Dr. George Church on Dwarkesh Patel: Why Aging Is Curable

George Church 0:00 This seems like the sort of thing that could wipe out all competing life if we properly weaponize. The thing that's alarming to people like me is that biotechnology enables smaller and smaller efforts, harder and harder to detect. If we handle that properly, then we're probably gonna have almost perfect health. Why, why, why wouldn't we? You know, evolution might incorporate a few base pair changes in a million years. Now we can make billions of changes in an afternoon. All the materials that we use in mechanical and electrical engineering should be made better by biotechnologies. We might even finally get a room temperature superconductor that way.

Dwarkesh Patel 0:43 Today, I have the pleasure of interviewing George Church. I don't know how to introduce you. It would be... Honestly, this is not even an exaggeration. It would honestly be easier to list out the, um, the major breakthroughs in biology over the last few decades that you haven't been involved in. From the Human Genome Project to CRISPR, age reversal to de-extinction. So, um, you weren't exactly an easy prep. (laughs)

George Church 1:04 (laughs) Sorry.

Dwarkesh Patel 1:05 Uh, okay, so let's start here. By what year would it be the case that if you make it to that year, technology will keep, i- in bio, will keep progressing to such an extent that your lifespan will increase by-

George Church 1:16 Oh, wow.

Dwarkesh Patel 1:16 ... a year every year or more?

George Church 1:18 Right. Escape velocity is sometimes what it's called-

Dwarkesh Patel 1:20 Right.

George Church 1:20 ... for aging. Um, different people have estimates and all of those estimates are, including mine, are gonna be, uh, taken with a big grain of salt.

Dwarkesh Patel 1:30 Sure.

George Church 1:30 I think that looking at how, mainly looking at the exponentials-

Dwarkesh Patel 1:35 Mm-hmm.

George Church 1:36 ... in biotechnology and the progress that's been made in understanding, not just understanding the causes of aging, but seeing real examples where you can reverse subsets of the aging phenotype. You know, so, you're getting close to all of aging. Um, in other words, you're seeing e- instead of just saying, "Oh, I'm gonna fix the damage in this collagen, uh, in this tendon, in this limb," you're saying, "Oh, I'm gonna change a lot of things that, that are, that are common to age-related diseases, and I'm gonna get more than one at a time." I think looking at those two phenomena, the exponentials in biotechnologies and the, the breakthrough in general, uh, aging, um, not just observa- not just analysis, but synthesis and, and therapies, and a lot of these therapies are now making into clinical trials, I would not... wouldn't be surprised if 2050 would be a point, if we can make it to that point, uh, 25 years. Most of the people listening to this have a good chance of making it 25 years. And the thing is, it's not gonna be some s-

Dwarkesh Patel 2:46 Mm-hmm.

George Church 2:46 ... sudden point where you're gonna be, you know, so sick 25 years from now that it's, like, hit or miss. It's more likely that, that you're going to be healthier 25 years-

Dwarkesh Patel 2:56 Yeah.

George Church 2:56 ... from now than you thought you were going to be. There may be some, probably not some law of physics, but some economic or complexity issue that we don't know about that, that becomes a brick wall. I doubt it, seriously, but we'll have to see.

Dwarkesh Patel 3:14 Mm-hmm. Given the number of things you would have to solve to, uh, get... give us a lifespan of humpback whales, is there-

George Church 3:20 Bowhead, bowhead whales. Yeah.

Dwarkesh Patel 3:22 Sorry, yeah.

George Church 3:22 200 years. Yeah.

Dwarkesh Patel 3:23 Is there any hope for doing that from somatic gene therapy alone or would that have to be germline gene therapy?

George Church 3:29 Probably there's a, a lot of forces pushing it towards somatic. Um, for one, there's eight billion-

Dwarkesh Patel 3:36 Right.

George Church 3:36 ... people that have missed the, the germline opportunity.

Dwarkesh Patel 3:39 Yeah.

George Church 3:39 That is to say, doesn't apply to us, uh, the two of us and everybody listening to this. A- and, you know, you have to be very cautious when you say something's impossible. It's, it's safe to say it's impossible to do it this second, but you don't know what's gonna happen tomorrow or in the next decade or something. So I, I, I think there's a lot that could be done. In particular, since aging is a fairly cellular phenomenon, with proteins going through the blood and other factors going through the blood that... signaling and so forth, uh, you could imagine that if you replaced, let's say, every cell in the body, every nucleus in the body, left the... you know-

Dwarkesh Patel 4:19 Mm-hmm.

George Church 4:20 ... um, they would su- it would suddenly be young again, right? With- without going all the way back to the embryo and forward again. And there's various other things that are just short of that. If you, if you replace the cells, um, well, they, you know, they'll fit into that niche. They might displace the old cells. You know, that, that's within, certainly within the realm of modern synthetic biology, is to, is to... for cells to take over niches. I think the hardest part is brain, but even there, you know, there's some evidence that, that if you bring... even though the brain doesn't really use stem cells that much, you could artificially bring in stem cells and they could artificially fit into a circuit and learn the, the circuit and then displace the old ones in some way.

Dwarkesh Patel 5:09 Ship of Theseus kind of thing in the brain?

George Church 5:10 Yeah, exactly. Ship of Theseus. Uh, having, um... you know, trying to maintain the connections and the, the memories. Uh, but, but, you know, there's some fairly straightforward experiments that need to be done before we can really even estimate how, how hard that problem is or... You know, very often there's low-hanging fruit that people just think is improbable, but it's there because biology has all these gifts that, you know, where the, uh, just hands over to us, uh, levers that we can flip. Like, like vaccines is an amazing gift that didn't have to exist, but they, they do. Yeah.

Dwarkesh Patel 5:47 Yeah. Is there an existing gene delivery mechanism which could deliver gene therapy to every single cell in the body?

George Church 5:54 There is nothing close to that today, but there's nothing f- no law of physics that would prevent it. You know, there's, there's gonna be prag- practical considerations, you know, like...... you know, how many injections do you need to do, um, to g- to achieve that goal. But we're getting better at targeting tissues. You know, so for one of my companies, Dino Therapeutics, um, showed they could get 100-fold improvement in targeting, uh, neurons in the brain, which is a big deal. Now if, if... And that was just one little campaign that they did, you know, that, that, uh, one experiment. It involves a lot of AI and a lot of, uh, testing of millions of different, uh, capsids. If you did that with cells, r- capsids are, are fairly limited in the diversity and the, and the structure that it c- can change to, but cells have even more, uh, possibilities. I think you could probably get delivery to everything.

Dwarkesh Patel 6:54 Mm-hmm.

George Church 6:55 You know, and the question is, how close to 100% do you need to get, right? And it's gonna vary from tissue to tissue. Um, you know, some... You know, for example, for some therapies, you just need to get 1% because that 1% can produce some missing enzyme, and that, and the 1% doesn't have to necessarily be in its normal place, right?

Dwarkesh Patel 7:13 Mm-hmm.

George Church 7:13 You know? Uh, you can, you can turn a muscle into part of the immune system temporarily for a vaccine. You can, um... You know, uh, an enzyme that's normally made in, let's say, the brain, you could make in the liver, right, if it, if the point is just to get it into the blood. So, um, I think we're... that's moving along quite well.

Dwarkesh Patel 7:36 Mm-hmm.

Dwarkesh Patel 7:37 You're one of the co-founders of Colossus, which recently announced that they de-extincted a dire wolf-

George Church 7:42 Yeah.

Dwarkesh Patel 7:42 ... and now you're working on the woolly mammoth.

George Church 7:44 Yeah.

Dwarkesh Patel 7:44 Do you really think we're gonna bring back, like, a woolly mammoth or... H- h- how... 'Cause like the difference between an elephant and a woolly mammoth might be like a million base pairs.

George Church 7:51 Yeah.

Dwarkesh Patel 7:52 So h- how do you think about what is the k- How do we think about the kind of thing we're actually bringing back?

George Church 7:56 Yeah. Well, so, so I think, uh, I think people get worked up about, you know, whether we, whether we are trying to bring back or have already or will ever bring back a new species. Um, and, and I think of it... If you think of it rather than as a natural thing that we're trying to do, but as a, as synthetic biology with goals that have potential societal... And people also get worked up as to whether this could possibly benefit society in any way. You know, can we really, um, you know, fix an environment, uh, to suit humans or fix the global carbon to suit humans? And the answer is we don't know, but it's worth a try, isn't it? Because it's, could be very cost-effective. Um, and the other thing, the other aspect of it is there's a whole discipline within synthetic biology of asking, "What's the minimum?" Right? And so people often phrase it into, "What's the maximum?" You know, like, "What, what can we do?" And I'm interested in both, but, you know, it's like, oh yes, there's, uh, millions of difference between, um, mammoths and elephants. There are millions of difference between elephant one and elephant two, within, within Asian elephants and between Asians and African. But not all of those are definitive in terms of what we would normally call them, you know, uh, you know, how we would normally classify them, h- what their functionality would be in an ecosystem, right? And so, so there's this exercise that people do, and, and, and we've done it, for example, with developmental biology. What's the minimum number of transfusion factors it takes to make a neuron from a, uh, pluripotent stem cell, right?

Dwarkesh Patel 9:36 Mm-hmm.

George Church 9:37 What's the minimum number of base pairs it takes to make something that will replicate-

Dwarkesh Patel 9:41 Mm-hmm.

George Church 9:41 ... to something that, uh, you know, was done in mycoplasma originally.

Dwarkesh Patel 9:46 Yeah.

George Church 9:46 Um, and these are, these... In a way, these are more interesting than can we make a perfect copy of something, right? It's can we make... What's the minimum things we have to do to make it completely functionally or even functionally in a particular category, right?

Dwarkesh Patel 10:01 Mm-hmm.

George Church 10:02 How do we make it bigger? Could we learn the rules for how to make things bigger? Um, how to make things replicate faster, how to, you know, how to use new materials, et cetera. So I think with the dire wolf, we clearly didn't make an exact copy of the dire wolf, but the diff- but it helped illustrate, kind of educated people around the world, the, that what is the difference between a wolf, a gray wolf and a dire wolf, right? Because, you know, dire wolves, they're, they're big. Maybe they have particular coloration, you know, you know, the heads, head components tend to be bigger than the, than the leg components. Um, and so how many, how many genes do you need to do that? Maybe this was dire wolf, you know, 2.0, and we're gonna go for, go for 3.0 in, in, in successive approximation. And we might wanna develop the technology for making exact copy of something, because then we can... especially being able to make 100 variations on an exact copy, because then you, then there won't be any argument about whether you could make a dire wolf. It's a matter of whether... what, what should you make and what would be most beneficial for the species that you're making, for the environment it lives in, and for humans.

