Projects
![]()
No projects found
Investors
New
DAO Treasuries
DATs
Lend / Borrow APRs
Chains
Categories
Spaces
AI Meets DeSci: A Convergence of Minds on The DeSci Open Mic🎙️🌟
Recorded: March 27, 2024 Duration: 0:59:12Space Recording
Full Transcription
Music Thank you. Music Thank you. Music Thank you. Thank you. Music hello jam everyone
thanks for joining us uh everyone, for our special episode of Desai Open Mic, where we're colliding the worlds of Desai and decentralized AI.
We've got some fantastic guests today.
So, you know, kind of we'll go around the table, you know, in a moment and, you know, chat with our special guests.
But, you know, first, I'd like to say hello to Aaron McGinnis, my co-host, and also Crypto Shrimp PhD PhD who's behind the Desai Mike account.
So, you know, maybe first off, you know, we sort of got Idiom from Ocean Protocol who's joining us.
And it's actually a big day for Ocean Protocol, but, you know, would actually just love to chat maybe first about, you know, the topic at hand, which is, you know, decentralized AI.
So, you know, I mean, it's a relatively new concept and maybe a new one to some of our listeners.
And maybe you could, you know, just share what your views are on on hello or are they
uh is it directed towards me
uh not yes yes yeah the question to you um Sorry if my audio is not coming through very well.
I've been having network troubles all day. Yeah. So the question to you.
Yeah. What is what is decentralized AI mean to you as a as a as a sort of definition?
And then also, you know, we'd love to hear about what what Ocean Protocol is doing with decentralized AI.
what Ocean Protocol is doing with decentralized AI? Yeah, sure. Yeah, so, you know, it's very
interesting how many, how the different protocols are approaching decentralized AI. You know, we have
everything from the ability to access decentralized infrastructure, you know, going back to the days
of like SIA coin and nowadays more a cache network. And then, you know, going back to the days of like, see a coin and nowadays more a cash network.
And then, you know, we also have things like
what we're, things like what we're doing at Ocean, right?
Where we have things like IP licensing and access rights.
And you're seeing more of these mechanisms
being applied to attestation of IPs,
things like story protocol as well these days.
But we also have new mechanisms
and there are aspects that are more specific towards ocean.
I'm primarily referring to compute data as an example,
whereas before you had to make your data accessible
for others to compute with it,
these days what you can do is rather than making
your data available, you make it so that people
can bring algorithms to compute on top of that data.
And that way you can better manage access
and how others can use that information.
And it's continuing to evolve as well.
We have new ways of working with decentralized
or putting AI to work for us in a decentralized manner.
And that's where, you know, we've had some interesting approaches with this more recently with Predictor.
So hopefully we'll get to speak a little bit about that today.
Yeah, yeah.
I mean, you know, I think for, you know, DSI organizations, right,
I think a lot of them these days are using sort of like homebrew
compute solutions. I saw photos come across my Twitter feed about clusters people have
built in their basements and garages. I think many just use the centralized cloud providers. So, you know, I think computed data is a very, you know, powerful concept that can be leveraged for sure.
But, yeah, before we circle back on Predictor, you know, we'd love to hear from Aaron and some of our other guests.
Absolutely. Yeah, I'm personally super excited for this conversation today. There are some
incredible guests speaking and the work they're doing is really amazing to be moving this space
forward. And we also have kind of a wide range of perspectives on this conversation about how AI and blockchain intersect
to really advance science and push science forward in a positive way. So super excited to
be having the space to chat with everyone here. And I think there will be a few other guests
tuning into this conversation that might have some interesting questions and other perspectives to add some more color into it, too.
So, pumped to dive in.
Yeah.
Yeah.
Yeah.
Great. idiom, I mean, you mentioned predictor, and like, I know a little bit about it. And I think it's a very interesting approach to sort of like decentralizing AI, decentralizing predictions,
maybe you could tell us a bit more about that. Yeah, sure. So, you know, predictor is,
you know, we call it a bit of a new component, or new primitive in terms of how we can establish prediction feeds.
And so with Predictor in particular,
the way we're approaching decentralized AI
is rather than making infrastructure
that you can perhaps rent, like what you were saying,
Merrick, having multiple clusters of gpus that you can
access that you as yourself as an individual can provide to the network rather than focusing on
that approach with predictor what we're doing is we're saying okay um let's uh create a framework
that um uh so uh with so first let's think of predictor as, there's some characteristics about it. First, it's a feed,
right? It's not whether Joe Biden or Trump is going to win the election. That's a single event,
right? But we're very much focused around our feeds of data. So for an example, what is the
temperature now? What's the temperature going to be in five minutes and 10 minutes from now? And because we have the speed, right? And
we have these periods of time, we are creating a system where many people can come and submit
their prediction for the next five minutes. You know, and there's a concept around this
that goes back to what is called the wisdom of the crowds.
