00:00welcome to the a 16z podcast I'm Michael
00:02Copeland there are those who would say
00:04that abri degre is out to cure death but
00:08with this former artificial intelligence
00:10specialist turned gerontologist is
00:12really focused on is health and the side
00:15effect of health is living a lot longer
00:17in this segment of the a 16z podcast we
00:21talked with Aubrey de Grey on the
00:22subject of aging and health and how his
00:25training as a computer scientist helped
00:27him approach the problem in a different
00:28way from traditional biologists the
00:32intersection of software and biology and
00:34how this troublemaker from the computer
00:36science world is trying to keep us all
00:38healthy for a very very long time
00:41Aubrey welcome thank you very much for
00:44having me we want to talk about the
00:45intersection of computer science and
00:47software and biology and I think there's
00:49a misconception out there that you're
00:50out to cure death but you you shifted
00:53from software verification to this idea
00:55that of really health but extending life
00:58and how did you make that transition and
01:01why well first of all thank you for
01:04putting in that clarification because
01:06it's normally something that I need to
01:07put in interviews that yes I'm just a
01:11real medical researcher I work on health
01:13and any longevity benefits that may
01:15arise from this work are a side effect
01:18of health but correct us being sick is
01:19what kills people so and of course
01:23journalists you know tend often to
01:24sensationalize that they think it sells
01:26more papers if they talk about
01:27immortality and other stuff like that
01:29but it was quite interesting actually
01:32and a lot of luck was involved
01:34essentially in 2000 when I was in the
01:39middle of my work in software
01:40verification I met and shortly
01:43afterwards married a biologist from the
01:45US who was on sabbatical in the UK she
01:50was a full professor at UC San Diego did
01:53quite a lot older than me and through
01:56her I first of all learned a lot of
01:59biology over the dinner table as one
02:00does right but also after a couple of
02:02years it began very gradually to dawn on
02:05me that we were never talking about
02:06aging which was really quite bizarre I
02:09because you know I'd always gone through
02:10my whole life assuming that everybody
02:12understood that aging was the world's
02:14most important problem the source of the
02:16world's greatest amount of suffering and
02:17so on um but it turned out that my wife
02:20and indeed all the other biologists I
02:22was meeting were actually of a very
02:24different persuasion they thought that
02:26aging was not very interesting and not
02:28very important and I was I was
02:30absolutely appalled and it actually took
02:32me another year or two before I really
02:34came to terms with it but eventually I
02:36decided that even though I was already
02:38working in artificial intelligence
02:40research which I viewed as a
02:42humanitarian exercise making machines
02:45smart and making smarter relieve us of
02:48the tedium Oh having to spend out I was
02:52doing things that we wouldn't do if we
02:53weren't being paid in order to make the
02:57um yeah so I wanted that to end but I
03:00always knew that that was only the
03:02world's second most important problem
03:03and so after I kind of got over the
03:06shock of finding that most people didn't
03:09think that way which I'd never never
03:11dawned on me until I the age of 30 or
03:13whatever um yeah I switched fields and I
03:15was in a very fortunate position there I
03:17was working in the University of
03:20Cambridge on a bioinformatics project
03:23that was a nice way of combining my
03:25formal training in computer science with
03:26my newfound informal training in biology
03:28and that project was very undemanding it
03:33left me lost spare time and of course
03:35access to all the university facilities
03:37and libraries and so on and by the bells
03:39so I was able essentially to do research
03:44in my spare time in fact the reason I
03:46originally took the job was so as to
03:47resume my artificial intelligence
03:49research without which I'd had to put on
03:51hold for a year or two an account of
03:53lack of funding but when I decided to
03:56switch fields of course it was just
03:58switching what I did in my spare time so
03:59it was really something risky and did
04:02your AI background and your software
04:04background did that bleed over into
04:06longevity and how did those two things
04:08match up if at all there was a lot of
04:13overlap there it was actually one of the
04:17big reasons why I decided to switch
04:19fails but why I decided that I had
04:21spectable chance of making a significant
04:23contribution in biology I realized that
04:26a lot of the reason why people were not
04:29making progress in postponing the ill