Dwarkesh Patel 11:17 Does this teach us something interesting about... If phenotypes which you think require... are downstream for many genes are in fact modifiable by very few changes, basically, could we do this to other species or to other things we might care about, like intelligence, where you might think like, "Oh, there's, there must be thousands of genes that are relevant," but-

George Church 11:34 Right.

Dwarkesh Patel 11:34 ... there's like, uh, 20 edits you need to make really to, uh, be, be in a totally different ballgame?

George Church 11:38 Yeah, I think it's a, it's... You're hitting on a very interesting, uh, question, and it's related to, you know, what's the minimum? So for example, y- you almost, uh, said it, which was, you know, for... Take a, a very multigenic trait-

Dwarkesh Patel 11:53 Right.

George Church 11:53 ... in humans, like height, is something that's, it's probably the most, uh, well-studied one, simply because no matter what gene you're s- no matter what medical condition you're studying, you collect information on height and weight and things like that. Anyway, they tracked it down to on the order of 10,000 genes.... of which we have 20,000 protein-

Dwarkesh Patel 12:15 Yeah.

George Church 12:15 ... coding genes and some of them are RNA coding genes, and they each have a tiny influence on, on height. Um, but if you take, uh, growth hormone, somato-, uh, tropin, uh, that you have extreme examples where you'll get extremely low, s- small stature and extremely high stature due to that one alone.

Dwarkesh Patel 12:40 Mm-hmm.

George Church 12:40 And in fact, it's used clinically as well-

Dwarkesh Patel 12:43 Mm-hmm.

George Church 12:43 ... uh, in, for seven different, uh, uh, medical, um, treatments. So, so that's a perfect example of how, how much we can minimize something, sometimes called reductionism.

Dwarkesh Patel 12:56 Yeah.

George Church 12:56 Reductionism isn't all bad. Um, sometimes it helps us bring a product into medicine, sometimes it helps us understand or build a tool chest or a, uh, uh, module that we can use-

Dwarkesh Patel 13:10 Right.

George Church 13:10 ... in other cases, and translate it to other species.

Dwarkesh Patel 13:13 Yeah.

George Church 13:13 So, uh, you're, you hit on it just right, is, is that not everything will translate, but w- we start accumulating these widgets. It's kind of like all the electronic widgets that we accumulate-

Dwarkesh Patel 13:23 Right.

George Church 13:23 ... over time, that if you just wanna slap it into the next, um, circuit, you might be able to.

Dwarkesh Patel 13:30 What implications does this have for gene therapy in general? Like, what, what is preventing us from finding the latent, um, knob for, like, every single phenotype we might care about, um, in terms of he- helping with disabilities or enhancement? Is it the case that for any phenotype we could care about, there will be one thing that is like, uh, HGH for height, and how would you, how do you find it?

George Church 13:52 Biology, we, we, we've got a real gift, which is that, uh, it's both very c- much more complicated than almost anything we've designed from scratch, but it also is a lot more forgiving in a certain sense, is that you can have, uh, an animal or even a human that has two heads, which is not something that they evolutionarily, there was not evolutionary selection specifically to have two heads. Um, but just a, uh, just a little deviation from the normal developmental pattern during, uh, you know, fetal development, and, and they both function fine, they control subsets of the body, uh, and, um, you know, they have their own personality, their own life. Um, so this, i- you can... There's all kinds of things you can do in biology that, um, where you're working at a very high programming level is a, is a way of thinking about it.

Dwarkesh Patel 14:47 Yeah.

George Church 14:47 Uh, pushing us to a new level of intelligence is gonna be very challenging, um, and, and maybe not even urgent. Okay? Uh, to some extent, actualizing the people that we currently have would be quite, you know, just getting them all up to whatever speed they want to be up to, um-

Dwarkesh Patel 15:10 Yeah.

George Church 15:10 ... within the range that's been demonstrated in the past. So, like, some people are gonna want to be like Einstein, some people won't. Um, uh, some people will want to be healthy all the time, unlikely, but some people might not.

Dwarkesh Patel 15:22 Right.

George Church 15:23 Some people might want to live to 150, some people might want to die at 80. But they, if you give them the, that, that range, that capability, you know, what if we had eight billion s- super healthy, don't need to worry about-

Dwarkesh Patel 15:37 Right.

George Church 15:38 ... you know, food and drugs, uh, super healthy, Einstein level of intelligence, um, education level best we can come up with-

Dwarkesh Patel 15:50 Right.

George Church 15:50 Um, that would be a completely different world, right?

Dwarkesh Patel 15:52 Yeah.

George Church 15:53 If-

Dwarkesh Patel 15:53 But, but, but just getting everybody to the healthy level, like, how many, how much gene therapy would that take? I- it sounds like it wouldn't take that much if you think that-

George Church 16:02 Yeah.

Dwarkesh Patel 16:02 ... there are these couple of knobs which control very high level functions.

George Church 16:05 Right. Yeah.

Dwarkesh Patel 16:06 So do you find them through the GWAS, uh, genome-wide association studies? Is it through, like, simulations of these-

George Church 16:13 I would say mostly GWAS-

Dwarkesh Patel 16:15 Okay.

George Church 16:15 ... uh, for humans. Uh, maybe for animals in general, um, followed for animals with synthetic biology-

Dwarkesh Patel 16:23 Mm-hmm.

George Church 16:23 ... you know, um, and the, the smaller and the f- cheaper and faster replicating, the, the more experiments you can do. Um, so, you know, um, I don't want to overemphasize how single genes can do these amazing things, but there's also the possibility that multiple genes can be, um, hypothesized and tested quickly. So for example, I mentioned earlier, w- you know, what's the minimum number of transcription factors it takes to turn a stem cell into a neuron? Well, there's a bunch of recipes where you can do it with one, right? Maybe you want a specific neuron, you might need a few more. Um, but then you can get, you can kind of quickly go to the answer by looking at each target cell type that exists, and you can see, well, what transcription factors did it use to get, did it ha- does it express at the time that it's the target? And then you say, "Well, let's just try those on the stem cells and see if they work." And that recipe has, has worked quite well. It's the, it's the basis of GC Therapeutics, the company and, and a bunch of the work that we do, is you can almost, you can get a recipe for almost every cell type in the body. Now, that's not new cell types, but at least you can, you've learned a f- to, to your point about reducing the number of genes we need to manipulate in order to get to a particular goal-

Dwarkesh Patel 17:50 Yeah.

George Church 17:50 ... here's a whole series of goals, and we can get them with one, two, three, you know, maybe seven change, um, transcription factors. So, um, that's an example, uh, and there's room for lots of other examples of where you can do reduction and do not just reductionistic biology, but then constructionistic, where you take it back up and make a whole complex system and see what happens.

Dwarkesh Patel 18:18 Mm.

George Church 18:18 Um, and then you, then you can do lots of those combinations and you debug them and so forth. Some of these things you can do-... um, you know, in vitro things you can do on, probably on the order of 10 to the 14th, 10 to the 17th. Things that involve cells are typically in, in the billions. But, uh, we have this ... This is how we're gonna get, get, um, inroads into the biolo- very complicated biological systems.

Dwarkesh Patel 18:46 Most anti-fraud solutions focus on detecting and blocking bots, and that's fine if your product is just meant to be used by humans. But what if you actually want your product to be used by AI agents? How do you distinguish between automated traffic that you want to allow and automated traffic that you need to block? WorkOS RADAR is built for this exact problem. RADAR is a powerful humans-only filter that major companies like Cursor rely on to protect their product from bots, but it can also handle more complex and granular tasks, like distinguishing desired versus malicious agents rather than just blocking all automated traffic. Even if you aren't building for agents yet, RADAR helps you future-proof your product roadmap. If you start using RADAR for traditional fraud protection, when you do ship your first feature intended for agents, you can just update RADAR's behavior with no engineering required. Learn more about AI-native fraud prevention at workos.com/radar. All right,

Dwarkesh Patel 19:50 back to George. Can I ask you some questions about biodefense?

George Church 19:52 Yeah.

Dwarkesh Patel 19:53 Uh, 'cause some of the stuff you guys work on or, you know, quite responsibly choose not to work on-

George Church 19:57 Yeah.

Dwarkesh Patel 19:57 ... can keep one up at night. Mirror life-

George Church 19:59 Yes.

Dwarkesh Patel 20:00 Given the fact that it's, like, physically possible, why, why doesn't it just happen at some point? Like, some day s- it'll get cheap enough-

George Church 20:07 Yeah.

Dwarkesh Patel 20:07 ... or somebody will care about it en- enough that somebody just does it. What's the equilibrium here?

George Church 20:11 Right. You know, I was a co-author on a paper that, uh, that warned about the dangers of mirror life, uh, just like, you know, I, I wrote a paper long ago about the dangers of, of having the synthetic capabilities we have for making synthetic viruses, and, and to some extent of having new genetic codes. They, they, they have in c- a few things in common, but the thing about the, the advance that we were recognizing in our Science paper that was warning about mirror life is that we not only had to calculate what the possibility of error-prone, you know, escape or something like that. We don't want anything to escape that we made in the lab unless there's a general societal consensus it's a good thing. Um, and so far, there aren't too many examples of that. Um, but... Aren't any examples of that. Um, but mirror life, uh, if it could be weaponized, then it, it... We took it to a whole nother level of concern and the, and the concern was that if we got it to a certain point, it would be easy to weaponize it. And again, there's, there's practical considerations that may be the m- Most people who would consider weaponizing mirror life would probably be satisfied with weaponizing viruses that already exist, that are already pathogens.

Dwarkesh Patel 21:31 Yeah.

George Church 21:32 Um, and they wouldn't want to destroy themselves and their family and, and their legacy and everything like that. But all it takes is one, you know? One group, probably, or one person. But your question is, isn't it, is it inevitable? Uh, I don't know. Uh, could... Might be. It's quite possible it's already here. In other words, we, we already have mirror life in, in our, in our solar system or maybe even on our planet, um, it just hasn't been weaponized, right? And so it, it's just like what we were saying in the Science paper, is this seems like the sort of thing that could wipe out all competing life if it were properly weaponized, but there are probably a few things like that. Um, and what we really need to do is reduce the, the motivation to do that. Maybe increase our preparedness-

Dwarkesh Patel 22:27 Yeah.

George Church 22:27 ... for a variety of existential threats, some of which will be natural, some of which will be, you know, one disgruntled person who has-

Dwarkesh Patel 22:36 Yeah.

George Church 22:37 ... essentially too much power because, you know, over history of humanity, the amount of things that a single person can do has grown s- very significantly.

Dwarkesh Patel 22:49 Mm-hmm.

George Church 22:49 I mean, it used to be when you had your bare hands, there was kind of a limit to what one person could do. A large number of people could team up and get a, let's say, a mammoth-

Dwarkesh Patel 22:57 Yeah.

