Why We're Living in a Biological Singularity With Ben Lamm | MOONSHOTS Live #297

0:04I was a little terrified backstage when Palmer's like, there's something on the screen that says I'm not supposed to say this. It's like, for the love of God, whatever it is, please don't make it about Colossal. Please don't say it. So I remember at Abundance 360 last March, just before you came on, Elon was there. and he said, yeah, I want a pet woolly mammoth. How are you doing on that project? We get that request. Yeah, but there was actually an amazing, I think it was either American Dad or Family Guy episode where they explain, did you see this? I don't know if you guys saw this, but it was like explain Chris Burnham was amazing.

0:45And it was amazing because then at the end, they're like, yeah, and the woolly mammoths came out and they're about the size of a dog. That's actually how big they were. The fossils were just wrong, right? It was amazing because it explains CRISPR, but I was like, we get the request, the number two request we get is teacup mammoths. Teacup? Teacup mammoths. Everybody wants a teacup. They want to take a mammoth and put it in your purse. We can engineer a new melanin and make it pink, and then Paris may well end. So we are not working on that currently, but we're making good progress on the mammoth project.

1:22Amazing. So let me kick this off. So you've started like four successful companies before Colossal and gaming and defense and AI and mobile. And so you knew nothing about synthetic biology. Yeah. Before you started. Now a company worth over$10 billion. So what made you make the extinction your moonshot? and having no background in synthetic biology, was that a hindrance or an advantage? Oh, I think it's a massive advantage. You know, I think that it affords me the opportunity to go into rooms of, you know, Palmer talked a little bit about this too, like hiring people to replace you and hiring people that are much smarter than you, right?

2:06And so I get what's great about this from my vantage point is like I get to deal with a lot of the same bullshit, but at the same time, I can go in and like sit down with like best Shapiro and learn about ancient DNA extraction and how not to do things. But then the next meeting, I can go learn about where we're pushing the boundaries of multiplex editing and how many edits that we can make at once, right? And so I would say that I knew how to, and I guess in all of my career, I've known how to ask the right questions because I'm really curious. But I really do kind of subscribe to that old adage of putting the top smartest women and men around you, just asking them the questions, right?

2:44And so I get to go to meetings to just ask questions. And, you know, 90 % of the time, I think people are like, okay, if you knew more about biology, this meeting would go faster. But then 10 % of the time, but then 10 % of the time, they're like, we never thought of it that way, right? Because... By the way, this is such an important lesson for all of us as entrepreneurs, right? Just because you're not an expert in an area doesn't make it an area that you shouldn't go into if you love, if you attract the best talent around you. Yeah, George Church, and this is the only brag that I will say, which I'm very proud of.

3:18George Church says that I am the best student he's never had. Because I will literally just pepper him and be like, hey, my favorite times of the year are during holiday seasons when no one's working. Because I will just get on calls for hours and hours and hours with George Church and just talk about the possibilities with synthetic biology. Who's George Church for? So if you don't know George, George is arguably the father of synthetic biology. He's the head of genetics at Harvard. and a lot of the next-gen read-write technologies that were invented came out of the church lab. And his lab's prolific.

3:49There's been numerous multi-billion dollar companies that have spun out of the lab. It's probably the most active startup biology lab in the world. And he's also like 6 '7 with narcolepsy and hilarious. And such a sweetheart guy. Yeah. And I mean, he is the most collaborative person ever, right? And so he's also like hardcore, even though it doesn't come from software, He's hardcore into open source and just like trying the democratization of technologies like genome sequencing He wanted that to go from billions to a hundred dollars Well, she was very active in that category and so he is literally just the most collaborative Co-founder I've ever you know had the pleasure of working with Dave

4:29Peter Diamandis:Yeah, really curious about the business model of biology This is like I think we all know that of all the use cases of AI that are imminent solving all disease, curing all pain is just like highest on the priority list. The business model in biology is just always... Yeah. I mean, and it's great to get your perspective as a serial entrepreneur who came into biology from the outside. So my first question on like the business model is the foundation models. Like we just saw that like out of the box Astra can drive a car. Yeah. Is it going to do biology out of the box? Or do you say, no, no, no, no.

5:02Peter Diamandis:We need to build our own. So the foundation models now, and we've been very fortunate to work with both OpenAI and Anthropic on getting early access to some of that stuff, which has been great. For a long time, we were leveraging it for kind of like thoughtful middleware, right? Coming from a software perspective. Where we were connecting lab notebooks and Jira. I mean, one of the hardest things that we did at the company was retrain scientists to work in Jira. So that's harder than SimSol reprogramming at times. And so we actually, for a long time, you know, the LLMs before the Frontier models were really good at writing term papers, right?

