说3到5年而不是几十年 速度这么快是因为
665the patients themselves have organized the resources that a scientist or a clinician might病人自己组织着科学家或临床医生可能掌握的资源
666need and it's incredible.需要,这是不可思议的。
667But I think to some degree you're going to need something like this但我觉得你一定需要这样的东西
668because there are going to be many more new things that can get created, but that doesn't mean that因为还有很多新的东西可以创造 但是这并不意味着
669for like the general population that you're not going to want the same level of vetting that和普通民众一样,你不希望 同样的审查水平
670we've had historically, but making it so that people who want to be on more of the frontier我们曾经有过历史,但做它 所以人们谁想 在更多的边界
671have the ability to do that is I think also going to be pretty helpful.有这种能力,我想也会很有帮助
672Yeah.对
673Letting people opt in让人们选择加入
674to be part of trials I think is one of the big shifts that is starting to happen, but could我认为,这是开始发生的重大转变之一,但有可能
675really help accelerate biology in general.有助于加快生物学的发展
676All three of you have mentioned at different points你们三个都在不同的地方提到过
677like the power of open ecosystems in such a large space.就像在如此广阔的空间中 开放的生态系统的力量
678I think some of that logic around我认为一些逻辑围绕
679open source and the breadth or diversity of data collection that you were describing,开源和您描述的数据收集的广度或多样性,
680it should also apply in the language model world and the multimodal AI world.它还应适用于语言模型世界和多式联运AI世界。
681Do you think that's你觉得那是
682right?对吧?
683Does any of the work you're doing here change how you think about AI and meta?你在这里做的工作 有没有改变你对AI和meta的想法?
684I mean,我是说...
685it's a similar philosophy overall.这是一个类似的哲学整体。
686Priscilla was talking about this, that a lot of our focus普瑞西拉在谈论这个 我们的焦点很多
687is building tools that empower individuals to do things.正在建立工具 赋予个人做事情的能力。
688That's a common theme across a lot这是一个常见的主题 跨越很多
689of the things that I work on is just putting the technology in individuals' hands.我的工作就是把这些技术 掌握在个人手中
690We don't believe我们不相信
691in this very centralized future where there should be a small number of institutions that在这个非常集中的未来,应该有少数机构
692basically are advancing all of the stuff.基本上所有的东西都在推进。
693Our vision is not that there's going to be我们的愿景是不会
694some central superintelligence that solves all of science.一些解决所有科学问题的中央超级情报
695I think people are really important我觉得人很重要
696and I think will be more important in the future and giving people more tools to be more productive我认为将来会更加重要 给人们更多的工具来提高生产率
697is going to be a critical part of any positive future.这将是任何积极未来的一个关键部分。
698That's how progress has always been made进步总是如此
699historically.历史上
700It's not through centralization, it's through empowering individuals to try things不是通过中央集权,而是通过授权个人尝试事物
701that are somewhat out of the mainstream that other people didn't think were good ideas because they这有点脱离主流 别人认为不是好想法 因为他们
702thought they were good ideas that already have been done.认为这些想法是好的 已经做了。
703I think that that's very central to我觉得这对
704the whole ethos.整个风格。
705To some degree, it's like why you create something like social media to give在某种程度上,这就像为什么 你创建的东西 像社交媒体给
706people a voice.人们的声音。
707I think a lot of the stuff that I care about in terms of empowering people with我想了很多我关心的东西 赋予人们权力
708individual AI, open source is one instantiation of it.个人AI,开源是它的一个即时。
709It's not the only way to do it.非唯此义.
710It certainly当然
711is one way that you basically are saying we're going to take this technology and put it in一种方式是,你基本上 是说,我们要 采取这一技术 并把它
712everyone's hands.每个人的手。
713In terms of science, I think it really makes sense and we're deeply committed在科学方面,我觉得这真的有道理 我们深深地致力于
714to open source.改为打开源。
715There are obviously interesting considerations on this that are important too关于这一点,显然有一些有趣的考虑也很重要。
716because there's a lot of considerations around biosafety and things like that that we're going因为有很多关于生物安全的考虑 和类似的东西,我们要去
717to need to balance and think through how to handle.需要平衡和思考如何处理。
718But I think overall, this is very deep in the但我认为,总的来说,这是非常深刻的
719ethos of the work that we're doing both at Biohub and probably a theme for a lot of the stuff that我们在Biohub的作品的特质 也许是很多东西的主题
720we do.и穦
721We believe that a positive future is one where you build a technology as a tool,我们相信,一个积极的未来 是一个你建立技术的工具,
722you put it in individuals' hands, and that's how society makes progress.社会就是这样进步的
723You have this incredibly ambitious mission at Biohub and yet the AI scientists that work here你在Biohub有个雄心勃勃的任务 然而在这里工作的AI科学家
724could also go work in commercial enterprises.也可以去商业企业工作。
725How do you think about the talent and how to如何看待天赋和如何
726do that?这样做?
