Latest news
Episode

A World Model of Life: Biohub's Alex Rives on Virtual Cells, Measurement, & the $1.8B Data Push

Biohub Head of Science Alex Rives discusses efforts to create predictive virtual cell models and simulate biology. He details the data requirements, modeling breakthroughs, and measurement tools driving Biohub's 1.8 billion dollar push to eradicate disease.

Nathan Labenz
A World Model of Life: Biohub's Alex Rives on Virtual Cells, Measurement, & the $1.8B Data Push
45 min read

Watch Episode Here

Listen to Episode Here

Show Notes

ℹ️
The following notes are AI-generated, based on the episode transcript. Please listen to the episode for the full conversation.

Alex Rives is Head of Science at Biohub. He started the ESM protein language model project at Meta's Fundamental AI Research lab and then co-founded EvolutionaryScale, whose team joined Biohub in November 2025. Nathan opens with Biohub's mission to cure all diseases. Does it mean "solving biology," or curing aging, or living forever? Alex calls himself "a practical person" and sets a concrete target: "a world without disease," meaning no cancer, no infectious disease, and no chronic or degenerative disease. Getting there, he argues, takes fundamental science to understand how disease works. It also takes technology that moves medicine past pills and small molecules toward programmable interventions that can "read and write the programs of biology."

Nathan offers his own two-part split of the challenge: first find the causal target, then design something that hits it. Alex calls that the classic drug-development framework and says it still has plenty of room to run with AI's help. He prefers a more information-theoretic view, though. Life is a program, cells execute rules encoded in the genome, and disease "unfolds in very regular ways" as part of that program. That framing explains his frank assessment of the current field: "the models that are being called virtual cells today, I just don't think they meet that definition of a model that could actually simulate the cell in a meaningful way." He says progress needs three things at once. The first is a better modeling paradigm, since the transformer (Attention Is All You Need) lacks the multi-scale structure a cell demands. The second is data at scale. The third, and most fundamental, is new tools to observe and measure what matters.

Nathan's naive outside view is that today's techniques would work if the data were good enough. Alex mostly agrees, in the limit: with infinite data and compute, all architectures become function approximators and the inductive bias falls away. Protein biology, with gene sequencing at scale, is close to that regime. That is why ESM and its successors (ESM-2/ESMFold, ESM3) worked. Cell biology is far from that regime, so inductive bias still matters. The long-term goal is "a world model of life," able to project a cell's state down to atomically resolved structures and let development unfold in silico, and Alex says plainly that this is far off. What he thinks is attainable within about five years is an information-level model of the cell: a probabilistic transition function from a cell state plus inputs to a distribution over next states.

The data for that model is the point of the Virtual Biology Initiative, which Biohub launched in April with a $500 million commitment. Two days before the recording, Biohub announced a $1.8 billion expansion. The new partners are the Department of Energy, which is putting in more than $500 million over five years through its Genesis Mission; the NIH, which will coordinate datasets from more than $500 million in prior federal investment; and Google DeepMind, Isomorphic Labs, and Meta, which are investing $300 million collectively (coverage: Axios, Endpoints). Alex expects the roadmap to emphasize contextual maps of cellular biology: spatial biology that places cells in tissue and links molecular measurements to imaging "at the scale of hundreds of billions or more cells."

Asked how to balance expensive, rich modalities against cheap, plentiful ones, Alex turns to the laser phase plate. It brings phase contrast to electron microscopy by using an intense standing-wave laser to shift the phase of the electron beam. Alex calls it "the most powerful standing-wave laser in the world." Biohub's own release describes it more conservatively as "one of the brightest lasers of its type in the world," first proposed more than 15 years ago by UC Berkeley's Holger Müller and Robert Glaeser, with Biohub backing it since 2021. The results appeared in Science in June (see also Berkeley News and C&EN). The main application is cryo-electron tomography, where the extra contrast makes individual macromolecules visible inside the cell. For Alex this is Biohub's core belief in action: many paradigm shifts in science have come from new measurement technology, and the biggest difference between biology and other AI domains is how limited our ability to observe it still is.

That leads to the problem of scale. A single point mutation can cause disease, so a useful model has to bridge molecules, cells, tissues, and organisms. Alex insists that a real model of the cell includes the cell in context, not just cells in culture, and he describes Biohub's imaging work at each level. That includes whole-organism imaging of transparent zebrafish, where microscopes track every cell from the single-cell embryo to a developed organism (see Zebrahub). On the old "in mice" joke, Alex grants that model organisms mislead in both directions, producing false cures and hiding human mechanisms. His answer is transfer, not a perfect proxy. ESM and AlphaFold worked because they were trained across all of evolution, not just on human proteins. Across vertebrates the building blocks are conserved even though "the programs are completely rewired."

On the language-model recipe, Alex expects biology to follow the same principles. These are the self-supervised representation learning that Hinton, LeCun, and Bengio championed, made concrete by scaling laws, plus an elicitation stage and eventually RLVR (reinforcement learning with verifiable rewards). The catch is that biology is empirical. Math and code can be verified computationally, but biology needs "empirical AI" that proposes experiments, runs them, and learns from them in a recursive loop with the lab as the verifier. On timelines, he expects first drafts of cell-level atlases of health and disease within one to two years and a fairly comprehensive map within five. On electronic health records, he is cautious: without the right basis "what you're gonna learn is a bunch of correlations." On fusing biology and language into one latent space, he is openly undecided. Bridging data is scarce today, but "probably" eventually, and protein language models already learn features for biology "that we don't yet have language or words to describe." Asked about "a million Bryan Johnsons," he calls deep longitudinal human data immensely valuable but sees Biohub's lever at the technology layer. His own self-tracking? "I have a watch that tracks my sleep. That's about as far as it gets."

Topics covered

No timestamps are listed. Ad placement in the edit will shift every time, so topics appear in conversation order.

