A Deep Dive into Synthetic Biological Intelligence with Hon Weng Chong from Cortical Labs | #6 - The Strange Attractor

Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Samuel Wines (00:00):
Hello and welcome to the Strange Attractor, an
experimental podcast from CoLabs, a transdisciplinary innovation
hub and biotechnologyco-working lab based in
Melbourne, Australia.
I'm your co-host, Sam Wines,and alongside my co-founder,
Andrew Gray, we'll delve deepinto the intersection of biology
, technology and society throughthe lens of complexity and

(00:22):
systems thinking.
Join us on a journey ofdiscovery as we explore how
transdisciplinary innovation,informed by life's regenerative
patterns and processes, couldhelp us catalyze the transition
towards a thriving future forpeople and the planet.

(00:47):
This week, we sat down with theCEO of Cortical Labs, hon Wang
Chong, and just had a chat abouteverything they're doing.
It's a crazy cool company doingsynthetic biological
intelligence, which is just ascrazy as it sounds, and a really
, really interesting story abouthow he transitioned from being

(01:10):
a medical doctor through to CEOof this company.
So I will not say too much moreand just jump right into it.
Alright, let's go, let's getthings started so you all might
loop back to some of the thingswe've had a chat about.
So yeah, hon, welcome to ourpodcast.

(01:31):
Thanks for having me on theshow.
Yeah, no stress.
Thanks for being able to helpus co-create all of this sort of
together.
It's been a really excitingjourney in the past year.

Hon Weng Chong (01:42):
I know it's been actually how many years has it
been?
Like five years?

Andrew Gray (01:47):
Well, since you first stopped by the little
shipping container lab.

Yes.

Yeah, I'd say that was about five years 2018, 2019.

I think it was 2018, yeah, 2019.

Try to figure out how to fit you what was your
initial thoughts.

Oh shit, this is way smaller than I anticipated.
No, I loved it.

I'm still thinking about container lab
idea, except it's just notreally practical.

Maybe it would be a fun way to do modular server
racks in shipping containersthat are called, that can be
transported around the place.

Exactly, it's possible.
You could definitely modulatelike.
It just would be a big upfrontinfrastructure cost.
But once you've figured thatout, then you've got your labs
and you can just plug and playthem like Legos.

I did see this at one of the longevity startups
headquarters in San Francisco,in River City they actually had
built out an entire likelaboratory space.
I think it was.
What is it?
I think they had 10 or 12different containers just
stacked with HVAC and so forth.
But then again I realized thatBSL2 is actually less stringent

(02:55):
than the PC2 here.

Well, yeah, the US BSL2 is self-regulated.
So you just say yeah.
I'm meeting the requirements.

I know.

Which is why, like you can like Garage, biotech is
a legit thing in the US.

Yeah, I walked in and I saw a sign that said
BSL2.
I was like okay, it's BSL2.
But wait a minute, all theyhave is a wall with those
ceiling in a warehouse.
I'm like that's not reallycontained.

And then even some of the things you can do in the
US are pretty wacky compared tohere, like you can have, I think
, every Halloween.
I see this like image ofsomeone doing a jack-o'-lantern,
but instead of having a candlein there, they have like
bioluminescent bacteria thatthey've modified like outside,
which would like freak out anyof our regulators here in
Australia.

They saw that.
Is that literally in yourgarden, in your pumpkin patch?
No, no, it's on the porch, it'sin the lab it's contained.
It's within my property.

It's in a pumpkin.
Yeah, oh gosh.

Yeah, so okay, you loved it.
I'm glad that you thought itwas pretty cool.

Oh, and for reference, which there was
bioquisitive that you visitedback in the day it was
bioquisitive.

yeah, and for reference, it was Meow Meow who
got us onto bioquisitive.
Ah yeah, I don't know how Icame across Meow, meow or
something like that, butsomebody, I think everyone just
comes across it.

Yeah, it's a Disco Gamma Ludo Meow Meow.

And so I think he was like doing some
biohacking stuff and people werelike, oh, you should talk to
him.

Was it the?
I think he went in the news.
I remember like back then hewas in the news for doing a
sub-dermal injection of the Opalcard.
Yeah, he did that.
Yeah, he did that he did hiscourt case.

He went to court over that.
I think he I want to say he wonLike he.
He still had to pay the fines,but there was like it was still
a victory for him and forcyborgs.

What a fascinating human.

Wow, so yeah.
And then from there you Iremember because you had an
advisor somewhere else that youwere sort of talking to around I
think some early research thatwas sort of underpinning what
you guys were doing, and he wassort of calling in.

Oh yeah, yeah, it was a guy by the name of oh
gosh, Steve Potter Steve.

Potter yeah.

Steve Potter.
He was actually a really bigdeal in the early 2000s when
what we were doing was kind ofhot, and he retired, I think, in
the early 2000s.
So in the late 2000s, moved toIreland and had an interesting
life as a maker, in themakerspace kind of thing.

(05:32):
And interestingly enough, Ithink in 2018, I went on a bit
of a break and caught up withfriends that I had done my
research year in between mypreclinical years at Melbourne
Uni where I had spent time atJohns Hopkins doing medical
informatics.
And it's so interesting how thethreads kind of are interwoven,

(05:56):
because my friends there, henryand Sonia, both did their
undergrads at Georgia Tech atSteve Potter's lab.
Oh, wow.
And I told them what I was doing.
They're like hey, we did that60 years back or something like
that.
You can connect with StevePotter.
So that's how I ended upconnecting with him and his

(06:17):
advice to us was if you're goingto do this, you can't be purely
just on the tech side and justrely on somebody doing the
wetwear.
You got to own the entire stack, you got to do everything clean
the wetwear and so once I heardabout that, I was like, oh,
okay, I guess I'll just have togo set up a wet lab.
I'll find some space.
And I was like how can it be?

(06:38):
Like Melbourne, the Biotech Hubof Australia, I start calling
around and spoke to MelbourneUni.
Floria was like actually, we'reout of space.
And I called up Biotech 21 outof space.
And I was like, oh, my god, whohas space?
And I spoke to Weehai and theysaid well, there is actually a
space, but you'll need to takeup like 2,000 square meters or

(06:59):
something like that up in LaTrobe, la Trobe Uni, or
something like that.
Anyway, at that point in time Ialmost gave up.
I was like, oh, this is toohard, until I came across you
guys and I started speaking toyou, but then the container was
too small and there was notenough space, and I still
remember your auto club wasactually in the rice cooker or
something like that.

I still have that thing, and sitting in storage.

And we're not for the fact that we need a
rodent culture Because westarted out doing primary rodent
cultures.
We might actually have doneactually, no, we might not have
done it as well, because if wedidn't, if we had to do stem
cells, we would need a PC2anyway and that was only PC1.
So that led me down tocontacting the Alfred, right,

(07:41):
and somehow along the line,somebody you know is just
referral by referral, referralthey said hey, you know, there's
a new lab being built by Monashof the Alfred.
You should talk to TerryO'Brien.
So I did and there was a spaceand we negotiated something and
that was where we kind of gotstarted.
So, yeah, that was like what?
2019 or so, yeah.

So what I mean just to double loop back on that
, like can you tell us a bitabout the journey, like how did
you go from you know studyinglike medicine and obviously
going over to Johns Hopkins, tothen flipping the switch and
jumping in a sea of corticallabs?
Like what was the, what was themoment where you're like I'm
going to pursue this instead of?

oh, this goes back even further.
So I graduated medicine in 2012and in 2011, I went back to to
Johns Hopkins for my clinicalelective.
So I did my research year.
That was in 2008.
And boy was that a real weirdtime.
Because you know, it was postGFC in Baltimore, which is like

(08:43):
the roughest, one of theroughest cities in America.
And I still remember, you know,taking the bus down from New
York and I'm like whoa, allthese houses have no doors,
they're all boarded up, kind ofthing.
And then I came back in 2011 todo my clinical elective and it
was about that time that Icaught up with a mate of mine,
si Hormé.
So Si was working as adeveloper evangelist for

(09:08):
Microsoft and he was like, hey,you know Microsoft runs this
competition called the ImagineCup.
You know it's pretty sweetbecause you know they have the
competition all around the worldin different locations and if
you win they fly you there andall that stuff.
And he's like, given thatyou're doing medicine, you're
pretty good and pretty smartwith tech.
If you combine these two, youknow you likely have a winner in

(09:28):
your hands.
And I was like, yeah, maybeI'll think about it.
So that seated the thought inmy head and came back to, to
Melbourne.
I think this was in.
What was it 2012?
Yeah 2012 and I had just donelike a unit in global health in
pediatrics, and what wasinteresting was that I had

(09:52):
learned that pneumonia childhoodpneumonia, and I think it still
is the single largest killer ofchildren under the age of five
in the developing world.
And now in childhood pneumonia,it's actually very easy to treat
if you can differentiatebetween two types, right, the
two types of pneumonia there'sviral pneumonia and there's

(10:13):
bacterial pneumonia.
Viral pneumonia you don't doanything, you just, you know
supportive care, oxygen and soforth.
It's the bacterial one that youreally need to give antibiotics
, because if you don't A, youknow they're 68 and secondarily,
they also get sepsis and theydie.
Identifying, you know, theearly symptoms of pneumonia and
getting care was quite critical,but that was really poorly done

(10:34):
, because one of the ways we doit and I still think this is the
case is the most sensitivemeasure is respiratory rate, and
we, what we do is we watch therise and fall of a child's chest
within 15 seconds and countthat and multiply by four, which
is prone to a lot of error, asyou can imagine, because kids

(10:56):
are like twitching everywhereand doing all sorts of stuff.
So the other way that we do isalso using you are using a
stethoscope, so you plug yourstethoscope in, you listen, you
count inhalation, exhalationtime, it and so forth.
But that's a hassle because, ayou've got to recognize what a
breath sound is.
B you've got to take out thestethoscope and all that stuff.

