Powering AI-Driven Drug Discovery with President of Isomorphic Labs Max Jaderberg

Elad Gil: The large part of the industry makes medicines through a very empirically driven process. Where things really fail is in the clinic because we never could see the problems until we get to the point where we put these drugs in patients and we say, "Oh no, there's toxicity that we didn't account for." The really big opportunity is to have much more of that understanding before you go into patients. How can we bring all of that drug discovery work onto a computer? How can we get out of the real world? I'd be really disappointed if we fast forward even five years and the predictive understanding of disease biology has not changed in a massive way. Do you have a prediction on when you think, for example, 80 percent of human cancers will be treatable?

[Music fades out]

Elad Gil: Today on No Priors, we're joined by Max Jaderberg. Max is the president and part of the founding team of Isomorphic Labs. Isomorphic is backed by Alphabet and a variety of other investors, and its goal is to solve all disease, which is a very exciting topic. We're going to talk about the role of AI in the future of biopharma, human health, and longevity. Max has a background in machine learning, with a PhD in the topic, working on machine vision. He eventually sold a company to Google DeepMind, rose up the ranks there, and now is president of Isomorphic. Very excited to have this conversation. Max, thank you so much for joining us at No Priors.

Max Jaderberg: No, it's such a pleasure to be here. Thank you.

Elad Gil: Isomorphic is one of the more fascinating companies happening in technology right now. Alphabet is a major shareholder and sponsor. Demis, who pioneered a lot of models at DeepMind and other places—which I believe you sold your company to before—really helped with the inception and setup and now runs the company. You folks are tackling one of the biggest problems left for humanity: human health. Could you tell me a bit about how you came to work on this company and what was the motive force for starting it?

Max Jaderberg: I was part of the founding team of Isomorphic Labs. I was at DeepMind for a long time beforehand. Personally, I was working on the early days of generative models, language models, deep reinforcement learning, all of the big game playing challenge problems. Those were my jam. AlphaGo, AlphaStar. It got to the point where we were going after these "impossible" problems at that point in time. This was probably 2018, 2019.

Elad Gil: Gameplay. Can AI beat video games?

Max Jaderberg: Exactly. At that point, we were still trying to prove to the world that AI was even a thing. We were going after these impossible problems, and I could see firsthand how to systematically solve them. It got to the point for me personally where I thought, "Okay, sure, we're not at AGI yet, but we clearly have this amazing set of technologies in deep learning and reinforcement learning. What are the problems we really need to be focusing this stuff on?" I was thinking all of this. Of course, Demis was thinking all of this. At the same time, we were seeing the AlphaFold results, the first embers of that starting to emerge. For us, this was the first proof of concept that you could get a deep learning model that generalizes in a meaningful space in biology. That gave us the conviction that now is the right time to go after the big mission of changing the way we think about human health, going after solving all disease. That's going to take a substantial effort, and it warrants its own company, its own culture, specific people of different disciplines coming together, and the ability to run really fast. That was the genesis of Isomorphic Labs in about 2021.

Elad Gil: Amazing. Your background started off more on the deep learning and AI modeling side. I believe you sold a company to DeepMind in 2014?

Max Jaderberg: Yeah, that's right. Quite early. I was at this stuff before people realized how important it was, or many people realized how important it was.

Elad Gil: Could you tell me a bit about the journey you took and how you got there? What got you interested in it to begin with?

Max Jaderberg: I come from an engineering background. I ended up almost by accident doing a PhD in computer vision. This was just at the turning point. I started my PhD pre-AlexNet.

Elad Gil: Because AlexNet was 2012, was it?

Max Jaderberg: Yeah, exactly. Then I saw that emerge and I thought, "Wow, this is awesome." I remember speaking to Yann LeCun about convolutional neural nets and thinking we needed to translate this into the core computer vision space we were working on at Oxford. We ended up developing all of these methods. We had the ImageNet winning network of 2014, best object recognition, text detection, all of that. I built a company in this space—Computer Vision as a Service—and was trying to raise money. At that point, Demis and DeepMind had just sold to Google. I ended up speaking to Demis Hassabis, asking for angel investment. He said, "Well, why don't you just come to DeepMind to solve AGI?" I said, "What's AGI?" But that began the journey into DeepMind and everything there.

