Making Sense - Sam Harris - #394 — Bringing Back the Mammoth

00:00:00.000 welcome to the making sense podcast this is sam harris just a note to say that if you're hearing

00:00:12.500 this you are not currently on our subscriber feed and will only be hearing the first part

00:00:16.900 of this conversation in order to access full episodes of the making sense podcast you'll

00:00:21.800 need to subscribe at sam harris.org there you'll find our private rss feed to add to your favorite

00:00:27.020 podcatcher along with other subscriber only content we don't run ads on the podcast and

00:00:32.500 therefore it's made possible entirely through the support of our subscribers so if you enjoy

00:00:36.520 what we're doing here please consider becoming one welcome to the making sense podcast this is sam

00:00:49.800 harris today i'm speaking with ben lamb ben is a technology and software entrepreneur

00:00:56.300 who has been featured in many publications the wall street journal new york times forbes discussing

00:01:02.820 topics related to innovation and technology he's also the co-founder and ceo of colossal biosciences

00:01:09.700 a company he started with biologist george church for the purpose of resurrecting extinct species

00:01:17.060 like the woolly mammoth and the tasmanian tiger and the dodo and they aim to reintroduce them into

00:01:23.780 the wild ben is also a fellow of the explorers club and serves on the scientific advisory board

00:01:29.580 of the planetary society but we focus on his work at colossal we discuss the difference between their

00:01:35.980 approach and jurassic park the details of resurrecting the mammoth and other species the relevance of

00:01:43.240 this work to human health the role of artificial intelligence here what it would take to reintroduce

00:01:48.920 mammoths and tasmanian tigers and dodos back into the wild the environmental and business case for

00:01:54.780 doing this and other topics anyway the future appears to be almost here and now i bring you ben lamb

00:02:03.180 i am here with ben lamb ben thanks for joining me thanks so much for having me so um we're going to

00:02:15.160 talk about some amazing stuff that you're doing over there at colossal your biotech company but before

00:02:21.700 we get there how do you summarize your career and interests at this point or how did you um give me

00:02:28.160 the potted bio that gets us to the topic at hand well i'm i'm definitely insatiably curious and so i'm

00:02:35.460 always you know i'm not really a technologist i'm not really an engineer i try to look at things from a

00:02:40.500 systems design perspective and i'm always fascinated with how things work and how things can be improved

00:02:46.380 and i always like to find new interesting projects and so i've been in everything from mobile gaming

00:02:51.800 before that was quite big i built some precursors to large language models that we were actually

00:02:57.380 calling conversational operating systems at the time my last company was actually satellite software

00:03:02.400 and defense so we actually built a common operating picture to understand and track everything

00:03:08.000 in the sky all the way actually lower the orbit all the way down to the surface of of the sea uh and

00:03:14.700 work closely with the u.s air force and space force and some of our global partners on that and then i met

00:03:19.460 george church and you know i actually kind of fell into de-extinction i reached out to him because i'm curious

00:03:26.180 and i thought that the intersection of synthetic biology and ai and computational biology and you know

00:03:32.240 quantum which i hear is only two years away every two years um will eventually you know kind of give us

00:03:37.280 dominion to engineer life and and do directed evolution on a scale that you know is unprecedented

00:03:43.680 for you know human advancement and so i got massively excited about the opportunities there and and then

00:03:50.480 i asked george the question and i said if you had one project with unlimited capital that you could

00:03:56.180 focus on for the rest of your life you know what would it be george and you know didn't know what i

00:04:01.600 would get out of george is it going to you know another star system or what and his feedback was i would

00:04:06.400 bring back woolly mammoths uh and help reintroduce them back into the ecosystem to help biodiversity

00:04:12.120 in the ecosystem as well as develop technologies for both human health care and species preservation

00:04:17.780 and and at that moment i was pretty hooked hmm yeah george is a very impressive scientist i've met him

00:04:23.840 i think it might have only been once maybe maybe twice at a conference but he's is he still at harvard

00:04:30.160 he's still at harvard so i do get to monopolize a decent amount of his time but i do we do share

00:04:36.160 him with harvard and a handful of other initiatives he's co-founded so the company is colossal biosciences

00:04:43.280 is that the the full name correct and uh so what are you doing over there at colossal yeah so we decided

00:04:49.560 that we wanted to build the world's first de-extinction and species preservation company because