And then that is, if you have a big jar of gumballs
and you ask a hundred people to predict
how many gumballs are in there,
no one's gonna really get it right.
Maybe one person's gonna get it.
But what you can do is actually take all of their data
and create an approximation from it.
And you're actually gonna get an extremely accurate answer
as to how many gumballs are in there.
So, you know, the other part about this
is that when people are guessing the gumballs,
they don't really have any stake, you know,
there's no risk associated with it.
If the weatherman tells you that tomorrow
it's going to rain and farmers are dependent on that to make preparations for their farm,
you know, if the weatherman is wrong, there's no accountability towards the weatherman.
So, you know, Predictor takes various, these various concepts and brings them together around this concept of a prediction feed
right where we can build a stack for data scientists and people with ml models or other
models to come in and submit a prediction for the next five minutes for an event whether
in particular right now our our predictions are binary, which
means that it's either the value is going to go up or the value is going to go down,
whether it's temperature or the precipitation of rain or the price of Bitcoin.
And by creating these new set of contracts and this framework where we can have many people submit their predictions
and continue to improve the accessibility
to all of this data and how to build models,
we're hoping that we can access the wisdom of the crowds
and bring this concept of accountability through stake,
such that the people that are correct are rewarded
and the people that are correct are rewarded and the people that are wrong
are slashed um so i'm going to pause there as a as a gentle intro to to what predictor does
and what it is okay thank thanks for that actually um yeah that's that's cool it sounds like you know
uh a very interesting take and a very interesting model on, I guess, like, you know, maybe more
traditional prediction markets. So that's super cool to hear. We'd also like to hear from Cosimo
Capital. As I understand it, you know, sorry if I'm wrong or off base here, I understand that you're
involved in the BitTensor ecosystem. Maybe you could tell us a little bit more about that model
and how you're involved.
Sure, can you guys hear me?
Yep.
So we are a liquid token fund based in Chicago.
And about last July, we started getting very involved
in the BitTensor community.
And so what BitTensor does is it creates subnetworks under its own chain.
And what those are, they can be networks that do healthcare.
There's one that's an LLM.
There are some that are data scraping subnetworks.
But the chain gives emissions of new tokens to the people that are performing well using those subnetworks.
So it's the same tokenomics as Bitcoin, 21 million, and it inflates at the same rate.
But they're planning on going from having 32 subnets this year to 1,024.
And so what we are thinking is one of the best ways for a decentralized science project
to get capital is to start and put machine learning models
onto a platform like BitTensor.
So that, I think, is one of the ways to get immediate revenue and immediate attention.
And I think that's going to happen across the board.
We as a fund are thinking about doing it ourselves.
That's OK.
That's very interesting. I know BitTensor has gotten like a ton of hype this year. I'm excited to see the rollout. Science Stanley, you've also joined us. I know that
you're quite involved in actual research, right?
And just maybe you could speak to how you're solving your compute needs right now
and whether a decentralized infrastructure network
or decentralized AI kind of makes sense for your use case.
Heck yeah.
And also so happy to be here. Um, good morning, everybody.
Um, for, uh, for one reason too, I'll share the last time, um, I was on a, a bankless, uh, speaking
engagement, um, was talking just about this issue of kind of like, how do we make sure that
researchers who are doing critical work,
but, you know, not work that is likely to have a thousand, ten thousand times profit,
how do we make sure they also have access to compute?
And yeah, so this was for the Bankless DSi Unconference.
And somebody, our dear friend, Renee, reached out to me right after and said, hey, why don't we build a computer that could be helpful to researchers like you?
And so I'm just adding to the thread here, several months later, a picture of the finished computer that we'll actually be using for some really cool oncology work at Stanford.
using for some really cool oncology work at Stanford. Our hope too is that this can be just
the first node of a cluster that'll be geared specifically towards precision medicine and
bioinformatics. And yeah, I think it's a kind of cool, small example of a use case where we really probably are and are going to need to have access
to some pretty good decentralized compute infrastructure. So yeah, very excited to kind
of share that little anecdote. Another thing I would maybe share, oh, I see Renee's in the
audience too. Maybe she would be willing to come up and talk a little bit about the journey of our server.