Elad Gil: Amazing. Your focus now is a vertically integrated company doing everything from models to biology. The idea is to find drugs that can help people with different aspects of human health. If you look at the pharma industry, one thing that's really striking—I've graphed this out in the past—is that the number of new drug approvals per year is roughly flat, despite an explosion in knowledge. The number of scientific papers has gone up 10 to 20 times during that period. The number of new modalities—antibody-based drugs, small molecules, proteins, peptides—has gone up. The price has gone way up, from a few hundred million dollars per drug to 1.5 billion. It went from eight or nine years to 15 years. Every metric has gotten worse while the knowledge base has exploded and the types of molecules have increased. Why is that? Why has it been so static despite all that progress?

Max Jaderberg: It's crazy to see how it keeps getting harder and harder to make new medicines—more expensive, longer. I think there are probably a number of reasons. The thing I think about a lot is that the way we actually make medicines in general, as an industry and as a species, has forever been a very empirical process. At the very beginning, a long time ago—a hundred years ago—the medicines we came up with came about by chance. We stumbled upon something that people realized had physiological effects, and people adopted it. Things got a bit more sophisticated, but it's still empirical trial and error.

Elad Gil: A lot of medicines we just guessed. Aspirin, for a long time, we had no idea how it worked.

Max Jaderberg: Exactly. Aspirin, we didn't know how it worked. Paracetamol, we didn't know how it worked. The way the large part of the industry makes medicines is this very empirically driven process. You have a hypothesis on how a disease might work, a hypothesis on what a good molecule design could be, but you have to make it and test it in the real world and do this very iteratively. That's slow. It's really difficult to get signal out of that. The space of biology is really big and complex. The cell is a big, complex, chaotic system, super high dimensional. Molecule design—whether small molecules, antibodies, peptides—is again really high dimensional, combinatorially large design problems. The only tools we have are trial and error screening processes. That limits the complexity of solution you can find in a finite period of time or with a finite amount of money. Over time, the low-hanging fruit of medicines that can be found with our empirical process are being found. But it's getting harder and harder to ratchet up the complexity and find things that are out of reach from this empirical process.

Elad Gil: That's interesting. But if I look at any other industry, often what happens is the age of the industry impacts the rate of change. The older the industry, the slower it moves. Things become more bureaucratic, there's more regulatory capture. The last time there was a biopharma company worth over $50 billion—besides Moderna, which was almost an accident of COVID—was the 1980s with Regeneron. It's been almost 40 years since we've had a new tens-of-billions-of-dollars biotech built.

Max Jaderberg: Regeneron, Vertex. Amgen was that era too, the early 80s.

Elad Gil: How much do you think is just regulatory issues, bureaucratic organizations, or natural aging? That happens everywhere—automotive until Tesla.

Max Jaderberg: I think there's a big cultural aspect of how you can keep innovating as an industry, how you can keep moving really fast. Related to the science and technical parts, what are the mechanisms that actually give you economies of scale on the innovation side, on the research and development side? If you go after a particular drug target and you're trying to make a particular medicine, and you're a pharma company or biotech going all in on that, what do you learn from that that helps you work on the next medicine much quicker, and then the next medicine even quicker after that? If you could get that compounding of learnings over time, you could get the benefits of scale and build something really amazing. But I think a lot of the techniques and the way the industry goes after medicines treats each one as a segmented, bespoke frontier science project. You learn so much about that project, but maybe not so much is transferable to the next. This is where things can shift quite a lot in the world of AI and machine learning. If you have this platform underneath going after these medicines that allows you to really suck up and understand the learnings and impact your core foundational understanding of disease biology and how to design good molecules and exquisitely engineer these medicines, then you can hope to get compounding improvements the more medicines you go after.