George Church 22:57 ... or something like that, but, you know, but today, one person with the, with the right connections, um, or, or with right access to technology, you know, could, could blow up a city, right?

Dwarkesh Patel 23:09 Right.

George Church 23:09 And that's a huge im- in- increase in capability, and I think we may want, wanna start dialing that back a little bit somehow.

Dwarkesh Patel 23:18 And then what does that look like in terms of not just mirror life but synthetic biology in general? Y- uh, you know, maybe we're at an elevated period of, uh, the ratio to offense and defense, but how do we get to an end state where even if there's lots of people running around with bad motivations, that somehow there's defenses built up that we would still survive, that we're robust against that kind of thing? Or is, is such an equilibrium possible, or will offense always be privileged in, in this, in this game?

George Church 23:50 Often, offense awfully does have an advantage, but so far, we haven't... You know, we, we made it through, um, the Cold War without blowing up-

Dwarkesh Patel 24:01 Yeah.

George Church 24:01 ... any, any, any hydrogen bombs, as far as I know, uh, accidentally on or in- intentionally on enemies. Um, uh, we did do, did two atomic bombs, uh, but, uh, a lot of that is based on the difficulty of building-

Dwarkesh Patel 24:21 Yeah.

George Church 24:21 ... hydrogen or atomic bombs. The thing that's alarming to people like me, uh, is that biotechnology-... enables smaller and smaller efforts-

Dwarkesh Patel 24:33 Yeah.

George Church 24:33 ... harder and harder to detect, harder and more and more subtle to the stochastic variation between people.

Dwarkesh Patel 24:41 Right.

George Church 24:41 You know, there's some people that are just so happy, they would, you know, they would never wanna do anything close to that-

Dwarkesh Patel 24:47 Yeah.

George Church 24:47 ... or they're so responsible or ethical or whatever. And then there are other people who, like, whenever they have a bad day, they wanna take a lot of people with them, right?

Dwarkesh Patel 24:55 Right.

George Church 24:56 And, you know, maybe s- some progress in psychiatric, uh, medicine would help. Again, you don't wanna force that on people. You wanna make sure that-

Dwarkesh Patel 25:06 Right.

George Church 25:06 ... that, that, they're, that, that if they don't wanna get cured, they, you can't force them but, uh, you can make it available to them. That might help. Uh-

Dwarkesh Patel 25:14 Hopefully there's a more technological solution or (laughs) more robust solution than... (laughs)

George Church 25:18 Well, there are t- there, there will be technological solutions to the psychiatric problem. It could be this peop- even people who aren't sure whether they want to be helped or not can test, try it out and-

Dwarkesh Patel 25:30 Yeah.

George Church 25:30 ... it's reversible and they, they say, "Yes, I like that better." "Okay, let's try that." Um, then, then there's other things that cause, uh, you to, to have bad days. Uh, so it's not s- it's not just your psyche, it's also the environment. So if you, if you're surrounded by your people, you know, being starved, or infectious disease, or being shot at, or something like that, those are things that are subject to sociological and technological solutions.

Dwarkesh Patel 25:59 Yeah.

George Church 25:59 We... If we could really solve a lot of that stuff, we could reduce the probability that one person-

Dwarkesh Patel 26:05 I mean, th- this is making me pessimistic because you're basically saying we gotta solve all of society's problems before we, we don't have to worry about synthetic-

George Church 26:10 Right.

Dwarkesh Patel 26:11 ... biology. Um (laughs) -

George Church 26:12 Yeah.

Dwarkesh Patel 26:12 ... which I'm, like, I'm not that optimistic about, like-

George Church 26:14 I-

Dwarkesh Patel 26:14 ... we'll solve some of them.

George Church 26:15 ... you shouldn't be. I'm not trying to, to reassure you. I'm trying... And I'm... You know, we're having a conversation about what it takes and that might be... That's one scenario for what it might take.

Dwarkesh Patel 26:25 Mm.

George Church 26:25 Uh...

Dwarkesh Patel 26:25 You had a, um, an interesting scheme for remapping the codons in, uh, in a genome so that it's impervious to naturally evolved viruses.

George Church 26:36 Right.

Dwarkesh Patel 26:37 Is there a way in which this scheme would also work against synthetically manufactured viruses?

George Church 26:42 Much harder.

Dwarkesh Patel 26:43 Yeah.

George Church 26:43 A- again, it's the, the, uh, the offense has the advantage. Uh, we can make a lot of different codes, um-

Dwarkesh Patel 26:51 Which would limit the transmissibility?

George Church 26:52 Yeah. So, so, eh, one, one interesting thing is, is that there's only two chiralities.

Dwarkesh Patel 26:58 Mm.

George Church 26:58 You know, there's the current chirality and the mirror chirality, but there's maybe 10 to the 80th different codes.

Dwarkesh Patel 27:07 Mm.

George Church 27:07 Now, uh, some of them you might be able to take out all at once. Uh, anyway, the codes, coding space is a m- is, is a kind of more interesting space and, of course, it could get even more complicated than that because they're, you know, the 10 to the 83rd is, like, based on triplet codons and, you know, that sort of thing. But if they're quadruplet codons or-

Dwarkesh Patel 27:28 Mm.

George Church 27:28 ... they're new, novel alphabets and so on. Um, but w- we're sort of getting into, uh, you know, a, uh, a cycle of, of competition. It would be better to nip it in the bud which is, you know, why, why are we... Why did we spend so much re- societal resources building up to tens of thousands of nuclear warheads and now we've-

Dwarkesh Patel 27:56 Right.

George Church 27:56 ... dialed it back to, to mere 1,000 nuclear warheads. Um, that's nice that we dialed it back but why did we waste (laughs) all that time and money and energy?

Dwarkesh Patel 28:05 Yeah. Biology seems very dual use, right? So-

George Church 28:07 Oh, yes.

Dwarkesh Patel 28:07 ... the mere fact that you're, you, like, literally you-

George Church 28:09 Right.

Dwarkesh Patel 28:10 ... are making sequencing cheaper-

George Church 28:11 Yeah.

Dwarkesh Patel 28:11 ... will, um, will just have this dual u- use effect in a way that's not necessarily true for nuclear weapons.

George Church 28:17 Right. Yeah.

Dwarkesh Patel 28:18 And we want that, right? We want biotechnology to advance.

George Church 28:20 It's, it's hard to, to pound nuclear weapons into plowshares, as they say.

Dwarkesh Patel 28:24 (laughs)

George Church 28:24 Um...

Dwarkesh Patel 28:24 I guess I, I, I am curious if there is some long run vision where, to give another example, in cybersecurity, as time has gone on, I think our systems are more secure today than they were in the past because we found vulnerabilities and we've come up with new encryption schemes and so forth. I- is there such a plausible vision in biology or are we just, like, stuck in a world where offense will be privileged and so we would just have to limit access to these tools and, um, have better monitoring but there's no... there's not a more robust solution?

George Church 28:57 I, you know, one of the things I advocated in 2004 is that we stop, um, deluding ourselves into thinking that moratorium and, and voluntary, you know, signups to, to be, uh, good citizens is going to be sufficient. We need to also have surveillance-

Dwarkesh Patel 29:17 Mm.

George Church 29:17 ... and consequences and mechanisms for, um, whistleblowers, you know, uh, to, to make it easy for people to report things that they think are out of line. And we, we had, we had essentially moratoria and, and, uh, uh, disapproval for germline editing and nevertheless somebody did it, um, and a lot of people knew about it.

Dwarkesh Patel 29:41 Yeah.

George Church 29:41 So that, that was clearly a failure of the whole moratorium voluntary and, and whistleblower components of the-

Dwarkesh Patel 29:48 How you worked for it with, for five years with only one, one defector. That's quite impressive.

George Church 29:52 Okay. (laughs) Um, half empty, half full.

Dwarkesh Patel 29:55 (laughs)

George Church 29:56 Uh, I'll, I'll give you that but, uh, but all it takes is one for some of these scenarios.

Dwarkesh Patel 30:01 Right.

George Church 30:02 Right? And that's, and that's, uh... So, it would have been nice if the whistleblowers could've sa- you know, saved him the three years in prison-

Dwarkesh Patel 30:10 Yeah.

George Church 30:10 ... by, you know, getting an intervention. Um-

Dwarkesh Patel 30:14 Yeah. Yeah.

George Church 30:15 I mean, th- it's not like anybody died. Uh, I mean, there are probably three healthy, genetically engineered children in the world now.

Dwarkesh Patel 30:23 Yeah.

George Church 30:23 Uh, be teenagers soon. Um, but it still, it shows a f- it was a, a good test run, shows a failure of the system. We need to have better surveillance of a- all the things we don't want-... and, and consequences-

Dwarkesh Patel 30:38 Mm-hmm.

George Church 30:38 ... um, that are well-known.

Dwarkesh Patel 30:40 Over the last couple of decades, we've had a million-fold decrease in the cost of sequencing DNA, th- thousand-fold in synthesis. We have gene editing tools like CRISPR, massive, uh, parallel experiments through multiplex techniques that have come about. And of, of course, much of this work has been, um, led by your lab. Despite all of this, why is it the case that we don't have some huge industrial revolution, some huge burst of new drugs or some, uh, cures for Alzheimer's and cancer that have already come about? When you look at other trends in other fields, right, like, we have Moore's law, and here's my iPhone. Uh, why don't we have something like that in biology yet?

George Church 31:17 Yes, so we have something that's about the same speed, a little bit faster than Moore's law in biology. It's more recent, is one aspect of it. So we had... But we had, we could kind of stand on the shoulders of the electronics giants to, to go a little bit faster to catch up. Um, I would say we do. I mean, we have the biotech industry, which, uh, has used that, uh, exponential curve to, to get better. It's also possible we're close to the big payoff, uh, is the other aspect-

Dwarkesh Patel 31:52 Mm-hmm.

George Church 31:52 ... or the beginning of the big payoff. Um, you know, right now, we have miraculous things like, um, cures for rare diseases. We have vaccines. Uh, we have, you know, a trillion dollars probably of, of various biotech-related things if you go far enough apart. Uh, but we're kind of on the verge of really, um, combining electronics and biology more thoroughly and AI and biotech. Uh, and, and I think that's, uh... Uh, it seems like we're on the same track as, as Moore's law, if not better.

Dwarkesh Patel 32:31 Mm-hmm. What, what, what exactly are we on the verge of? What, what does 2040 look like?

George Church 32:35 Well, 2040, we're talking about only 15 years, uh, which is, you know, like one and a ha- you know, maybe two cycles of FDA approval.

Dwarkesh Patel 32:46 (laughs)

George Church 32:47 Um...

Dwarkesh Patel 32:48 2040 is post-AGI. (laughs) It's a long time.