(11:17):
And then it occurred to me whatif we could remove what's this
interobserver reliability issuewhere somebody might hear a
sound and classify it as abreath in and a different person
will hear it as a breath out orso forth?
And then that causes thediscrepancy in picking up child
pneumonia by building a digitalstethoscope that you could plug
into your smartphone like areally low cost one.

(11:39):
And I was like you know whatthat could work.
So I spoke to Professor JimBlack, who was a lecturer at the
I don't know if they stillexist the Nostal Institute of
Global Health at Melbourne Unipitched him the idea.
He's like this is really good,why don't you go ahead and try
and do this?
So I went back home.
I got an old stethoscope not myfancy Lippman 3M one and I

(12:04):
snipped off the head, pulled thesnipped off the tube, pulled
out the head, saw how it was alldone and then got, went to
J-Car, picked up an electricmicrophone, soldered the wires
into a stereo jack, threaded itthrough to through the tubing
and shoved it at the head of thebase of the head of the

(12:25):
stethoscope, put it into thephone and record it while I was
on my chest and I was like oh my.
God, I can't actually hear myheart sounds and record it.

You are a maker at heart.

Yeah.
So and then I did the lungsounds.
I was like, oh wow, I canactually also hear lung sounds.
So that was the genesis of thewhole thing.
I came back to Jim.
I was like, hey, I got this.
Would you like to be a mentorat this ImagineCup stuff?
And he's like sure.
So I cobbled together a team.
Andrew Lin, who's my co-founderat Clinic Cloud, joined in the

(12:55):
initial team.
Kim Ramchin, who's a made fromhigh school, was like a Matts
and Limpy genius.
He was supposed to be doing thealgorithms.
And then Master Solay, he wasalso a computer science like,
algorithm algorithms person.
So we banded together.
We named ourselves TeamStethoscope Cloud because we
were like what's hot in techright now?
Oh.
Microsoft has this Azure thing.

(13:16):
Let's just have a cloudcomponent where you beam the
recording up into the cloud andall that stuff.
So, anyway, we did that,pitched in the Australian
competition and we won theAustralian competition.
So we got to representAustralia.
Unfortunately, however, forthat year we were the final
competition.
The global competition washosted in Sydney.

(13:36):
So instead of a trip somewhereexotic.
We just went up to Sydney andunfortunately we didn't actually
win that.
We came third in the globalcompetition.
But what was interesting in2012 was that Microsoft decided
to put an extra thing in for theImagineCup, which was the
grants.
So and I think this wasactually more important than

(13:58):
actually winning the thing wherethey invited participants who
made it to top three to submitgrant ideas.
And we submitted a grant ideawhere we were going to take the
Stethoscope and collect a wholebunch of data to see if we could
train a machine learningalgorithm to classify lung
sounds for the respiratory ratestuff.
And we actually won $75,000 fromMicrosoft, so took that money,

(14:22):
put it into a research programat the Royal Children's Hospital
and I think it was a what wasit?
I hired two research nurses.
I had a, a maid of mine who wasa pediatric registrar help out
as well and did the experiment.

(14:44):
So we we ran the trial, wecollected close to what like 200
participants recording lungsounds, and then the idea also
evolved a little bit where wewere like not just doing
respiratory rate, but we wantedto figure out how, if we could
build an algorithm that couldclassify wheeze or determine

(15:04):
like severity of wheeze so thatthe emergency department would
not be clogged up with people,kids with asthma Right, because
if you go to the RCH you go toany pediatric hospital like.
I guarantee 80% of the timeit's somebody with asthma.
Just undiagnosed asthma orUndiagnosed asthma or diagnosed

(15:24):
asthma that you know, parentscan't tell whether it's getting
better or worse because, they'vedone the protocol and they
don't see any change or theycan't hear the change right
Because they don't.
They can't, they're not trainedto listen with a stethoscope.
So that was the idea behind it.
So, anyway, ended up doing that, I did my internship year at
Frankston.
At the same time I was hackingand building back end mobile app

(15:46):
services and all that stuff tosupport that research program.
And then midway through 2014,or was it?
No, it's not 2014 or somethinglike that my, actually no.
Midway through my internship, myresearch nurse came up to me
and said hey, do you know thatI'm getting a lot of questions
from parents about how much thisdinky little stethoscope device
costs and where can they buy it?

(16:08):
And I said why would they beasking something like this?
And she said I'll let me askthe parents.
So she asked one of the parents, and one was like a parent of a
child with a congenital heartdefect and they lived all the
way up in Mildura.
So they were like well, I haveto fly all the way down to
Melbourne for a 30 minuteconsole.
That's like a three hour trip.
If only there was a way for meto do this in Mildura and not

(16:30):
have to come down.
This was way beforetelemedicine became a thing.
So you know, with that in 2014,after I finished my internship
and did a couple of months ofresidency, I told Andrew about
this and I said I think there'ssomething in this, let's go
check it out.
So we both left our jobs.
I mean put it on hold.
He was at Bain, I was doing mysecond year residency.

(16:53):
We put it on hold, get thewhatever savings we had, packed
our bags and went over to SanFrancisco with a very simple
prototype.
You know met a whole bunch ofpeople.
You know pitched the idea.
They gave a couple of pointersand one person I can't remember
who he was was like saying youreally should go to South by
Southwest.
Mm.

(17:14):
So we looked at it as like howmuch does it cost?
Oh, it's within budget.
So we just bought a ticket fromSan Francisco to Austin.
Wow.
And we just, like you know, justwalked around randomly and we
were told the advice was don'tgo for any of the day sessions
because all the deals happen atnight.
So we didn't do any of the daysessions, we slept, and then it
was good for us because it waslike Australian, like nighttime

(17:35):
or whatever, and then we justwent to all the parties, came
across a whole bunch of people,that's where everything happens.
It's where it all happens andpeople were like, oh, this is
really cool here, have a checkas an angel investor.
And we collected a small sortof angel round, came back to
Australia, worked on a prototypeand then we launched a
Kickstarter or a self-starter,whatever they call it now a

(17:56):
Kickstarter thing and thatcaught the attention of Ping Un
Ventures and Tencent Venturesand they flew.
They were like they told Andrew, can you come up, because
Andrew was CEO and I was CTO atthat time Came up, he pitched
the thing.
They're like we really likethis, we want to fund it, so
sort of the bad.
He's like yeah, we got fivemillion US to make this a thing,

(18:17):
so that was a lot of moneyactually back in 2014, 2015.
Yeah.
So, anyway, that was like thestart of my first journey with
Clinic Cloud took the money,build a product.
What was it?
We took the prototype, made itinto an FDA C certified device

(18:41):
and started selling to Best Buy,Amazon and so forth.
But the problem with one of thethings when you do take venture
capital money is that you'resomewhat, you know, on the hook
because you have to grow.
And in order to raise the nexttrench, you have to grow two
acts or three acts, whatever.
We had massive difficultiesscaling the business side of the

(19:04):
equation because what we foundand this was really the case in
2014 before COVID and so forthwas that MedTech was not a thing
and in many cases, a lot ofpeople don't think about buying
medical devices if they'rehealthy.
It's like being in Vegas sayingyou should buy an umbrella,
like why it doesn't rain here.
I was like, yeah, on theopportunity, it does.

(19:25):
So we couldn't really scale thebusiness, tried many different
ways of doing it.
Eventually, you know, whatreally sank us was in 2016, when
Trump got elected and we hadsecured a really big deal with
the VA to supply them with thethermometers and the set of

(19:46):
scopes, and that was our firstbig break.
But then, along with the VA andeveryone else, trump comes in
and says I'm gonna rip upObamacare and I'm gonna bring in
Trumpcare.
And so everyone who had adiscretionary expenditure budget
, which was where the money wascoming to pay for our stuff,
suddenly looked around and saidwe shouldn't spend the money
that we've allocated for thisdiscretionary expenditure

(20:09):
because we have no idea wherethe money is gonna come from
next year.
So that was put in hold and Isaid that's okay, maybe Trump
will come in and he'll figureout how to get Trumpcare through
.
Long debates ensues in monthsand months of negotiation.
He proposes Trumpcare like, Ithink, mid 2017 or so, gets
rejected and then we're back tosquare one.