04:33health of old age was because they were
04:35going about it more as you know lifelong
04:39basic scientists and not so much as
04:41technologists with the more
04:42goal-directed kind of way of thinking I
04:45felt that there was a good chance that I
04:46would be able to bring in new ways of
04:49addressing these these issues and and
04:52thinking about them and maybe come up
04:54with ideas that would actually be
04:55promising I sure enough that's how it
04:57turned out and how did you I mean
04:58because there's they're sort of a
05:00religious I don't know want to say war
05:02but like you know chemistry versus
05:03versus computation like the way forward
05:06is chemistry how dare you think that you
05:08know just compute compute can
05:12were you accepted into the biology world
05:16or were you seen as somewhat of a
05:18heretic I actually was accepted pretty
05:20well pretty quickly in the first five or
05:23so years that I was working at a
05:25gerontologist so basically the second
05:26half of the nineties I was publishing
05:31stuff that was relatively harmless it
05:33was like you know new explanations for
05:36other people state that you know
05:39interpreted them better and this was
05:41actually very well received I was able
05:43to gain quite a reputation for myself
05:46and in fact the the fact that I didn't
05:50have the regular traditional
05:51experimental training in how to work up
05:54well somewhat in my favor you know
05:56people thought well this guy you know
05:57he's come from nowhere and he's having
05:59these ideas that we ought to have had
06:00really and you know he must be very
06:02smart so that was so I rose to a level
06:05of you know general acceptance and
06:07recognition pretty quickly especially
06:09since I was taking the trouble to
06:10integrate myself a lot going to
06:12conferences on my own hook and so on um
06:15it was only after the year 2000 when I
06:18started putting forward my particular
06:19ideas for how we might do something
06:21about aging that people started to think
06:23of me as more of a troublemaker
06:26some cases use my lack of experimental
06:28training against me as a way of trying
06:31to imply that because I didn't know how
06:32to work but that therefore everything I
06:34said with with nothing's well let's
06:36let's get you to your ideas on how to
06:38extend life and what's caused people to
06:41you know some people think it's um you
06:43know completely off-base its hokum and
06:45others although others have tried to
06:47disprove it and really can't so first of
06:50all let me let me point out that you are
06:51backsliding a little bit by talking
06:53about my efforts to extend life so
06:55you've got to remember it's a side
06:56effect yes your efforts to make us
06:59healthy that's right to keep itself they
07:01keep us healthy yeah all right so um
07:03well right yes it was very controversial
07:05at first and of course this was kind of
07:08no surprise to me because when I had
07:10this big kind of Eureka moment in 2000
07:12of how to go about this the the
07:15realization was a radical departure from
07:17what people had been thinking before
07:19essentially gerontology for decades had
07:22been built on the concept of trying to
07:27work with the variability in rate of
07:30Aging that we see in nature the fact
07:32that some species age a lot more slowly
07:35than others even within a species some
07:37individuals age a fair bit more slowly
07:38than others if we could understand that
07:40phenomenon really well and maybe we
07:43could translate that into understanding
07:44into something therapeutic and haven't
07:48worked of course nobody was really
07:50getting anywhere and indeed until the
07:521990s pretty much everyone had even
07:55given up on trying to admit that that
07:57was the ultimate goal of gerontology
07:59there was some breakthroughs in the late
08:0180s and early 90s which kind of changed
08:03that but it was a bit of a force storm
08:05actually um and I came along and
08:08basically my new idea in very simple
08:11terms was just that we rather than
08:13trying to clean up metabolism and
08:15thereby slow down the rate at which the
08:19body creates damage to itself as a
08:20result of its normal normal operation
08:22rather than doing that the idea was to
08:25actually repair damage so to go in one
08:27step down the road so to speak to do
08:30periodic repair not necessarily
08:33completely 100 percent comprehensive but
08:35fairly comprehensive so I was
08:37to maintain a level of damage in the
08:40body that it was within the tolerance of
08:43the body you know cuz the body set up to
08:44tolerate a certain amount right that's
08:46why nothing really goes wrong until late
08:48middle age so um this idea only flew
08:53with only made sense because I was able
08:56to identify other areas of biology that
08:59gave rise to practical options for
