December 17, 2024 • 39 mins
Tech Meets Humanity: In-Depth with Sultan Meghji
In this episode of The Human Code, host Don Finley converses with tech visionary Sultan Meghji, who brings over 30 years of experience in AI and emerging technologies. They explore Sultan's personal journey and professional milestones, from early experimentation to advising governments and leading Frontier Foundry. Key topics include the transformative potential of decentralized AI and blockchain, the critical role of ethics in technology, and the tangible impact of quantum computing on various industries. Sultan also shares insights into personal health optimization using AI and prevailing trends in data privacy and cybersecurity in the quantum age. Tune in for a compelling discussion bridging the gap between advanced technology and human growth.
00:00 Introduction to The Human Code
00:49 Meet Sultan Meghji: A Visionary in Tech
01:47 Sultan's Early Experiments and Career
04:15 Frontier Technologies and Real-World Applications
05:41 The Evolution of Networking: From X.25 to TCP/IP
09:22 AI, Crypto, and Blockchain: The Future of Tech
13:13 Personalized AI and Data Privacy
20:04 AI in Regulated Environments
24:24 AI and Longevity: A Personal Journey
35:15 The Importance of Cyber Hygiene and Quantum Readiness
39:11 Conclusion and Final Thoughts
Sponsored by FINdustries
Hosted by Don Finley
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Don Finley (00:00):
Welcome to The Human Code, the podcast where
technology meets humanity, andthe future is shaped by the
leaders and innovators of today.
I'm your host, Don Finley,inviting you on a journey
through the fascinating world oftech, leadership, and personal
growth.
Here, we delve into the storiesof visionary minds, Who are not
only driving technologicaladvancement, but also embodying
(00:23):
the personal journeys andinsights that inspire us all.
Each episode, we explore theintersections where human
ingenuity meets the cutting edgeof technology, unpacking the
experiences, challenges, andtriumphs that define our era.
So, whether you are a techenthusiast, an inspiring
entrepreneur, or simply curiousabout the human narratives
(00:44):
behind the digital revolution,you're in the right place.
Welcome to The Human Code.
Today on the human code.
We're thrilled to welcome SaltonMeghji.
A true visionary at thecrossroads of technology and
humanity.
With over 30 years of experiencein AI and emerging technologies,
Salton has been at the forefrontof innovation, building
(01:04):
ventures, advising governments,and shaping the next wave of.
technological progress.
As the co-founder and CEO offrontier Foundry.
He leads a team tackling globalchallenges with cutting edge
solutions.
In our conversation, Saltonshares, captivating stories of
his early experiments withfrontier technologies, dives
into the transformativepotential of decentralized AI
(01:25):
and blockchain and explores thecritical role of ethics and
alignment and creating AI thatserves humanity.
From the exciting world ofquantum computing to his
personal journey and leveragingAI for health and longevity.
This episode is packed withthought provoking ideas and
actionable insights.
Let's dive in and learn fromSultan's unparalleled
perspective on how technologycan truly empower and transform
(01:46):
our lives.
I'm here with Sultan Meghji andSultan absolutely loved our pre
talk.
And I think we're going to havea fantastic conversation, but
what I would love to start outwith is what got you interested
in the intersection of humanityand technology?
Sultan Meghji (02:02):
it's a fun question and thanks for having
me on.
it's been a great chat.
I think you and I went longer inthe pre chat than we were
supposed to.
Don Finley (02:08):
It works.
Sultan Meghji (02:09):
So I was really fortunate when I was a what we
would call a tween.
My dad was a geneticist workingon the first generation of
genetically modified crops andso he was You know, a very
frontier scientist in his ownway, but he was an African
refugee basically got kicked outof Tanzania by the socialists
who seized all our land and allour property because he wasn't
the right color.
(02:30):
And, And so I had theopportunity to run around Paris
and Basel and Berlin and placeslike that, like when the wall
came down, like I was literallythere when they tore down
checkpoint Charlie, it washilarious.
Don Finley (02:42):
that is awesome.
Sultan Meghji (02:44):
And it was great.
But what it.
What I did was combine twodifferent things in my brain at
a very early and formative age.
One was technology and howimportant it is to keep pushing
on that frontier.
And then the second is, becauseI was living in a bunch of
different cultures, my mom wasfrom the middle of a cornfield
in Illinois, my dad was fromEast Africa, I grew up mostly,
(03:04):
bouncing around, Illinois andParis and Basel and places like
that, that I grew up in amulticultural environment that
fundamentally was focused onmacro events.
So the 1st, I was overseas whenthe 1st Gulf War happened.
so when I came back to the U.
S.
The kind of normal high schooljust wasn't the right fit.
And so we found a way to get meinto an accelerated program in a
(03:28):
high school that was part ofmajor university that allowed me
to do things like researchprojects and to go take classes
at that university and to, applyfor and win an NSF grant.
And in all of those cases, it's,A combination of pushing the
edges of frontier technology inreally applicable ways, not as a
(03:48):
theoretical experiment, not as apress release, not as something
that's very shiny and pretty,but something that actually has
meaning, but that also has amacroeconomic impact.
And so that's really been thethesis for most of my career.
Don Finley (04:01):
You and I have so much in common on this because
I've also been at that forefrontof like, how do we actually
apply this technology?
How do we bring it to helpingsomebody out?
What have been some of yourfavorite, applications of this
frontier tech?
Sultan Meghji (04:15):
Oh, there are some really cool ones.
very early on, I think I was 15.
I was at NCSA at the Universityof Illinois, and I got to do two
things that I didn't think Ishould have been allowed to do.
The first one was, Getadministrative access to a
supercomputer so that I couldupgrade a bunch of the libraries
on it because that vendor wasn'tkeeping up with all of the
(04:39):
advances of the internet age, Sowe were trying to do things like
have a, connect to the internetmore efficiently and, eventually
put a web server on it.
So I was the one, actuallytrying to make that.
The work that guys on the Mosaicproject and CERN were trying to
do.
I actually build a bunch of thaton supercomputers, which I
didn't think they should havelet a 15 year old do it, but
they did, That was awesome.
Don Finley (05:01):
It's honestly probably the best person for the
job as well.
Sultan Meghji (05:04):
it's interesting.
I'm far older than that.
I have to say, going through andfiguring out, reading all the
news groups and learning how GNUAutoconf worked and then trying
to, grab the CVS repository anddo this and do that and just
whack away at this early stagecode to make it work was
absolutely fascinating.
But what it did do, and it wassomething that I very quickly
picked up is.
Understanding what we call thefull stack now, everything from
(05:24):
the CPU to the network device,all the way up to what we call
layer seven, where theapplications live and protocols
and all of that kind of stuff.
And once you understand thatwhole stack, it makes it really
easy for you to, kind ofdisassemble.
What's a good project and what'sa bad project.
So one of the very firstcommercial projects I worked on
was removing an old networkingtechnology called X.
(05:46):
25 and replacing it with TCP IP.
You might not have heard of X.
25, but I'm sure you've heard ofTCP IP,
Don Finley (05:53):
Absolutely.
Sultan Meghji (05:54):
So X.
25 was one of the first computernetworks and it first.
you would know it if I showedyou a video of all those grainy
pictures from a hundred yearsago about people plugging in
the, to connect phone calls,they'd plug in the cords into
different holes.
Don Finley (06:08):
Yeah,
Sultan Meghji (06:09):
that was the original idea of X25 was to
automate that.
And so we pulled that out andgot TCP IP to work in its place
for a bunch of the marketsystems, which was really, super
interesting.
Don Finley (06:20):
absolutely fascinating because I remember
reading about some of thattransition from the old
telephone systems into TCP IPand how most networks actually
fought that for a bit.
Sultan Meghji (06:30):
Oh yeah, we had these modems and we were using
these Sun, the Sun pizza boxes,if you're old enough to remember
Don Finley (06:37):
Oh, yeah.
Sultan Meghji (06:37):
and the X.
25 driver only worked up to acertain version of the operating
system.
And the current versions, the,going from Sun OS to Solaris,
which happened I think in themid nineties.
We wanted to buy Solaris boxesbecause they were infinitely
faster and you could actuallybuy parts instead of going to a
garage sale and things likethat.
(06:58):
And so we bought the Solarisboxes, but then we couldn't
figure out how to get thelibraries to move over.
And we were getting no supportfrom the modem vendors.
The guys at Sun were like, theyreally wanted us to be
successful.
It was great.
I got to know a bunch of themand they were like, we don't
know how to do this.
and so I got access to a bunchof like internal development
stuff.
It was great.
that was really fun.
And it took the better part of ayear.
(07:20):
to get a SunPizza box with amodem attached to it that could
do about a hundred transactionsper second on the New York Stock
Exchange to replace it with amodern Solaris box at that era,
which is basically a four CPUdevice.
And it was well north of the, Iwant to say it was 15, 000
transactions per second orsomething like that, and it took
(07:40):
up three times as much space.
So 300 versus 15, 000.
I think that was a good win.
Don Finley (07:45):
That's a nice little win there.
And I think one of the thingsabout the previous systems of
connectivity with the X25 wasthere was very little
resiliency.
If you lost the wire, you lostthe whole communication where
the main benefit of TCP IP isthat it just finds its way home.
Sultan Meghji (08:01):
Yeah, x.
25 is prescriptive.
I'm trying to get from point Ato point Z.
I have to lay out B, C, D, E,all the way down.
And if any one of those goesdown, you just get an error
message, TCPIP, you just pointit at the destination and
assuming the routing and BGP andeverything works.
Don Finley (08:16):
Routing gets you there.
It finds out how to go closer.
It's like the old days of whenwe'd print out stuff from
MapQuest, if you ran into a roadthat was closed, X25, you would
not get to your destination.
TCP IP will find you aalternative route to get you
there.
Sultan Meghji (08:32):
Now, comparatively there is a
challenge.
It is slower, TCP IP is, andthat's why we use UDP so much.
So the video that you and I aretalking over is not being
transmitted on TCP IP.
It's been transmitted on UDP,and it's fundamentally because
TCP IP will guarantee that itwill get you there as long as
the physical wires are there andthat the network is configured.