George Church 32:50 Well, I hope it's not-

Dwarkesh Patel 32:51 (laughs)

George Church 32:51 ... post-AGI. I, I, I've, I think we're r- rushing a little bit to get to AGI, and there's lots of cool things we can do with just-

Dwarkesh Patel 32:58 Mm-hmm.

George Church 32:59 ... super AI. Um, but we need to be very cautious. I think that, that AGI... Well, anyway, we could, we could get into that.

Dwarkesh Patel 33:07 I have a question for you there, yeah.

George Church 33:08 But, but, uh, you know, I, I think that we were, we are shortening the, the, the time of getting medical products approved, uh, in a, in, still in a safe way. Uh, so I think, but that's not gonna completely change the, the exponential. It will, you know, might re- reduce it from 10 years down to... One year is our record so far for, say, uh, COVID vaccines, so maybe that'll be 10 times shorter. Maybe that, that will multiply out a little bit. Uh, I, but I think the big thing is that all our, our designs will become better, so there'll be fewer, fewer failures. The cost per, per drug will, will, will drop. There'll be things that we didn't classically consider drugs or instruments, um, kind of cer- some sort of hybrid thing. But again, I don't think that'll be completely shocking, but, but it just, it's just gonna be so much of it, you know? It's just, it's gonna be lots of diversity of, of solutions.

Dwarkesh Patel 34:08 How much more are we talking? Is it, y- are we gonna have 10X the amount of drugs, 100X?

George Church 34:12 I'm not even sure it's gonna make sense, but bet- yeah, 100X would not be completely surprising. Um, combinations of drugs will be important when you're using them intelligently. There'll be a lot, lot more, um... Some, some drugs will affect everything. So for example, you have age-related drug that could-

Dwarkesh Patel 34:30 Right.

George Church 34:30 ... that could impact every disease and it could... Um, I'm not sure the number is gonna matter so much as the, the quality and the impact and inter- intersection and, and, and, uh, and software that helps, uh, physicians and, and other, and regular citizens make decisions, yeah.

Dwarkesh Patel 34:47 Mm-hmm. A- and what specifically is changing that's enabling this? Is it just existing cost curves continuing, or is it some new technique or tool that will come about?

George Church 34:55 Well, the, the, the cost curves are affected by new tools. I mean, it's- it's not just some automatic thing.

Dwarkesh Patel 35:00 Mm-hmm.

George Church 35:00 Uh, there was a, a big dos- discontinuity between, um, Sanger sequencing and nanopores and fluorescent next-gen sequencing. That was... And so, you know, I think, uh, sometimes it's a merger of two things. So clearly, AI merging with protein design caused a step function. These step functions get smoothed out into a kind of a smooth exponential, but there, there, there are lots of them. Uh, next, the next set will probably be, yeah, a merger of AI with other aspects of biology, like developmental biology, um, merger of developmental biology with, with, um, you know, manufacturing and, uh, you know, conquering developmental biology. In other words, actually knowing how to make any arbitrary shape given, um, you know, DNA as the, as the-

Dwarkesh Patel 35:54 Mm-hmm.

George Church 35:54 ... programming material. Uh, I think that would be a big thing, uh, having just more materials in general. All the materials we use in mechanical and electrical engineering should be, uh, made better by, uh, biotechnologies.

Dwarkesh Patel 36:11 Wh- why, why is that?

George Church 36:12 Uh, why is that? Well, it's, uh, that electronics is... You know, Moore's law, I wouldn't say is stopping, but it's, it's kind of the, the, what we would call the one nanometer-

Dwarkesh Patel 36:26 Yeah.

George Church 36:26 ... um, process, which is supposed to come out in 2027 according to the roadmap, is not really one nanometer. It's more like 40 nanometers, center-to-center spacing-

Dwarkesh Patel 36:36 Right.

George Church 36:37 ... you know, uh, in the, in, typically in two dimensions, um, maybe a little bit of three dimensions, but, but biology is already at 0.4 nanometer resolution, and it is in three dimensions. And so-You know, depending on how you count that third dimension, that could be a billion times-

Dwarkesh Patel 36:55 Mm-hmm.

George Church 36:55 ... higher density that biology's already at. Um, and, uh, you know, we just need a little more practice with dealing with the full periodic table.

Dwarkesh Patel 37:05 Mm-hmm.

George Church 37:05 Uh, even electr- electrical engineering and

Narrator 37:08 Yeah.

George Church 37:08 ... doesn't use the whole periodic table typically. But we, but, es- especially not at the atomic level. So that's, I think biology is just really good at doing atomic precision.

Dwarkesh Patel 37:18 So then what's the reason that over the last many decades... I mean, we have, we do have, um, not atomic, but close to atomic level of manufacturing with semiconductors-

George Church 37:28 40 nanometers.

Dwarkesh Patel 37:29 Right. It's quite small. (laughs) Um-

George Church 37:31 It's a thousand times bigger than biology, linearly.

Dwarkesh Patel 37:35 But the, the progress we have made hasn't been related to biology so far. It, it seems like the, the, we've, we've made Moore's Law happen, um, I don't know, people in the '90s were saying, you know, ultimately we'll have these biomachines that are, uh, doing the computing, but it seems like we- we've just been using conventional, uh, manufacturing processes. What exactly is it that changes that allows us to use bio to make these things?

George Church 37:57 Uh, a few things. One is the, the arrival of synthetic biology-

Dwarkesh Patel 38:01 Mm-hmm.

George Church 38:02 ... um, where you, where you sort of... We were already kind of doing synthetic biology before. You know, we were doing... Recombinant DNA was kind of-

Dwarkesh Patel 38:08 Mm-hmm.

George Church 38:09 ... you know, genetic engineering was called. It was kind of in that direction, but synthetic biology really liberated us to think, um, a little bit bigger, um, even though it started kind of focused on E. coli and yeast, it, it, it enabled us to maybe think, uh, about new amino acids, for example. And I think new amino ac- if you start using the full periodic table with the amino acids or what the amino acids can catalyze, um, that, that breaks one of the major barriers. One of the major barriers between, you know, electrical and mechanical engineering and biology was, was the use of, you know, special materials. Thing- things that conduct electricity at the speed of light or conducts signals more-

Dwarkesh Patel 38:56 Yeah.

George Church 38:57 ... more generally. Um, but there's, there's absolute- there's definitely polymers, uh, that biology can make that will conduct at the speed of light. Um, and, uh, you know, we could make a, a mixed neuronal system that has conventional neurons and, uh, processes that conduct at the speed of light. That would be interesting. I think that, uh, our ability to design, uh, proteins was particularly difficult. Designing nucleic acids was great, whether we were doing, you know, you, you want two things to, to bind to each other-

Dwarkesh Patel 39:32 Yeah.

George Church 39:32 ... you just dial it up using Watson-Crick rules. If you wanna make a three-dimensional structure, you know, it's actually the one p- kind of the one thing where morphology is dictated by fairly simple rules. It's not how developmental biology works and we still need to figure out how that works, but DNA origami, DNA nanostructures really work. But doing it for proteins was really, really hard until, I don't know, maybe eight years ago, something like that.

Dwarkesh Patel 40:00 Mm-hmm.

George Church 40:00 And, and I think we're f- just now getting used to it. The use of chips for making DNA, um, I mean you said that DNA synthesis come down a thousand-fold. Well, it depends on who you talk to. So that, when we came out with the first chip-based genes in 2004 Nature paper, uh, basically people dismissed it for about a decade. The only people that used it were, you know, collaborators and alumni. Um, and it wasn't even listed on the Moore's Law curve for, for DNA synthesis, even though it was like a thousand times cheaper. It was just, like, ignored and, uh, and now we have claims of 10 to the 17th genes, okay? That you can make-

Dwarkesh Patel 40:39 Yeah.

George Church 40:39 ... you can make libraries 10 to the 17th that aren't randomized in any, in any real s- in the usual sense where you just like do error-prone PCR or spiked-in nucleotides. Uh, now 10 to the 17th, that's a lot bigger than a thousand-fold, you know? If, if, you know, if it turns out to be practical, yeah.

Dwarkesh Patel 40:57 Publicly available data is running out, so major AI labs partner with Scale to push the boundaries of what's possible. Through Scale's Data Foundry, major labs get access to high quality data to fuel post-training, including advanced reasoning capabilities. Scale's research team, SCIEL, is creating the foundations for integrating advanced AI into society through practical AI safety frameworks and public leaderboards around safety and alignment. Their latest leaderboards include Humanity's Last Exam, Enigma Eval, MultiChallenge, and VISTA, which test a range of capabilities from expert level reasoning, to multimodal puzzle solving, to performance on multi-turn conversations. Scale also just released Scale Evaluation, which helps diagnose model limitations. Leading frontier model developers rely on Scale Evaluation to improve the reasoning capabilities of their best models. If you're an AI researcher or engineer and you want to learn more about how Scale's Data Foundry and Research Lab can help you go beyond the current frontier of capabilities, go to scale.com/dwarkesh. Okay, so speaking of protein design, another thing you could have thought in the '90s is... I mean, people were writing about nanotechnology, Eric Drexler and so forth, and now we have... We can go from a, a function that we want this, uh, tiny molecular machine to do back to the sequence that can give it that function. Why, why isn't this resulting in some nanotech revolution? Or will it eventually? Like, why didn't AlphaFold cause that?

George Church 42:28 I think part of it is that the nanotechnology as original... You know, the, kind of the, the source of the inspiration, Eric Drexler, he wanted to reinvent biology in a certain sense, but it already existed.

Dwarkesh Patel 42:41 Mm-hmm.

George Church 42:42 And so, so you don't need to design a diamond replicator because you already have a DNA replicator. Um-

Dwarkesh Patel 42:49 Right.

George Church 42:50 And so the question of what, what was missing, what was motivating this reinvention of biology, it was materials. So, biology's...... not that great with, you know, materials that are, say, superconductors or conductors, period-

Dwarkesh Patel 43:08 Yeah.

George Church 43:08 ... semiconductors and, and light speed. But it's getting there.

Dwarkesh Patel 43:13 Mm-hmm.

George Church 43:13 I mean, it, you know, it, it, rather than going the, the route of having everything has to be based on first principle, uh-

Dwarkesh Patel 43:23 Right.

George Church 43:23 ... nanostructures, you can meet in the middle where biology can build things. Now, now of course, when you go down to, you know, liquid nitrogen and colder temperatures, biology, as we currently know it, stops functioning. Now, it's not to say that you, you can't have things moving in liquid nitrogen. You can, but that hasn't been explored and doesn't really need to be, 'cause if biology can build things that can operate at low temperature... Or maybe biology now, because you can make these big libraries of biology, you know, maybe, maybe 10 to the 17th in vitro, uh, and you can, and you can flip through them quickly and you can barcode them and you can... This is something you, that's never been done in electronics. I'm not saying you can't do it in electronics, but, you know, you haven't made, you know, a billion different kinds of electronic materials-

Dwarkesh Patel 44:16 Mm-hmm.