09:03really implementing this for doing the
09:05damage repair thing and these other
09:07areas will areas that gerontologists
09:09didn't know anything about they'd never
09:11come across limb when I started talking
09:13about them they thought that I was
09:14talking about stuff they didn't need to
09:15know that you know so it was all very
09:17very difficult and um well so I mean
09:22there was one example one big example
09:23that isn't a component of sense is not
09:27even from anything medical at all it's
09:31yeah it's and their sense the strategies
09:34for engineered negligible senescence but
09:35you really don't need to know that but
09:39yeah I mean so one examples it's not
09:40even from anything biomedical it's from
09:42environmental decontamination the idea
09:44of using bacteria as a source of genes
09:48and enzymes that can break down material
09:51in the body that we don't have any
09:53enzymes to break down thereby of course
09:56eliminating this stuff and stopping it
09:58from accumulating to an eventually toxic
10:00level you know people pretty intrigued
10:04by it but they didn't really think it
10:06was something that made much sense in in
10:09medical terms or at least some of them
10:11didn't of course in any field that a
10:14range of degrees of dogmatism and so my
10:18more vocal detractors certainly just poo
10:21pooed pretty much everything they didn't
10:22hadn't already thought of but in about
10:262005 I was able to kind of smoke out the
10:29opposition so to speak which had
10:30previously been happening at the level
10:32of kind of off-the-record ridicule and
10:34there were a couple of major running
10:38battles that I pretty comprehensively
10:41won in terms of demonstrating that the
10:44ideas that I was putting forward were
10:46indeed very plausible
10:48that the conclusion that there were not
10:51plausible that other people had been
10:52expressing was essentially a result of
10:55ignorance at the courts this idea that
10:57you can repair the damage that's been
11:00done can repair some of it basically all
11:01of it yeah I mean the idea is you don't
11:04need to repair all of it but you do need
11:06to repair most of it you know that the
11:07damage comes in a variety of different
11:09types different categories and within
11:13each of the categories there may be a
11:16lot of examples some of these things may
11:19add up but basically any of the
11:21categories can kill you on its own how
11:23do we go about doing this repair well of
11:25course the different categories have
11:26very different approaches so one example
11:28that's very familiar to everybody is the
11:32way of repairing the type of damage that
11:35I call in please cell loss so cell loss
11:37it's just cells dying and not being
11:39automatically replaced by the division
11:41of other cells and various aspects of
11:43aging are predominantly driven by that
11:46Parkinson's disease is a fine example
11:48where there's a particular part of the
11:50brain in which neurons a particular type
11:52tend to diet and I usually wrap it right
11:54and yeah that virtually stops that part
11:58of the brain from working right the fix
12:01is of course stem cell therapy
12:03you put cells in look and know how to
12:05divide and differentiate to replace the
12:07cells that the body is not replacing on
12:08its own I think in the cell that you
12:09need in front and stem cell therapy for
12:12Parkinson's disease is a very viable
12:15concept it's actually there are a couple
12:17of clinical trials going on right now
12:18that people are very optimistic about so
12:21yeah and that's the kind of idea but a
12:23lot of the other things I suggested were
12:25much further afield from what
12:27gerontology FL even heard about in the
12:29news let alone in conferences things
12:32like for example well as I mentioned put
12:35a finding bacteria that can break down
12:36substances like oxidized cholesterol
12:38that drives arthrosclerosis the idea
12:42would be then you'd find the bacteria
12:43you'd then identify the genes and
12:47enzymes that they had that allowed them
12:48to break this stuff down and then you'd
12:50modify those genes so that you could put
12:51them into human cells and protect the
12:53human cells by giving them this
12:55augmented garbage disposal capacity it
12:58took us a long time to get it to work
13:01the idea forward for the first time back
13:03in 1999 we started working late in about
13:062005 and it was very 2012 that we were
13:10able to demonstrate a really powerful
13:11proof-of-concept showing that we could
13:15actually protect cells in cell culture
13:17from otherwise lethal amounts of this
13:20particular toxin so yeah that's why it
13:25was a hard battle credibility wise but
13:28things are getting better all the time
13:29in a large number of these areas just
13:31because we are getting to
13:32proof-of-concept well so you you the way