  • Defining "cure all diseases": a world without disease, and programmable medicine
  • The classic drug-development framework vs. an information-theoretic view of life as a program
  • Why today's "virtual cells" fall short: modeling paradigm, data at scale, and measurement
  • Inductive bias vs. the infinite-data limit; protein biology vs. cell biology
  • A "world model of life" (far off) vs. a probabilistic cell-state transition model (about five years out)
  • The Virtual Biology Initiative and the $1.8B expansion with DOE, NIH, Google DeepMind, Isomorphic Labs, and Meta
  • The laser phase plate, cryo-electron tomography, and measurement as the engine of paradigm shifts
  • Bridging scales: cells in context, Biohub's imaging work, whole-organism zebrafish imaging
  • Model organisms, the "in mice" problem, organoids, and transfer across evolution
  • Pre-training, elicitation, RLVR, and "empirical AI" with laboratory verification
  • Long time horizons, and cell atlases of health and disease within one to five years
  • Electronic health records: valuable for care, but "a bunch of correlations" without causal grounding
  • Fused latent spaces for biology and language, and protein features with no names yet
  • "A million Bryan Johnsons," continuous monitoring, and Alex's minimal self-tracking

Resources

Guest

Biohub and the Virtual Biology Initiative

Measurement and imaging

Models, papers, and ideas

Quotes worth pulling

"What we wanna do is really live in a world without disease."

— Alex Rives
"The models that are being called virtual cells today, I just don't think they meet that definition of a model that could actually simulate the cell in a meaningful way and actually really serve as an oracle for experimental biology."

— Alex Rives
"So in that sense, you could think of it as a world model, right? Kind of a world model of life, a world model of biology. I think we're very far from that."

— Alex Rives
"The cell is an information-processing machine, which has a genome which encodes the programs that run life."

— Alex Rives
"ESM and AlphaFold did not come from training a model just on human proteins. They really came from training a model across all of evolution… If you trained a model just on the human proteins, you wouldn't get an accurate predictive model."

— Alex Rives
"The programs are completely rewired, but they're built with the same building blocks."

— Alex Rives
"The things that we most want to know about biology, they're not represented in text. And they cannot be just derived from first principles or from what we already know."

— Alex Rives
"With biology, we're gonna have to build the systems that actually allow verification in the laboratory."

— Alex Rives
"I have a watch that tracks my sleep. That's about as far as it gets."

— Alex Rives

Sponsors:

Parallel: Parallel provides enterprise-grade web search APIs for AI agents, offering the optimal balance of quality, speed, and cost. Get started for free at https://parallel.ai/tcr

OutSystems: OutSystems is the leading agentic systems platform that enables enterprises to engineer, orchestrate, and govern AI applications on a single unified platform. Learn more and see how it works at https://outsystems.com/tcr

Tasklet: Tasklet empowers your business with AI agents that connect to your tools and automate recurring workflows with no code required. Visit https://tasklet.ai and use code cog rev for $50 in free credits

CHAPTERS:

(00:00) About the Episode

(03:58) Biohub's grand mission

(09:18) Limitations of virtual cells

(13:37) World models of biology

(19:35) The $1.8B data initiative (Part 1)

(19:40) Sponsors: Parallel | OutSystems

(22:58) The $1.8B data initiative (Part 2)

(29:29) Bridging multiscale biological systems

(35:44) Learning across model organisms (Part 1)

(35:49) Sponsor: Tasklet

(37:26) Learning across model organisms (Part 2)

(42:24) Pretraining and empirical AI

(52:04) Unifying biology and language

(56:32) Continuous monitoring and health

(59:30) Episode Outro

(01:03:11) Outro

PRODUCED BY:

https://aipodcast.ing

SOCIAL LINKS:

Website: https://www.cognitiverevolution.ai

Twitter (Podcast): https://x.com/cogrev_podcast

Twitter (Nathan): https://x.com/labenz

LinkedIn: https://linkedin.com/in/nathanlabenz/

Youtube: https://youtube.com/@CognitiveRevolutionPodcast

Apple: https://podcasts.apple.com/de/podcast/the-cognitive-revolution-ai-builders-researchers-and/id1669813431

Spotify: https://open.spotify.com/show/6yHyok3M3BjqzR0VB5MSyk

Transcript

This transcript is automatically generated; we strive for accuracy, but errors in wording or speaker identification may occur. Please verify key details when needed.

Introduction

[00:00] Hello, and welcome back to the Cognitive Revolution! Today my guest is Alex Rives, formerly co-founder and Chief Scientist at Evolutionary Scale, makers of the ESM series of models, and currently Head of Science at Biohub, the Chan Zuckerberg Initiative backed moonshot with the inspiring mission to cure or prevent all disease, which just announced a $1.8B expansion of the Virtual Biology Initiative, a data-generation effort that brings together the Department of Energy, the NIH, Google DeepMind, Isomorphic, and Meta. I opened by asking Alex what "curing all disease" means to him, and if I should expect to live much longer than the standard human lifespan. Calling himself "a practical person" in response, Alex described, not immortality, but a world without cancer, without infectious disease, and without chronic or degenerative conditions. From there, we get Alex's perspective on why today's virtual cell models haven't revolutionized medicine. The heart of the issue, Alex explains, is that our ability to observe what matters in biology is still very limited by the measurement technology we have available, and that as a result, we simply don't have the quality of data needed to represent the fine-grained causal structure that models presumably would be very happy to learn. Better architectures built on clever inductive biases can make the most of scarce data, such as in the extremely thoughtful use of physics in AlphaFold style models, and we do see real progress in next-token modeling of the biological code, but the available data is failing to capture critical information about how cells and tissues work. It's on this basis that Biohub is investing heavily in data, by opening new modalities, as they did with laser phase plate, an advance in microscopy that is allowing scientists to begin to directly see how individual proteins and other macromolecules interact inside the cell, and also by systematically building cell atlases that link molecular and imaging datasets across hundreds of billions of cells, representing both healthy and diseased states. The goal, over the next few years, is to build an information theoretic model of the cell, which effectively encodes the physical programs that run life, starting with a transition function that takes a cell or tissue state and a set of inputs and predicts a distribution of cell states at the next timestamp, ultimately building toward what Alex calls "a world model of life." This, of course, would support the use case I always dream of, which is answering the question: how do I push a pathologic state back to a healthy one, or in other words, effectively treat disease? Along the way, Alex describes how the LLM training paradigm applies to biology, his vision for "RLVR for biology", with lab verification, reacts to my proposal for "a million Bryan Johnsons", and ponders whether biological and language latent spaces will become deeply fused or remain separated by tool-calls, noting that the things we most want to know about biology are "novel concepts for which we don't yet have words and names." Finally, I asked Alex how he tracks his own health. He answered that he's so focused on Biohub research that he barely thinks about it, which surprised me, until I considered that, if he and the Biohub team are successful in creating the data layer needed to train highly generalizable biofoundation models, a little bit of timely, relevant personal health context might indeed be … all you need. With that, I hope you enjoy this look ahead toward a world model of life, with Alex Rives, Head of Science at Biohub.