(20:29):
So we literally was justburning cash for the most of
2017.
And then in 2018, I had enoughof it.
I was like you know what, wecan't keep doing this, we'll
close down.
So, like what happened wasAndrew stepped away from the
thing.
I inherited their business.
I had no idea where to go withit and it was literally well, we
got some money left.
Let's just throw everything atthe wall and see what sticks.

(20:50):
And in 2017, Demis Hussab is thewriter of paper in Neuron that
I read in 2018, where he wascalling for machine learning and
AI people to re-engage withneuroscience.
And I did exactly that.
I was like this is really cooland we're trying to do something
in the AI machine learningspace.
I went over to the Flory and Istarted speaking to people there

(21:12):
and I said, hey, I'm a machinelearning AI guy.
Well, also from the doctor,tell me what's exciting in your
world.
And he's like, oh well, we havethis device where we can grow
in neurons and we can see theactivity on them and we can give
them a little bit of a jolt.
I was thinking about it Likethese neurons would have turned
into a brain and we know thatthe brain computes.
You now have a read and writeaccess into these neurons.
Why hasn't anyone gotten themto compute yet?

(21:34):
Did the background researchcame across Steve Potter's paper
once or twice, but nothing eversince then and I said, well,
this sounds pretty cool, maybewe should look into this.
So started working on this in2018.
That's how I got in contact.
There was a lot of groundwork,a lot of infrastructure that
needed to be done and thegroundwork infrastructure I

(21:56):
think that a lot of peopleforget about, and it was a big
portion of what I was doing in2018 was trying to figure out
how do I do this, what do I need, who do I need, kind of thing,
and came across Biquizidive andwas trying to figure out this
lab space thing.
Came across the lab at Monash,at the Alfred run by Monash, and

(22:24):
things were going OK until Irealized we were running out of
money and we would actually diein.
I think it was like mid-2019.
But that was OK because in theearly part of 2019, it was March
or so I was introduced to NickySkavak, who was the partner at
BlackBid, and he made on my medBenjamin, who's an advisor now

(22:49):
of the company.
I was like you should reallypitch your idea to Nicky.
So I was like, ok, fine, thisis so crazy, nobody would fund
this anyway.
Because what happened isinteresting, because in 2018,
while I was taking a break and Iwas just traveling around with
mates from Hopkins, I was inHong Kong and a friend of mine
was like, do you want tointroduce to John Tam from

(23:11):
Horizons?
And I was like sure, I'll talkto him about what we're doing.
So I pitched him the idea andhe's like this is really cool
stuff, but it's a bit too crazyfor us.
And, to his credit and to hispoint as well, it was too crazy,
like all I had was I had ahunch.
I think this could work if Ijust had enough funding to get

(23:33):
us across the line.
The rest is history, becauseHorizons came back in and now
they're actually our biggestbacker.
So they never say no to anymeetings.
You never know what happensdown the line.
So anyway, so where am I?
So in 2019, we actually ran outof money and we had signed the

(23:59):
term sheet and actually hadnegotiated everything with
Blackbird, but we needed to getsign off from all these people
to agree to taking this newtrench of funding and to, I
guess, recap and reincorporatethe company as cortical labs,

(24:20):
and we actually went intonegative and I had to put in my
own cash and it was so hairy.
At the point I was like I'm 40kin debt with this stuff Maxing
out credit cards.
Maxing out credit cards and mybank account was getting close
to the bottom part of.

Wouldn't be a startup if that wasn't a part of
the journey.
Oh, yeah.

And then I still remember taking walks around,
laps around Falkner Park becausethat was where I was living and
close by to the Alfred with mymate at Hooper and I was like
mate, I don't know if I can keepdoing this, I think I might
just have to throw in the towel.
He's like no, you're so close,you know, keep pushing, try to
talk to Nicky, and stuff likethat.

(24:58):
And I still remember he was.
I think they came down, theywanted to see the lab and so we
did the meeting and I think thiswas like in September or so
August or September, I can'tremember exactly now In 2019,
they came to see the lab.
They were really impressed withwhat they saw and I was like I
really need your help nowbecause I'm trying to keep this
afloat.

(25:18):
I have one signature left andthat was 10 cent.
And I said can you just help usout and just wire us the cash,
because I can't keep supportingthis?
And to their credit, they saidlook, we'll see what we can do.
They worked it out.
They said look, it seems likeyou have a verbal agreement
anyway with those guys that it'sonly a matter of time before
they sign it off.
We'll wire you the cashtomorrow.

(25:39):
What a relief, what a massiverelief.

I don't know.

It always comes too close to the wire with
Quarticle Labs.
It also happened to us lastyear where in December and the
reason why we were delaying it,we couldn't get started with
CoLab was Wasn't it a signatureyou were waiting on right?
It was also the same damnsignature that was holding us

(26:07):
back and we actually also wentto negative territory.
And this time I was supportingit again with my own cash, much
higher this time because therewas higher load, and one of our
investors, Atlanta Daniels fromRadar Ventures, was so great
because I said, look, I goteverything.
I just need this one signatureto come through.

(26:28):
Everyone else can't wire itbefore because of different
requirements and so forth.
Can you just swap around theordering so that you can buy the
shares first, wire us the cashwhile the money is in the bank,
I can work through to get therest of it done?
And she was so cool by saying,yep, no problems, I'll help you

(26:49):
out with that.

It's kind of one of those.
It makes me feel like there's afew themes there like leaning
in with curiosity and then alsoyou don't know if you don't ask,
and having the courage to belike, look, yeah, this is
probably going to happen.
It's with relative confidence,but can you help us out?
And sometimes thoseunconventional things and being

(27:12):
real and honest can kind of Idon't know.
It seems like it definitelyhelps and goes a long way.
We've seen that happen with usas well 100%.

I think the most important, one of the most not
the most, but quite possiblyclose to the most attributes a
founder needs to have is to haveno shame, like absolutely grow
the thickest, hardest, mostcallous skin that you can get.
And just ask I mean, I get alot of practice from Tinder, so
it's all right.
You get rejected a lot of times, but you don't know if you

(27:40):
don't do it right.
So don't be afraid to go up tosomebody and say, hey, I need
your help about this.
Or trolling through LinkedIngoing.
I need an expert in this space.
Can.
I just find somebody on LinkedInKind of thing.

And a lot of the time people at.
First of all, I must say Ithought you were going to say no
shame and it's the Javierna'sthat you're wearing.
That's no shame, Everyone'swearing Birkenstocks, it's 2024,
but no, I've got time for theJavies.
But yeah, I think the no shameand just asking, I think people
are more than willing to help,like people want to help.
If you give, if there's enoughmeaning and intent and it's a

(28:12):
genuine call out and askingsomeone.
It's not just like a copy paste, some narrative, like
everything that we've sort ofbuilt along the way has been
through that as well, being likehey, just like you know you're
someone whose opinion we value,could you provide us some
feedback, some thoughts?
And more often than not, peoplego over and above.

Yeah, 100%.
I think the key difference isasking for help in an advisory
way.

Like.

I don't know enough of this.
Can you tell me about this sothat I can go off and go do it?

It's not like I don't want to build a
codependency, it's just likethere's this period, there's
this bit that I feel like I knowthat you have a wealth of
knowledge and you could save ussix months of time and that also
gets them involved and theyfeel like they're important and
meaningful contribution to whatyou're sort of doing.

Correct right and it reinforces identity for a
person.
Right.
Let's say a person spent theirentire lifetime working on a
particular thing, and here'ssomebody coming up to them and
saying I need your advice aboutthis.
That's actually really like apositive identity thing for the
person giving the advice.
What people don't like, however, is I need you to do this for

(29:21):
me.
Yeah right.
Which is almost like 90% ofwhen somebody's like hey, I need
a favor.
You know it depends on thefavor right.

Yeah, yeah, yeah.

Right, and I think a lot of people walk away
from that going, oh, this guy'san asshole, he didn't help me
out and you're like, yeah, butthat's not because he, you know,
that's not because you'reasking for his advice for you to
go.
Then take that advice and go dosomething with it.
You're asking him to do thisstuff and people don't have time
for that right.

Can I ask you a favor?
Depends.
Okay, For those who don't knowwho cortical labs are like.
What is cortical labs and likewhat's your mission?
What are you aiming towards?

So it's a good question because it keeps it
evolves over time.
And initially I started outcortical labs as a bit of a
curiosity I don't have storiesof personal family or whatever
mission.
It was more of a how come noone's doing this kind of thing?

(30:26):
So, very naively, I just wantedto build the biological
computer Because I was like,well, if intelligence is a form
of computation and if the brainis the organ that does
intelligence, then it must beable to do some sort of
computation.
They're all interlinked.

(30:46):
No one was doing it so I wantedto do it and I was kind of
naive in thinking how easy itwas going to be.
It turns out it's much morecomplicated.
But that's evolved over time andactually it's through the
advice of important people whotend to advise us that the
mission and the vision hasevolved over time.
So it's no longer purely justto build a biological computer.