727It's a very hot market for AI researchers, but I think that part of what that means is that这对人工智能研究者来说是一个非常热门的市场, 但我认为其中一部分意味着
728there's a lot of demand.需求很大。
729They're very in demand and can work on the things that they want to work on.他们需求很大,可以研究他们想要研究的东西.
730I think this gets back to this point again about frontier AI and frontier biology.我想这又回到了 关于前沿AI和前沿生物学的这一点。
731Right.对
732I think the AI researchers who work here could go work on language models or things at any我觉得在这里工作的人工智能研究者 可以去研究语言模型或东西
733of the main labs, but those labs don't have the frontier biology part attached to it.但那些实验室没有边疆生物部分
734I think我觉得
735that there's also just a very large mission component of this, which is there's an ability还有一个非常大的 任务部分,这是一个能力
736to do this unique work here that you just can't really do at the other places.做这个独特的工作 在这里,你只是不能 真的在其他地方。
737If that's what如果是这样
738your focus is, then I don't actually think that there's any other organization in the world that's那你的焦点是,那么我不 实际上认为 世界上任何其他的组织
739doing both the frontier biology and the frontier AI.从事边疆生物学和边疆AI.
740Why are you here, Alex?你为什么在这里,亚历克斯?
741I think it's really simple.我觉得这很简单
742Our mission is to take care of the disease.我们的任务是照顾这个疾病。
743You say it with a straight face and a less than 100-year timeline.你说它有一个直面 和不到100年的时间。
744It's very serious now.现在很严重了
745There's no more.已无余.
746It's a really powerful mission.这是一个非常强大的任务。
747It's something people are deeply motivated by.这是人们深深的动机。
748We're at this moment in time where that actually seems like something that can be achieved.此时此刻,这似乎可以实现。
749We're building a really unique place where we're tackling that problem.我们正在建造一个非常独特的地方 来解决这个问题。
750We have the resources and the right things to actually really go after that and do that.我们有资源和正确的事情 真正去完成它。
751That resonates with me as somebody who talks to and hires a lot of research scientists.作为和很多研究科学家交谈并雇用他们的人,我对此有共鸣。
752They want他们想要
753to know if you have the data, if you have the tools, if you have the compute, if you have the如果你有数据,如果你有工具,如果你有计算,如果你有
754talent, and then what the mission is.天赋,然后是什么任务。
755I actually think that's super competitive.我觉得这太有竞争力了
756The other thing is that you don't need a very large team.另外一件事是,你不需要一个非常庞大的团队.
757I think it's an interesting thing我觉得这很有趣
758the world is that people care about different missions.世界是人们关心不同的使命。
759That's good.这是很好的。
760I think that's part of我认为这是一部分
761why building these tools and giving people the ability to explore what they care about,为什么要建造这些工具 并赋予人们探索自己所关心事物的能力
762whether it's across science or just across everything, is such a powerful way to make无论是跨越科学还是跨越一切 都是一种强大的方法
763progress in society is that people care about different things.社会的进步是人们关心不同的事物。
764In order to make progress in为了在以下方面取得进展:
765AI, you don't need many, many hundreds of AI researchers or thousands or anything like that.AI,你不需要很多, 几百个AI研究者 或数千或类似的东西。
766You can really make progress with a very strong group of a dozen or a couple dozen people.一个非常强大的十几个人或十几个人,你确实可以取得进步。
767Finding people who care about this mission is not a particularly hard thing.寻找关心这个任务的人并不是一件难事。
768This is a这个是
769super important thing in the world.超重要的东西 在世界上。
770It's just a cool thing about the world is that people世上最酷的是人们
771obviously are drawn to different missions.很明显,他们被吸引到不同的特派团。
772I think the simplest mental models that我觉得最简单的心理模型
773folks have, even if they're paying attention to the space, are essentially like, okay,人们已经,即使他们关注空间, 本质上就像,好吧,
774you know, structured prediction models for proteins and protein-protein interaction models.蛋白质结构化预测模型 蛋白质-蛋白质相互作用模型
775There's this one piece, which is fundamental understanding, and then there's this theory of有一个是基本的理解,然后是这个理论
776someday we're just going to be able to zero-shot things into either the clinic or the clinic with总有一天我们就能把东西 零射入诊所或诊所
777much better hit rate.更好的打击率。
778What needs to happen for us to go from ESMfold2 to this other piece?需要我们怎样才能从无害环境管理层面2 转向另一部分?