Main Episode

[03:58] Nathan Labenz: Alex Reeves, head of science at Biohub. Welcome to the cognitive revolution.

[04:03] Alex Rives: Thank you. Great to be here, Nathan.

[04:04] Nathan Labenz: Yeah. I'm excited for this conversation. So let's start with the big picture. Biohub has this grandiose mission, which I'm excited about, of curing all diseases Yes. Mean to you? Should I think of that as synonymous with another phrase people sometimes use as solving biology? Should I think of it as applying to aging? Should I... You know, are we living forever in your vision of the future? Like, how grand does this vision ultimately get?

[04:30] Alex Rives: Yeah. That's a great question.

[04:31] Nathan Labenz: How do

[04:32] Alex Rives: we define this? Well, I I I think I'm a practical person. Right? So I think about, you know, kinda what what is gonna be possible, I think, in a realistic time frame. And I think, you know, a good place to set our sights is to say that, you know, what we wanna do is is really live in a world without disease. Like, what would that mean? That would mean, you know, a world where there's not cancer, a world where there's not infectious disease, a world where there is not chronic disease or degenerative disease or, you know, all all the kinds of inflictions that have afflictions that have have have been, you know, really fundamental to to the human condition for a very long time. And, you know, I I think about that from a fundamental standpoint, like, you know, how do we how do we build the science that will, you know, really give us an understanding of of biology at a very deep and fundamental level that will under... Allow us to understand how disease works, that will allow us to to really identify the core mechanisms that will allow us to understand how how to detect disease, how to how how to how to intervene appropriately, and also kind of the the the technology. Right? So how can we go beyond kind of the current medical paradigm, which is, you know, kind of... And and I guess the therapeutic paradigm of of of kind of pills of of of molecules? You know, can we really go beyond that to far more programmable medicines that will will allow us to to really get to to the fundamentals of disease and really start to think about that as as as being able to to read and write the the programs of biology.

[06:03] Nathan Labenz: So sometimes when I've thought about this in the past, I've tried to break this grand challenge down into two slightly less grand challenges. One of them is just understanding the causal structure of what's going on inside our bodies so that we could know what leads to problems and therefore what our targets for intervention should be. And then the other one is like having identified a target, figuring out something that can actually have the desired effect. How would you... Know, is that a good taxonomy? How would you break that taxonomy of challenges down further if you want to? And which ones do we have kind of the most traction on, and and where do we need to catch up?

[06:42] Alex Rives: I mean, I think that's really the classic framework of drug development. So if you kind of think about, you know, how we think about creating a drug today, You know, we we think about physiology and disease. We try to identify really kind of a cause at the molecular level because that's where we can actually intervene today with the technology that we have. And then we really think about that problem of kind of, you know, finding and designing the intervention. And so, you know, that... That's that's kind of, I think, one paradigm. And I I think, you know, there's a there's a lot of progress that's that's still possible in that paradigm. So, you know, I I think that's that's kinda where we are today, of course. And and I think AI, you know, is is is is giving us, I think, really new potential to kind of look at each of the each of the parts of solving the... When you break the problem down that way, kind of each of the parts of solving it from that standpoint. But, I mean, I I think there's there's kind of also this this kind of, you know, more, I guess, like, information theoretic view on it that you might take where you can you can kind of think about, you know, life as as a program. Right? You can think about cells as as building living organisms, and you can really think about, you know, that that is a system of of rules, of code. And and you can think about, you know, what what what are the programs. Right? And, like, you know, how how do we... How could we read those programs? How could we start to start to rewrite those programs? And I think, you know, kind of thinking in in that way opens up a a wider aperture to to solving these problems. And I think a big part of kind of where we're trying to get with Biohub is is to really, you know, think about biology in that way, kind of create the tools for programmable biology, create the models that will allow us to understand biology in that context, you know, to really really start to understand how the cell works, both both both down to the molecular level, but also, you know, kinda kinda zooming out the cell in context and the the the, you know, the programs that that it runs to to build a living organism. And I think, you know, you could... You can really think as... Of disease as kind of a part of that. Right? It's it's really a part of the program. And and and if you think about it, you know, disease often unfolds in in very regular ways. So if we could kind of really understand that, I think it gives us really fundamentally new new tools to think about how how to make progress.

[09:18] Nathan Labenz: So I know you guys are working toward virtual cell models. Of course, there have been virtual cell models for a while. My general understanding of the state of play right now is, like, they don't work super well largely, and people have kind of said... Or, you know, have advanced different hypotheses, I guess, let's say, for why that would be the case. The one that I have found the most compelling is basically the data we have is just not that high of quality such that it may not contain the signals that the ML models would really need to learn what's really going on. How would you react to that hypothesis? Like, where where do you think the... Everything that's happened so far, like, why hasn't it taken us further than it has today?

[10:05] Alex Rives: Yeah. The, you know, the cell is incredibly complex. So I think I think you have to start there. You know? It's it's it's it's kind of like the right maybe meso scale for thinking about biology. But, you know, if you if you think about the cell, it's it's it's constituted by this just incredibly complex dynamic collection of of of matter, you know, where where kind of down to the level of individual atoms and and and and the interactions between between molecules. You know, those are those are... Those events are are determining of of of what happens at this... The the larger scale of the cell. And so... And and then... So so there's that kind of perspective, you know, building up from from, you know, the atomic resolution, kind of the first physical and chemical first principles. And then you can also kind of then think about the cell as as embedded within within, you know, living system, right, within within tissue, within physiology, within an organism, and sort of the complex programs that cells are able to execute to sense and respond to their environment, to communicate with each other, to, you know, really do these incredibly complicated and dynamic things. And so it's it's an immense immense challenge. And I think there are, you know, a number a number of limitations. So so so one is kind of, I think, you know, the models themselves, right, and just kind of having the right the right modeling approach and the right modeling paradigm. And, you know, we we have the transformer in in other areas of of AI, but the transformer kind of doesn't have that kind of multiscale structure that could could could represent those those different levels of of complexity. So there's there's kind of a question of of what's the right modeling paradigm. But then I think even more fundamental, there's the the question of of observation, observability, and measurement. And, you know, so much about the cell, you know, we just really can't measure. Sometimes we can't see it at all, And in many cases, kind of generating data at at scale would be really hard. And so if you you think about kind of many of the advances in science and biology, you know, they they really come from kind of having a new technology that that, you know, that that changes what we can measure, what we can observe. You know, the the the new powers to to to to see and observe biology, you know, can can really open up a new understanding of it. And so I I really think there's kind of three parts to really solving this problem. There's there's the modeling part. There's the there's the the the part of generating data at scale that's going to really reveal the the information architecture and the programs of the cell in in its in its kind of statistical structure. And then it's really, you know, the tools that we have to actually observe the cell and and and actually measure the things that that that matter. And I think you need to see progress on all three of those. And so, you know, the models that are being called virtual cells today, you know, I I just... I don't I don't think they meet that that definition of a model that that could actually simulate the cell in a meaningful way and actually really serve as an oracle for experimental biology.

