Nov 6, 2025 · 53m · latent-space
Priscilla Chan and Mark Zuckerberg: Frontier AI + Virtual Biology To Solve All Diseases
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of the Latent Space podcast, Priscilla Chan and Mark Zuckerberg detail how the Chan Zuckerberg Initiative converges frontier AI models, high-throughput wet labs, and custom imaging hardware to build virtual cells. They outline their philanthropic strategy to empower the global scientific ecosystem to cure, prevent, or manage all human diseases by transforming biology into a predictive engineering discipline.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. How this is scored →
The hosts as informed peer 5.2 Guest teaching 4.3 Guest disagreement 0.6 The hosts pushing back 1.3
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 40:11 Chan rejects the longevity framing
Chan openly dissents from the longevity obsession, emphasizing that unlike the broader tech meme around curing death, her focus as a pediatrician remains strictly on pediatric health.
Hardest push from the hosts ▶ 4:06 Alessio pushes on tooling versus targeted disease curesAlessio questions why CZI chose abstract tooling and infrastructure over directly funding immediate disease eradication efforts like malaria in Africa.
Biggest teaching moment ▶ 16:20 Chan details multi-modal imaging beyond static slicesChan educates the hosts on the technological complexities of laser phase plates, dye-free live dynamic imaging, and the necessity of adding the temporal dimension to spatial biology.
The host holds their own ▶ 24:57 Alessio challenges digital benchmarking vs wet lab feedback loopsAlessio demonstrates technical depth by contrasting high-throughput automated ML benchmark sweeps with the physical throughput limitations of wet lab validation.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The hosts as informed peer | Guest teaching | Guest disagreement | The hosts pushing back | Why |
|---|---|---|---|---|---|---|
| CZI's 10-Year Evolution: Philanthropy and Basic Science | 4 | 3 | 0 | 0 | Swix introduces CZI's ten-year milestone and asks how their strategy differs from the Gates Foundation's translational focus. Zuckerberg and Chan explain their intentional focus on early basic science and iterative philanthropy. | |
| Tooling-First Strategy and the Scientific Innovation Funnel | 5 | 4 | 1 | 2 | Alessio asks why CZI focused on broad tooling infrastructure instead of directly targeting specific diseases like malaria. Zuckerberg explains the innovation funnel and how government grants underfund capital-intensive, 15-year tool-building efforts. | |
| Bridging AI and Biology: Aligning Timelines and Belief | 4 | 3 | 1 | 1 | Swix notes that curing all diseases sounds alien to outsiders, prompting Zuckerberg to clarify they are accelerating scientists rather than curing diseases themselves. Chan and Zuckerberg highlight the cultural clash between skeptical biologists and overly optimistic AI researchers. | |
| The Virtual Cell Vision and Interdisciplinary Biohubs | 5 | 5 | 0 | 1 | Swix asks about the timeline for achieving in silico virtual cells. Zuckerberg outlines their model of physically co-locating AI engineers with biologists across institutions, while Chan reviews the growth from the Human Cell Atlas to single-cell models. | |
| Multi-Modal Biological Imaging: From Atoms to Living Systems | 6 | 5 | 1 | 2 | Alessio asks whether physical microscopes remain the core bottleneck in digitizing biology. Chan and Zuckerberg describe overcoming spatial limitations with custom laser phase plates, high-intensity X-rays, and living zebrafish models. | |
| Frontier Biology Meets Frontier AI: Grounding Biological Models | 5 | 5 | 1 | 2 | Swix asks how AI models in biology can achieve physical grounding rather than hallucinating like language models. Zuckerberg presents the concept of synchronizing frontier AI labs with frontier biology labs to co-design measurement and modeling. | |
| Experimental Feedback Loops and AI Hypothesis Generation | 6 | 4 | 1 | 2 | Alessio compares machine learning benchmarks with wet lab validation latency. Chan and Zuckerberg explain that AI models will initially serve as hypothesis generators that de-risk bold wet lab experiments rather than replacing them entirely. | |
| Unifying Biohubs and Integrating EvolutionaryScale | 5 | 4 | 0 | 1 | Swix brings up the unification of the Biohub model, prompting Zuckerberg to reveal the integration of the EvolutionaryScale team led by Alex Rives to steer their frontier AI biology program. | |
| Precision Medicine: Resolving Variants of Unknown Significance | 5 | 5 | 0 | 1 | Alessio asks if releasing top biological models within a decade is sufficient success. Chan reframes the ultimate metric around clinical translation, explaining how virtual cell models can resolve variants of unknown significance and personalize drug efficacy. | |
| Transforming Healthcare: Proactive Medicine, Doctors, and Longevity | 6 | 5 | 2 | 2 | Alessio and Swix explore superintelligence in medicine and ask whether death itself should be classified as a treatable disease. Chan admits she focuses on pediatrics rather than longevity, while Zuckerberg discusses shifting medicine from reactive treatments to proactive prevention. | |
| Multiscale Hierarchies and Engineering the Virtual Immune System | 6 | 5 | 0 | 1 | Swix asks if all biological abstractions are leaky across physics, chemistry, and cellular biology. Chan describes how the New York Biohub is engineering immune cells to detect arterial plaques and repair tissue in vivo. | |
| Compressing the Timeline: AI Acceleration and Inverted Biology | 5 | 4 | 0 | 1 | Alessio asks what is required to compress the 100-year roadmap into 25 years. Zuckerberg details an inverted biology paradigm where wet labs specifically generate datasets designed to train foundation models. |
Statements from this episode (24)
Zuckerberg: Science and Biohub are CZI's primary focus going forward
“So, you know, for the next period, we really want to make science the main focus of what we're doing, and specifically the Biohub organization that we're really proud of this model that we've helped pioneer that we can go into detail on is really going to be l…”
Chan: Philanthropies lack the clear performance dashboards companies use
“And the thing that in running a philanthropy, I'm incredibly envious of people who run companies is that like you guys can have a dashboard and there's like financial results and people tell you if you're on the right track on the wrong track and there's clari…”
Zuckerberg: Major scientific advances are historically preceded by new tools
“If you look at the history of science, a lot of major advances are basically preceded by new tools or new ways of observing things, right? So the initial telescope allowed a lot of advances in astronomy. The microscope, the invention of that allowed a lot of u…”
Zuckerberg: Long-term scientific tool development is critically underfunded
“There's sort of been a hole in the ecosystem where tool development and kind of the 10 to 15 year runway that you need to do that and often hundreds of millions of dollars to build things like the microscopes and that you're, and imaging that you're seeing in …”
Zuckerberg: AI could help cure all diseases long before century's end
“And I do think that at the pace that, that AI is improving things, I mean, I think it might be possible significantly sooner than that. I mean, I don't think it's necessarily worth putting a number on it or a date”
Chan: Connecting AI researchers to wet-lab data collectors improves AI models
“Data is not just data. You guys know this, like you need to know sort of how the data was collected and from where and being able to connect the AI researchers to the folks who are actually gathering the data on a daily basis makes their work better.”