UDP is the, Pray and spray kindof idea.
(08:53):
It just shoots it out there.
And so I always love it whenlike the wifi gets wonky at home
or something.
And one of my girls is like, Oh,why is it all blurry and
everything?
And I started explaining it.
I get about 30 seconds into it.
And then I realized I justbought myself a trip to Sephora
to make up for nervousness.
Don Finley (09:11):
That is absolutely fantastic because I've seen the,
for some of my friends, the eyesglaze over when you get into
these conversations as well.
Now, I know we've been spendingsome time in the network stack
onto today.
I know you're a big proponent ofcrypto and blockchain as well.
And then also an expert in AItoo.
how do you see those playinginto, our technology stacks of
(09:32):
today and the future?
Sultan Meghji (09:34):
this is such an interesting question because
most people don't combine thosethree topics into one.
Don Finley (09:39):
Oh, they have to be.
Sultan Meghji (09:41):
you and I understand this.
There's probably, small millionsof people who feel the same way.
But, if you go back to the earlydays of the Internet, the
computer that you were browsingthe web on also was running a
web server.
And you would run both of them.
And that's how it worked.
It was decentralized, And so tome, one of the things about the
crypto discussion and thecurrent iteration of the
(10:01):
blockchain technologies, it's amodern way to get back to that
by moving things that werehistorically at the network
layer up to the applicationlayer.
The problem is, the use case setis so broad and the economics of
it of the current version of theinternet are so tied to these
highly centralized environmentsthat's where so much of the
friction and the angst comesfrom.
(10:23):
So, think about it like this inthe eighties and into the
nineties, every business had amainframe.
Or maybe two or three, And theywould have their terminals and
they'd have their thingsscattered, whatever.
And then they would put PCs onpeople's desks where they would
work on, Lotus 1 2 3 orWordPerfect.
I'm digging deep on old,
Don Finley (10:41):
You got some old tech going on there.
Sultan Meghji (10:43):
Like on the PS2, IBM devices, we call it IBM.
That's the, the first, thesecond computer I run was one of
those.
And, and then we moved forward15 ish, 20 years.
And now guess what?
Every company has clouds.
All the devices have theseendpoints and guess what?
People are still using Excel andWord and you're still
(11:07):
fundamentally trying to putthings in instead of a physical
centralized data store.
It's a logical centralized datastore.
And guess what?
IBM is still making a lot ofmoney.
And the fundamental thesis ofhaving a highly centralized.
Enterprise technology base isthere.
The thing about AI and cryptoright now is that it removes a
huge percentage of the use casesand needs of those highly
(11:30):
centralized environments.
And so I have a fairly modernMacBook Pro.
It's what I'm talking to you nowon.
I use this machine for exactlythree things and in any other
point in my life I would haveneeded multiple devices and a
variety of other technologiesthat I couldn't put in my
backpack, get on a plane andstill be as effective on.
Number one is I have a fullyinternet capable device.
(11:52):
I can do all of my collaborationtools.
I can do any email.
I can watch a movie.
I can do all the normal stuff wedo with technology.
Okay.
That's great.
Second thing is I can run anentire artificial intelligence
stack.
Not connected to the internet.
And that's what I don't think alot of people realize.
I think a lot of people thinkyou have to go to the name brand
AI firms to get access to thistechnology.
(12:13):
The fact is you don't, not atall.
I am running, as we speak rightnow, a analysis of a bunch of
bank regulatory data.
On a high quality generative AIplatform that doesn't have to be
connected to the internet.
It's entirely secure and it'sjust running on my laptop and
you and I are still having avery straightforward
conversation.
I'm not glitching on you.
Okay, so that's the second thingI can do with this device.
(12:34):
The third is the hard drive isbig enough that if I pick a
domain, I don't know, marketregimes or bioinformatics data
or bank regulatory data like Iwas just talking about, or any
of these others.
There is enough hard drive spacethat I can put everything I need
on that hard drive.
You can put me on a planewithout wifi for eight hours,
leave me alone.
And no matter what your skillset with nominal preparation,
(12:58):
you can go from zero to a fullyrunning application that will
replace a huge percentage ofwhat's out there in terms of the
existing application set.
The challenge that people haveis the economics of our society,
as you and I were talking aboutbefore, aren't really wired for
that.
And this is one of the reasonswhere, the crypto conversation
becomes so dangerous because byme doing all of this on my
(13:20):
laptop.
No big venture firm is making adollar off of me.
No big enterprise tech companyexcept Apple is making any money
off me.
And it was a single capitalexpense, I'm not paying tens,
hundreds of dollars a month,thousands of dollars a month for
this technology.
Now I'm unique because I'm stillable to do this myself, But
COVID in particular, the lastfive years, we have seen a
(13:42):
radical improvement in thetooling people can use to build
applications.
Whether it's the simplest thingin the world or taking an
existing code base likeWolfenstein or Doom or Tetris
and porting it onto somethingridiculous like your watch or
whatever.
All of that is readily availableat your fingertips.
And so the third thing that kindof the little cherry on top that
(14:03):
I don't think a lot of peoplerealize is that the confluence
of everything you and I aretalking about is around open
source and open sourcetechnology.
And I am on GitHub on a dailybasis, literally a daily basis.
There are projects I sponsor.
There are some projects that Itry to contribute to, but more
it's, thoughtfulness versuscode, probably safer for
(14:24):
everyone.
but the reality is, if you lookat what AI is giving us, where
hardware has come in the last 25years and the thesis behind
crypto, I could run an entirecentral bank on a laptop if I so
chose.
Don Finley (14:39):
Oh, that's a sexy argument.
Sultan Meghji (14:41):
And that is one of the reasons, and I can tell
you because I was a formerfinancial regulator, there are a
lot of concerns about wheretransactions exist and where
they go and the visibility ofthem, et cetera, et cetera.
And.
there are a lot of argumentsabout bad people.
crypto is mostly criminals andterrorists and stuff like that.
and yeah, absolutely.
They use them.
there's no doubt there.
(15:03):
Is it 90 percent or is it 20%?
There's a lot of arguments onthat.
And I've yet to see good datagiving me, saying anything other
than a floor of about 20%.
but it pales in comparison by,zeros, at least.
Toothpaper, US money, most ofwhich exists outside of the
United States.
Most of which is used for notthe best reasons, The other day,
(15:24):
I'm not sure if you saw thenews, one of the Israeli bombing
strikes blew up 1.
4 billion in cash and gold.
Don Finley (15:30):
Damn.
Okay.
Sultan Meghji (15:32):
that's a couple of cryptocurrencies entire AUM
right there.
It's
Don Finley (15:38):
interesting in the crypto space that you're talking
about the criminal side of theactivity.
And I think you're hitting on anumber of good points around
this.
But the main thing is the amountof criminal activity that is
happening in crypto today Likeyou're saying a small percentage
of the overall criminal activitythat we just can't even track
Because of the cash basis of it
Sultan Meghji (15:58):
Absolutely.
if you were to just remove NorthKorea's criminal activity from
Bitcoin, you'd see a non trivialdecrease in Bitcoin
transactions.
no question whatsoever.
If you got them away from theYuan and the US dollar, which
they use for all their graymarket and black market
acquisitions of technology,cars, products, et cetera, et
cetera, you're still talkingmany zeros more in, in, in fiat
(16:22):
currency versus in digitalassets.
Don Finley (16:25):
so playing off of this we're now at a point where
you can actually host a You knowthe llm on whether it's Llama,
Mistral, whichever one of theopen source ones that you're
actually going after,
Sultan Meghji (16:35):
Or ours.
We have a nice one too.
I have to put a
Don Finley (16:37):
Oh yeah,
Sultan Meghji (16:38):
here at Foundry has a really nice one too.
Don Finley (16:41):
And I do love the fact that you're actually doing
the work of many people on yourlaptop, running it right now.
We are having no lag in thisconversation.
I personally run, local LLMs aswell, mostly for the standpoint
of training it to be Carl Young.
And then I got a nice localtherapist cause I'm definitely
not putting that stuff out ontothe web as well.
(17:03):
and that's the privacy, the, myownership of data and not
getting it out.
Sultan Meghji (17:07):
this is a, this is where I think, if you and I
were to go down the vein oftalking about kind of the
beltway and kind of thestruggles on regulatory, on
regulation, on utilizationaround this, just a few years
ago, everyone was up in armsabout personal data around
social media, Especially withchildren.
And we seem to have forgottenthat argument entirely.
And we're letting so many ofthese big public cloud AI
(17:30):
companies just run the mark.
Grabbing all of this data,absorbing all the questions from
everyone.
And then using it to train theirmodels, most of on the frontier
founders that most of ourcustomers live in regulated
environments, there's some sortof external stakeholder they
have to be beholden to.
And in every single case, we gotthat deal.
When I fundamentally pointed outthat.
I have no idea what you areusing our technology for.
(17:51):
It is on your computers.
It is in your four walls.
It does not call home.
It does not need to be connectedto the internet.
I will sign in blood and you cantake it to any court in the
universe that we have absolutelyno visibility.
You can find the best hackers inthe world and they will be able
to demonstrate that there is no,Data leaving their four walls.
And the fact is a lot ofcompanies out there won't say
(18:13):
that they'll hand wave, they'lluse 800 different ways to say
it, but the fact is I don'ttrust them any more than I trust
a social media or a internetsearch company, not to take my
searches and use it to thenserve me advertising, because
that is the only economic modelthat's.
a couple of these big techcompanies has that actually
works.
Don Finley (18:32):
We give similar advice to our clients as well,
and we walk it into threedifferent buckets.
One is.
You don't care.
And so we say, Hey, go right toopen AI.
if you don't care, fantastic.
The second one, or on the otherside that you're talking about
as well, is that you want tohave control over your data.
You don't want to be handing itout.
And we recommend localimplementations as well.
(18:52):
So that for the exact samereasons that you're talking
about, but we also offer like amiddle of the road option where
you're like, you know what, youcare about it, but you're
actually okay with a legalframework.
That is beholden.
And so open AI does havecontractual obligations.