[13:38] Nathan Labenz: Let's do one more beat on how you imagine modeling evolving. I guess my naive sort of outside view, just based on all the success that we've seen in AI and language modeling and otherwise, is just like, if we had enough good data, feels to me like today's techniques would probably work. Maybe you would disagree with that. If you would disagree with that, what sort of future model do you envision? How do you think we sort of encode the right inductive priors and and sort of how how much of a, like, actual Rube Goldberg machine is the model? Is it, like, trying to match up the actual cell with, like, all the little proteins bopping around and interacting with each other and modeling all those, like, microstates? Or, you know, a kind of naive black box thing would be like, I have a cell. I perturb it. Like, what's the next state? You know, kind of the more of an input output without worrying about all the the detailed mechanisms inside. Where do you think we land on modeling when this really starts to work?

[14:41] Alex Rives: Yeah. That's a that's a great question. So I should start by saying that, you know, I really think that kind of in the limit of, you know, data and can... The limit of infinite data and and compute, if you're thinking in limits, kind of all mark... Model architectures probably kinda end up being being the same. Right? They're just kind of function approximators. Right? And so if you have the right inputs and the outputs and you, you know, you learn a very, very good general function approximator, then then then in that sense and those limits, kind of the model architecture itself falls away. But I think I think in biology, you know, we're we're very far from that point. We're very far with some exceptions. Right? Like, with with protein biology, you know, we have gene sequencing at scale. You know, we really can kind of, you know, get closer and closer to kind of that limit of of of of infinite data. But for for cellular biology, you know, we're we're just so far from that right now. And I think historically, when you're far from that limit, you know, inductive bias becomes really important. When I think about, you know, a model, I mean, I I think where do we ultimately want to get? You know, we we want to get to a model of the cells. Kind of what I was saying before. You know, it it has that complexity at the molecular level. It has a representation of that. You know, it could, you know, imagine kind of being able to project out the the the the state of the cell down to down to the level of, you know, atomically resolved structures of its components. You know, it would have that. It would also really be able to model the cell at the level of its programs, and it would be able to, you know, predict. It it would be able to to allow allow development to to unfold, to allow the the the the the the interactions within tissue to unfold and and and tell you what would happen. So in that sense, you could think of it as a world model. Right? Kind of a world model of life, a a world model of biology. I think we're very far from that. So what I think is possible now is an information theoretic model of the cell. And I think that the the... There are, you know, really these programs that run life. They're encoded in the genome. And I think it's it's historically... You know, there's there's been these kinda two dual ways to think about biology. One is this kind of bottoms up first principles approach kind of trying to build it from the basic chemical and physical principles. But I think the other view that's been really productive is to kind of really think about the the emergent level of the informational model of this... Of the cell. You know, the cell is an information processing machine, which has a genome, which encodes the programs that run life. And I think I think that is the... That's the level of understanding of the cell that could be attainable via a... AI in in kind of the the the next five years, let's say. I think I think that we really have... You know, we're in the age of information theory at at scale. And in a sense, you know, we have these incredibly powerful methods which can can can perform... You know, they do pretrained. They they use self supervised on these, you know, vast datasets. And from from statistical patterns that link inputs to outputs, they're able to capture really high fidelity representations of the the underlying latent variables that are that are driving the relationship between input and output. And so you can imagine generating data on the cell that could really start to reveal how information at the genetic level, at the transcriptomic level, at the molecular protein level would link to the programs and responses of the cell that can be observed through different molecular readouts, such as transcriptomics and gene sequencing and epigenetic measurements, but also through through imaging and the power of just being able to see how the cell responds. So I I think I think you were exactly right to kind of think about it as one one of the first things that we can really think about doing is is really learning this transition function that takes us from a state and a set of inputs, you know, to to the next state. And it's really a probabilistic function. It's not gonna be deterministic. You know, the cell the cell is going to transition, I think, to a distribution of of possible downstream state. So it's a very complex problem to learn. But I think today, you know, we kind of have the technologies now that are that are gonna make it possible to create that kind of model.

[19:40]Parallel: Parallel provides enterprise-grade web search APIs for AI agents, offering the optimal balance of quality, speed, and cost. Get started for free at https://parallel.ai/tcr

[20:59]OutSystems: OutSystems is the leading agentic systems platform that enables enterprises to engineer, orchestrate, and govern AI applications on a single unified platform. Learn more and see how it works at https://outsystems.com/tcr

Main Episode

[22:58] Nathan Labenz: So you guys just made some news this week with a... An announcement of a 1,800,000,000 initiative with some partners, BioLevine and a couple other major institutions. Maybe tell us a little bit about that partnership and what the money is gonna go toward. I understand it's largely about creating new data, but, you know, unpack that and and tell us what kinds of data are gonna be the top priorities.