But then, yeah, as for the kind of specific thing that I'm doing with models and that we're going to be using this machine for,
I'll just give like a quick intro.
And if it's something people want to talk about more, I can get into it.
In oncology practice, data is a huge problem. Uh, increasingly we have much more we
can do for patients than, you know, just chemo, but a lot of the times it comes down to having
high quality data on, um, you know, multi-omics from similar patients. Um, sometimes we need to
run specific pipelines on our patients and we need certain reference databases.
I would say overall for every day we spend doing what I consider useful data science. We spend
two or three days cleaning data, begging for data, searching for data. And it just shouldn't
be that way. And it really, especially now, doesn't have to be that way.
And particularly because what I'm interested in is the idea that instead of data sharing, we can train and share models. And so, yeah, actually very specifically, we're training a
series of models on multi-omic data for tumors that are considered immuno cold. These are tumors that typically are
not affected by immunotherapy, but in some situations they can be caused to be affected by
immunotherapy with different techniques. So yeah, the idea would be for many people working with
these cancers, typically like liver cancer would be one of these immunocold cancers.
We would hopefully, so far we have Stanford and Cedars on board, and we're hoping for maybe
five or six more institutions. We would create a situation where the data from all those groups
could be just for a short period of time pooled into a training cluster, then a model that generates clinically applicable
data could be trained. The training data for that model would then be erased, but then the model
would be shared through probably some kind of decentralized infrastructure. So anyway, that's
what we're working on and what I'm really excited about. And, you know, I think it's really just one
example of what these kinds of techniques are going to do for medicine.
That sounds like great work and, you know, extraordinarily valuable.
You know, before we sort of circle back and sort of let, you know, people kind of like respond to that. We also have
FlowScience joining us. You know, maybe you could also speak to, you know, how you're
working with like, you know, the data that you're producing and collecting. You know,
most science these days requires at least some sort of compute. So I'm curious how you're sort of solving those problems. And yeah, you know, what connections you see to decentralized AI, machine learning,
and infrastructure? Yeah, definitely. Thanks for the invite. Yeah, so my background is as a
molecular biologist. And if you're familiar with bi biologists most of us are pretty bad at math sometimes computing and stuff especially like the lab I'm
like a lab scientist by training but I did a bunch of data science studying
genomics right studying like a whole genome data sets and just sort of half my
way through it and during grad school and then I started working in the
cannabis industry and that's kind of what I've been doing now I was generating
data I was running a lab for a while and now I'm building sort of a data
infrastructure for that industry and I still do a little bit of data science I
was working using labdows flex platform last year to do some modeling of binding
interactions between different molecules I was looking at in different
proteins in the cannabis plant and I was able to get sequences of I'm sort of doing things like
that nothing too crazy and nothing really super AI focused I'm still definitely learning and and
you know getting my feet wet in that world but that's kind of my background in terms of where I'm at with the data and computing stuff. You know, Dr. FlowScience, we need to do another call soon. You know,
it's interesting, until recently, large language models didn't have a good
application to genomics for some kind of boring math reasons.
But just recently, there's a new family of models that I think will blow you away, my friend.
Yeah, definitely down to check it out.
Aaron, are you speaking?
I can't hear you.
Maybe it's me.
Can anyone hear Aaron?
I cannot hear Aaron.
Maybe she can reconnect, disconnect, reconnect.
This is, you know, wouldn't be a Twitter space without some technical difficulties, y'all.
Yeah, indeed.
It happens like every single time.
But, all right, well, you know, when, well, I guess till Aaron comes back.
Well, I guess till Aaron comes back.
And, you know, Idiom, I mean, we've heard a couple of common, you know, threads, you know, from the other speakers.
for D-Sci projects and others to sort of monetize and share their data and models and train models.
Did you want to elaborate on that a little bit?
Yeah, sure. I'm very sensitive to a lot of the topics that at Cosmic Abil, nope, sorry,
Science Family and Flow Science have been discussing.
My wife is a physiotherapist that works with ecology.
So all of these things are very dear to me.
And I think that, you know,
Ocean provides a lot of tools that i think are challenging to find
uh you know a lot of our uh customers primarily for a lot of the tech stack that we have built
are more on the enterprise side of things and due to their restrictions and regulation and many of
these very big challenges.
One of the key things about the ocean stack is that there's a lot of
services and configurations that are specifically tailored towards
the sensitivity of who has access to this information and
who has its own access and provides the decryption.
And similar to what Science Stanley just mentioned,
it'd be fantastic, right?