5:42They weren't really great at doing, like, ancestral state reconstruction or comparative genomics, right? Like, that's where they just didn't work, which would have been amazing. So for a long time, we were leveraging them for kind of, like, middleware layers of infrastructure so you don't have to go hire the Deloits or Accentures of the world to build all these systems and reporting, right? So I think we were very successful at deploying that for a long, long time. Now they're getting pretty far where you can run simulation experiments. There's a bunch of companies out there that are standing up, Lila and others, that are trying to do lab automation around it.

6:13We've been pretty thoughtful about how and when we invest in that category. But I think that what you're going to find is you're going to find interesting insights between connections that are really still language problems. So for example, if you go search the literature on everything on mice, if you just want to go build a new therapy company or whatever for humans and you go look at all of the published papers on mice, many of them, they'll call a gene a different thing, they'll classify it differently, they'll run that experiment slightly differently. And so if you go try to run the exact same experiment, you're going to get somewhere between 40 % and 60 % failure rate on published work, which is terrible, right?

6:55So where I think AI is going to be really helpful in the next kind of wave of outside of like small molecule drug discovery, I think it'll be very, very helpful kind of eliminating that gap and bridging kind of that nomenclature between work. Like there'll be times where we'll do work and we'll find out later that there was a peer-reviewed published paper on one of the things that we were trying to solve. But the way we were looking for it wasn't findable on PubMember 1. So I think in the short term, it's going to bridge the communication layer. But I do think you're going to, and you saw this, I think Anthropic announced this a couple days ago, right?

7:28Where they've, and you have Ginkgo and others that are now going to test it. You're still going to need to have a wet lab experiment to test and validate those. But running simulation design across a myriad of different experiments and helping creatively come up with the next experiment, I think that's where we're going to see AI and biology for the next, you know, three to five years.

7:48Peter Diamandis:What about this, like, if I think about text data and research results and going through, you know, thousands and thousands of prior tests, that's all LLM city. But what if my input vector is just a gene sequence? Like, if I dump that right now into Anthropic, pretty sure nothing good is going to come out the other side. Yeah, but at scale. So, thank you for the tee up. So that is what I fundamentally believe is the importance of our global bio vault system that we're rolling out, right? So we're trying to roll out this like Noah's art 2.0 model. Not that Noah didn't get it really right in the first stage, but like at least ours is slightly different.

8:29We're doing like ATAC sequencing and other things that he didn't do. And so we're going out and trying to work with governments around the world to stand up localized bio vaults to get all of that so that we can do T2T sequencing. And so, to your point, I do not think that a single genome is going to, like, you're not going to feed it into mythos and it's going to be like, oh, well, if you make these six changes, it's immortal, right? But I think that if you go look at, like, you know, thousands of genomes across all these evolutionary lanes in avian species and see that they are not susceptible to many of these diseases, you'll look at that, right?

9:05We're doing that on a small scale with things like p53, immortal jellyfishes, other things, looking at kind of like known outcomes of the species and then try to backtrack it down the tree of life. So we're doing some of that right now at one of our companies, which we're pretty excited about. And early indications are positive. But to your point, you're not just gonna be able to throw a genome in and it's gonna be like, oh, here's how you fix it and it makes it perfect. But I do think that that amount of data from a comparative genomics perspective at scale will get you there. But I also think that, so I would make the argument that, you know, the data set is more valuable than the model.

9:45Yeah, yeah, yeah. Because I have the data set. We've made that point so many times. Alex.

9:49Peter Diamandis:Yeah, I'd love to talk a bit about de-extinction. I'm cognizant that you now have four plus spinoffs, but nonetheless, de-extinction, I think, is still what you and Colossal are perhaps best known for. And certain things we're not supposed to talk about. We won't talk about those, though. So, de-extinction. There was a Russian philosopher, Nikolai Fyodorov, late 19th century, parent of a strain of philosophy called Russian Cosmism, that argued that the ultimate trajectory of humanity in developing science and technology would be essentially to develop the technology to revive every human who's ever lived.

10:26Peter Diamandis:Russian Cosmism, you can look it up. So, de-extinction. In some sense, you are... We are not doing that currently. Good to know. But in some sense, by de-extinguishing, de-extincting various model species, including the woolly mammoth, you're the first company, to my knowledge, on Earth that at least has a plausible business model or at least technical trajectory to try to go after the entire historical biosphere. So maybe not even just every human who's ever lived, but every non-human organism that's ever lived. In an era of superintelligence where grand challenges are falling left and right, do you think that humanity's common task, as the 19th century Russian cosmists thought, that we'll ultimately have the technology to revive every organism that's ever lived, or at least every human, do you foresee that becoming possible?