[23:21] Alex Rives: Yeah. We're incredibly excited about this. So in April, we announced the virtual biology initiative, which... With with a group of leading scientific institutions and and consortia and, you know, really said that that we're gonna go out and we're going to really build the datasets to enable virtual biology broadly. And, you know, really, the key goal there to start is to really have this kind of, you know, universal or or general model of of cellular biology. And, you know, so we've we've been working over the summer, and we we announced, I guess, was just the day before yesterday, you know, really a flagship partnership with the Department of Energy, with the NIH, and with with Google DeepMind isomorphic meta, really to come together and coordinate an effort to generate this data. So it's incredibly exciting. You know, we're we're we're collaborating with with with, you know, with a broad scientific community. You know, it's it's... We're... We've really put out a call. You know, we're inviting the worldwide scientific community to be a part of this for other funders to join it. But, you know, we're we're... There's a core group that's that's really focused on just, you know, how how can we how can we have a more coordinated effort across science to do this? And, you know, there... There's... The the the team is is convening, you know, to to to to kind of create that longer term road map. But I think some of the things that are that are really clear that are that are gonna come out of that is just, you know, dramatic expansion of these kind of maps, contextual maps of cellular biology. So things like spatial biology where you're really able to see the cell in the context of of tissue, of surrounding cells, be able to link molecular level measurements to imaging, and and really be able to do that at the scale of hundreds of billions or more cells. And, you know, the goal is really to to move fast here and to kinda create the first datasets that will allow us to start training these, you know, contextual information theoretic models of the cell.

[25:32] Nathan Labenz: In the context of self driving cars, there's, of course, this long standing debate of, like, fancy, you know, more expensive LiDAR type sensors versus just cheap cameras. Right? And, like, how much does it matter that the data will be richer versus we can get more of it? I know you guys also had a recent result in microscopy that I don't fully understand, but basically made the promise of, like, just being able to look inside the cell at higher resolution and even start to see macromolecules in position and to some extent, I guess, how they are interacting with each other. How do you see this in biology? On the one hand, we can do super expensive stuff, not as much of it. You know, alternatively, what are the cheaper techniques, and and how do you think about kind of building the right portfolio of different modalities of data?

[26:22] Alex Rives: Earlier this year, we announced the laser phase plate, and it's it's a really exciting project. It's decades in the making. Holger Holger Mueller at at Berkeley imagined it, you know, many decades ago. Biohub started investing in it years ago. And and, basically, what it is is it's phase contrast for electron microscopy. And so it's it's truly high technology. So so, basically, it's it's the most powerful standing wave laser in the world. So so a beam of light, which interferes with the electron beam to shift its phase so that you can have phase contrast in electron microscopy. And so what this really gets you is the ability to to to to look and and and so so the big application is cryo electron tomography, which is which is basically, you know, sectioning the cell and really now using electron microscopy to, you know, see see all the proteins and and other macromolecules in situ in in their living context. And so the contrast is really critical there because, know, the observations are are incredibly, incredibly noisy, and that contrast allows you to start to make out the individual molecules and and and understand what they are, how they're interacting. And I think it's it's just gonna open up a a totally new understanding of kind of the dynamic structural biology of the living cell. So I think this, you know, this gets back to really one of the core principles of Biohub, which is this idea that the the idea of the power of kind of new new technologies for for measurement and observation to to change how we understand the world, to change to change the science that we can do. And so many so many historic paradigm shifts of... In science have really come from new technologies for measurement. So I think, you know, biology is just so... You know, they're they're they're incredible parallels to other domains where we're we're developing AI, but there are also things that make biology fundamentally different. And I think I think the biggest difference is the the limits in our ability to to observe, see, and measure what matters. And and so so much of biology has been driven by the development of of new tools to to do that. And and as we have more and more powerful tools that allow us to to see more and more of biology, we have the ability to build richer and richer models for it. And so, you know, that's that's really kinda how we we think about this problem is how how can we drive up the number of parameters that we can measure in every cell? How can we expand the field field of view and the number of cells that we can observe simultaneously? How can we open up the temporal dimension to see cells over time, and how can we open up the molecular dimension to really see what's actually happening inside them down at the most fundamental levels?

[29:29] Nathan Labenz: So tell me more about that because I I think, like, a big question I had as I was preparing for this is, like, to what degree does the unit of a cell or even, you know, maybe a few cells tell us what we really need to know. Obviously, there's, like, know, quite a few orders of magnitude from cells to tissues to systems to whole organisms. And I could imagine that, like, you know, everything's connected would be kinda one story and maybe, one random cell in my body because it's all so integrated and information's falling everywhere. Maybe that has enough information to make a good prediction about my overall well-being or maybe it doesn't and probably sometimes it varies. How So do you think about kind of bridging these scales and, you know, what... And what new... Do we have the techniques we need to do that, or do we need new techniques? I was thinking about things like micro biopsies or, you know, what would it look like to have, like, a million Brian Johnsons, you know, who are, like, measured, you know, in all these different ways with all the data linked. Yeah. How do you think about that that challenge, and and what are the techniques you think are kinda most promising to get over those humps?

[30:37] Alex Rives: Yeah. It's a it's a really great question. I think it really gets to, you know, what what is so so hard and and so difficult about this problem. It's it's that there's... It's it's really a multiscale problem, and and complexity down to the molecular level can can influence events of the macro scale. Right? A single point mutation in a gene can can cause disease. And so, you know, you really you really ultimately are gonna need to have an understanding that bridges those those vast divides in scale. And it is a technology problem. I think I think it is very much a technology problem. I think in the near future, kind of the ability to identify the entire system down to the molecular level is is not possible. You know, maybe maybe someday in the future, you know, some something like that would be possible. But, you know, today, we can we can look at we can look at tissue. We can look at cells within tissue. And I think... So we can see that kind of mesoscale of the cellular biology. But what I what I do think is, you know, we can use we can use these informational principles to bridge the gap because if we... I think you have to be clear. You know? And I... When I talk about the cell and I talk about a model of the cell, I don't mean the cell in isolation. And, you know, biologists, they do often think about, you know, this is the cell in isolation and culture systems. They think about, you know, these kind of laboratory proxies for in situ biology. But, you know, I think that that a that a real complete model of the cell, you know, it would have to include the the understanding of the cell in its context in in these living systems because, you know, living systems are are created by the interaction between cells. And, you know, really in a certain sense, you know, the the interfaces that they have with with each other, the input output relations, and the and the transitions and and state that they can undergo.