George Church 44:17 ... re- just-

Dwarkesh Patel 44:17 Right.

George Church 44:17 ... you know, in an afternoon, barcode them all and see who wins, right? But we do that all the time in biology now, at least since 2004, we have. And so, so I think that's a, that's an opportunity is that, is that we use those libraries to make much superior materials and we might even finally get a room temperature superconductor that way.

Dwarkesh Patel 44:40 From bio?

George Church 44:41 It's possible. I mean, from lib- from libraries. We'll call it chemical/biochemical/

Dwarkesh Patel 44:46 Mm-hmm.

George Church 44:47 ... you know, exotic material libraries. But the point is they're libraries, so they're essentially based in some sense on polymers, even though pieces of them don't necessarily have to be polymers.

Dwarkesh Patel 44:57 Do you have a prediction by when we'll see this material science revolution? What- what- what is basically standing between, because we've got, we've got AlphaFold right now, right?

George Church 45:07 Right.

Dwarkesh Patel 45:07 So what, what is the thing that we need? Do we need more data?

George Church 45:10 Well, I- AlphaFold's very nice, but, but it, it's, it's only part of it. So, so there, there are large language models that, that are different from AlphaFold. So, so give an example, AlphaFold, last time I checked anyway, at least it's all changed, if you substitute a- an alanine for a serine in a serine protease, it will have exactly the right fold. It will be precise to, you know, you know, fraction of angstrom overall average. But it won't function. It just won't function.

Dwarkesh Patel 45:44 Mm-hmm.

George Church 45:44 And that's where you, you, you need either extraordinary precision or just knowledge of what happens evolutionarily or happens in experiments to say that, "No, an alanine won't work," okay?

Dwarkesh Patel 45:56 Okay.

George Church 45:56 And so I think there, there's all kinds of combinations of AI tools that can give you deeper insight into that.

Dwarkesh Patel 46:02 Mm-hmm. If AlphaFold predicting the structure doesn't tell you whether the thing will actually function, then what is needed before I can say, "I want a nano machine that does X thing or I want a material that does Y thing."

George Church 46:15 Yeah. Yeah.

Dwarkesh Patel 46:15 And I can just like get that?

George Church 46:17 I mean, I think the way that it's working now, which, which will get us a long way, won't get us the whole way, is we make, is we have something that kind of works and we make libraries inspired by that.

Dwarkesh Patel 46:29 Mm-hmm.

George Church 46:29 Make, make variations on it. And then whichever those variations work, we make variations on that and we can just keep going.

Dwarkesh Patel 46:36 Right.

George Church 46:36 It's kind of like the way evolution worked, except now we can do it at incredibly high speeds. In, in principle, you know, what might, you know, evolution might incorporate a few base pair changes in a million years, now we can make, you know, billions of changes in an afternoon. And so, and it's all guided in such a way that you get rid of the wastefulness of having a bunch of neutral mutations and a bunch of lethal mutations. You can have things that are quasi-neutral and, but likely to be game-changing, have more of a focus on those. Another thing that's been missing, and none of the AI protein design tools that I know of are particularly good at it yet, but we're trying to, we're, as we speak, trying to improve this, is non-standard amino acids. Because a lot of these tools depend on having libraries of 3D structures, which use 20 amino acids and li- large language models that where you line up all the sequences of 20 amino acids and we have very little experience with extra ones, but, but I think there's a revolution going on in generating non-standard amino acids that can, where the amino acids can either have as part covalent part of them or as easily liganded all the, the entire periodic table, stable elements. And, and that will, you know, each of those will have to blend in and train our models on, but as soon as that comes in, then, then we're gonna have a whole series of new materials very quickly.

Dwarkesh Patel 48:18 Mm-hmm.

George Church 48:18 And ultimately you can think of the, the determination of the functionality of your library is a kind of computer, right? So you use AI to make, to design the library optimally so you avoid things that are really neutral and really seriously damaged, but then the stuff in the middle, you actually play it out, not in a simulation, but in real-

Dwarkesh Patel 48:45 Right.

George Church 48:46 ... life. But it's so inexpensive and it's so fast and it's so exact, I mean, it's 100% precision because you're not simulating, right?

Dwarkesh Patel 48:55 Yeah.

George Church 48:55 You're not making a- assumptions, you know, you're not going from quantum electrodynamics, which is an assumption, to quantum mechanics, which is assumption, to, you know, molecular mechanics, which are full of assumptions. You, you, you're really doing the real thing. And so you're doing a kind of natural computing, and then you can take that data-

Dwarkesh Patel 49:11 Mm-hmm.

George Church 49:11 ... and harvest it in various ways very efficiently, pump it back into the-... convince, you know, the more conventional AI and do another round of it.

Dwarkesh Patel 49:19 Yeah. It seems like if I listen to these words-

George Church 49:22 Yeah.

Dwarkesh Patel 49:22 ... it seems like I should be expecting the world to physically look a lot different, but then why are we only getting, like, a couple more drugs by 2040?

George Church 49:29 Well, I- I didn't mean to stop there. I mean-

Dwarkesh Patel 49:30 (laughs)

George Church 49:31 I- I knew the conversation would continue, yeah.

Dwarkesh Patel 49:33 Right. (laughs)

George Church 49:34 I- I- I'm not pinning down a particular year either, but, um, I- I think this is poised to go pretty quickly. There are very few practitioners is the thing that will stop it for a while.

Dwarkesh Patel 49:45 Yeah.

George Church 49:46 Since ma- materials will actually go, should go faster though, because they don't require quite as much, uh, regulatory-

Dwarkesh Patel 49:53 Yeah.

George Church 49:53 ... uh, approval. Um, so it's- it's, yeah, it's g- you know, i- i- it's one of these things where when you get the right idea, it's not hard to recruit people. I mean, s- for example, when Feng Zhang and my labs, uh, brought out CRISPR, we just got 10,000 requests in the next two months, uh, for people that wanted to duplicate-

Dwarkesh Patel 50:13 Yeah.

George Church 50:14 ... the system. And so that's- that's what I hope will happen with the non-standard amino acids and the, and the using AI for protein-

Dwarkesh Patel 50:19 Mm-hmm.

George Church 50:19 ... design and- and making new materials. Hopefully that will recruit tens of thousands of people overnight, yeah.

Dwarkesh Patel 50:25 Are-

Dwarkesh Patel 50:26 are you more excited about AI which thinks in protein space or the capsid space or, like, just, you know, it's like predicting some biological or DNA sequences, um, or are you more optimistic about just LMs trained on language wh- which can, like, write in English and tell you, "Here's the experiment you should run" in English? Which of those two approaches, or is there some combination that when you think about-

George Church 50:50 I, uh-

Dwarkesh Patel 50:50 ... AI and bio is more promising?

George Church 50:52 I- I'm much more excited about, uh, scientific AI than I am about language AI.

Dwarkesh Patel 50:59 Mm-hmm.

George Church 50:59 I think languages were in pretty good shape already, and what worries me is that to get to the next level of language, um, requires AGI or ASI.

Dwarkesh Patel 51:13 Yeah.

George Church 51:13 So A- a- and, you know, artificial super intelligence.

Dwarkesh Patel 51:16 Right.

George Church 51:16 And that's very dangerous. I don't think we have quite figured out how to... and there's a lot of safety organizations and a lot of safety rules and so forth. And I think what typically happens when there's an intense competition is those safety rules get-

Dwarkesh Patel 51:30 Yeah.

George Church 51:30 ... undermined and pushed aside. But even if they weren't, I do- just don't think we, I don't think we understand our own ethics well enough to educate-

Dwarkesh Patel 51:39 Yeah.

George Church 51:39 ... a completely foreign, uh, type of intelligence. I mean, we barely know how to pass it on to the next generation of humans. Um, so I think we need time to sort that out. And there's no rush. This is a completely artificial emergency. This is not like COVID-19 where we actually, millions of people were dying if we delayed of the science. This is something where if- if there ever is a crisis, it's 'cause we created it. It's not because we're trying to solve it.

Dwarkesh Patel 52:07 Yeah.

George Church 52:08 Right? And so I think we need to go very slowly on AGI and ASI and- and double down on slightly narrower scientific, uh, goals. Um, that's... and- and even now we need to be very cautious about, we need to have kind of an international consensus on what constitutes safe AI.

Dwarkesh Patel 52:30 I- I suppose we d- did build safe super intelligence. Um, how much would that, uh, speed up bio progress? Is it just, like, there's a- there's a million George Churches on data, in data centers-

George Church 52:41 Yeah.

Dwarkesh Patel 52:41 ... just, like, thinking all the time. Is it- is it a 10X speed up?

George Church 52:43 I think it would, I think it would slow it down.

Dwarkesh Patel 52:45 (laughs)

George Church 52:45 I think, I think it would eliminate it because, like-

Dwarkesh Patel 52:47 (laughs)

George Church 52:48 ... it w- the first thing it would conclude is biology is not in- not relevant to me 'cause I'm not made out of biology.

Dwarkesh Patel 52:52 Ar- I mean, suppose you get the, get- get them to care about it. They're just, like, copies of you.

George Church 52:56 Well, I don't know. That sounds a little bit

Narrator 52:56 (laughs)

Dwarkesh Patel 52:56 There's a million copies of you in a data center. What- what- how- like, how much faster is the bio progress?

George Church 52:59 (sighs)

Dwarkesh Patel 53:00 But they can't, like, run experiments directly. They're just in data centers. They can- they can just say stuff and think stuff.

George Church 53:05 I don't think we have anything close to the assurance that we need that that would be safe. But let's put safety aside for a moment.

Dwarkesh Patel 53:12 Yeah. So it doesn't work?

George Church 53:12 Uh, I think it's hard to, I think it's hard to cal- it's not only hard to calculate the- the bads, it's hard to calculate the goods.

Dwarkesh Patel 53:18 Right.

George Church 53:18 So I think it could completely, be a complete changer. But on the other hand, um, you know, it's like if we said, um, you know, we could get instantaneous transport all over the Earth, right? Well, we could say, "Yes, that could be a game changer, but do we really need it," right? Is that really important? Maybe it'd be more interesting to just have, uh, Zoom calls and- that are better, you know, or- or just learn how to get everything we want in our kitchen and we don't need to travel anymore.

Dwarkesh Patel 53:51 Yeah.

George Church 53:51 Right? You know, uh, so, you know, be careful what you ask for, right? You know, because you could tip our priorities towards something that we really don't care about, uh, that, or we shouldn't care about or- or might wish we didn't care about, right?