779Is that feasible?这可行吗?
780I think that's a great question.我认为这是一个很好的问题。
781I mean,我是说...
782I would say that I'm really optimistic on that.我会说,我真的很乐观。
783I think on the one hand, these are problems that我认为,一方面,这些问题是
784historically people could spend an entire career working on.历史上人们可以花 整个职业生涯工作。
785How do you figure out how to effectively怎样才能有效
786optimize a drug?优化药物?
787How do you get it through preclinical?如何通过临床治疗?
788How do you do the early safety?早安怎么样?
789I think that when you have a new scientific paradigm, questions that were once hard我认为,当你有了一个新的科学范式, 那些曾经很困难的问题
790to answer become simplified through the new paradigm.答案通过新的范式简化。
791I'm very optimistic that many of these我很乐观,很多这些
792core problems will be solved in an emergent way through these models.核心问题将通过这些模式紧急解决。
793I think one great example我认为一个很好的例子
794of that is toxicity, whereas if you can really digitally simulate everything and be able to如果你真的能用数字模拟一切
795predict where a drug is going to distribute and bind across the human body, you have预测药物在人体的分布和结合
796the beginning of a solution to that kind of problem.开始解决这种问题。
797I think that once you have these我认为一旦你有了这些
798accurate representations at the molecular level, we're going to start to see really在分子层面上的精确表述,我们开始真正地看到
799rapid progress on a lot of these core problems.在许多这些核心问题上迅速取得进展。
800What is the most exciting use or experimentation什么是最令人兴奋的用途或实验
801with the models you've seen in the last week since release?和释放后最后一周见到的模特?
802Yeah, I mean, it's just been great to是啊,我的意思是,这是很好
803see it get integrated in all kinds of things.看着它融入各种事物
804I think one of the really interesting things我觉得其中一件很有趣的事
805we've been seeing is people connecting it with agentic systems to do automated design and just我们一直看到人们把它和代理系统联系起来 来进行自动化设计
806automate that whole process.把整个过程自动化
807It's really, I think, another example of how you can see bringing这真的是,我认为, 另一个例子,你可以看到如何带来
808together agentic and frontier AI with the ability to have a world model for biology and actually具有世界生物学模式的能力的代理和前沿AI
809reason about biology and really start to automate the entire design process.并且开始将整个设计过程自动化
810Are you taking, how do you decide what the next step in the research agenda is?你正在采取,你如何决定 下一步的研究议程是什么?
811It's like这就像
812world model for biology, and then I could, I'm just going to be very coarse here, like I could生物学的世界模型,然后我可以, 我只是要非常粗糙在这里, 像我可以
813scale it up.放大点
814I could add more data.我可以补充更多的数据。
815I could add, like adding data is a non-trivial thing in terms我可以补充,像增加数据 是一个非三角的事情,从术语
816of new methods and domains.新的方法和领域。
817Like what is, do you take input from the larger ecosystem about像是什么,你从更大的生态系统 输入关于
818how people are using it and what would make it more useful?人们如何使用它, 是什么会使它更有用?
819Or is it really like we understand还是我们真的明白
820the next step of structures or coverage that we're looking for?我们要找的下一步结构或覆盖?
821I think there's two things.我觉得有两件事
822We have a view on the next big challenge, which I think is the virtual cell and really being able我们对下一个大挑战有看法 我认为这是虚拟细胞 并且真的能够
823to ladder up the hierarchy of biological complexity to the cell.把生物复杂度的等级提升到细胞中
824Sorry, very basic question.对不起,非常基本的问题。
825This这个
826virtual cell model, what is the input and output I should expect?虚拟细胞模型,我期望的输入和输出是什么?