Elad Gil: One of the striking things that has emerged from deep learning and the foundation model era is the generalizability of some of these things with scale. It sounds like in biology as well, if you have sufficient scale and a sufficiently thoughtful approach, you can generalize some of these problems that were bespoke before. What approach are you all taking?

Max Jaderberg: I think that's the big shift we're taking with Isomorphic. We're not the first people to try and change the way we do drug design and drug discovery with AI. One of the things that's really fascinating about Isomorphic is the degree to which we have a very diverse base of people all working together towards a common cause, often coming from backgrounds that are very disparate.

Elad Gil: Can you tell me more about what sort of people you hire? What are their profiles? Who are you looking for currently?

Max Jaderberg: It's an interesting space because the mashup of frontier AI and drug discovery, chemistry, biology hasn't really existed for very long. You can't go out into the world and say, "I want to hire the world experts at AI for drug discovery," because they basically don't exist. If they do exist, they're not using our models or our techniques. So we think about going after literally the best people in core AI and deep learning who maybe don't even have a background in life sciences or biology or chemistry, but really want to see their work on deep learning translate into amazing impact for humanity. We hire core people in AI. We hire core people in biology and chemistry, maybe who haven't had much exposure to AI. Then we work hard internally to build that culture where everyone talks and learns from each other. Internally, we get these cross-disciplinary expert scientists emerging, which has been really exciting to see.

Elad Gil: How do you think about the ratio of those things? I started a company a long time ago called Color, and I was CEO for the first couple of years. We were using ML and data science applied to genomics, building genomic pipelines. We found you could have five ML people and then one research scientist from the traditional field. That was kind of the ratio we adopted. What is it like now? How do you think about how many people of each discipline you optimally have in a pod working on a problem?

Max Jaderberg: For some of these core AI superpower projects, the AI modeling projects, I think that's not too dissimilar. You have your core modelers, engineers, AI scientists, and then you just need some key domain experts who really understand the application, the nuances of the data—what the metrics mean, making sure we're not chasing smoke metrics and that these are real things that will translate into actually making medicines. Those ratios, one in five, resonate quite a lot for what our teams look like.

Elad Gil: Are there specific roles you're actively hiring for currently, or just across the board?

Max Jaderberg: We want the best AI researchers coming into Isomorphic to help us push on this mission. There's literally so much work to do. On the core simulation models like AlphaFold++, but also on the core reasoning models—taking all of the frontier models and open source models and adapting them to our space. That's a really important area for us to be hiring on.

Elad Gil: What are the components you tackle? What are the main areas you're focused on in terms of how you approach new drugs? Do you have your own labs? Do you run robots that do pipetting? What is the approach you're taking?

Max Jaderberg: If you go into our core AI team, our North Star and the thing we're really trying to achieve is: how can we bring all of that drug discovery work, all of that drug design work, understanding of disease biology, onto a computer? How can we get out of the real world? I know a lot of people are building ways to interface with the real world really efficiently so you can do lots of real-world iteration. That's great, and it's complementary. But fundamentally, the big time bottleneck, the big cost bottleneck, is synthesizing real atoms and then testing them out in real assays and experiments. Just like we've seen pretty much every other area of engineering move into a computer-aided design scenario—which changed what we can design and engineer—we want to do the same thing in this space. When we think about the core AI drug design engine development, this is really focused on how we can achieve experimental accuracy for our simulation models, our world models. If you can get a predictive model—let's use AlphaFold as an example. AlphaFold looks at structure prediction, the structure of proteins and how they interact with each other. Normally that takes months or years to do X-ray crystallography and get a structure out. If you can do that with a simulation model like AlphaFold, then you don't have to go into the real world and do that. You save months and months of time. Drug discovery is a combination of many different really hard scientific problems and engineering problems. If you can create predictive models instead of going into the real world, you're saving tons of time. Internally, we have a dotted line on all of our leaderboards that is experimental accuracy. The goal of all of these superpowers we're trying to create is to get to a point where it's experimental accuracy. At that point, you don't need to go into the real world. You can do so much more design and iteration on a computer so that when you do go into the real world—when you make your molecules, when you test your scientific hypothesis—the probability of success is much, much higher than it would have been traditionally. You don't have to do as many turns of the real world, and things become quicker and problems fall easier.