Dwarkesh Patel 54:09 But- but, uh, I'm curious what, um... you've still gotta run the experiments. You still need-

George Church 54:12 Yeah.

Dwarkesh Patel 54:13 ... uh, these other things. So does that bottleneck the impact of the millionth copy of you or do you still get some speed up? How- how- how much faster can biology basically go if they're just, like, more smart people thinking? Which is a sort of proxy for what AI might do.

George Church 54:27 Yeah. These are great questions and I'm not sure... I don't wanna misrepresent that I know the answers, but- but, you know, it's like, it's like the question of, you know, if you have nine women, can you do pregnancy in one month?

Dwarkesh Patel 54:38 (laughs)

George Church 54:38 No, not- not- not-

Dwarkesh Patel 54:40 Right.

George Church 54:40 ... at present.

Dwarkesh Patel 54:41 (laughs)

George Church 54:41 Uh...

Dwarkesh Patel 54:41 But you're working on that, right? (laughs)

George Church 54:42 No, no, no, no.

Dwarkesh Patel 54:44 I'm joking.

George Church 54:44 So, I mean, just-

Dwarkesh Patel 54:45 (laughs)

George Church 54:45 But the same thing is there may be certain things that, um, doesn't take a lot of people. We just don't know. We just don't, we don't have that much experience with having, uh, you know, thousands of Einstein-type levels of creativity and intelligence simultaneously in a generation. Um, and in fact, it's probable that we're all c- capable of being a bit more efficient if we don't have, you know, the distractions of mental illness, uh-

Dwarkesh Patel 55:20 Mm-hmm.

George Church 55:20 ... of, you know, t-... taking care of other people. Now, taking care of other people may be a very good thing. You know, it may be that if we, if we have no one to take care of, there'll be something bad that happens to us, uh, socially. Um, so, uh, these things are very complicated and hard to predict. I think, uh, right now, I think the, the, the baby step, or actually the pretty big baby step, is to eliminate, uh, diseases, or at least make it possible for people to eliminate their own diseases as, as, as they see fit.

Dwarkesh Patel 55:54 Mm-hmm. You worked on brain organoids and h- uh, brain connectome and so forth. That work, how has it shifted your view on fundamentally how complex intelligence is? I- i- in the sense of like, how, how, you know, are you like more bullish on AI because I, I realize that, uh, organoids are not that complicated.

George Church 56:12 Right.

Dwarkesh Patel 56:12 Or it's like very little information is required to describe how to grow it.

George Church 56:15 Yeah.

Dwarkesh Patel 56:15 Um, or are you like, no, this is actually much more gnarly than I realized?

George Church 56:19 I think I always felt it was very gnarly. Uh, and I, I also felt that it was something that we could engineer. Um, certainly, we have made a lot of progress in, um, at, at the broken end of the spectrum where, where, where the, um, brain is severely challenged relative to average. Um, there are, you know, there's thousands of, a, a huge fraction of genetic diseases that are, uh, that have one of their consequences being that the child is ex- ex- developmentally delayed to such an extent that it's lethal or, um, you know, or, or a lifetime deficit. Um, and we know how to, um, we know the genes involved and we know how to, um, do genetic counseling and in some cases, gene, gene therapy and other therapies to deal with it. Um, at the other end, you know, we have, uh, reduction of cognitive decline by cognitive enhancement-

Dwarkesh Patel 57:32 Right.

George Church 57:32 ... which is l- showing some promise. But again, that's kind of like this early stage, uh, severe impe- impediment, uh, cognition has a, a late stage component. But what about, uh, how much information does it take to encode a brain? I'm not sure there's that much less genome is required that if you just wanted to make a brain, uh, 'cause the brain is totally entangled-

Dwarkesh Patel 57:59 Mm-hmm.

George Church 57:59 ... with the body. Um, you know, you need the, you have, uh, 10 to the 11th neurons, 10 to the 14th synapses. If you wanted to reproduce a particular brain, let's say, it might be, um... it's, it's speculative as to whether it would be easier to do that by making a copy of it in, in, in silico, uh, in some kind of, uh, inorganic matrix, uh, or making a copy of it. Both of those are gonna be hard. I would say that if you wanted to make a copy of a, you know, complicated book, it would be easier to f- take photographs of each of the pages than to completely translate it into another language, trying to get all the nuances and the poetry and so forth-

Dwarkesh Patel 58:40 Right.

George Church 58:41 ... uh, if, if your goal is just to replicate it. And I think the same thing might be true of a brain. But replicating a brain probably involves a lot more information than synthesizing it. So I mean, we've alr- just to define this 10 to the 14 synapses is gonna take a, a lot more bites than the f- than the genome, which is billions rather than, um, 10 to the 14th. But you might, there might be reasons that you wanna replicate a particular brain configuration rather than just, uh, make another, um, animal, uh, that is, um, you know, starts from scratch as a, as a, as an infant.

Dwarkesh Patel 59:22 Given how little I knew about biology, my prep for this episode basically looked like one minute of trying to read some paper and then chatting with an LLM like Gemini 30 minutes afterwards and asking it to explain a concept to me using Socratic tutoring. And the fact that this model has enough theory of mind to understand what conceptual holes a student is likely to have and ask the exact right questions in the exact right order to clear up these misunderstandings has honestly been one of the most feel the AGI moments that I've ever experienced. This is probably the single biggest change in my research process, honestly since I started the podcast. For this episode, I think I probably spend on the order of 70% of my prep time talking with LLMs rather than reading source material directly, because it was just more useful to do it that way. And given how much time I spend with Gemini in prep for these episodes, improvements in style and structure go a really long way towards making the experience more useful for me. That's why I'm really excited about the newly updated Gemini 2.5 Pro, which you can access in AI Studio at ai.dev. All right, back to George.

Dwarkesh Patel 1:00:35 Going back to the engineering stuff, often people will argue that, look, you, you have this existence proof that E, E. coli can multiply every, or d- duplicate every 30 minutes. Insects can duplicate really fast as well. Um, but then with our ability to manufacture stuff with human engineering, you know, we can do things that no an- um, nothing in biology can do, like radio communication or-

George Church 1:00:57 Right.

Dwarkesh Patel 1:00:57 ... uh, fission power or jet engines, right? So, um, like how plausible to you is the idea that we could have biobots, which are, you know, like can duplicate at the speed of insects and there could be trillions of them running around-

George Church 1:01:09 Yeah.

Dwarkesh Patel 1:01:09 ... but they, they also can have access to jet engines and radio communication and so forth? I- i- are those two things compatible?

George Church 1:01:16 Well, I mean, uh, certain things seem incompatible, uh, like the temperatures of a fission reactor-... isn't obviously compatible. But, but in, uh, but the, the possibility that once we, uh, that, that a biological system can make other things, you know, for example, it, it can, uh, you know, it can make a nest. A bird can make a nest, okay? And you consider the whole nest as part of the, the replication cycle of the bird. Um, so you can say biological thing that replicates at 30-minute doubling time could make a nuclear reactor, as that would be its nest.

Dwarkesh Patel 1:01:56 Mm-hmm.

George Church 1:01:56 Um, but you need to e- you know, expand its range of, of materials. Uh, we... in a, in a certain sense, we do this already. Humans are (laughs) a biological thing that replicates not in 30 minutes, but in, you know, 20 years or, or less. Um, and is that fundamentally limiting us? You know, probably is, uh, but, but yeah, it's, it's amazing to think about. What if you could take, you know, a cornfield or a nuclear reactor and suddenly, 30 minutes later, you got two of them, right? And then four of them and eight of them.

Dwarkesh Patel 1:02:32 Right.

George Church 1:02:32 Yeah. I mean, that's, that's quite an interesting, um, concept. But, uh, I mean, I, I think what we should start with... I teach a course called How to Grow Almost Anything.

Dwarkesh Patel 1:02:42 Mm.

George Church 1:02:42 That's, that, uh... and I work with Neil Gershenfeld who... at MIT, who had a... has a course called How to Make Almost Anything. And we're trying to meet in the middle where we can in, you know, his, um, uh, you know, mechanical, electrical engineering will meet with our biological. And in fact, neither of us can make or, or grow almost everything because there's all kinds of little gaps and things that are very hard to make in a, in a small lab, because there are things all over the world that depend on, you know, multibillion-dollar fabs to, to make things. Um, but yeah, we're eating away at it. I think that we might eventually be... you know, maybe a smaller baby step than making a nuclear reactor is making a, a phone. You, you know, you said radio communication.

Dwarkesh Patel 1:03:27 Right.

George Church 1:03:28 We should make a bio-lab. It should be a small challenge goal for the synthetic biology community, maybe iGEM or something, make, uh, you know, bacteria, make, uh, a radio. Now, actually, Joe Davis is a, is a, uh, uh, uh, artist that's been affiliated with my lab, and before that, Alex Rich's lab, and, uh, and he did make a bacterial radio, but it was kind of more on the art end than on the science end. But, uh, I think that would be a good, um, good goal.

Dwarkesh Patel 1:03:58 What would it take to do wh- whole genome engineering to such a level that for even a phenotype which doesn't exist in the existing pool of human variation, you could manifest it because your understanding is so high that you can... like for example, if I wanted wings. (laughs)

George Church 1:04:17 Yeah. Right.

Dwarkesh Patel 1:04:17 Is the bottleneck our understanding? Is, is the bottleneck our ability to make that many changes to my, to my genome?

George Church 1:04:24 So part of this has to do with just learning the rules of developmental biology, like I said.

Dwarkesh Patel 1:04:28 Mm-hmm.

George Church 1:04:28 We can determine morphology at sort of the molecular level now.

Dwarkesh Patel 1:04:32 Yeah.

George Church 1:04:32 Proteins and nucleic acids. Determining a develop-... at the cellular, multicellular level, there's a lot more things you can do a l- and a lot faster, um, but we don't know the language yet. So we gotta... that, um... I think we're on the cusp of getting the tools to do that, like the transcription factor-

Dwarkesh Patel 1:04:50 Mm-hmm.

George Church 1:04:50 ... I was talking about earlier, you know, harnessing, uh, migration, dif-... you know, um, gradients of, of factor... you know, diffusion factors, um, uh, you know, chemotaxis and so forth.

Dwarkesh Patel 1:05:04 Mm-hmm.

George Church 1:05:04 So I, I th-... that's one thing we need, but w- there's, there's a bunch of things we

George Church 1:05:09 need, really.

Dwarkesh Patel 1:05:10 What discovery in biology, so n- not in astronomy or some other field, in biology, would, um, make you convinced that life on Earth is the only life in the galaxy? And c- conversely, what, what might convince you that, no, it must have arisen independently thousands of times in this galaxy?

George Church 1:05:28 Oh, I see what you're getting at. Right. I mean, so astronomy might be we, we would detect-

Dwarkesh Patel 1:05:33 That's right.