(31:10):
It's more about what do we havehere and how do we use it.
Is it even still a biologicalcomputer?
The more we learn about whatthese neurons are doing, the
less we are actually nowconvinced that they are a
computer per se.
And really I guess what we'vesettled on here now, as in 2024,

(31:32):
is that we want to be buildingtooling to help scientists in
this field explore theelectrophysiology of these
neurons and what it means toactually have closed loop
systems.
And what I mean by closed loopis that currently a lot of

(31:53):
electrophysiology is done in anopen loop, where stimulus is
given to the neurons and that'sit.
That's very unnatural.

That's not how a complex adaptive system works
Exactly.
It's feedback loops uponfeedback loops.

Exactly, and what we've shown with our work
is that you need that closedloop for more interesting
behaviors to emerge, likeplaying pong or the jumpy
dinosaur game.
And so that's now been ourmission, and the reason why
we're developing this CL1product here is to make it more

(32:27):
accessible for people to startexperimenting with neurons, be
it either they have a laboratoryand they can grow the neurons
for the system, or we grow theneurons for them in a remote
sort of cloud setting, but allowthem to experiment with code,
with systems, to probe and toget the results back and analyze

(32:48):
it.
So that's really our goal now,which is, I guess, in a sense,
to still try to build abiological computer, but now
with a lot more meat around thebone to say what exactly is this
biological computer?
How are we going to programthis, all the nitty-gritty,
specifics and details of what'sgoing into it.

Right, and just to riff off that, I've heard you
before say that you're kind oflooking at trying to build the
next navidia.
So that's the hardware side ofthings.
But what you just sort ofexpressed there is that you're
also looking at how do weco-create or collaborate with
others to make the software sideof things or the wetware side
of things, let's say so the waywe've started to define what a

(33:32):
biological computer is is thattake your traditional computer.

It's got two components hardware fixed at the
factory and what gives it itsspecialness?
Special sources of softwarewhere you can make it do a lot
of dynamic things.
Now a biological computer nowhas a third vertex.
The wetware the wetware and younow have more combinatorials
between your software and yourwetware alongside your fixed

(33:57):
hardware.

Because it's evolutionary, that's the.

Thing.

It's like when you're dealing with neurons or
something that's alive.
It's a dynamic process.
It's not fixed, it's not static.

Correct, and it's in real time, it's analog.
That's the key bit, which isthat analog is the real world
and that digital is really anabstraction of that.
And so, as you said, it'sevolutionary, it's happening in
real time, kind of thing.

(34:28):
And not only that the neurons,all these cells, all these
living organisms around us rightnow, we're a product of
billions of years of evolution.
Each organism has well, eachorganism exists today because
it's been able to survive indepth to that particular
ecological niche and with thatalso it inherits what I call,

(34:54):
personally, genetic priors, forinstance, a genetic prio is
another way of talking aboutinstinct.
How does a spider know how toweave a web if it's never seen
one?
It's because of the geneticprio, it's encoded in its genes.

It's the wetware.

Correct, it's the wetware that has learned
that.

And what's so fascinating, when I hear you say
that it's also then like whathappens if we do an octopus
brain Correct, or what happensif it's a whale brain.
How does this impact the way inwhich the neurons respond to
their environment?
Because, in a weird way as well, you're abstracting them from
the original context thatthey're from and putting them
into a new one, but you're stillhaving these priors which you

(35:37):
know are going to play an impacton it within this new space,
and it's just fascinating, it'sso interesting seeing what can
happen, right Like I remember wedon't 100% have to go there,
but there was neurons thatseemed to almost be reacting Of
the cardiac ones, yes.
That was fascinating.

Yeah Well, I mean, we know for a fact that
mouse neurons don't play a pongas well as human neurons.
That's something we've observed, so it'd be very interesting to
see what Kephalopod, like anoctopus one, does.
Having said that, octopusneurons really different right.
Apparently, even temperaturewill change this epigenetic like
gene expression.

(36:17):
Wow.

Didn't someone want to bankroll you to do
something?

Yeah, somebody did, but then I think this
person was a crypto bro withADHD, and so he kind of got
distracted by something else.
Yeah, so it's actually quitesad that we don't have more
discoveries in this space.

Well, thinking of just like, how can we simulate
other minds and will that helpus better understand or
potentially interact with them?
Yeah, it's just a fascinatingyeah.

And I imagine as well with as technologies
improve across, like within thisspace, not just with what
you're doing in the lab, likewhen you start messing around or
experimenting with organoids orthree dimensional structures,
probably, I imagine just keeplearning new and new things
about how these neurons work andbehave in a group.
What do you call cluster?

Yeah, yeah, I think that's definitely going to
be the case.
But for me personally and youknow, something I've set the
team to look into we think thatactually the bottleneck isn't in
the size of the neurons, right?
Because if you think about it,if fly only has 100,000 neurons,
we're going to like 500,000 to800,000, like 500,000 to a
million neurons.
It's kind of hard to pinpointexactly, but we certainly have

(37:29):
more neurons on the dish thanwhat a fly has.
You know, a fly can navigatethe world, it can avoid being
killed, find prey sorry, findpoo and reproduce right or a
thing that makes a fly happy.
Yeah, I mean, flies are actuallyamazingly like powerful for the
amount of energy that they use.

(37:50):
Imagine a drone with a flybrain.
Right, it could probably justkeep flying forever.

Yeah, it wouldn't get shot down either, it
wouldn't get shot down.
You know all of these things,but then what that makes me
curious about is the geneticpriors that you're speaking
about, so correct me if I'mwrong.
I don't want to be anticipatingsomething that you might be
looking at exploring, but it'salmost like at what context
would a certain type of brain beapplicable for a certain type

(38:18):
of technological device?
Is that kind of a direction oran area that are looking at,
leaning into the future?
Yeah, definitely.

I think that's something to look into.
I mean, the most frequentquestion is, like you know, if
you've got a pro gamer's cellsand play this against some
random person, would the progamer cells actually outperform
them in particular games, right?
We don't know.

Got to get that tetris, the kid who finished
Tetris' brain, onto a DSC.
What happens?

Exactly Get some , you know, e-spots champions,
you know, get them in the gameof Doom or something like that.
I don't know, but you do raisea good question right About that
.
However, there is also a schoolof thought in machine learning
and AI about the embodimenttheory, that you cannot get AGI

(39:09):
without embodiment andintelligence is actually shaped
by the particular embodimentthese systems are in.
Which is to say that if you hadhuman brains in the mouse body,
you wouldn't get the sameintelligence, and you might
actually get better intelligencewith the past brain and like

(39:29):
human embodiment.
That's it.

Because I have heard, like is it the four E?
Cognitive science?
Like the embodied enacted, likethere's all of these things.
I can't remember them off thetop of my head and I haven't
been able to get chat GBT upquick enough to get it.
It feels like, from what I cantell by reading the literature
on cognitive science, that is areally big element that when
you're, and because you'reobviously taking them out of
that context, like what does andas you're saying as well, like

(39:56):
there's might be up to a millionneurons, right, so there is
something that it is like topotentially be one of these
dishes.

Yeah, and this is the really cool thing about
like the, the, the, thebio-ledge computing system,
right, If you want to change it,the embodiment, you can do it
in software.
The software gives theseneurons embodiment because now
they have the ability to to takein sensory information from a
virtual world, process it andthen action on that simulation

(40:29):
world based on how you'veencoded the embodiment.
So, in the case of Pong,essentially a bat.

Right, so you, yeah, this and this is your body
part.
And then because, becauseplease go on how, how you manage
to make it, I guess, how youtaught it to play Pong through
the reinforced feedback loops.
Yeah, That'd be really cool tohear.

Yeah, so essentially the where we got it
to play Pong was we use thetheory developed by Professor
Carl Friston at UCL called thephantry principle, or that was
one of our guiding principles.
There might be other things.
I mean, we also put somethought into heavy emplasticity
and learning and so forth, andwhat we said was look, you know,
we didn't have the ability touse dopamine because we had a

(41:14):
purely electrophysiologicalsystem, no chemical reward and
so forth.
And we said, well, based onthis phantry principle, which is
very complicated I'm going totry and boil it down to the
brain Instead of the traditionalway of thinking where photons
hit receptors, gets processedand an action then happens, it's

(41:35):
actually the inverse, wherewe're actually predicting ahead
of time, using a generativemodel of what the world is and
the senses are actuallyproviding us input, where we
then determine how closely ourgenerative model is reflecting
the real world.
And the whole point ofbiological systems is to get

(41:56):
really good at predicting theworld.
And, if you think about it,that's a prerequisite for
survival, right?
If you predict the worldaccurately, a you can avoid
being prey and find prey better.

What goes to those instincts that you were talking
about before?

Exactly right.
So somewhere in thesebiological systems there is a
genetic prior, the veryimportant genetic prior for
intelligence, which is trying tocreate a generative model of
the world and to accuratelypredict that world, based on the
senses and also our motorfunctions in changing them.
So if those are the propertiesthat we are looking for, we

(42:34):
figured that if the desirableoutcome was to move the paddle
to hit the ball, the neuronswould need a positive signal for
that, and the positive signalwas in the form of a sine wave,
and if they missed the ball, wewould give them a random, noisy
signal.