Zuckerberg: Virtual cell modeling requires a hierarchical multi-scale architecture
“I think like you want to kind of build up these models hierarchically so You give them a lot of data about specific proteins, and they can model specific proteins in the cells, and then you can model different cell behavior, and then eventually you get, you ki…”
Chan: CELLxGENE corpus reached 125M cells with 75% community contribution
“It took us about 10 years to get to a place where we had, we now have one of the largest corpus of RNA transcriptones, a hundred and twenty-five million cells. Cost a lot of money. And the really cool thing we discovered through that process was if we could se…”
Chan: CZI Billion Cell Project takes months at fraction of historical cost
“Now we're doing the billion cell project and that is taking months and at a fraction of the price.”
Zuckerberg: Direct molecular imaging inside living organisms remains far off
“It's probably going to be a while and people don't have great hypotheses on how you'd actually do like molecular imaging, like of a cell deep inside a living organism.”
Chan: Science has examined only a fraction of billions of human cell types
“We know there are billions of cell types in a human, and we've only truly looked at a fraction of them, and we looked at it in largely healthy cells.”
Zuckerberg: Synchronizing biological tool design with AI model building outperforms using legacy datasets
“Part of what we're trying to unlock here with Biohub is the idea of what actually, what happens If you do frontier biology and frontier AI in sync together and you're designing the tools on the frontier biology side in order to specifically collect and be able…”
Zuckerberg: AI biology models will assist hypothesis generation long before replacing wet labs
“Pretty soon if you have these models, you're just going to be able to run experiments with the models without even having to go to a wet lab. And it's like, no, I mean, I think that that's kind of like, I think that that's sort of the biological version of lik…”
Chan: Predictive AI will de-risk bold, high-risk biological hypotheses
“Right now, because the wet lab is so expensive and relatively slow compared to sort of Computational experimentation. Like people are choosing like, I need something to hit. So people are going for hypotheses or ideas that are like, you know to use a sports an…”
Zuckerberg: EvolutionaryScale is the most talented AI and biology team
“I mean, this is like probably the most talented team working on AI and biology, right? And like at the intersection of doing of like basically good biology background and also, you know, they've just been working on ESM three.”
Zuckerberg: EvolutionaryScale team joins Biohub under Alex Rives
“We're doing that by basically combining the team that we have, that's already put out all the models that we're talking about today, plus having the evolutionary scale team, which is just like very renowned join. And Alex is basically going to be running the p…”
Zuckerberg: Biohub was probably first to build large-scale biology compute cluster
“I think we were probably the first to build out a large scale compute cluster for biological research.”
Chan: AI models will evaluate genetic variants of unknown significance
“And what you really want to do, and I think these models will be able to do is look at those variants and actually model out what is the impact in the different cells, how it influences cellular behavior and whether or not that is Tied to a pathway to disease …”
Zuckerberg: Specialized virtual cell models will merge into a biological Omni model
“I would imagine you're taking these different types of virtual cell models and eventually merging them into the equivalent of like a biological Omni model, kind of like how on the language model side, you had people that did language and then, you know, people…”
Zuckerberg: CZI avoids specific diseases, focuses on accelerating overall science
“We're basically choosing to not focus on any specific disease and like verticalize. Our strategy is one of trying to accelerate scientific progress overall.”
Zuckerberg: Average life expectancy rose ~0.25 years annually over past century
“Since that happened, the average life expectancy has basically increased by, I think it's about a quarter of a year every year over the last hundred years.”
Chan: CZ Biohub NY engineers immune cells to detect heart plaques
“And our New York biohub, we're doing cellular engineering to say like, Hey, can you go in to this person's heart? Check if they have plaques that are causing problems, read it into your DNA, self lice, And then we can read out the signal as cell-free DNA and g…”
Zuckerberg: Timelines to cure diseases depend more on AI than biology
“I guess if we're, you know, predicting whether it's going to take. 10 or 20 or 40 years, that is probably more a function of the pace of AI development than it is a pace of the pure biology side.”
Zuckerberg: AI models cannot solve biology purely from first principles
“You could probably have the smartest AI model in the world, but if it doesn't actually have the data to understand this stuff, it's like, okay, you can't just like reason from first principles about all these things. I mean, a lot of human knowledge comes empi…”