Their license basically saysthat they're not going to use
your data.
If you call it through the API,you can flip off some things,
(19:15):
but there's a trust factor thatgoes into that.
So we usually recommend thatpeople go to Microsoft because
Microsoft has more to lose byviolating your trust than open
AI does, and you'll get the samemodels.
Sultan Meghji (19:26):
it's really interesting you say that because
we have the exact same threedeployment methodologies.
We have fully local, it's usedfor, a variety of people who
maybe aren't connected to theinternet because they're in
places that doesn't have it orwhatever.
We have a bunch of people whojust run it inside their own
data centers or a closet intheir office or whatever.
And then the third for us,though, actually us running
something for them.
(19:48):
and could it be hosted on,Azure?
Sure.
No problem.
Could it be hosted on AWS?
Sure.
No problem.
The challenge, and here might bethe difference between the two
companies, which is this is afascinating tangent for us to go
on, is we're quantitativelyfocused and deterministic.
We find that the vast majorityof LLMs out there are just too
(20:09):
probabilistic and 2 plus 2equaling for 99 percent of the
time is usually okay if you'redoing the equivalent of a Google
search, which is what mostpeople use, chat GPT for the
minute you do something thatgoes into a regulated context
and you have to go to, let's saythe SEC or Congress and say, the
AI did the following work cameto this decision.
We took its advice and did thefollowing things.
(20:31):
Two plus two equaling Kumquat,one out of every thousand times
was really bad,
Don Finley (20:35):
It's not good.
Sultan Meghji (20:36):
right?
And so we've spent two and ahalf years basically building
highly quantitative, oriented,high accuracy technologies, and
that's why we don't compete withany of the companies we've
really talked about because ourcustomers have very narrow use
cases that they want veryspecific solutions for, but they
have to be accurate.
every single time you run it,and no matter how often you run
(20:59):
it, you need to get the sameexact answer.
And that really does shape howyou use them.
And so we only use the LLMs forhuman communication.
It's the thing you interactwith.
We use entirely differenttechnologies for everything
behind it because LLMs don'tthink.
Don Finley (21:15):
No, there are probabilistic machines.
And basically, even if you havethe temperature turned way down,
you can't get a deterministicsolution from them, but they can
help communicate, summarize,translate for the most part.
and I agree with you that whenyou're in a regulated
environment, you need thatdeterministic approach.
Plus the side of it, where youwant to understand how it came
(21:37):
to that conclusion.
The LLM isn't the solution forthat, but more narrow focused AI
is in that
Sultan Meghji (21:44):
Well, the example I give a lot of people is let's
just say you are a hedge fund,And you're using a bunch of this
technology and you make a fairlyradical investment two years ago
and it maxes you out for themonth.
You can take the rest of themonth out and just sit on your
rear end and go to Barbados orwhatever, You need to, at that
moment, be able to capture thedata you used, the code that was
(22:05):
used against that data togenerate it, the things that you
gave it.
All that entire process all theway down to, and this is why I
hit buy or sell or whatever, Andto print that as a PDF because,
I'm young enough that I knowwhat a PDF is.
And then hand it to theappropriate regulatory body or
internal chief risk officer oryour internal control mechanism,
And you might not need to do itfor every transaction, but at
(22:26):
any moment you're using any ofthese transit, any of these
technologies in a regulatedenvironment, I tell everyone.
It doesn't matter who wins thepresidency.
It doesn't matter what controlof Congress looks like.
At some point between now andfive or 10 years in the future,
there will be some new rule inplace, even if Chevron deference
isn't overturned, or even if itis, you're right.
You will have to go back andthere will be someone
(22:48):
antagonistic to you who willtake your action and want to
know why you did it.
do you want to be in a situationof saying, Oh, it was a magic
black box and I asked it thisand it told me that.
Or do you want to be able to sayhere's 137 pages that explains
exactly why we made the decisionwe made.
Thank you.
Now I'm going to go get a coffeeand you guys come back in two
weeks once you've read it.
(23:08):
I like the latter.
Don Finley (23:09):
I think, the latter is a much safer opportunity, and
also, it's grounded in sometruth.
And so we understand that LLMsbeing probabilistic machines
that are basically just pickingthe next word that makes sense,
yet do have some emergingqualities of finding the
truthiness of things.
it's not guaranteed.
Sultan Meghji (23:28):
I want to be careful here because I'm not
criticizing LLMs or any ofbecause they're doing great
stuff.
this is foundational movement.
This market has exploded.
I started working on backpropneural nets in like the early
nineties and I got to tell you,we've seen more in the last,
movement in the last five yearsthan anything else, And it's
great.
It's absolutely fantastic thatpeople are actually embracing it
versus being afraid of it.
(23:48):
in most quarters, anyway.
The challenge though is it's notone, it's like Lord of the
Rings, right?
It's not one AI to fit them all,we will all be touching hundreds
of different AIs every day, andit will be entirely transparent
to us.
We're just starting to see that,and it will, and my thesis, at
least for the next few years, isthat it will be highly
verticalized solutions withsecurity compliance and
(24:09):
explainability knobs.
Based on the comfort of the enduser, the comfort of the person
paying for it and the regulatoryor lack thereof environment they
operate.
Don Finley (24:20):
so this cause he just gave me the perfect
transition.
There's an industry that I thinkwe're both passionate about
longevity
Sultan Meghji (24:29):
Yeah,
Don Finley (24:30):
how the technology can play into the longevity
aspect of this.
And really what I see is, wehave centralized repositories,
we have centralizedcorporations, social media
companies are selling ourattention.
And so therefore they want ourattention.
And my big focus for AI and LLMsright now is how do we get the
(24:50):
AI to be aligned with theindividual instead of aligned
with the corporate interest?
And maybe there is an alignmentbetween everybody, but at the
same time, I want the AI thatI'm working with to want to see
me achieve my goals incomparison.
And I'm wondering how that playsinto longevity or like where
you're focused in that
Sultan Meghji (25:09):
Well, you're hitting on a really cool thing
that I don't talk aboutincredibly publicly, but over
the last Two and a half, threeand a half years, I've lost
basically 50 pounds.
I'm, I am the healthiest I havebeen since I was probably 15, 16
years old, something like that.
And IU did a whole bunch ofdifferent stuff, but the anchor
(25:32):
of it was having access on mylaptop.
Not connect, not calling home tosome corporate overlords or an
insurance company that would tryto raise my rates or aim me at a
drug that they got a specialdeal on or all this kind of
stuff where I don't know whattheir motivations are because
their motivations most of thetime aren't actually having us
live longer healthily.
They're not focused onhealthspan.
They're focused on just purelength of life.
(25:54):
And the sooner you have achronic issue that requires a
prescription, the better offthey are.
And so I have done every genetictest imaginable.
I've done exome, whole genome.
I've done all of that.
And a decade ago, I worked at alittle biotech company and it
took us a, like a baby cluster.
to analyze a single exome in 45minutes.
With the same computer I'mtalking to you on again, I can
(26:16):
take my whole genome, do theentire analysis if I so choose.
It might not be done in 45minutes but definitely be done
in a day.
And I can go in and I can say,oh, I have my own opinion about
this database of genetic databecause I know something about
that university or I knowsomething about that research
group or hey, I already knowthat I'm genetically predisposed
for X.
(26:37):
And it allows me to be aninformed consumer of that
medical ecosystem.
And so I've built a program thatI've run myself through and I'm
still running myself throughthat.
I'm eventually just going tostart posting on Substack or
something, but it's acombination of molecular
diagnostics.
Nutrition, Exercise, andSupplements.
(26:57):
And I've probably put 40 peoplethrough this.
None of it is clinical, don'tthink for a second that I'm a
doctor or anything insane likethat.
But everything is over thecounter, everything is something
you can buy retail.
And every single one of them issomething that as long as you
are compliant with what it istelling you.
You'll see a positive impact.
So before we started recording,you and I were both talking
(27:18):
about how we both have, we wereboth fans of acupuncture, in no
place in this program would Ihave tried acupuncture and
unless and until I had datasupporting its use.
guess what?
If you have inflammation, one ofthe few things that's actually
proven about acupuncture is thatit really releases nitrogen in
the blood, in the essentialnervous system.
which makes room for white bloodcells.
Don Finley (27:41):
okay.
Sultan Meghji (27:42):
no matter what, it's actually not a bad thing
because you're always producingwhite blood cells.
And that's the first thing torush into a gaping hole in, in
part of your bloodstream becausethat's how it, how you scab and
stuff like that.
but it's a, I'm not one of thesecrazy people who spends like
millions of dollars a year and47 pills a day or anything like
that.
I literally, between acombination of three different
(28:03):
genetic tests, including I didwhole genome, I did an exome,
and I did a couple of differentmicroarray panels.
I did a whole bunch of differentblood work.
And basically, I looked at howmy body was responding to food,
what kind of athlete I was, I'vebeen doing all this really long
distance running and bicycling.
And I learned through mygenetics that I'm actually not
(28:23):
wired for that at all.
Like not at all.
I'm a burst athlete.
I used to play tennis.
I should be boxing.
I should be weightlifting, stufflike that.
Weightlifting is great.
Everybody should do it anyway.
But like I'm especially wiredfor that.
but that's me, I've probablyinvested, let's call it rough
order of magnitude, 20, 000 overthe last four years getting
educated, understood, buildingthe program and all of that.
(28:46):
And I can tell you at least halfof this, the result of this is
because of modern artificialintelligence built on top of
open source technology.
Don Finley (28:54):
that is fascinating.
And so basically the open sourcetechnology that you're talking
about that you've been using hashelped you to process all of
this information and put it intoa form that is actionable for
you.
You
Sultan Meghji (29:08):
year there's a certain battery of tests.
I get every quarter, there's acertain battery of tests.
I get, some stuff is reallystraightforward.
like a lot of people who don'tknow this.
and you'll notice it in people,who they have, deficiency.
and it's, you'll notice it whensomebody perks up after they eat
a steak.
and their color gets like betterif they eat a steak.
That's a B12 deficiency.
and especially if you're, veganor vegetarian, it can be really
(29:32):
hard to make sure you get allthe B12.