(42:54):
Now, there is a reason for that.
If you took two signals, a sinewave and a white noise signal
of equal length, you canactually summarize the signal,
at least with a sine wave, intoone function.
Right, Because from one periodyou can predict n number of

(43:16):
periods going forward.
So that has what we call verylow information entropy.
And a random signal isunpredictable because from one
segment you can't predict nnumber of segments.
Right, so that has a very highinformation entropy.
So entropy is a really importantmeasure, right, so it's used in
everything in computing,particularly in things like file

(43:38):
compression.
Right, Like your zip file.
How does it know how tocompress something that is like
a meg down to like sometimeslike 10 kilobytes?
Well, it realizes that thereare sequences that repeat over
and over again.
So, rather than encoding all ofthese things like, the simplest
way I can explain to you islike saying, if I had the
sequence of ABC, ABC, ABC,rather than writing out you know

(43:59):
the characters, nine charactersall the way through, I can
summarize it by saying take ABCmultiplied by three in a
sequence right?
So that is a compressible pieceof information and therefore a
low information entropy.
So, yeah, it turns out thatthese neurons are seeking low
information, sorry, seeking lowinformation entropy targets.

(44:22):
So by being low entropy, theyare predicting the world more
effectively.

So if I was going to interpret that, what you're
essentially saying is that theylike an ordered signal and that
kind of and that allows them tolike coherently communicate as a
collective.
So you might get a whole bunchof random signals, but if you
start picking up on a concert ofsine waves, then it knows okay,
great, that's when I you knowI'm firing or something.

Yeah, exactly, and we haven't done any formal
thing around it, but it's kindof intuitive.
You think about that and haveimpassivity, which says that if
two neurons fire together, theywire together.
Right, your chances of acoincidental event happening
increases if you actually have apredictable signal as opposed

(45:15):
to a random signal.

So what does a happy neuron look like?

Well, I don't know what ha like.
I mean happy is a, is a is avery used it very yeah.
Anthropomorphic term but we doknow what the healthy looking
one is Right and healthy cellsare ones that have good
electrical activity, that has toadhere to the surface of the,
of the chip and and have verygood synaptic connections.

(45:44):
So morphologically we can, wecan tell what they, what a
healthy cell is happy?
Not quite sure.

So when you, when you give them a sine wave, for
example, is there like anobservable behavior, is the
behavior firing versus therandom noise?

They start to coordinate their firing patterns
.
Cool yeah.

So they, they sync up, which to me it makes sense
Like, given that I know as welllike neurons also sync up
through like multiple things,like they pick up on vibration,
they pick up on likeelectromagnetism.
There's so many differentthings and I'm wondering if,
like, yeah, so fascinatingthinking about how, how all of

(46:29):
this sort of works, and imaginebeing able to ask them like, how
are you feeling today?

I mean, maybe one day we could do that At the
end of the day.
You know, if you think about it, it's kind of what we're
working towards as well.
Right, Can we build thesethings where you can actually
start probing them?
And they give you responsesback and then you can teach them
to do things Right, because themoment you can start teaching
them to do things, you now havea programmable system.

On that note, I remember chatting with you would
have been maybe about sixmonths ago.
You actually had some prettyinteresting progress on that
front with programming ofneurons, something that might
have been a I don't know if itwas a world first, or at least
it was the first time that Iheard about it.
Yep.
Do you remember thatconversation about the encoding
information into neurons?

Oh right, yeah, so we've been looking at.
We've been looking at how doyou, how do you take digital
information right, which is?
Take, for instance, a JPEGimage.
Jpeg image is your width times,height times three, right,

(47:36):
because you have RGB channelsand your RGB channels are zero
to 255.
Let's just say you know that'show you encode information thing
.
So all digital information ispresented in the form of a
vector, matrix or a tensor.
The JPEG is a tensor of widthtimes, height times three, of
numbers that go from zero to 255.

(47:56):
People have said why don't youjust connect these neurons to a
neural net and see what happens?
I'm like, yeah, I could.
But then again, what the helldoes a tensor mean to these
neurons, and let alone a tensorthat is just these numbers?
Because that's not how theneurons work, that's certainly
not how a vision works, right.

It's like trying to play a vinyl, like on, like
you don't just plug yourheadphones into a vinyl, or
something right.
It's like a different way oflike different operating system
100%.

I mean, it's just.
It's just.
It's just not how the physicalworld works, right?
So if, like, take your examplewith vinyl, vinyl is essentially
all the grooves and you havethat little head and you know,
every time though it hits agroove it vibrates a little bit.
On the other hand, a CD is justbits, right, like you know say

(48:46):
so, you hear these terms like 16bit, 44, 100 hertz, right.
What that actually means isthat when a microphone or
something that is picking up therecording that is put into the
CD, it's getting a number thatis 16 bits in integer, right.
So it can go from zero to, like, I think, 16,384 number or

(49:07):
something like that, or maybemore.
That's probably eight bit.
So double that number, so32,000 or so, and each of those
numbers represents some point inthe sound the audio spectrum.
In the spectrum, and the 44, 100hertz is how fast this
microphone is splitting up time.

(49:29):
So and this is a veryinteresting thing, because time
is a bit of a weird dimension incomputing space it doesn't
really have a concept of that.
But let's just say you have 44,100 hertz.
It means you take one secondand you're slicing the thing
into windows and at the end ofone second you're getting at
44,100 windows of a signal, andthat's just how audio works

(49:51):
right.
And so when you take 16 bitaudio, 44, 100 hertz, and you
feed it back into what we call aDAC, a digital analog converter
, it's like an audio interface.

To keep the metaphor going.

Correct, yeah, you take that and that will take
each of those points right andplay it back.
And it takes that and itvibrates a membrane and that's
how you get audio back fromrecording to that.
Now, if you took those numbersand you put it on a vinyl

(50:23):
gramophone kind of thing, itjust has no idea what to do with
it.
Same thing with the neurons.
Right, we can get thesematrices in tensors.
You put it in neurons, like Ihave no idea because neurons
it's got no context for it.
Exactly, neurons speak the samelingo like a vinyl record, that
it does it as a time encodedthing.
And so what we've been workingon and we've had some

(50:46):
breakthroughs here is trying toconvert the digital
representation of, say, mnist,the Henrydon digits, into spike
trains that we can feed intoneurons.
Now we're building new hardware, so once the new hardware is
online, we can then starttraining the neurons by feeding
these spike trains in and see ifthey can recognize the Henrydon

(51:07):
digits.

So in a way, if I was going to interpret that,
it's like you're trying tocreate a language of coherence
that they will be able tounderstand, like that audio
interface layer.

Yep, it's the translational layer and this is
the really exciting thing aswell, because it has not just
impact in our field, where theend, like one of the North Star
goals, was to still try to getto the play doom, but real doom,
like pixel-based doom, not likethe top-down view of doom, the

(51:40):
pixel-based one.
This is also really importantthe brain-computer interface
space.
Right now, brain-computerinterfaces are limited to read.
Only.
The only real right-capablebrain-computer interface today
that we have is the cochlearimplant.

And just for context, because as soon as I
hear right I just think oh gosh,brainwashing.
But I assume that your meaningmore so like hypothetically
someone could have epilepsy andyou could send a right code to
sort of essentially like a stopprogram for I mean, we already
have that today, but more so,not really brainwashing, but

(52:25):
think about it.

How could it be if you could just think a
thought.

Learn French.

Yeah, or you know.
Essentially what we have is oursmartphones today.
Right?
Someone like wants to show youwhat the view of the world is.
They take a photo, they send itto another person's phone and
that person sees it on Instagramas what the representation was.
But imagine if you could encodeinformation.

It's just removing a layer, so it's almost like
literal telepathy Telepathy.

Wouldn't it be cool if I could see something
from the same perspective?

Yeah, I mean that actually might solve a lot of.

I think most of our problems in humanity come
from back communication and theinability to perceive another's
perspective, or from a holisticperspective, rather than an
individual, isolated atomicreductionist sort of randomly in
the world.

It is a little bit black-mirror-ish.
We don't know where thistechnology could go.
But what if you could actuallytap into the amygdala right and
read the activity of someone'semotions?
Like you know, somebody couldbe struck with grief.
You just don't know about thisand it's encoded and you're
feeling the same thing becauseit's transmitted of the wire.

(53:37):
And now there's a decoder thatwe can then decode it back into
what you would feel the samething in your brain.

That's fascinating to think that you could feel
someone else's emotions.

Right.
I mean those are kind of thethings that an encoding-decoding
bit for the neuron to digitalsystem, one they could unlock.

And I imagine that you could also potentially,
depending on the bandwidth ofknowledge transfer, you could
potentially like this is goingto go crazy.
You could potentially screensomeone's visual field and you
could potentially observe theirvisual field from another
location.
Yeah, theoretically.
That's yeah.

Theoretically, that could be a thing where, if
we could, you know, tap into thevisual cortex and we could
stream information back out andinto a digital system we could
see on the screen, or if I couldwrite back into your brain.
I could write what you saw intoyour brain and go oh my God,
I'm actually what's the word?

(54:37):
Vicariously living through aperson own eyes.

It's like a whole other Twitch platform right
there.