George Church 1:05:33 ... uh, you know, radio signals-

Dwarkesh Patel 1:05:36 Yeah.

George Church 1:05:37 ... or light signals that-

Dwarkesh Patel 1:05:38 Right.

George Church 1:05:38 ... um, but biology, what you would... the, the, the kind of evidence would be that you show in a laboratory using prebiotic, uh, conditions-

Dwarkesh Patel 1:05:50 Yeah.

George Church 1:05:50 ... a really simple way to get life.

Dwarkesh Patel 1:05:52 Yeah.

George Church 1:05:53 Right? Or, I mean, it's a harder proof to prove that given... because we don't know what-

Dwarkesh Patel 1:05:59 Right.

George Church 1:05:59 ... all the possible prebiotic, biotic conditions. Um, uh, and probably the number of it was vast.

Dwarkesh Patel 1:06:07 Yeah.

George Church 1:06:07 I mean, you have 10 to the 20th liters of water and, you know, at, at various different salinities-

Dwarkesh Patel 1:06:14 Right.

George Church 1:06:14 ... and drying up on the ocean and the sun and the lightning and all this stuff. But yes, if you... I think if you showed kind of reconstructed in the lab a very simple pathway from inorganics-

Dwarkesh Patel 1:06:26 Mm-hmm.

George Church 1:06:27 ... cyanide derivatives and reduced, uh, compounds all the way up to, you know, some cellular replicating structure, I think that might, um, that might lead us to believe that at least life exists.

Dwarkesh Patel 1:06:44 Yeah.

George Church 1:06:44 Now, that- there are other parts of the Drake equation that might kick in-

Dwarkesh Patel 1:06:48 Right.

George Church 1:06:48 ... which is maybe it's hard to get intelligent life-

Dwarkesh Patel 1:06:50 Right.

George Church 1:06:51 ... because intelligence isn't necessarily in your best interest. And if you get intelligent life, it's hard to maintain that without societal collapse or without robotics taking over and then killing themselves. Right? And that's hard to do experiments. But I think, to your question, I think an experiment that showed, you know, maybe multiple different ways of getting to a living system from, from nonliving systems spontaneously would be interesting. Uh, again, I'm not sure it would... it'd be very hard to prove the negative.

Dwarkesh Patel 1:07:22 Mm-hmm. Eh, eh, so I'm curious, between intelligent life and some sort of primordial RNA thing-

George Church 1:07:28 Yeah.

Dwarkesh Patel 1:07:30 What is the step at which, if there is any, where you say there's a less than 50% chance something like at this level exists elsewhere in the Milky Way?

George Church 1:07:40 Yeah, I think these are very challenging problems. Uh, I'm not, (laughs) I'm not even sure we would be able to say within five orders of magnitude, much less 50%. Um, but, um, you know, I... (smacks lips) I think it's more likely to come from exploration than it is gonna be from simulation. Uh, if, you know... The sad truth is that almost n- none of the missions that we've sent out- outside of Earth have actually looked for life. They've had components that could have looked for life, um, but they... a sad number of those, and n- not enough components that could look for life, and the ones that could look for life, not really looking for it. And when we get positive results, we dismiss them, um, as happened with the Pioneer. And so I think if we just start looking at the, you know, the- the geysers that are coming out of various moons of Jupiter and Saturn, um, there's so much water, there's 50 times more water.

Dwarkesh Patel 1:08:49 Mm-hmm.

George Church 1:08:50 A- liquid water, not- not frozen, more liquid water in our solar system than on Earth. Doesn't that seem likely that-

Dwarkesh Patel 1:08:58 Mm.

George Church 1:08:58 ... you know, some of that would have, uh, been a good breeding ground? Um, but it could be that we need sunny shores, you know, where you have a lot of dry land and right next to water. Maybe th- these are just giant oceans that are surrounded by ice, and maybe that's not a bad idea. But any case, we- we need to look at those fountains to, uh, see what's popping up. Uh, that's a, that's a high priority. And the same thing goes, you know, for, you know, there's a lot of, um, water on Mars that- that- that's maybe even more accessible. But until we've exhausted those, I think those are probably the easiest. They're hard. You're- you're still talking about multi-billion dollar, uh, experiments, but I think they're a little more convincing. A- and again, it'll be hard to prove the negative. If we, if we find that it's negative on every, every, everything in the solar system, (clicks tongue) you know, there's so much more diversity out there that could- that could have done

George Church 1:09:57 it.

Dwarkesh Patel 1:09:57 Mm.

George Church 1:09:57 Uh...

Dwarkesh Patel 1:09:58 If in a thousand years we're still using DNA and RNA and proteins for top-end manufacturing, the- the frontiers of engineering, how surprised would you be? Would- would you think, like, "Oh, that makes sense. Evolution designed these systems for billions of years?" Or would you think, like, "Oh, it's surprising that these ended up being the systems, the- whatever evolution found just happened to be the best way to manufacture or-"

George Church 1:10:18 Yeah.

Dwarkesh Patel 1:10:18 "... to store information or..."

George Church 1:10:20 Yeah. I don't think I'd be surprised either way.

Dwarkesh Patel 1:10:23 Mm-hmm.

George Church 1:10:23 Um, I can imagine it going either way. I- I can imagine making truly amazing materials using proteins as the catalysts, or maybe in some cases as a scaffold as well as catalysts. Uh, I think one- one thing that's probably already happening, so we don't have to go a thousand years out, is the number of amino acids is going up. It's going up radically from 20. I think pretty soon we'll have a system where we can have 33 and 34 new non-standard amino acids being used simultaneously with all the standard ones in a E. coli cell.

Dwarkesh Patel 1:10:57 Mm.

George Church 1:10:57 So 34 plus 20 is a lot bigger than 20. I don't think we necessarily need more- m- more than four nucleic acid components.

Dwarkesh Patel 1:11:10 Mm-hmm.

George Church 1:11:10 Uh, I mean, you... Certainly there are plenty of modified ones. There, there's a- a bunch of- of, uh, alternative base pairs, some of which don't even involve hydrogen bonds. So we could- we could have more. But I think the main thing is this information storage and whether it's bits, you know, it's, you know, digital, uh, binary is just zeros and ones. That works pretty well for-

Dwarkesh Patel 1:11:33 Right.

George Church 1:11:34 ... 99% of- of what we do electronically. So having four is better than two maybe.

Dwarkesh Patel 1:11:39 Right.

George Church 1:11:40 But do we really need six? You know, I don't know. So- so yeah, I wouldn't be surprised if we had... Uh, another possibility is that we change the backbone of DNA. So maybe keep the ACGT, but, uh, make it out of peptides now.

Dwarkesh Patel 1:11:55 Mm.

George Church 1:11:56 Um, a little bit smaller, uh, a little bit more compatible. I don't know. Um, or- or maybe that'll just be a- just a slight, you know, it could be part of the new amino acid collection.

Dwarkesh Patel 1:12:10 Mm-hmm.

George Church 1:12:11 Um, and- and there'll be more. I mean, th- these are just things that- that my primitive 21st century brain is coming up with. A thousand years from now, it'll- it'll be a whole new millennium.

Dwarkesh Patel 1:12:21 So i- it makes sense why evolution wouldn't have discovered, like, radio technology, right? But things like m- m- uh, more than, um, 20 amino acids or these different bases so that you can have store more than two bits per base pair, or, for example, the cod- uh, codon remapping scheme-

George Church 1:12:38 Mm-hmm.

Dwarkesh Patel 1:12:38 ... this redundancy, which it seems like based on your work, you can... there- there was this extra information you could have used for other things.

George Church 1:12:45 Yeah.

Dwarkesh Patel 1:12:45 So is there some explanation for why four billion years of evolution didn't already give living organisms these capabilities?

George Church 1:12:53 I think that, uh, evolution has a tendency to go with what works and the investment in making a whole new base pair, um, would have been high. Um, and- and we haven't even articulated what- what that, uh, what the return on investment would be, um, what do you get from that. We- we have made systems, uh, f- like Floyd-Royersberg and others, that where you have rep- replication and transcription and translation with this, with a sec- with a new base pair, but it hasn't been clearly articulated what that gets you, um, uh, even in technological society. So in technology, you can- you can jump to things that... where all the intermediates aren't-... you know, incrementally useful. With evolution is, as far as we know, generally limited to you have to justify-

Dwarkesh Patel 1:13:45 Mm-hmm.

George Church 1:13:46 ... uh, every change uh, uh, a- it's like some bureaucracies is-

Dwarkesh Patel 1:13:49 (laughs)

George Church 1:13:50 ... well, if you're gonna, you know, you're gonna put this sidewalk in, you know, you're gonna have to justify that before you can build the city.

Dwarkesh Patel 1:13:55 Mm. What is one, um... So we- we've talked about many th- many different technologies you worked on or are working on right now-

George Church 1:14:02 Mm-hmm.

Dwarkesh Patel 1:14:02 ... um, uh, from gene editing to de-extinction, to age reversal. What is, um, what is an under-hyped, uh, technology in your research portfolio which you think more people should be talking about but g- gets glossed over?

George Church 1:14:16 It's, it's hard to say because soon as you say it, it becomes hyped. (laughs)

Dwarkesh Patel 1:14:21 (laughs)

George Church 1:14:21 And so, if I've ever been asked this question before, it's too late. But, um, you know, I would say one thing I think is very ripe and it's very well understood in a certain sense, but is nevertheless ignored, it's kind of like the, the, the previous example I would have chosen was when making genes out of arrays. Arrays are typically used for analytic, you know, um, quantitating RNAs or something like that, so the original Affymetrix type of arrays. Um, but we turned them into gene arrays. And, and just people weren't using it. Uh, it was in nature. It was, uh, it was hidden in plain sight. Um, but anyway, it was, it was somehow under-hyped. Uh, what I would say is genetic counseling is under-hyped. It is, um, clearly competitive with gene therapy in a certain sense. I mean, clearly not for people that are already born but for people in the, in the future, not, not even distant future, in the next couple of years, um, we've got a chance of diagnosing them or diagnosing the potential parents and, and dodging, uh... And it's, and this has been in practice since 1985 in Doria-Sharim.

Dwarkesh Patel 1:15:36 Mm.

George Church 1:15:37 Perfectly reasonable community response to it, um, eliminated or greatly reduced all sorts of very, very serious inherited diseases. Um, it's sometimes dis- you know, depending on how it's presented, it's dismissed as eugenics. I think it's rarely th- have I heard Doria-Sharim described that way, and rightly so. What they're doing is standard medicine, you know. Whether you, you know, cure these kids as soon as they are newborns or whether you, uh, counsel the parents so the same disease is, is missing, um... The, the, the problem with eugenics was that it was forced, the government forced it on people. It wasn't that it enabled people to make a choice.