11:24I do not. I think through DNA synthesis, prediction models, and some components of synthetic biology, we will be able to get pretty close approximates. I don't think, though, with two big caveats. Number one, just like I have those other, Colossal does not use any of our technologies at Colossal for humans. So we won't do that, even though I think our technology... You want me to spin off for that?

11:50Peter Diamandis:Is that what you're saying? I think that when you start working with humans, you need different, sometimes different investors, sometimes indefinitely different governance, right? And you go through a different process. And more patients. And more patients, right? That was actually really good advice I got from Bob Nelson. Bob Nelson in the early days was like, hey, don't apply any of this to humans for a while because you effectively go into like code freeze with the FDA and it's just a, like, you're just going to be in this monotony forever. Go get to the technologies until they plateau before you then take them out.

12:21But back to your question, I think that fundamentally, most organisms, remember, these are the most organisms that we know. Very few people realize this, but there's like less than like 100 T-Rexes that have ever been. I just want to go down the path that will scare you. There's less than like 100 T-Rexes that have ever been found, right? But with that, you know, there's billions that allegedly were on Earth at different points in time, right? And so very few things leave a fossil record. So I think it's more likely that we will be able to engineer life from a programmable life perspective in the way that we want it than just bring back everything that ever existed because I don't even think we know, right?

13:02I think it's probably highly likely based on AI superintelligence at some point where we get to the point where we're saying, oh, we're going to engineer this thing. And to your question, maybe it did exist, but we just never know because it left no fossil records, right? So I think it's more likely that we will be engineering life to our advantage than trying to bring back things. And even if colossal or a subset of our technologies, both cloning and genome engineering, can bring back things, as you know, environmental factors, epigenetics, all these other things, like if we could clone Mozart, that doesn't mean that he's going to come out and be like, oh, I'm going to solve where the terrible trajectory of music has gone.

13:39Peter Diamandis:Well, just maybe a follow-up question. So for Mozart specifically, we do know quite a bit about Mozart's life. So our arguendo, if we did want to resurrect Mozart, we'd have no problem at all with reconstructing his childhood environment. To a point, right? Like, to a point, right? And so I'm not in, and by the way, just to be very clear, I am not encouraging in any way this line of questioning, but we are taking humans and growing them. It's like, you're going to make, like, you're going to go down this, like, Michael Jordan LeBron super basketball team in a second and scare me. So I do think it's highly likely that if we were able, or it's not highly, it's 100 % accurate to say that we understand their genetic disposition and aptitude towards these certain traits.

14:25And under the right environments, if you want to go like, you know, if you want to go all simulation design on them and put them in the right environments, then your Mozart 2.0 could probably be better than the Mozart 1.0. But Ben, let's bring it back to the animal kingdom, because right now the whole thesis is that the AI systems can design the genome sequence that reflect a phenotype. So if you want an animal that's bigger... It's getting better. Is that 100 % there yet? But that's the objective, right? If you want an animal that's got a longer snout or an animal that has wings. So, like, I asked you on stage at FII, you know, could you make a Pikachu?

15:08Yeah, that seems to be a weird, like, fan favorite. You know, people are more accepting, I think, of Pikachu than these large genetic human camps that you're thinking. Branded species. Yeah, yeah, it is. Yeah, I think people are more open to that model of genome engineering than...

15:25Peter Diamandis:But it's not just maybe to tie a bow on this. It sounds like it isn't a technical objection, per se. It's more worries of social, political, regulatory concerns. Correct, yes. It's inethical. So this is where I'd like to jump in. You know, in Stuart Brand made this famous comment, Peter, the name of the title of your book, he said, we are as gods, we might as well start acting like that, right? And he said that in 1968. We are at a point where you can de-extinct or bring back any species. How do you think through the ethics of what should we bring back? Which ones, which ones shouldn't we bring back?

16:02Peter Diamandis:How do you think to that? I'm a huge Stuart fan. You know, I love him, love Ryan. That quote has been out there kind of like, you know, Minnie from a famous dinosaur movie. But I don't know if I would characterize it like Stuart and then get like on a plane. But from my perspective, I would say that we spend a lot of time. It is not plausible yet to bring back everything or engineer everything from a synthetic biology perspective. So we try to be very thoughtful, and people ask us, is there a checklist? How exactly do you go about selecting your species? Because there's a species that we've been very public about, and then there's species that we have not yet yet.

16:40Peter Diamandis:Hold on. I'm not talking about what you're doing specifically. Oh, you're saying philosophically. Generally philosophically, right? If we could bring back any species, who gets to decide? What evidence do you use to say, oh, bring this back or that? Just from a societal ethics perspective, how should we be thinking about that? Speciesism, right? Yeah. Yeah, and so it's a great question. And because we're kind of the, you know, guinea pig in this world, the way that we think about it is what was their contribution to the environment? What was their contribution to the food web? What was their, why did they go extinct?