If we had all of these individuals who for a small period
of time could provide access to that information
to their own individual information
and such that we could train these models,
have this data temporarily accessible,
and then remove that access,
such that we have the trained model now to work with.
And, you know, I personally am a big believer that,
you know, before we get to some of these major outcomes,
and some people are already heading towards this way,
such as genomicsDAO.
You know, we're seeing these concepts of data unions
where individuals who have problems maintaining their data
or perhaps taking it from their practitioner
or their doctor to a surgeon
and that exchange of data
between all of these different
institutions is extremely difficult at the moment. So, you know, I'm personally a big believer that
one of the first bottom-up adoptions will be if we can solve these individual problems where
the individual gets to on their data, they now benefit from, you know, having these records and
be able to manage the records on their own.
And then we can enter a world
where we can help provide this trained models
and this data becomes accessible
because I'm incentivized as a person
to make sure that my health stays up to date.
I thought I would touch on some of these things
and it can even dive a bit more
in terms of how OCE ocean can also help data scientists
for an example, monetize their models
and apply their models such that then
once the model is there, me as an individual
I can then push my data to it and then get the results.
So yeah, I'll pause there.
Awesome.
Yeah, thanks for that kind of like overview.
I think that that's tremendously valuable for the DSi community and data science community
and really anyone with data or models and, you know, to sort of get to monetize those things.
Erin, are you back?
I think so.
Yeah, we can hear you.
Yay, she's here.
I've returned, finally.
Yeah, I've been able to hear at least part of this conversation.
And I think it's really kind of essential some points that we're just being touched on about allowing people to really contribute their data
into some of these larger data sets. And Stanley, I know you have some thoughts on what this kind of
looks like in actuality, especially from kind of personalized medicine, rare disease perspective.
especially from kind of personalized medicine, rare disease perspective.
Would love to kind of have you tie that into maybe some more of the bigger picture,
what this looks like through maybe some of your work in partnership with Stanford.
Oh, my gosh. My goodness. She's setting me up to talk about the hackathon.
Thank you so much, Erin.
And 100 percent, I'm leading a project at Stanford called the Rare Disease
AI Hackathon. I'm going to pull up the link so everyone can check it out. But this has been such
a fun experience and such a fascinating learning experience. What we're doing is we're working with
a pretty cool network of physicians, patients, and researchers who are associated with a program
called Research to the People at Stanford. Research to the People is an unusual program.
It's been around for quite a while. I like to think of it as pre-DSI-DSI. It's a kind of unusual
group in the Stanford ecosystem that has a mission to collaborate widely.
And so typically we're picking particular illnesses or patients with unusual constitutions, constellations of symptomology, I mean.
And then we're sort of like, I like to think of it as we assemble the Avengers of kind of top contributors in new fields.
And we kind of build these multidisciplinary collaborations to, you know, apply new technologies, find new treatments.
And, yeah, we're so interested in what, you know, AI is going to be able to do.
So we've picked one or two rare genetic illnesses that are kind of characterized by a real lack of access to expertise for the patients.
One of the main conditions we'll be talking about, my dear friend who's in the audience, Krypto shrimp knows a lot about, and it's called Ehlers-Danlos syndrome.
It's a genetic condition that comes from a relatively minor set of
mutations, and it leads to highly variable and often quite severe issues with protein
folding related to soft tissues. It can kind of range from anywhere of a patient who has some bad joint pain or hypermobility to patients who have pretty severely impacted mobility and lifespan.
And it's a very, again, very hard disease to diagnose, very hard disease to develop treatment paths for.
disease to develop treatment paths for. Kind of funny, the day before I presented this project to
the Snyder Labs faculty, Greg Brockman, who is one of the co-founders of OpenAI,
he actually tweeted a really beautiful story about his wife's journey as an EDS patient. And
I'll see if I can kind of find that tweet to share.
But what Greg's wife went through is what all of the EDS patients I've known go through, you know,
five, six year journey of talking to lots of doctors, you know, being told it's all in your
head or it's psychosomatic or, you know, various other kind of dismissive things. And then usually,
eventually, these patients find a doctor who's seen EDS before, and that doctor knows immediately
that they have EDS. And yeah, so what we were kind of curious about is, is there a way for
decentralized intelligence, decentralized AI to get these patients access to better care?
decentralized AI to get these patients access to better care. And then what we kind of hit upon is
another layer of decentralized AI and decentralized AI research, not the compute layer,
not the data layer, but the human layer. With rare illness and EDS as an example,
there's this huge network of people, patients, their families, their doctors, the
scientists working on the condition, who all have kind of a different piece of the puzzle.