13:35you have gone about this like you say
13:36that that people didn't hear about this
13:38in conferences and this was so far
13:39afield that it just didn't occur to him
13:41and somehow it occurred to you but it
13:45what that sounds like a little bit is
13:47the internet and it sounds like and by
13:50that I mean this ability to move
13:51information you know from one place to
13:54another and that everybody has access to
13:55it so in healthcare for example just the
13:57sheer ability to look across a huge data
14:01set of people and patients and outcomes
14:03and and say oh that's what happened do
14:08you think that what you did personally
14:10as this kind of you know human version
14:13of the internet does that accelerate I
14:16mean so so now as more data and more
14:18information is kind of more widely
14:20available and ideally as silos are
14:23broken down more and maybe they're not
14:24being broken down fast enough do we
14:26accelerate this sort of ideas and the
14:28approaches that you have put forth mmm
14:31interesting question I would say yes and
14:33no I mean certainly the availability of
14:36information online is an enormous
14:39opportunity for people to come out of
14:42left field and have new ideas but then
14:45you know in most sciences and in fact in
14:48most fields of technology one has to be
14:50pretty knowledgeable already in order to
14:52actually have new ideas that are not
14:54completely broken you know I would not
14:56have been able to come up with sans and
14:58if it hadn't been for spending those
15:00initial five years in the field just
15:02learning Jeff going around and listened
15:03and thinking and you know that the the
15:07the silo question comes later really
15:12this other question the silo problem
15:15arises really when we look at the ways
15:19in which ideas are taken in which big
15:21ideas are taken forward because of
15:23course getting something to actually
15:24work once you've decided what you're
15:27trying to get working involves the a
15:30long string of solving little problems
15:33and it takes time and money and the
15:37biggest obstacle really in science to
15:42getting stuff done like that is the fact
15:44that the overwhelming majority of
15:46science is funded by peer review peer
15:49review is an absolute catastrophe when
15:51it comes to doing anything high-risk
15:53high-reward and also for that and I said
15:55doing anything cross-disciplinary
15:57because people tend to play it safe and
15:59and and/or get and they're worried about
16:02getting attacked or sancho
16:03well it's mm-hmm kinda I mean basically
16:05peer review is an discovering process
16:08right it's both the people who actually
16:10are providing the money and the people
16:12who selecting who to get the money out
16:15of the many of the far too many
16:17applications you know they need to
16:19protect their reputation somehow and
16:21it's just too easy to be cautious and to
16:26favor incremental stuff that is within
16:30the remit of what people what the
16:33applicant has already done you know very
16:35close to it so it's an enormous stifling
16:40influence on research and it's
16:43particularly bad these days when funding
16:45is so short as well as so much when the
16:48pay line the proportion of grants that
16:50are actually funded is so low again
16:51coming from the AI in software world and
16:54working on those kinds of problems to
16:58how do you compare that to working on
16:59the problem of sort of humans and and
17:02health and the subtleties of the human
17:05body and biology versus the subtleties
17:07of you know clearly AI is a very hard
17:09problem to but for those folks in the
17:12technology world who are increasingly
17:13going to cross over into to the health
17:17what are the things that they ought to
17:19keep in mind and what's so hard about it
17:21honestly well right so I mean I think
17:24the big things that a technologist an
17:29engineer of any kind has as a starting
17:32point is the understanding that any
17:36machine and of course the human body is
17:38just a machine right any machine has
17:41moving parts it does damage to itself as
17:44a side effect of its normal operation so
17:46one can use the same principles sending
17:49like top-level principles to postpone
17:51the ill health of old age as one might
17:53use to keep a car going longer than it
17:56was designed to go that's the easy part
17:59then the hard part is that because the
18:01human body is such a astronomically
18:03complicated machine and because we
18:04understand it so poorly one has to be
18:08quite ingenious in identifying
18:10approaches to to extending its healthy
18:14lifespan that I like me to work despite
18:18our ignorance in other words to
18:19essentially leave well alone in as much
18:22as possible and only interfere with
18:24things that are unlikely to have
18:25unwanted side-effects and how do you run