17:14How do indigenous people feel about it, right? Like some of the species that we work on, they have a deep spiritual connection to the indigenous people side of it, right? So I don't think you're going to walk in and quote Stuart Brando, right? And so I think that it's very important to kind of like factor all those and weight those. And we're probably going to get it wrong. And I think society is going to get it wrong, but we're going to continue to try. But then we also look at like, is there an educational benefit to it? Right. Like one of the things that we did when we did the dire wolves, right, after working with the indigenous people groups and the Red Wolf coalition teams and all these different components is we did.

17:45We did weigh in the pop culture nature to it. We thought, is there a way that we can bring all these people that focus on sci-fi and Game of Thrones and Magic the Gathering, can we bring them to wolf conservation? And can we bring them, can we teach them about genome engineering because we brought something back that they thought was only a mythological creature in their fantasy universe, right? So we try to weight all of these things differently. There's not a perfect kind of internal algorithm for how we look at it. But I do think that as the technologies proliferate and more governments start deploying our technologies, they will have to weight that on an individual basis.

18:22Yes, they will. And I'm sure that all of us are going to get it wrong at some point.

18:27Peter Diamandis:Well, because there's so many different factors involved, right? Like I'm some indigenous tribe and I worship the Tasmanian devil. That doesn't mean you should or shouldn't bring it back. There's all sorts of other factors. We're gonna have to think through that at a very because the power that you're bringing to the table here is something that we've never seen in the history of humanity Yeah, yeah, I mean it is it is you're like you're like a walking singularity. Yeah, we are We we do I I think that society doesn't fully Understand I mean they also address part and I think that they've all occurred like Ian Malcolm like talk about it But I do think he was really right like I do think that the technologies that are being developed with the kind of intersection of synthetic biology, compute, AI, and then eventually quantum, will be more powerful than any weapon system that's ever been created.

19:20Ben, we talk about solving everything. We talk about, you know, math is cooked, physics is next, chemistry and biology. Is there an inflection point, a singularity in biology, where all of a sudden there is a complete knowledge base of all the ingested DNA, all the ingested phenotypes, and you can literally design an organism. You can have like the CAD software for biology. You can design, you can prompt, give me an animal that does this. Yeah, I think that DNA synthesis isn't quite there yet. But project for me here. Yeah, yeah, yeah. So I think that world is less than 10 years out. Okay. All right, so you can prompt your favorite Pikachu or animal in 10 years.

20:02It doesn't mean whoever owns this technology should do it for you. But I do think that the technologies of being able to engineer key phenotypes on base-level organisms in different clades and be able to synthesize them or multiplex engineer them and then grow them ex utero is within a decade. Okay, ex utero. Within a decade. Talk about your... Within a decade. Within a decade.

20:25Peter Diamandis:Holy shit. This is unbelievable. Well, let me give you a current curve, and this doesn't obviously mean that it's going to continue on this curve. But, you know, we were taking victory laps at 20 edits. And I think that 99 % of biopharma and academia would do that today. We are delivering 300 plus at 90 plus percent efficiency consistently. Right? That was a year ago. Wow. Right? We're now testing 1 ,000. So that's an exponential growing curve. That doesn't mean it's going to continue, right? We're testing 1 ,000 curves. We're working over 1 ,000 edit deliveries right now. It doesn't mean it's going to work.

21:07We're working on it.

21:08Peter Diamandis:Tripling year-over-year base edits? We don't know if it's going to work, but we are having to go. But you think it might. Early indicators have low efficiency, but it's working. But there is a point where large cargo swaps with DNA synthesis is just better. Unfortunately, the people in that category don't really have a business driver to synthesize DNA after a certain scale. So we just started doing that internally. So doing the math, tripling year over year, you're at a thousand base edits right now, that gives you like 12 years? We are consistently north of 300. We're testing a thousand. I think it's highly likely we will get that working.

21:46Peter Diamandis:What's the point at which it kind of really goes crazy and you can do whatever you want? I think the synthesis is gonna, I think synthesis will replace multiplex editing faster. Right, so basically a machine that generates the, the gigabase code you want. I think that has a higher likelihood of success faster. So Ben, you've come up with the artificial egg, not the kind you eat, but the kind that gives birth to an avian species. How far are we from a lady here in the audience, a young lady or older lady in the audience, having a baby in an artificial womb? A human. Well, from a technology perspective, I think that's a very different answer than a societal and acceptance and ethics and regular.

22:30Okay, okay, so let's talk about an artificial womb for a mammal. Yeah, I think within 24 months, we will colossal birth animals fully ex utero. From gestation through delivery. Yeah, that never went into a surrogate. Wow. Wow. That's pretty amazing. Yeah. That's wild. Hopefully sooner, but I think that 24 months is highly likely. Wow.