Elad Gil: When I look at drug development, it's about a 15-year process now, costing $1.5 billion. My sense is the small molecule and in vitro work, the tissue culture and mouse work, is about two to three years and $50 million, order of magnitude. The other $1.45 billion and 12 years is the clinical trial and the different phases and safety. How much of that do you think you can actually encapsulate in a computer? You have the discovery phase where you go from idea to thinking you've got a good molecule you could put in patients. Before that point, you have to do preclinical animal studies to make sure it's safe and working as intended. That takes another year. Then you go into clinical studies—phase one, phase two, phase three. When you look at the timeline and the failure rates, that's what really exacerbates the costs on average. Why do things cost so much? Because so many things fail. Where things really fail is in the clinic. It's not because these are problems—they fail in the clinic because we never could see the problems until the rubber meets the road and we put these drugs in patients. We say, "Oh no, there's toxicity we didn't account for," or "We have a molecule that hits the thing we designed it to hit, but it hits something else with a really negative effect," or "The thing we hit, we thought it would solve the disease, but it actually has a really negative effect in and of itself." People only learn about those problems when they start introducing it to real patients. The really big opportunity is to have much more of that understanding before you go into patients. If you have a better understanding of disease biology, how cells work, you can understand how the thing you're trying to hit and modulate is really implicated.

Max Jaderberg: So you're actually modeling all these other components, not just what molecule will bind to what.

Elad Gil: Exactly. We're modeling all of these components.

Max Jaderberg: How do you get the data to model those things? Where does that data set come from?

Max Jaderberg: We have a ridiculously comprehensive data strategy. Data is all over the place. But fundamentally, what we've learned over the last five years is there's a lot of opportunity to generate this data yourself. There are big repositories of data. Sometimes that's really good for deep learning. Oftentimes, these repositories—whether public or in pharma or in companies—have been the byproduct of many years of passive data collection from the activity of making medicine. They can be very biased, which often doesn't help if you're trying to make a general solution to drug discovery. If you really care about generalization out of these models, there's a different distribution of data you need. So we go out and generate this ourselves. We've seen some really amazing data scaling laws as a result.

Elad Gil: How do you measure whether some of these things will have side effects? There's the old saying—I have a PhD in biology and used to work in cancer, neuroscience, etc. On the cancer side, we cured cancer in mice hundreds of times, and those molecules wouldn't translate into people. A lot of the data you actually needed had to be generated in human clinical trials. How do you go about generating that data that actually informs it? If you're using animal models, often the data that comes out is incorrect relative to human biology.

Max Jaderberg: Exactly. It sounds like you experienced a lot of this firsthand. It's very painful. If you think about toxicity, the gold standards you need to get through for regulators is proving your molecules aren't toxic on rats and other animals. But even when things are approved and proven on rats, they end up being toxic in humans. Why is that? A rat isn't actually a human. It's not very predictive. Choose your study, but it's about 30% predictive of human clinical tox in a rat. If one of my AI scientists or engineers came up to me and said, "We've got this amazing model, it's 30% predictive," you'd say, "Okay, maybe do a bit better than that." There's a lot of opportunity to rethink what those predictive endpoints are for human clinical tox. Forget animals—they've been a proxy for the last 50 or 70 years. We can do better by bringing together in vitro endpoints—things you can measure quite easily in the lab—together with AI predictive models, and also with some amazing new experimental technologies coming out: organoids, organs on a chip, synthetic human tissues that approximate and create microcosm environments of human organs. There's probably an opportunity to bring all of this together and directly try to model human clinical tox. There's not much data on that side because there are only a finite number of clinical trials that have been run. But we have some really good methods now.