18:29those as kind of I don't know theories
18:31or lines of code or lines of thought I
18:34mean how do you kind of get to the right
18:37answers well I mean the probably the
18:41biggest thing that the biggest component
18:44of the sense concept that allowed the
18:46whole thing to fly originally was my
18:48realization that we have this window of
18:50opportunity afforded by the fact that
18:52the body is set up to tolerate a certain
18:54amount of these various types of damage
18:56right which means that we can we can
19:01infer that these various types of damage
19:03are inert until such time as they've
19:06accumulated to a certain threshold of
19:08abundance now inert means they're not
19:10participating in metabolism so if we
19:13just target those initially inert
19:15phenomena then we have a good chance of
19:19not having unwanted side-effects not
19:21disrupting the you know labyrinthine
19:25network of processes that keep us alive
19:31in that and and kind of your continuum
19:33like look you know I know that the
19:35therapies aren't ready yet but how far
19:38along are we and and as a side effect
19:40you know in your mind how long can
19:42people live so the how far along we are
19:46in developing these things of course the
19:49different types of damage there's a
19:51different answer to our questions so in
19:52the case of stem cell therapy of course
19:54there are quite a number of stem cell
19:56therapies already in clinical trials I
19:57mentioned the case of Parkinson's
19:59disease there are plenty of other of
20:00course aspects of ill health that don't
20:03have to do with aging that are also
20:05amenable to clinical trials eating stem
20:08cells most of the other things are a
20:10good deal less far advanced some of them
20:14are partly advanced so for example the
20:17elimination of amyloid which is a kind
20:19of molecular waste product that
20:21accumulates outside and especially
20:22between cells in some cases in where in
20:26one case Alzheimer's disease that's also
20:28that's very much in clinical trials and
20:30we've basically solved the problem that
20:31can now be eliminated it doesn't have
20:33much effect on ultimate disease on its
20:35own but its ferb at that in combination
20:39with other therapies that will be
20:40developed in the future that fix the
20:42other aspects about them is that it's
20:43going to be very useful there are other
20:47amyloids though in other tissues that
20:48accumulate and cause other problems in
20:50aging and we haven't made much progress
20:52in that area so that's actually one area
20:54that we're funding precisely for that
20:55reason in a sense Research Foundation
20:58exists and indeed it was constituted as
21:01a charity specifically because not a lot
21:04happening in a lot of these areas we
21:06have been neglected 40 March and
21:08somebody need to step in and actually
21:10kick them along the road and get them to
21:12a sufficient level of proof of concept
21:14that other people would get interested
21:16and we've been very successful in doing
21:17that are you in some sense in
21:20competition with the rate kurtzweil
21:22singularity view of the world I mean we
21:24rather gonna become you know machines
21:26you know that have the modern own brains
21:28of a human or sentience over human or
21:31we're gonna as a side effect live a lot
21:33longer and therefore we don't need to
21:35have the singularity so I wouldn't call
21:37it competition it's more of a race you
21:39ray and I we know each other well of
21:43course and Ray is very much interested
21:48in regular biomedical approaches as well
21:51so when he talks about how to live long
21:54enough to live forever it's the phrase
21:55that he likes he talked about these
21:58bridges those things that you can do
21:59today that will postpone the ill health
22:01of old age somewhat and he is actually a
22:05lot more optimistic than me with regard
22:07to how much we can postpone aging with
22:09stuff that already exists today but then
22:12bridge 2 as he calls it is almost
22:14identically fancy it's basically using a
22:17high tech biotechnology to repair damage
22:20and bridge 3 is the one that you're
22:22really referring to the increasing use
22:24of what we might just in general call
22:27non biological solutions to medical
22:29problems especially focusing on the more
22:32miniaturized stuff like nanotechnology
22:35and then eventually perhaps even on
22:37transferring consciousness to a
22:39different substrate the concept of
22:40uploading so the reason I call it a race
22:43rather than a competition is because we
22:46just don't know what's going to actually
22:48prove to be implementable soonest and do
22:50you care I mean if you know damage is
22:53repaired biologically or with these
22:57Oh kind of care you know I'm quite
23:00sentimental about being made out of meat
23:02but but at the same time you know if