827I think there's a different我觉得这不一样
828is a system that can really model each of the levels of complexity.是一种能够真正模拟每个复杂程度的系统。
829So the proteomic layer,所以蛋白质层,
830the genetic layer, the transcriptomic layer, and connect that to the phenotype.基因层 转录层 连接到苯基
831And you need enough你需要足够多
832generality so that you can ask the model questions about a new intervention in a context that it概括性,以便您可以问模型中的新干预的背景问题
833hasn't been trained on and kind of get an answer from it.还没有接受过训练 也没有得到答案
834And the gap that we need to close as a以及我们需要填补的空白
835field is being able to really make those predictions that can generalize.字段能够真正作出那些可以概括的预测。
836So that's这么说吧
837going to require enormous effort to generate data.需要巨大的努力来生成数据。
838Yeah.对
839And then in terms of what you decide to do然后你决定做什么
840next, I think this is a pretty normal process of constraint management.接下来,我认为这是一个相当正常的 约束管理的过程。
841I think every lab我想每个实验室
842in every field across the world probably feels compute constrained.在世界每一个领域,都可能感到计算受到制约。
843I think that that's probably我认为,这可能是
844true here too.这里也是真的
845There's always questions like, okay, should we double down more on advancing总是有问题 比如,好吧,我们是否应该 加倍推进
846the protein piece?蛋白质?
847Should we do more of the cellular stuff?我们该多做点手机的事吗?
848I think that those are kind of我认为这些是种
849ongoing debates in terms of how you sequence that.持续的辩论 如何排序。
850And then within that, there's kind of being at the然后在其中, 有一种是在
851Pareto frontier about how much you want to train the different models in order to... And the size帕雷托边疆 关于你有多想训练 不同的模型,以便... 还有大小
852of the models is also dependent on the scale of the data that you have for obvious reasons.的模型还取决于数据的规模。
853So yeah, I mean, I think there's some of that is just where you want to be on the curves and所以,我意思是,我认为有一些是 只是在那里你想 在曲线上
854the normal constraints.正常的限制。
855But I think that this is probably the same process that any research但我认为这个过程 可能和任何研究都一样
856organization goes through of you want to go in all these different directions and you're just组织要经过所有这些不同的方向 你只是
857trying to constraint optimize and make enough progress to do world-class work at one thing at试图限制 优化和取得足够的进展 做世界级工作 在一件事上
858a time while planting some seeds that can blossom over the next couple of years as well.播种一些种子 在未来几年也能开花
859This has been the most dynamic period of technology at least I've seen over my career.这是我职业生涯中最活跃的科技时期
860I mean, it's so exciting in terms of everything that's happening with AI and every week there's我的意思是,它是如此令人兴奋 在一切 发生在AI和每个星期
861something new that's changed.新的东西改变了。
862Are you tired or invigorated?你累了还是有活力?
863I'm both.我两个。
864I feel like everybody's我觉得大家的
865in this manic phase.在这个疯狂的阶段。
866Yes, it's a combination of invigorated and exhausted.是的,这是活力和疲劳的结合。
867Yeah, it's wonderful.是啊,这是美妙的。
868And so I guess things are very unpredictable right now.所以我猜现在事情很不可预测。
869It's really hard to know what's coming.很难知道接下来会发生什么
870We have this almost like early signs of exponentation on the model side with agentic我们几乎有 早期的征兆 在模型方面 与代理
871flows that we're starting to see in really interesting ways.我们开始以有趣的方式看到流动。
872Models starting to help模型开始帮助
873more and more with models.越来越与模型。
874That's still very, very early days for that.现在还很早呢
875If you're thinking如果你认为
876back five years from now and you were to define what success was relative to your efforts,五年后,你要确定 相对于你的努力,什么是成功的,
877and I know things are very dynamic, things changed a lot, but you have this common thread我知道事情是非常动态的, 事情变了很多, 但你有一条共同的线
878of tooling for the biohub.用于生物壳的工具。
879You have a common thread of empowering scientists at scale.你有一个共同的线 赋予科学家规模。
880You're looking back five years from now, is there a specific thing that you really want to make sure五年后,你回头看, 有一件特别的事情,你真的想确定
881that you've accomplished or achieved or a primary goal?你是否已经实现或实现了主要目标?