So fascinating, but so many ethical questions
100% which is the thing, right.

So there was a post that came up recently Tom
Oxley posted on LinkedIn aboutbringing computer interfaces.
And how do you think about it?
And I say it's very simple.
The simplest way to think aboutbringing computer interfaces is
that every great leap incomputing comes with a change in
human-computer interaction.
We had the terminal to beginwith.
The first big computingrevolution was the GUI.

(55:12):
That's how Windows became athing.
That's how Microsoft became themost valuable company on the
planet.
What happened for them to bedisrupted?
Touchscreens, the smartphone.
Apple won that war and I guessGoogle as well.
The question is what is the nextthing?
Some people think it's AR, somepeople think it's VR, whatever

(55:33):
right.
What we don't know is what's inbetween.
But we know that the end of theline is a brain-computer
interface Directed the brain.
So if you just think about thatand just go, you know what, I'm
not gonna bother with anythingand be in the middle and I'm
just gonna go straight for theend of the line.
If you own that technology, youhave a technology that is
indisruptible, very powerful,but also very scary kind of

(55:56):
control.

And so it's really heartwarming to see that you
guys are already writing paperson ethics.
That's gonna be my next thought.
Did you have an ethicist inresidence, don't you?

We don't really have an ethicist in residence,
but we do work very closely withpeople like Julie and
Sevillescu and so forth, who youknow.
At the end of the day, ethicsis and I keep saying this is a
conversation.
What I mean by that is thatethics, just like values, change
with time and with societiesright.

(56:31):
So, for instance I give you agood example would be IVF very
acceptable today, even I guessgovernments are now subsidizing
it.
But it was only just, I thinkwhat is it?
The 70s I think it was only 40years ago where it was looked on
as an ethical technology?

(56:51):
You know, we're meddling withsomething that God could only do
or something like that right,and a very religious base.
That's completely changedbecause of time, because of
society becoming more used to it.
We're seeing positive outcomescome from it happier families
and couples and so forth.
I mean we don't really talk somuch about a negative thing like

(57:14):
potentially screening forparticular traits or so forth
the Gallagher style.
But, as I said, that was one ofthe things that people thought
about and why it was not a goodthing to allow Other things,
societal things like gaymarriage right, tell that to
anybody in the 50s.
They'd be like get out of here.
That would never happen.

(57:35):
Values change, so ethics alsohave to change, and I think this
is the reason why we've engagedwith the biathesis, because we
do it as a way of saying what isthe current temperature today,
what are red lines that weshould and shouldn't cross?
But what are the things thatare informing for these red
lines, you know, and then, astime progresses, as people get

(57:57):
used to things, it's like theAlverton window you keep moving
it up.
You then sample it again.
You say what is the temperaturenow in the room?
What are acceptable andunacceptable things?
Right, because, for instance,what if and this is a very
interesting hypothetical thepathway to creating a cure for
things like dementia orAlzheimer's, dementia or
Parkinson's requires thecreation of potentially

(58:18):
semi-conscious organoids.
What is the risk?
What's the trade-off?
We really have conscious beings, or partially?
It depends on what yourconsciousness, your definition
of consciousness is, but I thinka mouse is pretty conscious.

Like right.

And we sacrifice a lot of mice in laboratories
to find cures for diseases.
Right, we as society havedeemed that an ethical use of
animals.

It is getting phased out, I will admit but I
mean you're actively looking atfinding ways in which you could
use organoids to address thatright Exactly.

But then there are these red lines that people
are saying what about aconscious organoid?

What about all the other humans that we?
Everything you're saying islike I agree, but it's almost
like.
To me, it feels like the layerthat we need to be working on is
the human-to-human interactionsand how we relate to one
another and then how we can bewise stewards of the technology
that we're co-creating.
And I think this is happening.
Yeah, as you said, there's aconversation that happens

(59:22):
adjacent to, but yeah, it'sstill a.

Because it's a very interesting ethical
question, right, like, let'sjust say you're very let's just
say with animal studies, right?
And you are very much againstthe use of animals for
experimentation and testing.
Yet, let's say, your bestfriend's grandfather is dying of
dementia a slow, horrible wantto watch death.

(59:47):
Who's more unethical now?
The person doing theexperiments to find a cure but
is killing these animals forthat, or you, preventing that
person from doing that work?
That could find a cure?
Right, but there's now, I guess, leaving that person to die the

(01:00:09):
death of dementia, right.
It's these very difficultquestions that need to be
thought out.

Samuel Wines (01:00:15):
And it's not a binary, yes, and, and I think
that's a really important thingwith this sort of stuff, but I
don't even know what a yeah,it's such a fascinating.
Like I feel like we couldliterally speak for hours just
on this topic.
Because if we are like, can Iimagine, like if you are
programming them and ask themhey, how are you today?
Like if that's something thatyou bake into it, just as like a

(01:00:37):
quick way to update.
Oh yeah, you know, maybe giveme some more media, but like I
can instantly see that peoplemight find that like that could
be a very big red flag.

Hon Weng Chong (01:00:47):
Oh, yeah, right.

But in a way, doing that could then lead to
the cells not suffering and notdying, because they can actually
communicate with you what theyneed or want.

Yeah, so, yeah, it's such a.
I mean, how do we know thatwhat we're doing in the
laboratory, like all around theworld, many laboratories, is not
even causing suffering forthese neurons?
We don't know, right, like, forinstance, maybe we grow them
and we just leave them therewith no stimulus?
That could be a bad thing, wedon't know.
So you know, it opens up, see,like a whole bunch of counter

(01:01:19):
worms, and this is what sciencedoes, right?
Science finds the discovery andthen now the world has to look
at it and go, well, what are wegonna do with this?
Right?
Yeah, and this was, like, Iguess, the big whole thing about
, like the Oppenheimer, you know, he was like yes, let's go do
this.
And then later he was likeactually I didn't really think
it was that well, picture thisyeah.

Whoops.
Yeah, at least this isn'tdropping a nuclear bomb, but in
a way it's an internal nuclearbomb.
Should Should it be used inthat Gattagablack mirror sort of
style thing?
But I think no.
It's exciting to see thatthat's not the direction I feel
like I think it's important tothink about it and it has to be,
but realistically for you, likeI see so much of this as having

(01:02:01):
, like, given your background,like the implications for
healthcare for this could beimmense as well.
Oh yeah, 100%.

I mean, like, one of the things that we're
very excited, and the directionsthat I personally wanna take as
well, is can we use thesesystems, these dish brains or,
you know, cr1 systems withneurons on them, as a analog for
a person that we can then useit for drug development and

(01:02:28):
testing?
Right.
Because if you say take bloodand transform that into stem
cells and stem cells throughneurons or whatever organs,
those organs or organoids shouldtheoretically inherit the same
genotypical diseases andconditions and states.
And we've seen this wherepatients with epilepsy, where we

(01:02:52):
take their blood and we turnthem into neurons, also exhibit
epilepsy in a dish.
Wow, yeah, and the because theyhave the same genome, same
genotype, they would have thesame drug profile resistance
side effects 渡 implications forpersonalized medicine 100%, and
so I think this is the path toactually doing personalized

(01:03:13):
medicine.
Alongside, you know, there's awhole like big genomic screen
with, like you know, big dataand so forth.
But why not just grow an analogof that and just test it?

This is the fascinating thing about what
you're doing, because you'reessentially if I'm reading
between the lines you'rebuilding a platform and along
the way towards, like I knowwe've spoken about this before
like you know, one of the likethe constellation goals I guess
you're aiming towards is likewell, maybe AGI needs to be
biobased, so maybe this can be away of actually getting towards
like a true AGI.
But that aside, you know thatmight be a star that you are

(01:03:45):
aiming towards, but on the way,you might be making some really
cute constellations which couldhave an even bigger impact in
the short to medium term, Iguess.

I think you know you're absolutely right and I
think the way to think aboutthis is that this is far too big
and undertaking for just us orany one company.
Right, different people willhave different goals.
Some may want to do more drugdiscovery that's fine, we're
here to support you.
Some may want to do more AGIthat's fine, we're here to

(01:04:14):
support you.
Right, it's hard enough tryingto do the hardware, let alone
the software interface, withthat, plus the wet wear.
So I think what we should focuson is to say what are we good
at?
We're good at building thetools that will help other
people take that and implementwhatever they see fit.
Right, hopefully within ethicalreasons as well, you know,

(01:04:38):
within the ethical realms.

But then again, it's really hard if you sell
these things right and that'swhy, if you do it as a service,
I guess there is that Servicelevel agreement that you can put
in.
Yeah.

Andrew Gray (01:04:49):
And that's what we're looking for and yeah, I
mean everything you've spokenabout today and everything that
we've seen at least suggeststhat you'd be able to offer it,
you know, probably betterservice if it's maintained you
know to some level here or withyou guys 100%.

I mean, at the end of the day, the AWS model
right.
Like you know, people want towrite applications, they want to
solve business problems.
They don't want to solverunning servers.

Yeah.