[32:56] Nathan Labenz: So I think you can

[32:57] Alex Rives: learn a lot kind of at this this mesoscale that's going to be really, really informative to kind of the complete picture of the system. But you really have to think about the complexity levels. And so you can think about the complexity of the cell in isolation, And that's gonna be really valuable because, you know, so you can learn a great deal about cellular biology from, you know, a a cell and culture, from cells that are that are transformed. They they they may be very different from the states of cells in in situ and tissue, but they still, you know, they still have those... That core machinery and those core programs. Right? And so you can learn a lot about them, you know, even even if you're looking at the cell kind of in a very poor poor proxy like a like a culture system. But but we can learn even more if we start to really look at the cell at at that next level of context where we can see it in in tissue or systems. And, you know, we're we're we're really working to address those those kind of gaps. Right? So at Biohub, we have an imaging institute that's really kind of building technologies at each of these complexity levels. So we're... You know, the the the the laser phase plate is is kind of opening up the atomic resolution view, but you can really only see that for for a part of the cell. You know? So it's not gonna give you the picture for an entire tissue. We're building technologies that allow us to see kind of fields of of cells to be able to perturb those cells, understand their responses, and measure that statistically across many cells in culture. And then we're also, you know, really developing whole organism level imaging, which is which is really incredible. And so we do a lot of work with with with the zebrafish, which are, you know, transparent fish, which means you can really image... You can image them, and you can actually see every single cell. And so we have microscopes that allow us, for example, to track a a developing zebrafish embryo over time and and actually see and track every every single cell, you know, how how that system goes from from one cell all the way, you know, to a fully developed organism. We have systems where, you know, you can label different cell cell types. So for example, all the immune cells. So you can look in a, you know, living a living fish and and track and follow all the immune systems and see how they how they move and and and traffic and and interact with with each other and the surrounding tissue. So, you know, I think it's possible to get information at each of these different scales. And then what you really need to do is kinda build that model that can integrate integrate across the the hierarchies of of biological complexity.

[35:49]Tasklet: Tasklet empowers your business with AI agents that connect to your tools and automate recurring workflows with no code required. Visit https://tasklet.ai and use code cog rev for $50 in free credits

Main Episode

[37:26] Nathan Labenz: To these model organisms, I'm interested to hear your perspective on the role that they will play. Obviously, the the old joke is most of the time when you read a exciting headline from biology or medicine, you should mentally append in mice to the end of it.

[37:42] Alex Rives: Yeah.

[37:43] Nathan Labenz: And it seems like, you you look at these headline stats of just like how much DNA do we share with mice and, you know, how much of the kind of core mechanisms of the cells are conserved, and it's like an awful lot. Right? So it seems like should be the case that the treatments that we develop to, like, you know, treat mouse cancer or mouse Alzheimer's or whatever should work in humans more often than it seems like they actually do. And I'm a little confused by that gap and how you think about closing that gap. Is it something where, you know, we somehow need linked datasets between the model organisms and humans or... You know? Because the other the other story I could imagine telling is, like, in the end, we're just gonna need a ton of human data and, you know, there's, no amount of zebrafish data that's ultimately going to add up to the level of insight that we really wanna have into humans to get us where we wanna go. But what's your take on all that?

[38:45] Alex Rives: Yeah. I mean, I think I think it's important to start by by, you know, recognizing that kind of most of the fundamental advances and kind of our molecular understanding of disease and and, you know, the biological pathways that are that are core to that have have come from experiments and model systems and model organisms. You know? And even experiments in in yeast, which is the simplest eukaryote, you know, have have revealed really important mechanisms that are that are relevant to to human biology and human disease. So model organisms have played, you know, really a fundamental role in the development of molecular biology and and just our understanding of biology. But I think you're you're right to point out that they're they're not perfect proxies by by any means and, you know, over... Overly relying on them. I mean, it can lead to two things. Right? It can lead to, like, you know, you think this is a cure to cancer, but it only works in mice. It can also lead to, you know, you really missed a mechanism that's relevant in humans because you don't have a good proxy for it in mice. So you can you can really, I think, make that mistake in in both ways. And, you know, there's there's no simple solution to this problem. Right? I mean, I think there's there's kind of a a constant push in biology to have kinda better and better experimental proxies for for the human biology. And I think, you know, kinda the latest advance there is really organoids, which which allow you to kind of recapitulate more and more of kind of the level... The complexity of tissues in a, you know, in a culture system. But I think I think this is also where AI can can play a role. And so I kind of think about... And I guess I... You know, my... I come from this this background of thinking about protein biology and building models of proteins. And I think one one thing that's really interesting to observe there is, you know, ESM and AlphaFold did not come from training a model just on human proteins. They really came from training a model across, you know, all of evolution. And what that says is there's there's really information across evolution that's that's relevant to making predictions for and understanding human biology. And it's not even that it's relevant. It's essential. If you train a model just on the human proteins, you know, you wouldn't you wouldn't get an accurate predictive model. And so there's there's, I think, something really powerful in this idea of generalization and transfer. And so much of biology is conserved across, you know, all all across life, across all the different domains of life. And then, you know, as you get closer and closer to humans and even if you just think about the vertebrates, so much of the core molecular machinery there, the transcription factors, the way that the way that they interact with enhancers and and promoters, just kind of that language is preserved. And what's happened is the programs have been rewired. So the programs are completely rewired, but they're kind of built with the same same building blocks. So I think you can really, you know, you can really look broadly, and there's a lot to learn from all of these different systems. And I think it's very different to kinda think about

[42:02] Nathan Labenz: building a model

[42:03] Alex Rives: that's going to capture and integrate and learn from all of that information versus trying to create a perfect experimental proxy, which really just isn't isn't gonna be possible. But I but I think I think that there... There's a there's a there's a power in kinda trying to build a predictive model that can use all of that information to make far better predictions for human biology.

[42:24] Nathan Labenz: So in the context of language modeling, there was a time, and we've kind of grown out of it now, where the overwhelming amount of compute was just going into simple pretraining on data of, like, kind of unknown quality and relevance. And then you do a little, like, instruction, you know, following at the end, all of a sudden you have, like, a a pretty helpful assistant. What do you think that recipe is gonna look like in biology? Like, can we just take tons and tons of raw but kinda not that great data and then use that to build up these representations that we then sort of elicit what we want at the end, or is it gonna be sort of a different path that you think for for whatever reason?