So take a B12 supplement.
Nothing wrong with taking toomuch.
Your body will get rid of anyexcess.
but that's, there are geneticmarkers for that.
There are a lot of geneticmarkers that are in the gray
area.
We don't yet know What they comewith.
So just eliminate those, thereare certain research
institutions that I'm not thebiggest fan of.
I just, I don't, there arecertain researchers who I'm just
(29:53):
like, your research is neverrepeatable.
if you write a research paperand you don't put all the data
up there and the code up thereand show somebody how to
duplicate the exact result yougot, especially if it's
government funded, It's hard forme to take you seriously and
there's an entire cohort ofpeople who do that and because
they're trying to protect IP andthey want to launch a company
and they want to be You know,the next George Church or
(30:14):
someone like that, which isfine, but I don't think that's
useful.
I'm very strongly on the side ofthe, if the government pays for
the research, that informationshould be made public.
Don Finley (30:21):
and I share the same opinion on this.
I am a huge proponent of anygovernment funded research needs
to be reproducible, but alsothat we have an entire
reproducibility problem acrossthe sciences and that we don't
promote that aspect of it.
So
Sultan Meghji (30:37):
it's slow moving, before we started recording, we
were talking about.
how long it takes legislation topass, how long it takes
regulatory change to pass.
And, I think NSF, FDA, NIH,Department of Energy, DOD, those
are the biggest research fundersout there.
I do think that they should alldemand of all research that It
(30:58):
is at minimum given back to thefunder and then the funder can
make that decision.
for example, if the DoD isfunding you how to make UFOs,
maybe don't post that on theinternet.
Maybe, but like the DoD shouldbe able to take that, drop it in
Area 51 and show the otheraliens and know that it's okay.
Don Finley (31:16):
that's a great example.
Sultan Meghji (31:18):
The topicalness of UAPs is just hilarious, isn't
it?
Don Finley (31:22):
I ended up becoming really spiritual after 2018.
Kind of went from full agnosticto the other side of the
spectrum and dove in.
And one thing that I was alwaysreally surprised about is how
the spiritual community initself was very accepting of any
bit of information that kind ofvalidated their ideas.
Yet at the same time, there's alot of interesting stuff that's
(31:44):
out
Sultan Meghji (31:44):
Well, we live in an infinite universe, Like the
basic physics of what we know isthat we live in an infinite
universe.
So it is almost impossible toprove anything, but it is
impossible to disproveeverything.
I grew up in a household thathad every variety of Christian
you can imagine, from Catholicto waving generally on Easter, I
(32:07):
have Muslim in my family, I haveJain in my family, I have Jewish
in my family, I have Wiccan inmy family.
you just,
Don Finley (32:14):
cover the
Sultan Meghji (32:15):
Yeah.
And guess what?
99 percent of the time, it comesdown to being a better person,
Use the stuff that's helpfulthat makes you a better person
than not.
And I'm certainly a person who,in some cases, it took me longer
to get down that journey than Ishould have.
But I gotta say especially oncekids enter the conversation,
which is a trite way to say, butalso in the face of what we see
(32:35):
every day around us, see thatthe more extreme behaviors are
getting a lot of airtime.
They're getting a lot of time onthe social media platforms.
And frankly, I just don't.
Want any of that, I actuallystarted watching the evening
news again, not too long ago,and I got to tell you, it was
soothing because it's 22 minutesonce you exclude all the idiotic
(32:56):
political commercials and it'sreally just, Hey, this is what's
happening around you, all thiskind of stuff.
I'm sure you've got some app onyour phone that tells you the
global, international, importantstuff, oh, there's this, there's
this festival happening thisweekend or this road is being
closed or something just likethe stuff so that your daily
life is a little easier.
And so like watching the eveningnews and I like watching, and
(33:17):
this is a total name drop.
I really like watching CBSSunday morning, which is
probably a code that I shouldretire soon.
Don Finley (33:24):
You're moving towards that
Sultan Meghji (33:25):
I'm getting there.
I'm getting
Don Finley (33:26):
But I do love that because I think what you're
hinting at is the evening newsis definitely more local, Like
you are getting to the pointwhere you're in a sphere of
control that you have moreaccess to than when we talk
about the global nature ofthings and which are of the
extremes.
Sultan Meghji (33:43):
if I am a critical thinker who has access
to the internet, am I going tolearn fundamentally anything
differently about current crisisin the Middle East by watching
any of the major media groups?
No, because it'll be 5 percentnews, 25 percent people talking
about it, and then a bunch of,corporate viewpoints on it or
(34:03):
specific political brand.
Viewpoints on it, most of whichare silly, And so it's really
challenging, I think, in yourmind, to separate out, here are
the facts of what's going on inthe world that I need to pay
attention to.
And for me, it's entirelymacroeconomically relative to my
business, relative to the safetyof my family, things like that.
Then there's the other one,which is what's the stuff I
(34:24):
actually need to know?
Because as soon as I hang upwith you, I'm going to do school
pickup.
And Is the main road I'm goingto drive down going to be closed
or not?
Now can I open up an app and seethat?
Yes, but guess what?
I was watching the newsyesterday and heard that this
one road was closed, so I'mgoing to go the other way.
Don Finley (34:40):
It's actionable.
Now, I know we haven't touchedon quantum, but I got to wrap it
up.
And I think what we're going toend up having to do is bringing
you back.
So for the audience, let me knowif we should have Salton come
back and we'll make that happenso we can dive into the world of
quantum and what's going onthere.
cause you're also an expert inthat space, but in the meantime,
what would you recommend thatour viewers do to be.
(35:01):
On top of all the change that'scoming about in our environment
based on what's happening withAI, what's happening with
quantum, how can they preparethemselves?
Sultan Meghji (35:10):
it's a great question and we could probably
have an, like you said, there'sprobably an entire, episode on
that.
But the basics of cyber hygienethat are well known, don't reuse
passwords, use multi factorauthentication, all that, get
those systems in place.
And if your parents get yourkids to have those systems in
place, because at some point inthe next few years, the entire
encryption infrastructure thatwe're using is going to get
(35:32):
replaced.
Okay, all of it.
It's called RSA currently.
And it's not because it's bad.
It's because quantum computerswill have industrialized how to
break it and bad guys will haveaccess to those technologies or
people with who come from nationstates that are in competition
at minimum with us and want allof our data.
And there are examples but youdon't need me to tell you what
(35:53):
And so just know that it'scoming.
and just the other day, a paperwas released by a bunch of
researchers that showed that aquantum computer connected to
one of the big cloud providerswas able to actually break RSA
mathematically.
they didn't actually do it, butthey showed that there was a way
to do it.
And it's actually on a kind of aslower, older version.
Quantum computer.
(36:14):
the ones that are out there inthe real world and that, the
Chinese have put in orbit andstuff like that because they've
had them up there for a fewyears are far in a way more
advanced than this.
And so that the article thatcame out the other day really
just made me focus and be, yeah,you've got to be quantum ready.
And in doing so, you've got totake care of what you've got
right now.
And if you're a corporate personlistening to this, the easiest
(36:36):
way to think about this is ifyou have a piece of hardware in
your shop, that's older than2018, you're going to have to
replace it because that machinewon't be fast enough to run the
new math.
That's the easy answer becauselattice encryption, let's say
you decide to go down that one,which is one of the potential
replacements, 10 to the fourth,more computational needs.
if you're running it on an I3Intel or something like that,
(36:56):
you might as well cook an egg onthat.
It's
Don Finley (36:58):
Yeah.
Yeah.
you're not moving very fast onthat
Sultan Meghji (37:01):
Yeah.
and the cloud may or may not bethe right solution for you,
depending on what you're tryingto do, because there are going
to be places where theregulators are going to say you
can't use the public cloudenvironments, or you have to use
such a restricted version ofthem that they become cost
prohibitive.
And it's just easier for you tobuy a 3, 000 server and put it
in the closet in your office.
So I would say.
(37:22):
Being ready for quantum is firstentirely about cyber hygiene.
Second, it's about knowing theinventory of your technology and
what's old and what's needing tobe replaced and all that kind of
stuff.
And the third is fundamentallyaround making sure your people
are trained for good habits.
And, this isn't like retraininga workforce.
this is all stuff we've beentalking about for well over five
years, as many as 10 years, thecompanies that struggled to get
(37:43):
people virtual on COVID are thecompanies I worry about.
If one of the one things I'vegot is a little, baby investment
book.
And I went back and actuallyfound that list of all the
companies that got bad pressduring COVID because they had
trouble getting people virtual.
And I've gone in and had an AIlook at their 8Ks and look at
their 10Ks and start to see ifI'm seeing an uptick in
(38:04):
cybersecurity spend to replaceall their legacy technology.
and there are a bunch of marketswhere that hasn't happened yet.
Financial services, there's abunch.
Healthcare, there's a bunch.
Energy, there's a bunch.
And those are all companies thatI would be, if I were to work
with them.
Or if I was a customer oftheirs,
Don Finley (38:21):
Oh, that's a great analysis.
And I love the use of AI to helpyou put that together because
those are usually research jobsthat we would put, a young
analyst on.
Sultan Meghji (38:31):
that, that kind of 101 to 201 level primary
analysis of something issomething that I used to have
two or three analysts workingfor me at any one time.
Now I'm on the treadmill at thegym and I'm using my phone to
call into my private LLMenvironment.
And I'm asking it questions.
(38:51):
And each time I'm like, Hey, bean analyst, analyze this for me,
be an analyst, analyze this forme.
And then it reads it back to me.
Don Finley (38:58):
And at the same time, you're getting responses
in minutes compared to whatcould have been like days long
projects for an analyst as
Sultan Meghji (39:05):
sometimes seconds even.
sometimes it'll interrupt me,which I think is funny.
Don Finley (39:09):
That's awesome.
All right, Sultan, it has beenan absolute blast.
I really appreciate having youon today.
Thank you so much.
Sultan Meghji (39:16):
Thanks for having me.
This has been a lot of fun.
Don Finley (39:19):
Thank you for tuning into The Human Code, sponsored
by FINdustries, where we harnessAI to elevate your business.