And then they want to do patches and all that
stuff, changing media, yeahexactly.
So why not let us do the samething where we can do all this
less exciting maintenance workwhile you go off and think about
the applications?
And you know we can alwaysrevisit it and improve the
system to accommodate for newercapabilities and so forth.

(01:05:41):
So it is a platform and it'scontinuously evolving.

No, it's so fascinating even seeing, like
the renders of what it's allgoing to look like.
I can't wait to have them inthe lab.
Oh yeah, so there was anotherthing I wanted to potentially
have a chat about.
So like we speak a lot aboutthe need for transdisciplinary
innovation and how that's goingto be, and especially like
biology as a technology, I mean,you guys are like the epitome

(01:06:08):
of transdisciplinary innovationwith what you're doing.
I'm just curious, like how hasit been for you trying to find
ways to weave those multipledifferent disciplines together
to create?

like a coherent language across?

like your departments and teams, is there
been issues with trying to, Iguess, make sense of all of the
different ways of looking at theworld and the team members that
you've got?

Yeah, I think it's a really great question and
one that even we havedifficulty and we're still
trying to work our way through.
I think it really starts withthe leadership right.
Firstly, leadership has to becomfortable.
We're not knowing everything,and to asking for help or asking

(01:06:54):
for advice and listening,that's the first thing.
Once you have that and thosevalues are passed down the ranks
, it makes it a bit easier, andso a lot of the times what we do
we still try to work a moreefficient way of doing things is
a client, a client, a clientclient service model internally.

(01:07:14):
So, for instance, our engineersare now trying to build
hardware that our softwarepeople can write, the software
for our biologists to use.

Copy that.
It's like a fracterly linkedstack where everyone's kind of
like the UX and UI is going tobe designed in a way that makes
sense for the user, rather thanfrom the person doing the coding
or the creation.
Yes, exactly.

Right, and then the biologists will do the
experiments and then they willfeedback, you know, advice back
to engineering.
This is how we do it.
This is what we need Now.
Having said that, the languageused is different.
What we say for one thing isdifferent for another thing, and
so having the need to translatethat is critical.

(01:08:03):
Like, even within theengineering we had this issue
actually.
It's quite interesting.
So we came across this issue ofthe word crosstalk.
So in engineering, electricalengineering, crosstalk happens
when you have wires that are tooclose to each other and you
have a poor signalized ratio.
Turns out, our dishes also havecrosstalk, because when we stem

(01:08:25):
one electrode, because it's asalty media around it, the ions
carry the charge across to theother side.
So you end up with crosstalk inthe solute in the biology.

Right, that checks out.

So when we were talking now, we have to clarify
which crosstalk we're talkingabout the circuit crosstalk or
the dish crosstalk?

And this is the thing that I find about
transdisciplinary is sofascinating is that you know
you're gonna have to transcendand include both of them and
find new ways of communicatingas a whole and you're kind of
creating your own language to beable to address that in a way.

Yeah, 100%.
And what's really interestingwith transdisciplinary is you
may actually end up finding outsolutions that are completely
like unorthodox, based on theother disciplines.
You know view of things.
So we've had other people likeFinn, she's a chemist, right.
So we're like, hey, you knowwhat we observe.
There's like, oh, it makessense.
And it's like, what do you mean?
She goes on praddling aboutwhole chemistry stuff.

(01:09:23):
You're like, oh, I guess itmakes sense now, right?
Or when the engineers like lookat it and they're like, oh, it
makes sense, why you're gettingthis like biology problem and
then they go fix it.

Yeah, yeah, I love it.
I just think true, and realinnovation happens at the eco
tones.
So, like you know, that's likea ecology metaphor of like, eco
tones are the two areas wheremultiple different ecosystems
overlap.
That tends to be where the mostbiodiversity is.
Same with innovation, multipledisciplines coming together,
that's when you're going to haveall of these aha moments.
Oh, of course, like, yeah, andwe and we see this all the time,

(01:09:55):
but it's been reallyfascinating, like, because
normally that's happening fromone company chatting with
another company, but this ishappening internally internally
yeah cortical, so it'sfascinating to see that.

It's a double-edged sword because, well
, it's really great for us doingall of this stuff.
It's really hard when you haveto talk to say funders or some
of the next, yeah right, Becausethey love their buckets.
And then when they look at you,they're like what the hell are
you?
You are biotech.

Whatever you want me to be baby.

What is it today ?
What is it?

Quantum, I can be quantum if you want me to be
quantum, yeah.

I can be room room temperature quantum
computer.

Yeah, I don't know , but it's crypto backed.

It's really hard because and I think this is it
right why people complain about?
Oh you know, why do we not haveenough transdisciplinary?
You know, companiescross-disciplinary, like
startups and so forth.
Well, the reason is because thefunders make it difficult.
Funders want to bucket you intosomething.
If you're cross-disciplinary,you don't fit into one bucket,
you fit into multiple buckets.
And in that case, when thebureaucrats looking at which I

(01:10:53):
give a grant for this, they'relike which is funny, because the
bureaucrats love to putcross-disciplinary as a thing as
a criteria, but then they haveto, like, decide where the
bucket this thing goes into inorder to grant the funds.
So it's a catch 22.
Like well, I can be morecross-disciplinary, but then
that means my chances of gettingfunded goes down.
You know what?
I'll just be X.

And it's such a pain, but you've definitely
called out a massive like, likea root cause of a lot of the
reasons and this is everywherein everything in innovation is
like why are we not having moreof this?
It's like if you just look atthe root causes of it, so much
of it comes back to the way inwhich the funding is allocated
out.
To tick a box, yeah.

And VCs are guilty of this right as well
because, they like to bucketthings.
Are you a FinTech thing?
Are you a biotech thing?

That's just humans in general right.

It's a very left brain thing to do, correct To
categorize things and every timeyou go to a conference or you
go to something, there's alwaysa drop down.
You're like all these things,right?
Other Other Right, exactlyEvery time.
It'd be great if you haveanother.
Sometimes they don't even haveanother and you just have to
pick the closest thing, you know, I just have to update our ABN.
It's the same problem as Iwould.
What the hell are you right andyou have this list of finite

(01:12:01):
lists of what you can and can'tbe?
So?

Yeah, it's not a fun constraint and I think that,
yeah, honestly, I don't knowfrom a design perspective.
It's like all you can do istalk about it and have these
conversations and hopefullypeople hear and realize, but
realistically, like the feedbackloops for change in those
places are so we just needbetter brain computer interfaces

(01:12:23):
.

No actually no, it's actually even simpler than
that.
We just have we need betterdesign thinking in our forms.
So, the simplest thing isremove the drop-down box.
Remove, you know.

Just let people fill in what they are.

Yeah, make it a free text thing, what the hell
are you?
And that will allow for morecross-disciplinary startups.

And then just get GPT to put it into a spreadsheet
for you after the fact.
Maybe you could yeah, justcollate it.

Exactly.
But having like free textallows you to be more
descriptive, a lot more free tosay what you are, and I think
that will hopefully allow foryou know more cross-disciplinary
companies and stuff to emergeright.
Because you know it's really apain enough for us when we have
to put down something like Likepeople come up to me and they're

(01:13:09):
like oh, you're a bi-tech.
That's like not really, becauseI don't actually make a drug or
a device.
They're like what are you?
Then I'm like well, we are acomputing thing, but we also use
biology for that stuff.

You've kind of circled around deep tech.
Since I've kind of known you, Ifeel like that's a safe space
for you to.

It's a bigger bucket.
I guess it's the other bucket,maroon.
Yeah.
Really.

Everyone who doesn't make it in just falls
into it, right?
I mean, I guess, like culture,meet right before it became a
thing was a deep tech.
Other thing, Because what is it?
Are you an agricultural companyor are you a bi-tech company?

Yeah, and I know for a fact, both like Paul from
Magic Valley and then also theVOW team, when we had James in.
They were saying they're alwayshaving to deal with this.
Yes, getting put in buckets.
What's your hat?
Yeah, where do you fit?
Like, what house are you in atHogwarts?
It's like, no, I'm a muggle, Ijust Well, I think, probably how

(01:14:10):
much.
How are you feeling You've gotenough?
Like maybe 15, 20?

Yeah, yeah, I can do another 15 minutes Sweet.

Perfect, great Well, is there anything on your
mind, andrew, apart from havinga drink?

Did I miss anything when I had to step out?

No, you just missed the backstory, damn.

So you were part of the backstory, though, so so
what does this look like to youin the future, in the next?
Like it's deep tech, so we'llsay 10, 15 years, potentially,
if things go smoothly do youhave, or is it hard to tell
because you don't know how thisis going to develop?
I know that you mentionedcommunity, mm-hmm, so you want
to talk to potentially like howcommunity would grow like and

(01:14:55):
how that might start and whatpeople could look out for if
they want to.

Yeah, so I think community is really important.
I think we unfortunately havebeen a little bit delayed with
this, but we kind of jumped thegun on it too, which is we don't
have the platform ready yet,but we already have a community
of PD like dedicated followers,and they're not really that
visible.
I mean, there's a bunch ofpeople on Twitter and stuff like
that.
The real core people areactually on our Discord and

(01:15:24):
that's where we actually releasea lot of early stuff.
So we've had early videos aboutwhat we do in the lab and
there's a VODCUS, so episodething that Peter is working on,
where we're going through thebasics, we start with the paper,
and he's interviewing peoplelike Brad and Dad and asking
them can you explain to us whatyou do and how does it matter?