[43:08] Alex Rives: Yeah. I... That's that's also an amazing question. I I I really think that the principles in biology are gonna be shared with all the other domains where we're seeing progress in AI. And so I I actually think, you know, we're we're fortunate in biology because we have kind of the opportunity to see how those principles have played out in in different fields and really identify the parts there that are generalizable. But I think this core idea of the power of self supervision or representation learning, you know, the pretraining, you know, that... That's that's that's really critical because, you know, that gets you beyond. And, you know, this this was the dream of of of of Hinton and Jan Lecun and Bingio and and kind of the other deep learning pioneers is the idea that you didn't have to know all the relevant variables. You didn't have to encode all of them and know the settings for the parameters that that... You know, you could learn all of that with with a neural network. And so that's just, you know, proving out over over decades, and the scaling laws are really, I think, the latest or or or the most the most clear expression of that now. But... We we we have pretraining in biology. Certainly, we're seeing that playing out with with proteins right now, of course. And we... You know, I I think that will play out for cellular biology too. And I I think it's it's going to going to be similar. You know, there's there's there's a phase of of of really learning from statistical patterns to build a model of the the underlying system that's driving those statistical patterns, and then the the stage of elicitation where where you're where you're really now bringing out the part of that model which is relevant to the problem that you're trying to solve. And I I think that that it's going to be possible to have RLVR for biology and to have feedback. And, you know, ultimately, where where we need to get is to really build empirical AI and kinda build systems where where AI can can, you know, ask the question, perform the experiment, learn from the experiment, improve its understanding, propose a better experiment, and really complete that that cycle repeatedly so that you have a recursive improvement. And I I think I think all of that's gonna be possible in biology. It's... It just has... It's gonna need to be built differently. You know, we're we're gonna need... It's gonna need to work with models that have... You know, that that can that can accurately represent the the multiscale nature and the complexity of of biology in the relevant ways. And, you know, the biggest challenge for biology is it's it's empirical. Right? And so, you know, we're seeing so much rapid progress with our LVR in areas, you know, where we can really do fast verification computationally. So, you know, mathematics, coding, you know, re... Reasoning, and things things like that. With biology, you know, we have to build the systems that actually, you know, allow us... Allow allow verification in the laboratory.

[46:08] Nathan Labenz: How do you think about timescales? I mean, that's that's another way in which there's, like, orders of magnitude problems. Right? It's one thing to perturb a cell, look at it an hour later, a day later. But what I really care about is, like, how am I gonna live to be 120? Right? And that's a a very long term question. So how how do you think about this kind of, you know, prediction? This kinda goes back to the, like, you know, million Brian Johnson's idea. Do we have enough right now to go on to say, like, this cell state, you know, predicts good outcomes, or is that also, like, a, you know, big part of the climb of the grand challenge?

[46:50] Alex Rives: Yeah. I think that's still a really big big part of the climb. But I think at the same time, it's it's it's it's a scaling challenge. I think it's I think it's one that can be addressed with technology that we have today. I mean, I I guess what I really think about is kind of creating atlases of of cellular biology in the context of health and disease. And I think that we... You know, the technology that could really allow us to do that is is just starting to to really become scalable. This is a big part of what we're trying to do with the virtual biology initiative is is, you know, see cells across, you know, this this wide range of of physiological contexts and really really understand, you know, what what differentiates disease from health at a cellular level. And I think, you know, that's going to really start to show us, you know, both both, you know, what makes different diseases unique, but also shared common causes across different classes of disease. So I I think that's really within reach. I think first drafts of that within the next one or two years, you know, within five years with with a really focused coordinated effort, we can have, you know, a pretty comprehensive map there. And and that would have enormous value, not just for AI, you know, just just for kind of classical analysis. And I imagine a lot of analysis is gonna be agentic in the future. So, also, you can you can really imagine just kind of scaling scaling analysis to these very large... Classical analysis to these very large datasets. But, also, I think you can start to build a model from that that's going to to really identify precisely the different cell states that are relevant to disease. And if you have that model, then what you can start to do is ask the question, how do I intervene to reverse a state that's pathologic and bring it bring it into bring it into a healthy state. So I I think I think that... That's something... You know, those are questions that we'll really start to be able to tackle in the next five years.

[48:58] Nathan Labenz: How big of a deal do you think just like human health records will be as a part of that mix? Like, would you anticipate yourself maybe already been doing this? I don't know. But, you know, kind of going out to EHR companies and trying to do deals to get... And I I don't... The the rules around this, I know, can be quite complicated. But, like, you don't want personal information, but you do need to, like, meaningfully connect a trajectory of a person's health story to, like, you know, specific states of cells. Like, does the regulatory environment allow for that? Can... You know, are there are there other practical barriers that would stand in the way, or do you think that isn't in fact going to be part of the story as sort of all these cells but also linked to human health stories? Yeah.

[49:47] Alex Rives: You know, to be honest, I haven't thought about that question as much. And so maybe, you know, why why not? I think... I guess I guess, you know, for one, I've I've been more focused at kinda trying to get to that fundamental level. And I think what what I what I worry about if you don't have... I think about, like, kinda what's the right basis for solving the problem. And I think that... Because if if you don't have the right basis, basically, what you're gonna learn is a bunch of correlations. And so, you know, what what I would worry is that we're gonna... You know, we're we're gonna have, you know, tons of models of of of of patient health histories, but we're not really gonna have a way to link those back to kind of the deeper causal factors in the underlying biology. And that's the part, I think, that we really have to be able to do if we're gonna make this kind of step change advancement in both our understanding of disease, but also in our ability to treat it. So I think you can probably go a long way through health records and getting better care to to patients, kind of getting getting getting the right drugs and mechanisms that already exist to to the right people, maybe opening up people who might benefit from from drugs that exist, but we we kind of didn't know that that that mechanism could be relevant there. So so I I think there could be a lot of progress through that. I think that's a really important problem, but it it doesn't get down to the fundamentals where I think you really have to make progress to kinda get get this real step change. So, you know, Biohub is really focused on on that part of the problem, like the fundamental science and how how do we go to the point where, you know, there there there really is no disease anymore. And and that's... That... You know, the the only way to get there is going to be by by really focusing focusing on on on the kind of the deep biology and kind of building building the technology that can allow us to understand how that is is determining for for health and disease and and and, you know, how we can effectively intervene.