By improving operationalefficiency and accelerating
growth, we turn opportunitiesinto reality.
Let FINdustries be your guide toAI mastery, making success
inevitable.
Explore how at FINdustries.
(39:40):
co.
Welcome to The Human Code, the podcast where
technology meets humanity, andthe future is shaped by the
leaders and innovators of today.
I'm your host, Don Finley,inviting you on a journey
through the fascinating world oftech, leadership, and personal
growth.
Here, we delve into the storiesof visionary minds, Who are not
only driving technologicaladvancement, but also embodying
(00:23):
the personal journeys andinsights that inspire us all.
Each episode, we explore theintersections where human
ingenuity meets the cutting edgeof technology, unpacking the
experiences, challenges, andtriumphs that define our era.
So, whether you are a techenthusiast, an inspiring
entrepreneur, or simply curiousabout the human narratives
(00:44):
behind the digital revolution,you're in the right place.
Welcome to The Human Code.
Today on the human code.
We're thrilled to welcome SaltonMeghji.
A true visionary at thecrossroads of technology and
humanity.
With over 30 years of experiencein AI and emerging technologies,
Salton has been at the forefrontof innovation, building
(01:04):
ventures, advising governments,and shaping the next wave of.
technological progress.
As the co-founder and CEO offrontier Foundry.
He leads a team tackling globalchallenges with cutting edge
solutions.
In our conversation, Saltonshares, captivating stories of
his early experiments withfrontier technologies, dives
into the transformativepotential of decentralized AI
(01:25):
and blockchain and explores thecritical role of ethics and
alignment and creating AI thatserves humanity.
From the exciting world ofquantum computing to his
personal journey and leveragingAI for health and longevity.
This episode is packed withthought provoking ideas and
actionable insights.
Let's dive in and learn fromSultan's unparalleled
perspective on how technologycan truly empower and transform
(01:46):
our lives.
I'm here with Sultan Meghji andSultan absolutely loved our pre
talk.
And I think we're going to havea fantastic conversation, but
what I would love to start outwith is what got you interested
in the intersection of humanityand technology?
Sultan Meghji (02:02):
it's a fun question and thanks for having
me on.
it's been a great chat.
I think you and I went longer inthe pre chat than we were
supposed to.
Don Finley (02:08):
It works.
Sultan Meghji (02:09):
So I was really fortunate when I was a what we
would call a tween.
My dad was a geneticist workingon the first generation of
genetically modified crops andso he was You know, a very
frontier scientist in his ownway, but he was an African
refugee basically got kicked outof Tanzania by the socialists
who seized all our land and allour property because he wasn't
the right color.
(02:30):
And, And so I had theopportunity to run around Paris
and Basel and Berlin and placeslike that, like when the wall
came down, like I was literallythere when they tore down
checkpoint Charlie, it washilarious.
Don Finley (02:42):
that is awesome.
Sultan Meghji (02:44):
And it was great.
But what it.
What I did was combine twodifferent things in my brain at
a very early and formative age.
One was technology and howimportant it is to keep pushing
on that frontier.
And then the second is, becauseI was living in a bunch of
different cultures, my mom wasfrom the middle of a cornfield
in Illinois, my dad was fromEast Africa, I grew up mostly,
(03:04):
bouncing around, Illinois andParis and Basel and places like
that, that I grew up in amulticultural environment that
fundamentally was focused onmacro events.
So the 1st, I was overseas whenthe 1st Gulf War happened.
so when I came back to the U.
S.
The kind of normal high schooljust wasn't the right fit.
And so we found a way to get meinto an accelerated program in a
(03:28):
high school that was part ofmajor university that allowed me
to do things like researchprojects and to go take classes
at that university and to, applyfor and win an NSF grant.
And in all of those cases, it's,A combination of pushing the
edges of frontier technology inreally applicable ways, not as a
(03:48):
theoretical experiment, not as apress release, not as something
that's very shiny and pretty,but something that actually has
meaning, but that also has amacroeconomic impact.
And so that's really been thethesis for most of my career.
Don Finley (04:01):
You and I have so much in common on this because
I've also been at that forefrontof like, how do we actually
apply this technology?
How do we bring it to helpingsomebody out?
What have been some of yourfavorite, applications of this
frontier tech?
Sultan Meghji (04:15):
Oh, there are some really cool ones.
very early on, I think I was 15.
I was at NCSA at the Universityof Illinois, and I got to do two
things that I didn't think Ishould have been allowed to do.
The first one was, Getadministrative access to a
supercomputer so that I couldupgrade a bunch of the libraries
on it because that vendor wasn'tkeeping up with all of the
(04:39):
advances of the internet age, Sowe were trying to do things like
have a, connect to the internetmore efficiently and, eventually
put a web server on it.
So I was the one, actuallytrying to make that.
The work that guys on the Mosaicproject and CERN were trying to
do.
I actually build a bunch of thaton supercomputers, which I
didn't think they should havelet a 15 year old do it, but
they did, That was awesome.
Don Finley (05:01):
It's honestly probably the best person for the
job as well.
Sultan Meghji (05:04):
it's interesting.
I'm far older than that.
I have to say, going through andfiguring out, reading all the
news groups and learning how GNUAutoconf worked and then trying
to, grab the CVS repository anddo this and do that and just
whack away at this early stagecode to make it work was
absolutely fascinating.
But what it did do, and it wassomething that I very quickly
picked up is.
Understanding what we call thefull stack now, everything from
(05:24):
the CPU to the network device,all the way up to what we call
layer seven, where theapplications live and protocols
and all of that kind of stuff.
And once you understand thatwhole stack, it makes it really
easy for you to, kind ofdisassemble.
What's a good project and what'sa bad project.
So one of the very firstcommercial projects I worked on
was removing an old networkingtechnology called X.
(05:46):
25 and replacing it with TCP IP.
You might not have heard of X.
25, but I'm sure you've heard ofTCP IP,
Don Finley (05:53):
Absolutely.
Sultan Meghji (05:54):
So X.
25 was one of the first computernetworks and it first.
you would know it if I showedyou a video of all those grainy
pictures from a hundred yearsago about people plugging in
the, to connect phone calls,they'd plug in the cords into
different holes.
Don Finley (06:08):
Yeah,
Sultan Meghji (06:09):
that was the original idea of X25 was to
automate that.
And so we pulled that out andgot TCP IP to work in its place
for a bunch of the marketsystems, which was really, super
interesting.
Don Finley (06:20):
absolutely fascinating because I remember
reading about some of thattransition from the old
telephone systems into TCP IPand how most networks actually
fought that for a bit.
Sultan Meghji (06:30):
Oh yeah, we had these modems and we were using
these Sun, the Sun pizza boxes,if you're old enough to remember
Don Finley (06:37):
Oh, yeah.
Sultan Meghji (06:37):
and the X.
25 driver only worked up to acertain version of the operating
system.
And the current versions, the,going from Sun OS to Solaris,
which happened I think in themid nineties.
We wanted to buy Solaris boxesbecause they were infinitely
faster and you could actuallybuy parts instead of going to a
garage sale and things likethat.
(06:58):
And so we bought the Solarisboxes, but then we couldn't
figure out how to get thelibraries to move over.
And we were getting no supportfrom the modem vendors.
The guys at Sun were like, theyreally wanted us to be
successful.
It was great.
I got to know a bunch of themand they were like, we don't
know how to do this.
and so I got access to a bunchof like internal development
stuff.
It was great.
that was really fun.
And it took the better part of ayear.
(07:20):
to get a SunPizza box with amodem attached to it that could
do about a hundred transactionsper second on the New York Stock
Exchange to replace it with amodern Solaris box at that era,
which is basically a four CPUdevice.
And it was well north of the, Iwant to say it was 15, 000
transactions per second orsomething like that, and it took
(07:40):
up three times as much space.
So 300 versus 15, 000.
I think that was a good win.
Don Finley (07:45):
That's a nice little win there.
And I think one of the thingsabout the previous systems of
connectivity with the X25 wasthere was very little
resiliency.
If you lost the wire, you lostthe whole communication where
the main benefit of TCP IP isthat it just finds its way home.
Sultan Meghji (08:01):
Yeah, x.
25 is prescriptive.
I'm trying to get from point Ato point Z.
I have to lay out B, C, D, E,all the way down.
And if any one of those goesdown, you just get an error
message, TCPIP, you just pointit at the destination and
assuming the routing and BGP andeverything works.
Don Finley (08:16):
Routing gets you there.
It finds out how to go closer.
It's like the old days of whenwe'd print out stuff from
MapQuest, if you ran into a roadthat was closed, X25, you would
not get to your destination.
TCP IP will find you aalternative route to get you
there.
Sultan Meghji (08:32):
Now, comparatively there is a
challenge.
It is slower, TCP IP is, andthat's why we use UDP so much.
So the video that you and I aretalking over is not being
transmitted on TCP IP.
It's been transmitted on UDP,and it's fundamentally because
TCP IP will guarantee that itwill get you there as long as
the physical wires are there andthat the network is configured.
UDP is the, Pray and spray kindof idea.
(08:53):
It just shoots it out there.
And so I always love it whenlike the wifi gets wonky at home
or something.
And one of my girls is like, Oh,why is it all blurry and
everything?
And I started explaining it.
I get about 30 seconds into it.
And then I realized I justbought myself a trip to Sephora
to make up for nervousness.
Don Finley (09:11):
That is absolutely fantastic because I've seen the,
for some of my friends, the eyesglaze over when you get into
these conversations as well.
Now, I know we've been spendingsome time in the network stack
onto today.
I know you're a big proponent ofcrypto and blockchain as well.
And then also an expert in AItoo.
how do you see those playinginto, our technology stacks of
(09:32):
today and the future?
Sultan Meghji (09:34):
this is such an interesting question because
most people don't combine thosethree topics into one.
Don Finley (09:39):
Oh, they have to be.
Sultan Meghji (09:41):
you and I understand this.
There's probably, small millionsof people who feel the same way.
But, if you go back to the earlydays of the Internet, the
computer that you were browsingthe web on also was running a
web server.