882Well, I think we have a pretty clear view嗯,我想我们有一个相当清晰的视野
883of this hierarchical set of world models that we want to build around biology.我们想围绕生物学构建的 这一系列等级世界模型
884And the other part of that is that we want to do the highest quality work in the world.另一部分是我们希望 做世界上质量最高的工作。
885And I think we're basically set up to do that between having a world-class AI research team我认为,我们基本上被设置 做到这一点,有一个世界级的AI研究团队
886and this collection of biohubs, such as world-class life sciences research organizations.以及这个生物丘的集合,例如世界级的生命科学研究组织.
887That's fundamentally a setup that no other organization in the world has.这根本上是世界上没有其它组织有的圈套.
888But you can have a lot of great ingredients and that doesn't guarantee that you succeed.但是,你可以有很多伟大的成分,这并不能保证你的成功.
889To me, five years from now looking back, I'm sure other labs or efforts will try to produce things对我来说,五年之后,回顾过去, 我相信其他实验室或努力 将尝试生产的东西
890that approximate what we're trying to do.和我们想做的差不多
891And I just think that we should be able to do something我只是觉得我们应该能做点什么
892that is meaningfully better and a unique intellectual contribution to the world.这是对世界有意义的更好和独特的知识贡献。
893I think that that's kind of what you, whenever you do any kind of research,我认为这就是你, 当你做任何研究,
894that's what you're trying to do.这就是你想做的。
895So if we do that, I think we'll all feel very good.所以,如果我们这样做, 我想我们都会感觉很好。
896I would also expect that at some point we'll just start seeing a lot more idea generation我也希望,在某个时候,我们 刚刚开始看到很多更多的创意生成
897from the people using the models.从人们使用模型。
898But I have enough faith that that part will materialize但我有足够的信念,那部分将会实现
899that for me, it's more just about making sure that we do world-class work.对我来说,这更只是 确保我们做世界级的工作。
900And I think if we do, the rest almost will take care of itself.我认为,如果我们这样做, 其他几乎会照顾自己。
901Very last question for you.最后一个问题给你
902Snapshot of it's mid-2026.2026年中段拍摄的快照.
903What's the biggest update in your什么是你最大的更新
904own thinking about Biohub or the domain from the last year?自己思考Biohub还是去年的域?
905Well, from the last year, I mean, you joined in the last year.嗯,从去年开始, 我的意思是,你加入了去年。
906I mean, I think the biggest thing我的意思是,我认为最大的事情
907that we basically rotated, and I think in the last year, we basically kind of formalized that我们基本上轮流, 我想在去年, 我们基本上是一种形式化
908Biohub is the main focus of our philanthropy.Biohub是我们慈善事业的主要焦点.
909So I think this has been a very big shift.所以我认为这是一个很大的转变。
910But Alex and the team coming in, I think has been interesting, not only because it's a world-class但亚历克斯和队伍来了,我觉得很有趣, 不仅仅是因为它是世界级的
911group, right?组,对不对?
912I mean, you guys have worked together for a while.我的意思是,你们合作了一段时间。
913I think also, I mean,我也觉得,我的意思是,
914you talked about how stuff is changing so much in the field.你说过 田里的东西变了很多
915I think one thing that's underrated is我觉得有一点被低估了
916like, this is like an extremely talented group of people who also know each other and work well喜欢,这就像一个非常有才华的 群体,他们也 互相了解,工作很好
917together and are stable and good.在一起,是稳定和好的。
918And I think that that also is underestimated in terms of我认为,这也被低估了
919compounding benefit of people being able to work well in a stable environment over time.使人们能够长期在稳定的环境中良好地工作,从而进一步受益。
920So I think that's a really important piece.所以,我认为这是一个非常重要的作品。
921But part of what we wanted to do was prior to Alex但是我们想做的 一部分是在亚历克斯之前
922leading the effort, the previous leaders of the Biohub were basically primarily biologists之前的Biohub领导人 基本上主要是生物学家
923who were interested in technology, right?谁对技术感兴趣,对不对?
924And now I think this is the point where we really flipped现在我想这就是我们真正翻转的地方
925that, right?是吗?
926Where, I mean, obviously you have a background in biology as well, but like you are在哪里,我的意思是,很明显 你有一个背景 生物学,但喜欢你
927primarily an AI researcher who has a background in biology.主要是有生物学背景的AI研究员.