[52:04] Nathan Labenz: Do you anticipate a future where language and biological modalities will be co trained into joint latent spaces in the way that we now obviously have, like, text and image. You know, when you... I I say this all the time. So listeners have heard it, but just to motivate the intuition one more time, like, it's clear now that when I put an image into a model and I give instructions, the model both sees the image in, like, visual terms and understands my instructions, and and that creates a new kind of magic that, like, just having a language model call via, you know, sort of arm's length tool call to an image generation model, like, couldn't quite do. So do you think that reasoning through language and kind of biology models stay arm's length tool call relationships long term, or do we ultimately see, like, bio encoders that feed into the same reasoning space that we're developing in language?

[53:04] Alex Rives: Yeah. That's... It's really interesting question. I I think that the truth is I'm not sure. I think that language is is really powerful for reasoning. You know, it's a substrate for reasoning, for for thinking. And the power that language also gives us is to to tap into the model of biology that's been created through the... Through centuries of kind of human empirical investigation. And all of that is is represented in language. It's learnable through language. And I think, you know, this this very, let's say, ex... Expressive full system of of knowledge can, I think, really be be materialized and leveraged through reasoning systems? I think the challenge will be that the things that we most want to know about biology, they're not represented in text, and they cannot be, know, derived from first principles or from what we already know. So they're empirical facts that can only be derived by performing experiments. And so it's it's it's from thinking about this, I think, this, like, core aspect of what makes biology different, I think, than other other areas where we're building AI that I think will make predictive models of biology critical here. Now I think I'm not then strongly opinionated or convicted right now on, you know, is there, like, a latent fusion of these models? Are they gonna interact with each other in some system or, you know, kind of ultimately what the end state is of of how that works? Is there one latent space where where all of that is is fused together? I'm not I'm not sure, and I can see advantages to to doing it that way. I think at the beginning, that will be hard simply because you're gonna need a lot of kind of bridging data to to fuse those spaces together, and we just don't have a lot of it right now. But, you know, could could we get there eventually? Probably. So... And and that could be very interesting to have have a fused latent space. I mean, I think I think the other really interesting thing to think about is kind of all those features in biology that we don't yet have language or words to describe. I mean, I think protein biology is such a beautiful example of this. You know, we have billions of proteins that have been sequenced across life on Earth, and, you know, we don't know the function for the vast majority of them. And so we we don't even really have a language to to to describe all of that. So, you know, what's what's kind of interesting about training training a protein language model is that, you know, it develops these features that that represent protein biology at large. And, know, you there there are features that represent unknown protein biology. And you can imagine those are those are novel concepts that we don't have words and and names for yet. So, you know, there's there's there's a great deal, I think, to be to be learned just through through kind of the structure of these these spaces.

[56:32] Nathan Labenz: Okay. I know we're almost out of time. I'll do two more quick ones if we can. Okay. One, not to be a dog with a bone on this million Brian Johnson's question. You know, that would be pretty expensive. But if money were no object and, you know, for some, you know, rare philanthropist, money is maybe close to no object in pursuit of the cure of all diseases, is that, like, a promising paradigm? I mean, would it be really helpful to just have people who submit to micro biopsies on the regular and just like wear all the sensors and do all the things and, you know, track every dimension that they can track? Is is that the shape of data that you think ultimately will really move the needle and it's just too hard to get? Or do you think there is for more scientific reasons, like, just a more direct path through, like, scaling cells?

[57:24] Alex Rives: I mean, again, I... You know, I think I just come at it. We just come at it from this, like, kind of frame of of really thinking about the fundamental mechanisms. But I I don't... I think that data would be immensely valuable. I I just see that, I think, as as more more likely to to happen. I I think I think over time that will happen. I mean, there there are some really interesting, you know, things things that can move the needle there, which is developing the technologies for this kind of continuous monitoring over time. And, actually, having those technologies be able to measure more parameters and more molecular level parameters so that you can start to, you know, link link that kind of bottom up and and and kind of top down view. So I I think there's... You know, that that kind of work is is immensely valuable, and there's likely, you know, over time kinda convergence there. And, you know, we we just think about where where as an as an institution, we can really kinda move move the needle, and we we really see ourselves at that kind of technology layer.

[58:24] Nathan Labenz: Yeah. Interesting. Maybe less neglected than it seems to me from my outside position. Alright. Last one, personal health advice. What are you doing to, you know, build up the dataset that you need for yourself to be able to take advantage of these future technologies? And and what, if anything, you know, do you feel is, like, important for people who want to make sure they're in good position to take advantage of all the things that Biohub and others are gonna discover to be doing today?

[58:50] Alex Rives: You know, I'm not the right person to ask for for that kind of advice. And, you know, I... I'm just so focused on my research right now that I I I barely think about that at all.

[58:59] Nathan Labenz: So no... You're not tracking anything? You're not

[59:01] Alex Rives: No.

[59:02] Nathan Labenz: No.

[59:02] Alex Rives: No regular

[59:03] Nathan Labenz: blood tests?

[59:04] Alex Rives: I have a I have a watch that tracks my sleep. That's about as as far as it gets.

[59:08] Nathan Labenz: Alright. Well, we'll we'll check back in with you and see if maybe that's changed as we get closer to the promised land and we're climbing this monumental hill of figuring out how it all works and how we can cure all the diseases. This has been great. I really appreciate the opportunity. Alex Reeves, thank you for being part of the Cognitive Revolution.

[59:26] Alex Rives: Thank you, Nathan. Yeah. This is really fun.

Outro

[1:03:11] If you're finding value in the show, we'd appreciate it if you'd take a moment to share it with friends, post online, write a review on Apple Podcasts or Spotify, or just leave us a comment on YouTube. Of course, we always welcome your feedback, guest and topic suggestions, and sponsorship inquiries either via our website, cognitiverevolution.ai or by DMing me on your favorite social network. The Cognitive Revolution is part of the Turpentine Network, a network of podcasts which is now part of a sixteen z where experts talk technology, business, economics, geopolitics, culture, and more. We're produced by AI Podcasting. If you're looking for podcast production help for everything from the moment you stop recording to the moment your audience starts listening, check them out and see my endorsement at aipodcast.ing. And thank you to everyone who listens for being part of the cognitive revolution.

Published in: Episode

Read next

Great! You’ve successfully signed up.

Welcome back! You've successfully signed in.

You've successfully subscribed to The Cognitive Revolution.

Your link has expired.

Success! Check your email for magic link to sign-in.

Success! Your billing info has been updated.

Your billing was not updated.