00:04:55.860 if you look at some of these stats and kind of the trend line that we're seeing for biodiversity loss

00:05:02.300 and what the impacts to ecosystems can will and will be especially from a keystone perspective

00:05:07.600 it's pretty terrifying and when we started the company our original pitch deck all the data we

00:05:12.880 could find showed that if without massive human intervention or massive new technologies that we

00:05:18.640 could lose up to 15 15 of biodiversity between now and 2050 what's terrifying is in 2024

00:05:25.820 that number has been upped to 50 50 so that's not a very good trend line and so george had this

00:05:32.560 vision and i just feel like i'm kind of the steward and helper with it of we could go build a company

00:05:38.300 that could you know one build tools and technologies that could be capable of bringing back lost species

00:05:44.940 as well as applying those technologies and innovation to conservation giving that to the world for free

00:05:50.720 and all these species have direct applications uh those technologies like genetic engineering and

00:05:56.000 others to human health care so we really had this interesting opportunity to build a company that

00:06:00.620 hopefully could inspire people create true impact but also create massive value creation around the way

00:06:06.680 and which species are you focused on first so we've announced three species today the woolly mammoth

00:06:13.380 which george was actually working on uh for about eight years before i showed up collecting samples in

00:06:19.380 siberia working on computational analysis and elephants the tasmanian tiger also known as the

00:06:25.640 thylacine which went extinct in 1936 in tasmania and lower australia due to human hunting the australian

00:06:32.540 government actually put a bounty on eradicating the species and then you know we wanted a bird species we

00:06:38.280 wanted to recruit best shapiro who's our chief science officer so we did the dodo because there's

00:06:42.480 probably not a more iconic species than the dodo that symbolizes de-extinction so how is this

00:06:50.000 different from jurassic park i mean that you know that i don't think anyone would really associate it

00:06:55.240 with jurassic park until you bring in the mammoth and then all of a sudden we the we're talking about

00:07:00.040 charismatic megafauna and we're we're hoping for a t-rex to what degree does that vision account for

00:07:07.540 some of your enthusiasm around this and i mean obviously there's a difference between reintroducing

00:07:12.340 animals to the wild and and setting up a theme park are you i mean was jurassic park a formative

00:07:18.500 idea for you or is that or you arrived where you are by a different path so we get the jurassic park

00:07:26.440 question quite a bit as you as that may not surprise you yeah like occasionally when i go on stage to

00:07:32.000 speak they'll play the music you know we've seen every meme with like george's face on it or my face on

00:07:37.380 it so we we've heard this a time or two george will tell you so i think george and i have slightly

00:07:42.640 different perspectives on it george will tell you that in a weird way he thinks that michael

00:07:48.060 crichton was actually inspired and jurassic park was actually inspired by him because if you go look

00:07:53.080 in the original jurassic park uh novel there's actually a dna sequence early in the in the uh work

00:07:58.980 in in in the novel and it actually is george's work with only one letter changed and george will

00:08:05.600 argue that statistically um it's still plagiarism it's it's still and george loves you know many of

00:08:13.900 crichton's novels right and it's a very inspiring author that he was and but george will tell you that

00:08:19.240 you know he laughs and says maybe i inspired jurassic park because a lot of his original work in yeast is

00:08:24.360 actually shows up in the book i i will tell you from my perspective you know growing up uh you know

00:08:29.940 born in the 80s you know child of the 80s and 90s you know i think one you know i love science fiction

00:08:36.200 i love jurassic park that's not necessarily why i got into this but it sure makes it a lot easier to

00:08:42.040 connect with people because even though we have the memes and all the jokes that come around colossal

00:08:46.420 versus jurassic park at least you know jurassic park which was this dystopian movie at least it taught

00:08:52.600 people about there's this thing called dna and there's this thing called genetic engineering and

00:08:57.460 so like moms in iowa know that there's this ability to manipulate the genome because of mr dna right 0.99

00:09:05.900 and so we we also we a lot of times use jurassic park as an example of how we're doing it exactly

00:09:12.040 inverse meaning that we're not trying to fill the gaps in a ancient dna that with the holes that you

00:09:19.140 get from you know frogs or whatnot we're trying to truly understand the genomes so that we could

00:09:24.540 selectively choose the genes that we then want to engineer into that of a living species so it's