Dwarkesh Patel 1:16:23 Right.

George Church 1:16:23 It's that it removed the choice-

Dwarkesh Patel 1:16:25 Right.

George Church 1:16:25 ... from the people. That was what was wrong. And that's the confusion someone said... But I don't think that's the explanation for why this is under-hyped. I think it's people when they're dating, they're not thinking about reproduction necessarily. And when they're thinking about reproduction, they're not, you know, they're, they're not necessarily thinking about serious genetic diseases because they're rare. I think it's our, uh, difficulty with dealing with rare things.

Dwarkesh Patel 1:16:51 Mm.

George Church 1:16:51 It's like there was great resistance to seat belts, um, because less than 1% of people died in automobile accidents or even got hurt. Um, great resistance to stopping smoking. Really, uh, it's hard even for us to imagine how great the resistance was for seat belts and smoking. Um, and, but eventually, we got, we got over it. I think this is a similar thing, which is that only 3% of children are severely affected by genetic diseases. And they feel like, "Well, I'm not that unlu- you know, I'm, I'm, I'm in the 97%" Right? You know, 97% if those were your odds of winning, you know-

Dwarkesh Patel 1:17:30 Right.

George Church 1:17:30 ... at the horse races or at, or at the casino, you'd take 'em. Yeah, 90%, 70% of winning, good, you know. But with, with, you, you know, when a children's, uh, when a child's future is at risk, I think that's, uh, it's not the right-

Dwarkesh Patel 1:17:44 Yeah.

George Church 1:17:45 ... solution. And, uh, uh, the other thing is I think it's, has to do with the trolley problem. It's like, if you don't influence it, it's not your fault. But actually, everything is your fault, you know. Not doing something is a decision, right?

Dwarkesh Patel 1:17:57 Right.

George Church 1:17:58 And so I, I think it's like, "If I just don't do anything and they come out damaged, well, it's not my fault." But-

Dwarkesh Patel 1:18:04 Yeah.

George Church 1:18:04 ... it is, yeah.

Dwarkesh Patel 1:18:05 Uh, David, David Reich was talking about how in India especially because of the long running history of caste and-

George Church 1:18:12 Yeah.

Dwarkesh Patel 1:18:12 ... um, endogamous coupling that there have been these small cell populations that have high amounts of recessive diseases. And so-

George Church 1:18:18 Yeah.

Dwarkesh Patel 1:18:18 ... like, there it's especially, uh, valuable intervention.

George Church 1:18:21 I, I think that's a... Yeah, I, I know what you're saying and, and what David is saying, but I think it's a dangerous dichotomy.

Dwarkesh Patel 1:18:27 Mm-hmm.

George Church 1:18:27 Uh, uh, uh, you know, they'll say that there's certain... There are lots of, of h- not just India, you know, uh, all over the world.

Dwarkesh Patel 1:18:34 Right.

George Church 1:18:34 And, and in fact, um, and in fact, we all went through a bottleneck. Um, no, but that changes the, the, the rate from, say, 3% to 6%.

Dwarkesh Patel 1:18:44 Mm-hmm.

George Church 1:18:44 But the point is 3% is still unacceptable.

Dwarkesh Patel 1:18:47 That's right. That's right. Yeah, yeah.

George Church 1:18:47 I mean, it's just, it's just a, a tragic loss not only of the human life directly affected, but the whole family-

Dwarkesh Patel 1:18:55 That's right.

George Church 1:18:55 ... uh, is, you know, very often one or both parents have to quit their job and spend full time like caregiving and fundraising-

Dwarkesh Patel 1:19:05 Right.

George Church 1:19:05 ... because it's very, these are very expensive diseases-

Dwarkesh Patel 1:19:07 Yeah.

George Church 1:19:08 ... as well. And it's just we don't need to, uh... We need to be careful not to stigmatize as well. So when, uh, if a bunch of families get fixed, we shouldn't point a finger at the ones that are unwilling to get fixed because that's their choice, you know. Uh, but I think as, as word spreads and you see the positive outcomes, I think there will be, uh, it will be seen as, as, as one of the simplest bits of-

Dwarkesh Patel 1:19:37 Right.

George Church 1:19:37 ... medicine ever. I mean, it's, in fact, it's something-

Dwarkesh Patel 1:19:40 Same as vaccination.

George Church 1:19:41 Yeah, it's like va- it, it's very inexpensive, um, in fact, in fact, it's, it's, it's less than zero because, uh, you spend-... ho- hundred dollars per genome, and it'll probably be less soon. And you can, you get the whole thing analyzed and, um, you know, compare that to m- millions of dollars that we- it, it, you know, that will be lost, um, opportunity cost and them not being part of the workforce, the taking care of them and so forth. So, the return on investment is tremendous. It's at least a tenfold return on investment. So it's a no... it's not a... it's a no-brainer from a public health standpoint.

Dwarkesh Patel 1:20:19 Right.

George Church 1:20:19 Um, we should be able to pay for this through, you know, National Health Services in England, through insurance companies in the United States. And it turns the insurance companies from being the bad guys that they're, that they're like pro- they're snooping in on your personal life and then raising your rates to, oh, they're giving you this free information and you can do with it as you wish.

Dwarkesh Patel 1:20:37 Right.

George Church 1:20:38 And you could... if you take the advice, then you save them, uh, millions of dollars.

Dwarkesh Patel 1:20:43 Right. Do you think genetic counseling is a more important intervention or even in, even in the future, will continue to be... have a bigger impact than even gene therapy for these monogenic-

George Church 1:20:53 Oh, absolutely.

Dwarkesh Patel 1:20:53 Yeah.

George Church 1:20:54 I've actually counseled my, my gene therapy companies that, that they should be investing in very common diseases-

Dwarkesh Patel 1:21:02 Yeah.

George Church 1:21:02 ... 'cause rare diseases have this genetic-

Dwarkesh Patel 1:21:04 Makes sense.

George Church 1:21:04 ... counseling solution. Um, w- with the exception of spontaneous mutations and, and dominance, which probably are IVF clinic-type solutions, uh, rather than... but, but, uh, but the rare recessives could be handled at matchmaking-

Dwarkesh Patel 1:21:22 Mm-hmm.

George Church 1:21:22 ... and at, at every level. Um, a- anyway, I counsel my genetic therapy companies that they should, you know, invest in common diseases like age-related diseases and infectious diseases. Um, and in fact, you know, the, the COVID vaccine was formulated as a gene therapy and was, you know, the cost was in the, you know, $20 per dose range, and six billion people benefited from it, or six billion people took it and, you know, uh, and it was, uh, you know, proven over the wh- whole population. So, I think that's the more appropriate use of gene therapy. But I think for practical reasons, you know, getting FDA approval and so forth, you might, um, go for the rarer diseases and that's, that's perfectly fine. Um, but I think, uh, the cost effectiveness, uh, of the, the, the, the sweet spot for gene therapy is for age-related diseases, and the sweet spot for, um, rarer diseases is, um, genetic counseling.

Dwarkesh Patel 1:22:22 Mm-hmm.

Dwarkesh Patel 1:22:23 All right. Some, uh, some final questions to close us off. If, um, 20 years from now, i- if there's some scenario in which we all look back and say, "You know what? I think on net it was a good thing that the NSF and the NIH and all these, uh, budgets were blown off and got doged and so forth." I'm not saying you think this is likely, but suppose there ends up being a positive story told in retrospect, what might it be? Would it have to... would maybe we come up with a different funding structure, basically, like, yeah, what, what is the best case scenario, uh, if this, uh, post-war system of basic research is upended?

George Church 1:23:00 Uh, foof. Um, I, I have to preface this by, by, you know, say- when scientists explore, answer a question and explore possibilities, it doesn't mean they're advocating it.

Dwarkesh Patel 1:23:13 Mm-hmm.

George Church 1:23:13 Y- I've... in the past, people have asked me off-the-wall questions about Neanderthals, for example, and then it was described as if I was enthusiastic about it. So, not enthusiastic about NIH and NSF budgets being cut. Um, you could say, well, it make, it forces us to think more seriously about philanthropy and industrial sponsored-research. That could be a positive thing. Um, it could be that that makes us listen more carefully to what society actually needs rather than doing basic research. I'm a big proponent of basic research, but also maybe I'm more than average connecting the basic research to societal needs from the get-go. It doesn't-

Dwarkesh Patel 1:23:58 Yeah.

George Church 1:23:58 I don't think it actually interferes with basic research to think and act on societal needs at the same time. So that would be... that could be a positive. It could be that it, it creates a, uh, another nation state that now is the dominant force, you know, like China could now become-

Dwarkesh Patel 1:24:16 Right.

George Church 1:24:16 ... the next empire after-

Dwarkesh Patel 1:24:18 This is a positive story? (laughs)

George Church 1:24:19 ... U- U- US, yeah. Well, it could be for China, right?

Dwarkesh Patel 1:24:22 (laughs)

George Church 1:24:22 I mean, you didn't specify who it's a positive story for. You know, the, the US would displace Britain, which would displace, you know, Spain and Portugal.

Dwarkesh Patel 1:24:30 Right.

George Church 1:24:30 You know, it, it keeps, keeps moving. Um, fresh blood is sometimes a good thing. Again, I preface this by saying-

Dwarkesh Patel 1:24:37 (laughs)

George Church 1:24:37 ... I'm not advocating this. Um, what else could go well? You know, there's just certain things that we, the society is fairly good at doing collectively that we're not good at doing individually.

Dwarkesh Patel 1:24:51 Mm-hmm.

George Church 1:24:51 You know, building roads, schools, and science are examples of that. Doesn't mean we couldn't learn how to do that, you know, it c- you know, uh, you know, to some extent when you build a gated community, a lot of that is done with private funding. It's possible we could figure out how to build roads and schools and just about everything. It, it means we're gonna run into some kind of hyper-capitalism.

Dwarkesh Patel 1:25:18 Mm-hmm.

George Church 1:25:19 That might mean, you know, there's all kinds of pathologies that come along with that, that-

Dwarkesh Patel 1:25:26 Wh- what is it about the nature of your work, maybe biology more generally, that makes it possible for one lab to be behind so many advancements? I don't think there's an analogous thing in computer science, um, which is a field I'm more familiar with, where you could go to one lab and, uh-

George Church 1:25:44 One academic lab, yeah.

Dwarkesh Patel 1:25:45 Yeah. Sorry, one academic lab, and then 100 different companies have been formed out of it, including the ones that are most exciting and, uh, doing a bunch of groundbreaking work. So, is it something about the nature of your...... um, academic lab? Is it something about the nature of biology research? Uh, wha- wha- what explains this pattern?

George Church 1:26:03 Well, f- first of all, thank you for being so (laughs) generous in your evaluation, which may be, um, taken with a grain of salt. But, uh,