Elad Gil: For these approaches or techniques, do you tend to partner for them? Do you bring them in-house? I'm curious how you operate relative to—there's so much that could be done. How much do you want to emphasize the computer science side? In biology, there are things called CROs, contract research organizations, that people use. The quality tends to be pretty mixed in terms of output. How much do you depend on external parties versus internal?

Max Jaderberg: For making our drug design engine, all of the AI models, we do that all in-house. We see a big opportunity in generating proprietary data and owning that ourselves. But biology and experimental platforms are really complex and long-tailed. There are some amazing people working on fantastic platforms. So we work with the whole world—whether that's stuff we do ourselves, contract research organizations, CROs, or other innovative companies—to help us create the data that powers our models.

Elad Gil: That's cool. Anthropic recently announced they're doing a wet lab in San Francisco. Reading between the lines, it looks like there's some potentially robotic system or closed-loop system they're trying to build. In general, with AI models, as you know better than me, you want closed-loop systems to iterate quickly. If you can get rapid feedback—like writing code that can instantly get tested—you have a fast way to test if the model's working well. In biology, one issue is you set up an experiment, wait three weeks, get the results, interpret the results. There's a big loop. That slows down progress and the ability to iterate on data. Building these robotic platforms or other things may expedite that. How do you approach that problem? Do you think that's part of the solution you'll be pursuing?

Max Jaderberg: It's been really exciting to see how much better automated labs and robotic solutions for experimental labs have been getting. We've been working with these automated labs for a long time now. Everyone should, and we should keep driving down that iteration time with the real world. But fundamentally, I want to get out of the real world. You're going to hit the finite laws of physics at some point. How long does it take to synthesize a molecule? At some point, you have to wait for the reaction to happen, and that's how long the reaction takes. How long does it take to culture a cell? At some point, you have to wait those days for the cell to get to the right state before you can test things on it. We can do a lot of work improving automation and the way we interface with the real world. But I think we should also be driving really hard at bringing that in silico, onto a computer, so we can reduce the amount of real-world time. We can do much more predictive work on a computer so that when we do go into the real world, we go to validate, not necessarily to learn and iterate. We still need to go into the real world to validate because we're coming up with novel scientific hypotheses and we want to test them out on people. You have to make sure they work and are safe. The question is, how much can you reduce the amount of time you go into the real world to experiment and iterate?

Elad Gil: How much of your approach is focused on partnering with existing pharma companies versus developing your own drugs?

Max Jaderberg: We make the AI drug design engine in-house. We don't sell the model. We don't license out the model. We don't sell tokens in that way. We use it internally to make medicines. We do that in two ways. We have drug discovery programs that we fully own and fund ourselves and work on ourselves—our internal pipeline. And we have programs that we work on with amazing pharma companies and biotechs where we partner on these projects. Those are really exciting because often those companies have amazing ideas and expertise in specific disease areas. They come to us with the hardest problems—maybe problems they've been working on for 10 or 20 years without much progress—and together, using our drug design engine, we can make a huge amount of progress.

Elad Gil: What are the indication areas or areas of human health that you're focused on?

Max Jaderberg: Internally, we're focused on oncology, immunology, and inflammation. But the cool thing is the drug design engine and the models are general. They're disease area agnostic, target class agnostic, modality agnostic. They work across multiple modalities.

Elad Gil: So you're starting with cancer and autoimmune disease?

Elad Gil: Yes. So, immunology, inflammation, they're often quite related. So it's that sort of space. When we go after these things internally, whether it's oncology or inflammation, I think there's a big opportunity to go after some of these root node problems—the sort of things that could positively affect lots and lots of different patients for different downstream indications if you have a solution.

Max Jaderberg: That's an example of that. If you think about cancer, this is a really complex and long-tailed disease. It shows itself in patients in many different ways, but we also know there are some core driving mechanisms that are common to many different things.