(01:15:48):
How do you?
What is a stem cell?
How do you?
grow these things and so forth,and we're going to have Frank
come in and take some videofootage really to educate people
about what is going on here.

Is this content accessible after the fact, or is
it live stream only, likeyou're thinking of?

No, we're going to do it properly, like content
library.
Like there's going to be highproduction value and we're going
to make it interesting andengaging kind of thing.
Because people have alwaysasked me, like why don't you
have so much social media?
Or what's your content strategy?
I'm like, what do you mean?
Content strategy?
Which is, you know, if you're acontent creator, you have to
continuously pump up stuff topeople interested.

Tell me about it.

And we're not that.
We're a deep tech company andwe put out stuff that is purely
either A educational orimpactful.
So, like journal articles andso forth, I think having an
educational series about howstem cells, neurons are made,
how does the you know, theinterfacing of the system works,

(01:16:48):
and so forth, all these thingshelp with our brand, which is
credibility and seriousness.
So that's one thing how I seethis kind of pan out.
Really, I kind of hope thatthis would be like the Silicon
industry, but in a muchcompressed time scale.
So the Silicon industry we havetoday is like 70 years Well,

(01:17:14):
actually 1950, yeah, it's about70 years old now, and what we
have today is only like what?
Two years old or sorts.
But hopefully we can get tosomewhat at this point in a much
shorter time scale, which is ahaving people who understand how
these things work and havingcompanies build machines around

(01:17:34):
the technology.
So you know, we have, like, forexample, intel and video, amd
building the chips.
On top of that you havesoftware people building the
operating system layer right.
So you go Linux, you got Windows, and then on top of that you
have your app developers whothen build applications on top

(01:17:54):
of, you know, the hardware,subtracted by the software, the
operating system layer, and sothen you know, you have, you
know, your SaaS businesses, youropen AI and so forth.
What we're hoping is for thesame thing to happen with our
space, where there's a I mean, Iguess there's an extra layer
which is the wet wet, but itshould just follow the same
thing.
There will be people doing thewet wet.

(01:18:16):
They will have it withdifferent types of cells.
Some may be demented cell lines, some may be epileptic cell
lines, some will be normal cells, some may be enhanced cells all
these things that you can swapin and out and you can end up
with, like you know, better orworse performing systems.
Some people write differentsoftware that will take
advantage of these differenttypes of cells, and then, you
know, some people build thehardware that will then put them

(01:18:36):
all together, like what we do,and on top of that, other people
will then start thinking aboutthe applications that they can
apply to in their daily lives.
So that's what I really hopefor.
How it pans out, not quite sureyet, but we're trying to follow
the same playbook.

Right.
So it sounds like you're kindof hoping for the emergence of
like an interconnected, likecollaborative community of
people working on this sort ofspace.
Is there any element of thatthat will be open?
Or is it like so, for example,you mentioned the Linux, you
mentioned the Windows like areyou looking at going, you know

(01:19:12):
we might do elements of thisopen and we might do elements of
it closed?
Like, where are you sittingwith that?
Like is that something thatyou're considering as well?

Yeah, I think so .
I mean we're going to have amixture.
So there are some things thatare going to be closed because
you know it's going to be toohard anyway for most people to
like use any anyhow.
So things like specific timings, how do you like stimulate them
?
You know the pattern generators, all that stuff.
You know we're probably goingto keep that a bit closed.
On the other hand, the APIs aregoing to be open.

Yeah.

How do you interface with them at a higher
level?
How do you, you know, use them,which is going to be like, I
would say, 90% of the users outthere.
They're not going to care toomuch about the specific details.
For the other 10%, who reallyneed to know, we're happy to,
like you know, walk them throughas well and maybe you open it
up as well.
You know, if they sign yourspec, like deals with us or so
forth, but it's more so also, wejust don't want to give people,

(01:20:06):
like you know, tools that theymay potentially shoot themselves
in their foot with right,because the hardware we're
building has a lot of likecapability, but if you don't
know how it works, it canpotentially even fry yourselves.
So a lot of it is is sort ofabstracting that away.

Yeah, that checks out.
Yeah, I mean, that's prettymuch most of it from our end for
the conversation.
I'm just trying to think I'dlove to know, like one last
question in the context soobviously you know that's a
great vision of where we'retrying to aim and go towards,
but how do you think it willlike?

(01:20:48):
What are your predictions forhow the world will go over the
next sort of that five to 10year time frame?
And then how does this fitwithin that context, if that
makes sense, like what?

So I think there is a growing trend for
automation, right, regardless ofwhat we're going to say or do.
It is going to happen Becauseyou know capitalism and the fact
that we are trying to makemargin and we've really started
to see that with automationsoftware, you know, starting

(01:21:22):
with software as a serviceplatforms, that's now being
accelerated by things likeGenerative AI, but all of these
are happening in the white colorspace.
You know office desktops and soforth.
The one big area that hasn'treally been fully automated yet
is in the blue color space.
Right, the physical labors of,say I don't know tiling or

(01:21:50):
construction or whatever.
These kind of things are stillvery much in the realms of
humans because we do not havesystems that operate well in the
real world.
Potentially, a system like theDish Brain or biological
intelligence could start to dothat.

Integrated with something like a boss in
dynamics, correct?

And the next we think is going to be the next
big leap in automation, where weno longer just automate in the
digital realm for, likepaperwork, you know, clerical
tasks.
It's now starting to automatethe physical tasks and it
wouldn't be kind of great.
Like you know, nobody reallywants to, like, I don't know, do
garbage collection?
Get a robot to do it.
Nobody really wants to go minein a deep, dirty coal shaft, get

(01:22:37):
a robot to do it.

Yeah, I mean, I think, yeah, that that's sort of
crossing over of automation,and I guess what you would say
yeah, the wetware, yeah, I thinkthat will, because it'll have
to happen for it to be able tolike is like depth perception
and like this, like the sensoryawareness.
I just feel like it doesn'tfeel like robots are fully there
yet.
No and I think that the wetwareelement of that will be a real

(01:23:04):
game changer.

Yep, we think so as well, potentially so
fascinating.

So the next podcast will be on the ethics of
merging neurons with robots, socyborgs cyborgs.

Yeah, just go go full cyborg.
Yeah, I got, I got time forthat.

The clarification of are you more
human cyborg or more robotcyborg?

It's a spectra, it's like Robocop.

Yeah, the two, the two cyborgs.

Yeah, that's true.
Oh man, my brain's fried afterthat, yeah.

Always, always, always speechless.
But yeah, thanks, so much forcoming.

Thanks for having me on the show, always a
pleasure.

Yeah, now we really appreciate it and look
forward to seeing how thingscontinue to evolve on a merge
over the coming few years foryou.

Perfect.

Thank you, my gosh , what a conversation.
Thanks so much for tuning in.
We hope you enjoyed it just asmuch as we did.
Yeah, the stuff that'shappening at Cortical Labs is so
, so fascinating and so cool.
Like that intersection betweenbiological and computational
intelligence and ways in whichwe've biology into technology,

(01:24:19):
that wet layer being able tohelp people with potentially
personalized medicine, like theopportunities for this thing are
potentially endless, and it'sjust so cool to see this sort of
stuff happening here inAustralia and we're really proud
of all the work that they'redoing.

(01:24:39):
So, yeah, if there's any othercool innovations or folks you
think we should have a chat withon the podcast, please let us
know.
Drop us a line, give us somefeedback.
We always love hearing from ourlisteners.
And on top of that, I justthought to let you guys know
that we are also going to bedoing a bit of a special on
biomaterials and the builtenvironment this year.

(01:25:02):
So we're going to be releasing awhite paper or a report on, I
guess, on the industry withcollective fashion justice soon,
and then we'll also be lookingat collaborating with metabolic
and a couple of other folks onwhat it would look like to be
able to co-create a regenerativematerials economy here in

(01:25:25):
Australia.
So, yeah, you're going to behearing a lot more about this
sort of stuff.
We think it's a reallyimportant area that needs
support and help.
So if that interests you,please reach out, say hello.
This is a living, breathing,ongoing dialogue between
ourselves and everyone else inthe ecosystem and, yeah, we'd

(01:25:45):
love to hear from you.
Thanks so much and see you hereagain next time.

All Episodes

A Deep Dive into Synthetic Biological Intelligence with Hon Weng Chong from Cortical Labs | #6

Episode Transcript

Popular Podcasts

1. The Podium

2. In The Village

3. iHeartOlympics: The Latest

.css-15opob5{left:0;position:absolute;top:0.8rem;} All Episodes

.css-14f5ked{margin:0;word-break:break-word;display:-webkit-box;-webkit-box-orient:vertical;box-orient:vertical;-webkit-line-clamp:2;overflow:hidden;}A Deep Dive into Synthetic Biological Intelligence with Hon Weng Chong from Cortical Labs | #6