928I think that that's like a deep reflection我觉得这就像一个深刻的反射
929on kind of the way that we expect that this is going to kind of drive more value in the future.那种方式,我们期望 这将会 驱动更多的价值在未来。
930So those are probably the biggest updates in the last year in terms of the work that we're doing.因此,这些可能是去年最大的更新, 就我们的工作而言。
931It's a new leader, not just the leader, but a team that I think has been really good.这是一个新的领导者, 不只是领导者, 并且是一个团队 我认为是非常好的。
932And then,礛
933yeah, I mean, I think on the rest of the industry, it's on track.是啊,我的意思是,我想 在其余的行业, 它的轨道。
934I mean, I think it's kind of this我的意思是,我觉得它是这样的
935crazy thing because when you have an exponentially growing curve, I think the way that an exponential疯狂的事情,因为当你有一个指数增长的曲线, 我认为这个指数的方式
936curve feels is it's growing so quickly that the kind of emotional feeling is it can't possibly曲线的感觉是它生长得如此之快 那种情感的感觉是不可能
937keep going, right?继续走,对吧?
938But I mean, the nature of an exponential curve is it doesn't just keep going,但是,我是指,一个指数曲线的性质 是它不只是继续,
939it keeps accelerating, right?它不断加速,对不对?
940Exponential growth is accelerating.标志性增长加快.
941So I think that that has all of所以,我认为,这一切
942these emotions and psychology attached to it.这些情感和心理学 附属于它。
943But I think fundamentally when you look at the curve但我认为当你看曲线时
944in the industry, the kind of fundamental thing is it is on track.在工业中, 那种基本的东西是 它在轨道上。
945It has remained on that curve,它一直停留在这个曲线上,
946which I think has all these very profound implications for all of these domains.我认为这对所有这些领域都有非常深远的影响。
947But certainly it validates and makes one feel very good about making a very big investment in但它肯定肯定 并让人感觉非常好 做一个非常大的投资
948the things that will play out if you stay on that track and it seems like we are.如果你留在赛道上 事情会顺利进行 看来我们是
949So that I think所以我觉得
950is very good news.这是非常好的消息。
951I think the most important aspect of what you're doing there is you're我认为最重要的方面 你正在做的是
952actually closing the loop with the actual biology because with code and research, it's closed loop实际关闭循环 与实际的生物学 因为通过代码和研究,它是封闭循环
953systems.系统。
954And so they're very fast to iterate.所以它们速度很快
955This is an open loop system.这是一个开放的循环系统。
956So you're closing你快关门了
957a loop and that's really crucial to progress.一个循环, 这对进步来说非常关键。
958Yeah.对
959For me, one of the biggest changes对我来说,最大的变化之一
960with the strategy we're driving now and Alex at the helm is before we had amazing teams我们现在驾驶的策略是 亚历克斯在掌舵之前
961moving generally in the same direction and understanding the potential collaborations总体走向相同的方向,并了解潜在的协作
962and interconnectedness of our work.以及我们工作的相互联系。
963But now we are arms-length moving together.但是现在,我们正携手前进。
964It feels very directed.感觉很直接
965It's very directed and it's very exciting.其方向性很强,令人兴奋.
966It's a little bit scary,有点吓人
967but it's truly a team playing off each other and trying to make progress towards this goal.但是,它是一个真正的团队 互相发挥 并试图朝着这个目标取得进展。
968That has taken a lot of work, but also the maturity, our teams being able to have their这需要大量的工作,但也需要成熟,我们的团队能够拥有他们的
969work at a level of maturation where it actually does make sense to interlock.工作在成熟的层次 它确实有道理 交错。
970Amazing.厉害
971Well, to teams being on the curve, thank you guys for doing this.好了,对于队伍在曲线上, 感谢你们这样做。
972Thank you for joining us.谢谢你加入我们。
973Thank you.谢谢
974在NoPriorsPod的Twitter上找到我们。
975Subscribe to our YouTube channel if you want to see our faces.如果你想看到我们的脸,请订阅我们的YouTube频道。
976跟着Apple Podcasts的节目,Spotify,或者任何你听的地方.
977That way you get a这样你就能得到一个
978在No-Priors.com上,
该视频共有字幕 978 条。解锁更多字幕为会员功能,请移动到 价格