00:09:31.340 almost like reverse jurassic park and when we say that to the kind of average public and that we're

00:09:36.220 in in in in some journalists and whatnot when we're explaining the process and the science

00:09:40.420 they really resonate with it because i think that movie does have such a head was the right movie with

00:09:45.600 the right technology and the right story at the right time that really connects with people so

00:09:50.340 let's go over those details again so what was being proposed as the scientific you know bioengineering

00:09:57.140 basis for jurassic park and and what exactly are you doing with you know paleogenomics and going out into

00:10:05.300 the the wild and getting dna samples however imperfectly preserved and integrating them with living

00:10:13.460 species how how what what is your approach and how is it different from what was being i i it's been a

00:10:19.720 long time since i i saw the film i actually never read the novels i don't know if the films depart from

00:10:24.660 the novel in their logic and i i know nothing about any of the um you know errors that crichton might

00:10:31.900 have made and with respect to his molecular biology if he made any so what what was proposed there

00:10:37.120 and uh what are you guys actually doing so in drusk park they propose that you would go find pieces of like

00:10:45.140 amber uh which by the way is a very porous material it is not a good dna store not that we've tried but then

00:10:52.540 magically in amber you'd get insects and specifically mosquitoes that had been trapped for over 65 million years

00:11:00.500 uh and while that's true there isn't dna from that uh amber as i mentioned is a very porous material it

00:11:06.980 is not it's not a great dna store typically the best dna stores for us for ancient dna are cold dry

00:11:13.080 places so animals that passed away in a cave and a very dry cave that stayed consistent without other

00:11:19.900 animals in it that's kind of optimal for us and so then they would take this dna that they extracted

00:11:25.560 from a mosquito that lived you know 100 million years ago and and been a dinosaur and they would extract

00:11:31.700 in in the movie actual blood which also is impossible and then they would take that blood

00:11:36.700 use computers which is very similar to what we do which i'll get into and then fill in the holes of the

00:11:42.900 of the ancient dna because ancient dna is very very fragmented with that of in the movie frog dna

00:11:48.540 and amongst some other um many other things but the problem with that number one is there is an ancient

00:11:54.440 dino dna you know the oldest dna that we're able to collect is you know a little bit over a million

00:11:59.240 years there's some fragments and stuff that are older but you know for the most part we're working

00:12:03.860 in thousands and tens of thousands of years not you know millions of years because dna degrades very

00:12:10.360 very quickly it starts to break down the minute it leaves your body and so when you layer in like

00:12:14.800 radiation heat acidification other animals defecation other animals dying on it it starts to break

00:12:22.100 down and it also gets massively contaminated it's not truly endogenous at that point right and so

00:12:28.420 what we do is instead of going and taking a bunch of different pieces of a mammoth assembling it and

00:12:35.360 saying what's missing and how do we plug that with a frog or elephant dna we do it almost exactly in

00:12:41.320 reverse so the first thing that we did is we went out and we looked at phylogenetically so on that tree

00:12:46.680 of life that we've all seen some version of it you know in science textbooks and and today on the

00:12:51.460 internet we say what is the closest living relative to the mammoth in this case and and that's actually

00:12:56.920 the asian elephant it's 99.6 the same genetically it's actually closer genetically to an asian elephant

00:13:03.920 than an asian elephant is to an african elephant and that's kind of a fun party trivia for you and then

00:13:11.020 we we spend a lot of time trying to do comparative genomics truly use a bunch of software use ai some of

00:13:17.240 our custom models to understand what is the difference uh even from an african elephant to

00:13:21.820 an asian elephant what is the difference from a population level so we actually sequence a lot of

00:13:26.120 different asian elephants so what is truly asian elephant versus population diversity in those

00:13:31.840 genomes because not all genomes are obviously exact copies of each other and then how do we compare

00:13:36.620 that to the mammoth and and and then we can identify okay where are these regions of the genome

00:13:42.720 that are vastly different and what do we know about that from scientific research from other

00:13:48.320 peer-reviewed papers you know from actually doing molecular and functional assays actually growing

00:13:53.360 stem cells and testing our hypothesis so you have to do a lot of work to then kind of verify what we

00:13:59.820 think the core genes that made a mammoth a mammoth were so that then we can engineer them into that of

00:14:06.140 an asian elephant cell and that's not just taking pieces and pushing in there that's actually just