And you would run both of them.
And that's how it worked.
It was decentralized, And so tome, one of the things about the
crypto discussion and thecurrent iteration of the
(10:01):
blockchain technologies, it's amodern way to get back to that
by moving things that werehistorically at the network
layer up to the applicationlayer.
The problem is, the use case setis so broad and the economics of
it of the current version of theinternet are so tied to these
highly centralized environmentsthat's where so much of the
friction and the angst comesfrom.
(10:23):
So, think about it like this inthe eighties and into the
nineties, every business had amainframe.
Or maybe two or three, And theywould have their terminals and
they'd have their thingsscattered, whatever.
And then they would put PCs onpeople's desks where they would
work on, Lotus 1 2 3 orWordPerfect.
I'm digging deep on old,
Don Finley (10:41):
You got some old tech going on there.
Sultan Meghji (10:43):
Like on the PS2, IBM devices, we call it IBM.
That's the, the first, thesecond computer I run was one of
those.
And, and then we moved forward15 ish, 20 years.
And now guess what?
Every company has clouds.
All the devices have theseendpoints and guess what?
People are still using Excel andWord and you're still
(11:07):
fundamentally trying to putthings in instead of a physical
centralized data store.
It's a logical centralized datastore.
And guess what?
IBM is still making a lot ofmoney.
And the fundamental thesis ofhaving a highly centralized.
Enterprise technology base isthere.
The thing about AI and cryptoright now is that it removes a
huge percentage of the use casesand needs of those highly
(11:30):
centralized environments.
And so I have a fairly modernMacBook Pro.
It's what I'm talking to you nowon.
I use this machine for exactlythree things and in any other
point in my life I would haveneeded multiple devices and a
variety of other technologiesthat I couldn't put in my
backpack, get on a plane andstill be as effective on.
Number one is I have a fullyinternet capable device.
(11:52):
I can do all of my collaborationtools.
I can do any email.
I can watch a movie.
I can do all the normal stuff wedo with technology.
Okay.
That's great.
Second thing is I can run anentire artificial intelligence
stack.
Not connected to the internet.
And that's what I don't think alot of people realize.
I think a lot of people thinkyou have to go to the name brand
AI firms to get access to thistechnology.
(12:13):
The fact is you don't, not atall.
I am running, as we speak rightnow, a analysis of a bunch of
bank regulatory data.
On a high quality generative AIplatform that doesn't have to be
connected to the internet.
It's entirely secure and it'sjust running on my laptop and
you and I are still having avery straightforward
conversation.
I'm not glitching on you.
Okay, so that's the second thingI can do with this device.
(12:34):
The third is the hard drive isbig enough that if I pick a
domain, I don't know, marketregimes or bioinformatics data
or bank regulatory data like Iwas just talking about, or any
of these others.
There is enough hard drive spacethat I can put everything I need
on that hard drive.
You can put me on a planewithout wifi for eight hours,
leave me alone.
And no matter what your skillset with nominal preparation,
(12:58):
you can go from zero to a fullyrunning application that will
replace a huge percentage ofwhat's out there in terms of the
existing application set.
The challenge that people haveis the economics of our society,
as you and I were talking aboutbefore, aren't really wired for
that.
And this is one of the reasonswhere, the crypto conversation
becomes so dangerous because byme doing all of this on my
(13:20):
laptop.
No big venture firm is making adollar off of me.
No big enterprise tech companyexcept Apple is making any money
off me.
And it was a single capitalexpense, I'm not paying tens,
hundreds of dollars a month,thousands of dollars a month for
this technology.
Now I'm unique because I'm stillable to do this myself, But
COVID in particular, the lastfive years, we have seen a
(13:42):
radical improvement in thetooling people can use to build
applications.
Whether it's the simplest thingin the world or taking an
existing code base likeWolfenstein or Doom or Tetris
and porting it onto somethingridiculous like your watch or
whatever.
All of that is readily availableat your fingertips.
And so the third thing that kindof the little cherry on top that
(14:03):
I don't think a lot of peoplerealize is that the confluence
of everything you and I aretalking about is around open
source and open sourcetechnology.
And I am on GitHub on a dailybasis, literally a daily basis.
There are projects I sponsor.
There are some projects that Itry to contribute to, but more
it's, thoughtfulness versuscode, probably safer for
(14:24):
everyone.
but the reality is, if you lookat what AI is giving us, where
hardware has come in the last 25years and the thesis behind
crypto, I could run an entirecentral bank on a laptop if I so
chose.
Don Finley (14:39):
Oh, that's a sexy argument.
Sultan Meghji (14:41):
And that is one of the reasons, and I can tell
you because I was a formerfinancial regulator, there are a
lot of concerns about wheretransactions exist and where
they go and the visibility ofthem, et cetera, et cetera.
And.
there are a lot of argumentsabout bad people.
crypto is mostly criminals andterrorists and stuff like that.
and yeah, absolutely.
They use them.
there's no doubt there.
(15:03):
Is it 90 percent or is it 20%?
There's a lot of arguments onthat.
And I've yet to see good datagiving me, saying anything other
than a floor of about 20%.
but it pales in comparison by,zeros, at least.
Toothpaper, US money, most ofwhich exists outside of the
United States.
Most of which is used for notthe best reasons, The other day,
(15:24):
I'm not sure if you saw thenews, one of the Israeli bombing
strikes blew up 1.
4 billion in cash and gold.
Don Finley (15:30):
Damn.
Okay.
Sultan Meghji (15:32):
that's a couple of cryptocurrencies entire AUM
right there.
It's
Don Finley (15:38):
interesting in the crypto space that you're talking
about the criminal side of theactivity.
And I think you're hitting on anumber of good points around
this.
But the main thing is the amountof criminal activity that is
happening in crypto today Likeyou're saying a small percentage
of the overall criminal activitythat we just can't even track
Because of the cash basis of it
Sultan Meghji (15:58):
Absolutely.
if you were to just remove NorthKorea's criminal activity from
Bitcoin, you'd see a non trivialdecrease in Bitcoin
transactions.
no question whatsoever.
If you got them away from theYuan and the US dollar, which
they use for all their graymarket and black market
acquisitions of technology,cars, products, et cetera, et
cetera, you're still talkingmany zeros more in, in, in fiat
(16:22):
currency versus in digitalassets.
Don Finley (16:25):
so playing off of this we're now at a point where
you can actually host a You knowthe llm on whether it's Llama,
Mistral, whichever one of theopen source ones that you're
actually going after,
Sultan Meghji (16:35):
Or ours.
We have a nice one too.
I have to put a
Don Finley (16:37):
Oh yeah,
Sultan Meghji (16:38):
here at Foundry has a really nice one too.
Don Finley (16:41):
And I do love the fact that you're actually doing
the work of many people on yourlaptop, running it right now.
We are having no lag in thisconversation.
I personally run, local LLMs aswell, mostly for the standpoint
of training it to be Carl Young.
And then I got a nice localtherapist cause I'm definitely
not putting that stuff out ontothe web as well.
(17:03):
and that's the privacy, the, myownership of data and not
getting it out.
Sultan Meghji (17:07):
this is a, this is where I think, if you and I
were to go down the vein oftalking about kind of the
beltway and kind of thestruggles on regulatory, on
regulation, on utilizationaround this, just a few years
ago, everyone was up in armsabout personal data around
social media, Especially withchildren.
And we seem to have forgottenthat argument entirely.
And we're letting so many ofthese big public cloud AI
(17:30):
companies just run the mark.
Grabbing all of this data,absorbing all the questions from
everyone.
And then using it to train theirmodels, most of on the frontier
founders that most of ourcustomers live in regulated
environments, there's some sortof external stakeholder they
have to be beholden to.
And in every single case, we gotthat deal.
When I fundamentally pointed outthat.
I have no idea what you areusing our technology for.
(17:51):
It is on your computers.
It is in your four walls.
It does not call home.
It does not need to be connectedto the internet.
I will sign in blood and you cantake it to any court in the
universe that we have absolutelyno visibility.
You can find the best hackers inthe world and they will be able
to demonstrate that there is no,Data leaving their four walls.
And the fact is a lot ofcompanies out there won't say
(18:13):
that they'll hand wave, they'lluse 800 different ways to say
it, but the fact is I don'ttrust them any more than I trust
a social media or a internetsearch company, not to take my
searches and use it to thenserve me advertising, because
that is the only economic modelthat's.
a couple of these big techcompanies has that actually
works.
Don Finley (18:32):
We give similar advice to our clients as well,
and we walk it into threedifferent buckets.
One is.
You don't care.
And so we say, Hey, go right toopen AI.
if you don't care, fantastic.
The second one, or on the otherside that you're talking about
as well, is that you want tohave control over your data.
You don't want to be handing itout.
And we recommend localimplementations as well.
(18:52):
So that for the exact samereasons that you're talking
about, but we also offer like amiddle of the road option where
you're like, you know what, youcare about it, but you're
actually okay with a legalframework.
That is beholden.
And so open AI does havecontractual obligations.
Their license basically saysthat they're not going to use
your data.
If you call it through the API,you can flip off some things,
(19:15):
but there's a trust factor thatgoes into that.
So we usually recommend thatpeople go to Microsoft because
Microsoft has more to lose byviolating your trust than open
AI does, and you'll get the samemodels.
Sultan Meghji (19:26):
it's really interesting you say that because
we have the exact same threedeployment methodologies.
We have fully local, it's usedfor, a variety of people who
maybe aren't connected to theinternet because they're in
places that doesn't have it orwhatever.
We have a bunch of people whojust run it inside their own
data centers or a closet intheir office or whatever.
And then the third for us,though, actually us running
something for them.
(19:48):
and could it be hosted on,Azure?
Sure.
No problem.
Could it be hosted on AWS?
Sure.
No problem.
The challenge, and here might bethe difference between the two
companies, which is this is afascinating tangent for us to go
on, is we're quantitativelyfocused and deterministic.