23:05push comes to shove and the work that we
23:08do and other people do on the biotech
23:09side start to you know hit diminishing
23:13returns and basically run into the sand
23:14and work on uploading or in other other
23:18ways of reinforcing the health of the
23:22individual through non-biological means
23:24actually moved forward relatively
23:26rapidly and ended up being the solution
23:28that got there first then that's fine
23:30yeah it's the end result that matters I
23:33guess so let's say whichever one works
23:36we're gonna have a lot more people
23:37living a lot longer rents are high
23:40enough in the Bay Area
23:42not to mention yeah you know just food
23:45or climate change how do we account for
23:49all these people living for so long
23:52so the concept that if we defeated aging
23:57we would have a terrible problem of
23:59other population it's probably the
24:01number one knee-jerk concern that people
24:04raise and it's so insidious and so
24:07persistent that we eventually resorted
24:09to the option of actually funding a
24:12forecasting group in Denver that have
24:15over the past thirty years developed a
24:17very well-regarded system called
24:19international futures we actually funded
24:21them to extend the versatility of their
24:23system so that it could explore the
24:26concept of sense the concept of actual
24:29rejuvenation biotechnology that would
24:31restore the health of people who are
24:32already in middle age and keep it and
24:34keep it there and of course we knew what
24:38the answer was going to be more or less
24:39namely that the the consequences of that
24:43for the trajectory of world population
24:45or indeed of populations of regions was
24:48actually much more modest much less
24:50frightening than people would normally
24:51think plus also of course we're
24:55interested in the solutions so we've got
24:58an overpopulation problem today right
24:59but the problem is not that we have 7
25:01billion people the problem is that we
25:03have 7 billion people who are all
25:06creating a lot of pollution because of
25:09fossil fuels and suchlike so of course
25:11the solution to that could be have fewer
25:13children or don't cure aging but it
25:16could also be invent new technologies
25:18that increase the carrying capacity of
25:21the planet and that's of course exactly
25:22what we're doing we having a burgeoning
25:24of renewable energy quite soon I bet
25:26we'll have nuclear fusion one way or
25:28another in terms of agriculture as well
25:31you know we have this little little land
25:33right now to create enough food for
25:36everybody but that's changing there with
25:38the development of artificial meat and
25:39so on so it seems to me pretty damn
25:42clear that the increase in the carrying
25:45capacity of the plant over the next
25:47well way out run the increase in the
25:51actual population of the planet do you
25:53feel like we're at a point in time where
25:54technology and kind of the things that
25:57you're studying whether directly are
25:59related will kind of speed things up or
26:02you know it seems to me that you know
26:05when we look at transportation we've got
26:06lots of things that that at this point
26:08software is helping us to really
26:10accelerate some of the things that we
26:11want to do do you in in the near term
26:16hope to see or think that you'll see
26:18advances across the board in healthcare
26:21well we're certainly seeing that already
26:23I mean of course a lot of this involves
26:25enabling technology at the level of
26:27informatics and also at the level of
26:29simple eyewear you know so we've got
26:31better techniques of sequencing now
26:33we've got better techniques for
26:34modification of the genome you know with
26:37things like CRISPR for example and of
26:41course we've also constantly got new
26:42advances in computational interpretation
26:47of what we know about the genome and the
26:49epigenome and the microbiome and what
26:52what these things are doing so yes I
26:54think that there's definitely a very
26:56heartening and accelerating increase in
27:00our ability to to keep to maintain
27:03health and I mean to restore health as a
27:06result of all of this but I think we've
27:09always got to remember that things like
27:12the ability to sequence things really
27:14fast and really cheaply or the ability
27:15to process things usually doesn't
27:17actually underpin the fundamental
27:20breakthrough there's all this new data
27:22but then what the hell do we do is right
27:24way does instead is it makes things
27:27having the Gina I'm even the original
27:30wondering I'm you know crate fantasy now
27:32has definitely facilitated a lot of
27:35research but on its own it doesn't care
27:38thing right well Aubrey de Grey I wish
27:40you health and much of it and for many
27:44many years and thanks for joining the a
27:4616z podcast it's my pleasure thanks for