00:14:11.160 changing existing code so we fundamentally don't need long-term pieces of these dna we don't need

00:14:17.520 all these dead samples we just need the code in the computer so do we have the complete genome

00:14:23.500 of the woolly mammoth i mean is that something that's disputed or did we get enough samples of

00:14:30.360 sufficient integrity such that we just know we've got the full mammoth genome we have enough so we have

00:14:36.220 we have about 65 mammoth genomes most of those aren't published most of those are siberian and

00:14:42.360 russian mammoth samples we're now doing a lot of work with alaskan mammoths as well and we work with

00:14:47.720 about 17 universities across the world one of which is the university of stockholm and luva dallin's work

00:14:52.820 and luva is arguably the number one mammoth researcher in the world and so we've taken all of

00:14:57.400 his different samples and it's about a 700 000 year difference between all the different samples

00:15:02.360 to kind of fill that in but we have enough of the protein coding regions of it as well as colombian

00:15:08.820 mammoths step mammoths and others and we have a pretty cool paper that i hope will come out mid next

00:15:13.520 year about this that shows the comparative genomics that we know enough of the mammoth genome that we

00:15:18.840 can identify the core areas around cold tolerance fat hair curved tusks so we actually have enough to do

00:15:25.600 our work it is not as complete as our our thylacine genome which we recently announced is 99.5 percent

00:15:32.500 complete or no sorry 99.9 percent complete which is which is truly incredible for any genome let alone

00:15:37.520 ancient dna that's the tasmanian tiger correct so are you using crispr technology to insert mammoth

00:15:45.520 code into the an asian elephant zygote or or what what is the step there that would produce a living mammoth

00:15:53.140 yes we start with an asian elephant cell right and we actually had to spend a lot of time getting the

00:15:57.580 culture conditions right actually immortalizing those cells one of the things that you know before

00:16:02.180 we get into the genetic engineering side uh one of the things that's interesting about elephants and

00:16:06.740 blue whales and a handful of other species is they actually get cancer a fraction of what we do based

00:16:12.620 on age and body weight of which they grow to and the leading theory of that and we're seeing this also

00:16:18.300 being verified in our lab is they have an overexpression of a protein called p53 about seven

00:16:24.680 times more than we have in in mice have uh which i'm sure you're you're familiar with and what's

00:16:29.160 interesting is we've actually had to learn how to regulate that because anytime we went to go make

00:16:33.580 those changes which which we'll get into the cell would just senesce so not only do we have to build

00:16:38.140 immortalization constructs to keep the cells growing and living and healthy we also had to figure out how

00:16:44.400 we can quote unquote turn down p53 so that we could edit the cells and then be able to turn it back up

00:16:50.660 because you don't want to produce you know cancer and elephants right and so we had there's a lot of

00:16:55.920 prep work before we even get to the point that we can do the engineering itself and as you can probably

00:17:01.340 guess you know because you your deep background in science you know crispr has become a catch-all for

00:17:07.480 all genetic engineering they're like oh it's just crispr right we just we just crispr it but what's

00:17:12.400 interesting is we use a combination of tools some of which are proprietary some of which are have been

00:17:18.760 invented by other organizations and universities and then we layer new techniques on it so in some

00:17:24.080 cases we're changing the individual nucleotides the individual letters on on that double helix in

00:17:31.400 other cases we're knocking out certain genes and in other cases we're actually synthesizing big blocks

00:17:38.140 of dna where if there's like a bunch of changes along one kind of strand it's actually more

00:17:43.520 efficient for us to synthesize that block knock that block out and then insert this new block so

00:17:49.040 that you have less likelihoods of off-target effects or unintended consequences from your editing and i'd

00:17:54.900 say the last thing that we're doing that on the editing front that is is our kind of i i think the

00:18:00.420 thing that sets us apart from a from a genomics perspective is we're trying to become the biggest

00:18:06.460 pioneer of multiplex editing meaning editing all over the genome at the same time so instead of

00:18:13.180 making one edit maybe you can make 20 edits 50 edits a thousand edits all with a very high degree of

00:18:19.440 efficiency versus having to synthesize entire giant blocks i do believe that technology will get here

00:18:26.200 being able to synthesize even full chromosomes at some point but we're not we as humanity aren't quite