We find that the vast majorityof LLMs out there are just too
(20:09):
probabilistic and 2 plus 2equaling for 99 percent of the
time is usually okay if you'redoing the equivalent of a Google
search, which is what mostpeople use, chat GPT for the
minute you do something thatgoes into a regulated context
and you have to go to, let's saythe SEC or Congress and say, the
AI did the following work cameto this decision.
We took its advice and did thefollowing things.
(20:31):
Two plus two equaling Kumquat,one out of every thousand times
was really bad,
Don Finley (20:35):
It's not good.
Sultan Meghji (20:36):
right?
And so we've spent two and ahalf years basically building
highly quantitative, oriented,high accuracy technologies, and
that's why we don't compete withany of the companies we've
really talked about because ourcustomers have very narrow use
cases that they want veryspecific solutions for, but they
have to be accurate.
every single time you run it,and no matter how often you run
(20:59):
it, you need to get the sameexact answer.
And that really does shape howyou use them.
And so we only use the LLMs forhuman communication.
It's the thing you interactwith.
We use entirely differenttechnologies for everything
behind it because LLMs don'tthink.
Don Finley (21:15):
No, there are probabilistic machines.
And basically, even if you havethe temperature turned way down,
you can't get a deterministicsolution from them, but they can
help communicate, summarize,translate for the most part.
and I agree with you that whenyou're in a regulated
environment, you need thatdeterministic approach.
Plus the side of it, where youwant to understand how it came
(21:37):
to that conclusion.
The LLM isn't the solution forthat, but more narrow focused AI
is in that
Sultan Meghji (21:44):
Well, the example I give a lot of people is let's
just say you are a hedge fund,And you're using a bunch of this
technology and you make a fairlyradical investment two years ago
and it maxes you out for themonth.
You can take the rest of themonth out and just sit on your
rear end and go to Barbados orwhatever, You need to, at that
moment, be able to capture thedata you used, the code that was
(22:05):
used against that data togenerate it, the things that you
gave it.
All that entire process all theway down to, and this is why I
hit buy or sell or whatever, Andto print that as a PDF because,
I'm young enough that I knowwhat a PDF is.
And then hand it to theappropriate regulatory body or
internal chief risk officer oryour internal control mechanism,
And you might not need to do itfor every transaction, but at
(22:26):
any moment you're using any ofthese transit, any of these
technologies in a regulatedenvironment, I tell everyone.
It doesn't matter who wins thepresidency.
It doesn't matter what controlof Congress looks like.
At some point between now andfive or 10 years in the future,
there will be some new rule inplace, even if Chevron deference
isn't overturned, or even if itis, you're right.
You will have to go back andthere will be someone
(22:48):
antagonistic to you who willtake your action and want to
know why you did it.
do you want to be in a situationof saying, Oh, it was a magic
black box and I asked it thisand it told me that.
Or do you want to be able to sayhere's 137 pages that explains
exactly why we made the decisionwe made.
Thank you.
Now I'm going to go get a coffeeand you guys come back in two
weeks once you've read it.
(23:08):
I like the latter.
Don Finley (23:09):
I think, the latter is a much safer opportunity, and
also, it's grounded in sometruth.
And so we understand that LLMsbeing probabilistic machines
that are basically just pickingthe next word that makes sense,
yet do have some emergingqualities of finding the
truthiness of things.
it's not guaranteed.
Sultan Meghji (23:28):
I want to be careful here because I'm not
criticizing LLMs or any ofbecause they're doing great
stuff.
this is foundational movement.
This market has exploded.
I started working on backpropneural nets in like the early
nineties and I got to tell you,we've seen more in the last,
movement in the last five yearsthan anything else, And it's
great.
It's absolutely fantastic thatpeople are actually embracing it
versus being afraid of it.
(23:48):
in most quarters, anyway.
The challenge though is it's notone, it's like Lord of the
Rings, right?
It's not one AI to fit them all,we will all be touching hundreds
of different AIs every day, andit will be entirely transparent
to us.
We're just starting to see that,and it will, and my thesis, at
least for the next few years, isthat it will be highly
verticalized solutions withsecurity compliance and
(24:09):
explainability knobs.
Based on the comfort of the enduser, the comfort of the person
paying for it and the regulatoryor lack thereof environment they
operate.
Don Finley (24:20):
so this cause he just gave me the perfect
transition.
There's an industry that I thinkwe're both passionate about
longevity
Sultan Meghji (24:29):
Yeah,
Don Finley (24:30):
how the technology can play into the longevity
aspect of this.
And really what I see is, wehave centralized repositories,
we have centralizedcorporations, social media
companies are selling ourattention.
And so therefore they want ourattention.
And my big focus for AI and LLMsright now is how do we get the
(24:50):
AI to be aligned with theindividual instead of aligned
with the corporate interest?
And maybe there is an alignmentbetween everybody, but at the
same time, I want the AI thatI'm working with to want to see
me achieve my goals incomparison.
And I'm wondering how that playsinto longevity or like where
you're focused in that
Sultan Meghji (25:09):
Well, you're hitting on a really cool thing
that I don't talk aboutincredibly publicly, but over
the last Two and a half, threeand a half years, I've lost
basically 50 pounds.
I'm, I am the healthiest I havebeen since I was probably 15, 16
years old, something like that.
And IU did a whole bunch ofdifferent stuff, but the anchor
(25:32):
of it was having access on mylaptop.
Not connect, not calling home tosome corporate overlords or an
insurance company that would tryto raise my rates or aim me at a
drug that they got a specialdeal on or all this kind of
stuff where I don't know whattheir motivations are because
their motivations most of thetime aren't actually having us
live longer healthily.
They're not focused onhealthspan.
They're focused on just purelength of life.
(25:54):
And the sooner you have achronic issue that requires a
prescription, the better offthey are.
And so I have done every genetictest imaginable.
I've done exome, whole genome.
I've done all of that.
And a decade ago, I worked at alittle biotech company and it
took us a, like a baby cluster.
to analyze a single exome in 45minutes.
With the same computer I'mtalking to you on again, I can
(26:16):
take my whole genome, do theentire analysis if I so choose.
It might not be done in 45minutes but definitely be done
in a day.
And I can go in and I can say,oh, I have my own opinion about
this database of genetic databecause I know something about
that university or I knowsomething about that research
group or hey, I already knowthat I'm genetically predisposed
for X.
(26:37):
And it allows me to be aninformed consumer of that
medical ecosystem.
And so I've built a program thatI've run myself through and I'm
still running myself throughthat.
I'm eventually just going tostart posting on Substack or
something, but it's acombination of molecular
diagnostics.
Nutrition, Exercise, andSupplements.
(26:57):
And I've probably put 40 peoplethrough this.
None of it is clinical, don'tthink for a second that I'm a
doctor or anything insane likethat.
But everything is over thecounter, everything is something
you can buy retail.
And every single one of them issomething that as long as you
are compliant with what it istelling you.
You'll see a positive impact.
So before we started recording,you and I were both talking
(27:18):
about how we both have, we wereboth fans of acupuncture, in no
place in this program would Ihave tried acupuncture and
unless and until I had datasupporting its use.
guess what?
If you have inflammation, one ofthe few things that's actually
proven about acupuncture is thatit really releases nitrogen in
the blood, in the essentialnervous system.
which makes room for white bloodcells.
Don Finley (27:41):
okay.
Sultan Meghji (27:42):
no matter what, it's actually not a bad thing
because you're always producingwhite blood cells.
And that's the first thing torush into a gaping hole in, in
part of your bloodstream becausethat's how it, how you scab and
stuff like that.
but it's a, I'm not one of thesecrazy people who spends like
millions of dollars a year and47 pills a day or anything like
that.
I literally, between acombination of three different
(28:03):
genetic tests, including I didwhole genome, I did an exome,
and I did a couple of differentmicroarray panels.
I did a whole bunch of differentblood work.
And basically, I looked at howmy body was responding to food,
what kind of athlete I was, I'vebeen doing all this really long
distance running and bicycling.
And I learned through mygenetics that I'm actually not
(28:23):
wired for that at all.
Like not at all.
I'm a burst athlete.
I used to play tennis.
I should be boxing.
I should be weightlifting, stufflike that.
Weightlifting is great.
Everybody should do it anyway.
But like I'm especially wiredfor that.
but that's me, I've probablyinvested, let's call it rough
order of magnitude, 20, 000 overthe last four years getting
educated, understood, buildingthe program and all of that.
(28:46):
And I can tell you at least halfof this, the result of this is
because of modern artificialintelligence built on top of
open source technology.
Don Finley (28:54):
that is fascinating.
And so basically the open sourcetechnology that you're talking
about that you've been using hashelped you to process all of
this information and put it intoa form that is actionable for
you.
You
Sultan Meghji (29:08):
year there's a certain battery of tests.
I get every quarter, there's acertain battery of tests.
I get, some stuff is reallystraightforward.
like a lot of people who don'tknow this.
and you'll notice it in people,who they have, deficiency.
and it's, you'll notice it whensomebody perks up after they eat
a steak.
and their color gets like betterif they eat a steak.
That's a B12 deficiency.
and especially if you're, veganor vegetarian, it can be really
(29:32):
hard to make sure you get allthe B12.
So take a B12 supplement.
Nothing wrong with taking toomuch.
Your body will get rid of anyexcess.
but that's, there are geneticmarkers for that.
There are a lot of geneticmarkers that are in the gray
area.
We don't yet know What they comewith.
So just eliminate those, thereare certain research
institutions that I'm not thebiggest fan of.
I just, I don't, there arecertain researchers who I'm just
(29:53):
like, your research is neverrepeatable.
if you write a research paperand you don't put all the data
up there and the code up thereand show somebody how to
duplicate the exact result yougot, especially if it's
government funded, It's hard forme to take you seriously and
there's an entire cohort ofpeople who do that and because
they're trying to protect IP andthey want to launch a company
and they want to be You know,the next George Church or
(30:14):
someone like that, which isfine, but I don't think that's
useful.
I'm very strongly on the side ofthe, if the government pays for
the research, that informationshould be made public.
Don Finley (30:21):
and I share the same opinion on this.
I am a huge proponent of anygovernment funded research needs
to be reproducible, but alsothat we have an entire
reproducibility problem acrossthe sciences and that we don't
promote that aspect of it.