Elad Gil: Sure. We saw this amazing example earlier this year with Drax and Rasib, the KRAS inhibitor, getting a standing ovation at a big cancer conference. It drastically improved the lives of late-stage pancreatic cancer patients, from Revolution Medicines. Again, a great example of going after some sort of root node.

Max Jaderberg: I see. Yeah, Ras has been popping up in the cancer literature for 30 years. But these problems are devilishly hard from a drug discovery and design point of view.

Elad Gil: And it also has a natural function, so being able to modulate it is tricky.

Max Jaderberg: Well, yeah, exactly. And of course, in cancer, things mutate very quickly. So you think you've got a solution, but clinically it doesn't work out because it just mutates around the solution.

Elad Gil: That's an interesting point. For a lot of different drugs in different areas of health—this was also HIV, right? You end up with a cocktail of drugs you give eventually, and that allows you to suppress AIDS or HIV. In cancer, similarly, a lot of drugs were developed as individual drugs versus cocktails. To your point, often within a tumor you have a spectrum of mutations that you need to suppress simultaneously if you really want to cure it, because otherwise it'll evolve out given the substrata of mutations that exist in a distribution across the cells. Are you thinking about those more complex approaches of cocktails or multi-drug approaches? Why aren't we giving more cocktail drugs?

Max Jaderberg: Well, the problem is that often these molecules work, but they have some toxicity or they exhibit some effect on the liver. If you're taking multiple of these medicines at the same time, they have drug-drug interactions that can be really harmful for patients. Because we're not great at engineering exquisite molecules and medicines, we can't combine them together in a safe way for patients. So we're limiting the surface of applicability.

Elad Gil: I guess you only have so many patients and so many people who can enroll in a clinical trial. The combinatorics of it adds up on the clinical trial side, especially when there's a lot of risk and uncertainty about how these things will combine.

Max Jaderberg: That's where we see an opportunity. If you can much more accurately engineer these molecules to be safe, to not have these effects on the liver, to not interact with each other in bad ways, then you can feasibly come up with some quite complex combinations. Combinations that modulate not just a single target at a time, but a whole vector of targets that span these very complex biological pathways simultaneously, and have a better effect on disease.

Elad Gil: There's a very good YouTube video of this guy Janssen, who started Janssen Pharmaceuticals—it was acquired eventually by J&J. He was one of the big pioneers of drug development and pharma. I think they had one of the most productive pharma companies in terms of actual drugs launched. In it, his view that he expresses pretty strongly is that the main constraint on drug development is an overemphasis on safety, and that regulators curtail progress. He pointed out how during World War II, Winston Churchill tried to find a cure for soldiers for some venereal disease in the field. They rediscovered antibiotics because penicillin was discovered in the 20s by Fleming and then put on a shelf and forgotten. In a nine-month period, they rediscovered and commercialized penicillin from scratch specifically to treat this, because he removed all regulatory obstacles. One could argue the COVID vaccine was similar in terms of being able to rush it. How much do you think there's too much of an emphasis on safety? Janssen's point was it's about risk-reward. You need to make tradeoffs. You can't just monolithically say it's only safety. You should also consider benefit. In some cases, risk should be taken. How do you think about that?

Max Jaderberg: I think it's centrally important to always have patient safety in mind. But then you need to understand what the risks and the appetites of patients are, and that can be different for different indications. COVID is a good example where the risk appetite for patients, because it was so pervasive and so dangerous, just changed dramatically. That changed the demand and the force for doing things slightly differently. You even see this across different disease areas. If patients have late-stage cancer and not many options in front of them, there's a very different risk appetite compared to if you're trying to treat a chronic low-level inflammation indication. I think it's really important that we keep focusing on patient safety, but that we can modulate that depending on what the patient need is. Ultimately, we want patients who are living healthier lives and receiving treatment as quickly as possible. We speak a lot to regulators—the FDA here, but also all over the world. The one thing in common is that people really want to get patients the best drugs and the best solutions possible. I do think that over time we will see a difference in the success rates and the safety profiles of molecules coming out when they're designed with much more understanding about how they work. We've been able to search the design space to engineer something much better than you could from an empirically driven process. I hope to prove out over time that this is very statistically different, and that will change the way we think about regulation and getting drugs to patients.