00:18:31.760 there yet so editing is the most efficient kind of current modality that that we've been pursuing

00:18:37.520 so at what point did this actually become technically feasible i mean what what year would you say this

00:18:44.860 became something that you could actually start on and it seems to be just a a piece of science fiction

00:18:52.060 yeah so i i think you know people have been talking about you know crisper you know in in some

00:18:57.480 version of genetic engineering from from the 80s right but it was like it was i don't remember the

00:19:02.780 exact year but it was like what 2012 15 or 14 somewhere around there where we had like the true

00:19:08.920 kind of discovery around you know crisper and the idea that you could you know target a part of the

00:19:15.980 genome successfully knock it out and have it and have it repair itself and i think from there you've seen

00:19:21.820 work like david lu's work in in prime and base editing where you can change individual letters

00:19:27.320 you've seen kind of this like pre-cambrian explosion you know to use our jurassic you know use our 0.77

00:19:32.940 some of our jurassic fun terms of genetic engineering tools and technologies because

00:19:37.320 we've all been promised from the 80s and 90s gene therapies and genetic engineering capabilities

00:19:42.220 that allow us to do all kinds of stuff right that have never really manifested but i think that

00:19:47.720 that really um in the last you know 10 years has been where those technologies have been viable

00:19:53.660 i don't believe before that kind of 2012 2015 time frame of like that that crisper race with you

00:20:01.160 know fang and and shanford and dowden and all of them right that are just they're in george included

00:20:07.200 uh which were all incredible scientists i i don't believe that this would have been a viable undertaking

00:20:12.660 and and now after that it became viable but you know you still have compute you still have ai

00:20:17.880 there's a lot of other components to it and it just becomes very very costly the the goal to really

00:20:23.600 make this where it's possible and scalable i think is i think we're still a little bit early but we're

00:20:28.920 in kind of the right kind of five years to to truly be able to to deliver so is ai a necessary component

00:20:36.920 of the process it is and you know we're learning every day new ways that we can apply you know my

00:20:43.120 background has been mostly in software right and so you know we're finding every day new ways to apply

00:20:48.580 these technologies around it like we actually have a tool that we built internally that we've been

00:20:54.020 giving it this feedback loop so we built a cool little model that probably doesn't apply to most

00:20:59.120 people but for us we find it fascinating that will actually give us the right recommendation

00:21:03.420 that's over 90 accurate of what tool we should use for the specific edit that we're going after

00:21:09.600 and that's awesome when you think about biology because if you're going to make an edit you then

00:21:15.220 have to go see if that edit worked you then have to grow those cells those cells have to live then you

00:21:21.080 have to sequence those cells you got to wait a couple weeks in some cases if you don't have

00:21:25.140 sequencing cores internally to get that data back and so the feedback loop if you've make some if

00:21:30.320 you've made the wrong edit using the wrong tool at least the not the most efficient tool you know

00:21:35.300 can be months of lost uh scientific experiment time both costly in terms of go to market and

00:21:41.140 in terms of your research and in all the reagents and stuff that you had to go use in that right

00:21:46.220 and so we're we're now using ai not just for comparative genomics but even in what selection of

00:21:52.200 what editing tool we should use for the editing job that we're trying to go pursue so now how far have

00:21:59.700 you gotten and now i'm not asking just about the mammoth but you can talk about the dodo or the

00:22:05.520 the tasmanian tiger or anything else you've experimented with what what have you produced

00:22:11.080 in the lab and is it is it all still in vitro or what do you have a pregnant asian elephant that uh

00:22:18.360 has a name we there is there's no secret pregnant asian elephant mammoth unfortunately i i would be the

00:22:24.240 first i couldn't be more excited to share with you if there was so so de-extinction is a systems

00:22:30.360 problem right there's computational analysis or there's hdna if you'd like to continue listening

00:22:35.480 to this conversation you'll need to subscribe at sam harris.org once you do you'll get access to all

00:22:41.140 full-length episodes of the making sense podcast along with other subscriber only content including

00:22:46.320 bonus episodes and amas and the conversations i've been having on the waking up app the making sense

00:22:52.080 this podcast is ad-free and relies entirely on listener support and you can subscribe now at sam harris.org

Making Sense - Sam Harris - December 03, 2024

#394 — Bringing Back the Mammoth

Episode Stats

Summary

Transcript