So
Sultan Meghji (30:37):
it's slow moving, before we started recording, we
were talking about.
how long it takes legislation topass, how long it takes
regulatory change to pass.
And, I think NSF, FDA, NIH,Department of Energy, DOD, those
are the biggest research fundersout there.
I do think that they should alldemand of all research that It
(30:58):
is at minimum given back to thefunder and then the funder can
make that decision.
for example, if the DoD isfunding you how to make UFOs,
maybe don't post that on theinternet.
Maybe, but like the DoD shouldbe able to take that, drop it in
Area 51 and show the otheraliens and know that it's okay.
Don Finley (31:16):
that's a great example.
Sultan Meghji (31:18):
The topicalness of UAPs is just hilarious, isn't
it?
Don Finley (31:22):
I ended up becoming really spiritual after 2018.
Kind of went from full agnosticto the other side of the
spectrum and dove in.
And one thing that I was alwaysreally surprised about is how
the spiritual community initself was very accepting of any
bit of information that kind ofvalidated their ideas.
Yet at the same time, there's alot of interesting stuff that's
(31:44):
out
Sultan Meghji (31:44):
Well, we live in an infinite universe, Like the
basic physics of what we know isthat we live in an infinite
universe.
So it is almost impossible toprove anything, but it is
impossible to disproveeverything.
I grew up in a household thathad every variety of Christian
you can imagine, from Catholicto waving generally on Easter, I
(32:07):
have Muslim in my family, I haveJain in my family, I have Jewish
in my family, I have Wiccan inmy family.
you just,
Don Finley (32:14):
cover the
Sultan Meghji (32:15):
Yeah.
And guess what?
99 percent of the time, it comesdown to being a better person,
Use the stuff that's helpfulthat makes you a better person
than not.
And I'm certainly a person who,in some cases, it took me longer
to get down that journey than Ishould have.
But I gotta say especially oncekids enter the conversation,
which is a trite way to say, butalso in the face of what we see
(32:35):
every day around us, see thatthe more extreme behaviors are
getting a lot of airtime.
They're getting a lot of time onthe social media platforms.
And frankly, I just don't.
Want any of that, I actuallystarted watching the evening
news again, not too long ago,and I got to tell you, it was
soothing because it's 22 minutesonce you exclude all the idiotic
(32:56):
political commercials and it'sreally just, Hey, this is what's
happening around you, all thiskind of stuff.
I'm sure you've got some app onyour phone that tells you the
global, international, importantstuff, oh, there's this, there's
this festival happening thisweekend or this road is being
closed or something just likethe stuff so that your daily
life is a little easier.
And so like watching the eveningnews and I like watching, and
(33:17):
this is a total name drop.
I really like watching CBSSunday morning, which is
probably a code that I shouldretire soon.
Don Finley (33:24):
You're moving towards that
Sultan Meghji (33:25):
I'm getting there.
I'm getting
Don Finley (33:26):
But I do love that because I think what you're
hinting at is the evening newsis definitely more local, Like
you are getting to the pointwhere you're in a sphere of
control that you have moreaccess to than when we talk
about the global nature ofthings and which are of the
extremes.
Sultan Meghji (33:43):
if I am a critical thinker who has access
to the internet, am I going tolearn fundamentally anything
differently about current crisisin the Middle East by watching
any of the major media groups?
No, because it'll be 5 percentnews, 25 percent people talking
about it, and then a bunch of,corporate viewpoints on it or
(34:03):
specific political brand.
Viewpoints on it, most of whichare silly, And so it's really
challenging, I think, in yourmind, to separate out, here are
the facts of what's going on inthe world that I need to pay
attention to.
And for me, it's entirelymacroeconomically relative to my
business, relative to the safetyof my family, things like that.
Then there's the other one,which is what's the stuff I
(34:24):
actually need to know?
Because as soon as I hang upwith you, I'm going to do school
pickup.
And Is the main road I'm goingto drive down going to be closed
or not?
Now can I open up an app and seethat?
Yes, but guess what?
I was watching the newsyesterday and heard that this
one road was closed, so I'mgoing to go the other way.
Don Finley (34:40):
It's actionable.
Now, I know we haven't touchedon quantum, but I got to wrap it
up.
And I think what we're going toend up having to do is bringing
you back.
So for the audience, let me knowif we should have Salton come
back and we'll make that happenso we can dive into the world of
quantum and what's going onthere.
cause you're also an expert inthat space, but in the meantime,
what would you recommend thatour viewers do to be.
(35:01):
On top of all the change that'scoming about in our environment
based on what's happening withAI, what's happening with
quantum, how can they preparethemselves?
Sultan Meghji (35:10):
it's a great question and we could probably
have an, like you said, there'sprobably an entire, episode on
that.
But the basics of cyber hygienethat are well known, don't reuse
passwords, use multi factorauthentication, all that, get
those systems in place.
And if your parents get yourkids to have those systems in
place, because at some point inthe next few years, the entire
encryption infrastructure thatwe're using is going to get
(35:32):
replaced.
Okay, all of it.
It's called RSA currently.
And it's not because it's bad.
It's because quantum computerswill have industrialized how to
break it and bad guys will haveaccess to those technologies or
people with who come from nationstates that are in competition
at minimum with us and want allof our data.
And there are examples but youdon't need me to tell you what
(35:53):
And so just know that it'scoming.
and just the other day, a paperwas released by a bunch of
researchers that showed that aquantum computer connected to
one of the big cloud providerswas able to actually break RSA
mathematically.
they didn't actually do it, butthey showed that there was a way
to do it.
And it's actually on a kind of aslower, older version.
Quantum computer.
(36:14):
the ones that are out there inthe real world and that, the
Chinese have put in orbit andstuff like that because they've
had them up there for a fewyears are far in a way more
advanced than this.
And so that the article thatcame out the other day really
just made me focus and be, yeah,you've got to be quantum ready.
And in doing so, you've got totake care of what you've got
right now.
And if you're a corporate personlistening to this, the easiest
(36:36):
way to think about this is ifyou have a piece of hardware in
your shop, that's older than2018, you're going to have to
replace it because that machinewon't be fast enough to run the
new math.
That's the easy answer becauselattice encryption, let's say
you decide to go down that one,which is one of the potential
replacements, 10 to the fourth,more computational needs.
if you're running it on an I3Intel or something like that,
(36:56):
you might as well cook an egg onthat.
It's
Don Finley (36:58):
Yeah.
Yeah.
you're not moving very fast onthat
Sultan Meghji (37:01):
Yeah.
and the cloud may or may not bethe right solution for you,
depending on what you're tryingto do, because there are going
to be places where theregulators are going to say you
can't use the public cloudenvironments, or you have to use
such a restricted version ofthem that they become cost
prohibitive.
And it's just easier for you tobuy a 3, 000 server and put it
in the closet in your office.
So I would say.
(37:22):
Being ready for quantum is firstentirely about cyber hygiene.
Second, it's about knowing theinventory of your technology and
what's old and what's needing tobe replaced and all that kind of
stuff.
And the third is fundamentallyaround making sure your people
are trained for good habits.
And, this isn't like retraininga workforce.
this is all stuff we've beentalking about for well over five
years, as many as 10 years, thecompanies that struggled to get
(37:43):
people virtual on COVID are thecompanies I worry about.
If one of the one things I'vegot is a little, baby investment
book.
And I went back and actuallyfound that list of all the
companies that got bad pressduring COVID because they had
trouble getting people virtual.
And I've gone in and had an AIlook at their 8Ks and look at
their 10Ks and start to see ifI'm seeing an uptick in
(38:04):
cybersecurity spend to replaceall their legacy technology.
and there are a bunch of marketswhere that hasn't happened yet.
Financial services, there's abunch.
Healthcare, there's a bunch.
Energy, there's a bunch.
And those are all companies thatI would be, if I were to work
with them.
Or if I was a customer oftheirs,
Don Finley (38:21):
Oh, that's a great analysis.
And I love the use of AI to helpyou put that together because
those are usually research jobsthat we would put, a young
analyst on.
Sultan Meghji (38:31):
that, that kind of 101 to 201 level primary
analysis of something issomething that I used to have
two or three analysts workingfor me at any one time.
Now I'm on the treadmill at thegym and I'm using my phone to
call into my private LLMenvironment.
And I'm asking it questions.
(38:51):
And each time I'm like, Hey, bean analyst, analyze this for me,
be an analyst, analyze this forme.
And then it reads it back to me.
Don Finley (38:58):
And at the same time, you're getting responses
in minutes compared to whatcould have been like days long
projects for an analyst as
Sultan Meghji (39:05):
sometimes seconds even.
sometimes it'll interrupt me,which I think is funny.
Don Finley (39:09):
That's awesome.
All right, Sultan, it has beenan absolute blast.
I really appreciate having youon today.
Thank you so much.
Sultan Meghji (39:16):
Thanks for having me.
This has been a lot of fun.
Don Finley (39:19):
Thank you for tuning into The Human Code, sponsored
by FINdustries, where we harnessAI to elevate your business.
By improving operationalefficiency and accelerating
growth, we turn opportunitiesinto reality.
Let FINdustries be your guide toAI mastery, making success
inevitable.
Explore how at FINdustries.
(39:40):
co.
Popular Podcasts
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
Amy Robach & T.J. Holmes present: Aubrey O’Day, Covering the Diddy Trial
Introducing… Aubrey O’Day Diddy’s former protege, television personality, platinum selling music artist, Danity Kane alum Aubrey O’Day joins veteran journalists Amy Robach and TJ Holmes to provide a unique perspective on the trial that has captivated the attention of the nation. Join them throughout the trial as they discuss, debate, and dissect every detail, every aspect of the proceedings. Aubrey will offer her opinions and expertise, as only she is qualified to do given her first-hand knowledge. From her days on Making the Band, as she emerged as the breakout star, the truth of the situation would be the opposite of the glitz and glamour. Listen throughout every minute of the trial, for this exclusive coverage. Amy Robach and TJ Holmes present Aubrey O’Day, Covering the Diddy Trial, an iHeartRadio podcast.
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.