Elad Gil: When do you think you'll have your first clinical trial?

Max Jaderberg: I would love to give you a hard day. This is really hard frontier science. We've got some really exciting medicines in the works, and these are going really well. We're pushing really hard to get these to patients as quickly as possible. We've been building up our clinical development team, so we're getting ready for this.

Elad Gil: I know these are always impossible questions. Do you have a prediction on when you think, for example, 80% of human cancers will be treatable? Or some other metric of why this is so important for the world?

Max Jaderberg: I'd be really disappointed if we fast forward even five years' time and the base tools—the predictive understanding of disease biology—have not stepped changed in a massive way. Our ability to design molecules that are much more precise and much less toxic has not stepped changed in such a massive way that drug discovery and development still takes years. It will reduce over time, but it takes years now. We should have the tools in place to really start cracking these large portions of disease in a different way.

Elad Gil: If you remove cancer from society and heart disease, I think the stats suggest that the average person would live, I can't remember if it's five or seven years longer. The actual impact to longevity for two of the biggest killers is in some sense quite large, and from another point of view, kind of small. There are a number of companies that have tried to focus on aging or longevity as a more general process, which seems to clearly be genetically programmed. If you knock out certain genes in C. elegans or Drosophila or in mice, they'll live 20 to 30% longer. In the case of C. elegans, the little worms, they'll live three times longer if you knock out certain genes and pathways. I actually worked on those pathways as part of my PhD.

Max Jaderberg: Amazing. It was really fun.

Elad Gil: How do you think about that in the context of human lifespan? Should we just hit a point eventually where people live three times longer or five times longer? How do you think about longevity in the context of drug development and all these tools that will now exist?

Max Jaderberg: It's interesting. There are some of these major diseases that afflict so many people that if we could start to make an impact on those, that would have a concrete impact on how long people are living. Even outside of things like cancer and heart disease, you have the chronic afflictions of aging that start to add up. The root cause is literally aging. The fact that you're allowing these systems to degrade over time.

Elad Gil: Stem cell compartments.

Max Jaderberg: Yeah, and they combine in a way that means even if people live to a lovely long age of 80 or 90, maybe the last 10 years of their life, or even 20 years, isn't as great as it could be. Your bones start to degrade, your muscles start to degrade, you have a fall and suddenly you're not as mobile as you want to be. That has impacts on your mental health and your quality of life drops dramatically. You still might live another 10 years.

Elad Gil: In the aging community, they talk about that as lifespan versus healthspan.

Max Jaderberg: Exactly. We think a lot about that.

Elad Gil: But all the aging models, if you impinge on aging and you delay it, the healthspan goes with it. If C. elegans lives three times longer, it's actually a healthy adult for three times longer, and then it crashes out at the end. You've extended both simultaneously. I'm curious how you think about life extension and that kind of stuff.

Max Jaderberg: I think it's going to be really fun to see everyone working on this. We can attack this from every different angle. We can think about improving the things that afflict us and help maximize our healthspan, and also be trying to hit some of the root causes of aging and degradation of our bodies and cells and organs so that we can actually extend that over time. There's such exciting work happening in this space. It'll be interesting to see how this pans out.

Elad Gil: Amazing. Thank you so much for joining us today.

Max Jaderberg: No, it's been a fascinating chat.

Elad Gil: Find us on Twitter at NoPriorsPod. Subscribe to our YouTube channel if you want to see our faces. Follow the show on Apple Podcasts, Spotify, or wherever you listen. That way you get a new episode every week. Sign up for emails or find transcripts for every episode at no-priors.com.