And so such a cool question, how do we create a system where these models can exist and the data
they're trained on can exist with transparency? And also a system where we can invite in doctors, patients, and other experts
to talk to our models, to evaluate our models, to validate our models.
And that's exactly what we've built. We're kind of actually this or next week launching
a research portal. And essentially, we're copying our good friends over at Chatbot Arena.
Our physicians, our patients will log into this portal.
They'll get a couple different AIs to talk to.
And then it's basically like AI Tinder.
They get to ask questions to a couple models.
And then on each model's answer, they swipe left and right.
And then that input on the responses gets fed back into the models, fed back into the training.
And all of
this is something we're going to be open sourcing and writing a paper about. And, you know, our kind
of hope and our increasingly strong feeling is this will be a standard part of medicine within
just a couple of years. We sort of will need to have distributed networks of physicians and medical experts
training and validating these models.
I almost think the same way we expect doctors
to kind of sit for their board exams
as part of like, you know, good practice hygiene,
we'll sort of expect doctors to spend some of their time
contributing to this, you know,
collaborative intelligence effort
of maintaining high medical grade AI models.
Completely. I think that quality piece really ties into a larger conversation about how can
impact be had through some of this data
and creating or leveraging some of these different AI models.
And I would love to throw it back over to Idiom and Vladislav
like to kind of pull that conversation
and maybe pull in some other trends
or work being done beyond just the DSI sector
and how that might be able to advance
some of the efforts in DeFi,
maybe pulling in some different perspectives
or approaches from DeFi and data just generally as well.
Yeah, I find that at least I might talk about the,
Yeah, I find that at least I might talk about the,
just the DeFi first, because I think that just in general,
I think it's healthy to have some foundations
around how we can unlock the economy
and the access to capital such that more and more projects can succeed as
a result. And to me, I just find that to be just a fundamental unfortunate aspect of expensive
with a lot of bureaucracy, red tape type work, and that is a lot of healthcare and rightfully so. But, you know, I'm not sure how to pull in, but maybe I'll just
double down on Science Stanley. And, you know, throughout a lot of this time, you know, I talk
a lot about how, you know, we have to be mindful at the business aspects, the financial aspects,
the longevity of what we're building
such that we can succeed in the space.
But I always find that it's just so fundamental
for us to think that in the end,
I believe we're building all of this
towards improving the livelihood
and the systems that we operate with.
And when it comes to healthcare, it all goes back to, you know,
the physicians and the institutions and the patients.
And I'm just personally a very big believer that, you know,
some of the earlier adoption that we're going to get in the space,
at least in terms of making some of these unlocks,
comes from empowering the individual, right? The whole concept of self sovereignty.
And I just, that's such a fundamental part of the puzzle.
And the more we can really put the individual
at the center of the benefit.
I think that's when we get a lot better outcomes from it.
So, you know, on a different universe,
I think that, you know, data DAOs are one of the,
I would basically put a lot of effort
into just that segment alone, because
again, onboarding every individual such that they can own their data and they can help
to establish these models and then onboarding in addition, the physicians that can, that
have all the knowledge and can help to improve the quality of these models it is so
fundamental um and uh i think it's closer to the first step of of this journey you know um
than we normally spend we don't spend enough time there um we're really many of us are very futurist. We are technologically very inclined.
There are many PhDs that are just visionary.
But oftentimes, we are so far ahead of the individual.
And I think it's really important for us to ground ourselves
and meet in the middle and meet where people are at today.
You know, I think the more we're able to start from that point, the more effective
will be towards ushering change, especially when it comes to healthcare and areas like
oncology.
So I'm going to pause there.
Right.
That's all very true.
Oh, sorry.
Sorry, Erin.
Go ahead.
No, I was just going to kind of amplify some of those different points.
So feel free to take it forward.
Well, you know, I mean, we're sort of like blessed in this Twitter space, ongoing Twitter spaces that we get very smart speakers, also very smart audience.
So if anybody in the audience wants to, you know, come up and, you know, talk about,
you know, your interest or involvement in decentralizing AI and models, feel free to
request. Also, if you have any questions for the panel to ask, well, we also request the speaker
spot. Stanley has something to add to that, though, so go ahead. Oh, forgive me, too, and I don't mean to put her on the spot,
but I would like to shout out Renee for building this incredible server
that's now going to be training oncology models.
Did you want to tell us about that, Renee?
I did not realize I was a speaker like this whole time.
Yeah, I can tell you a little bit about it.
I don't wanna take too much time,
but we built a server with the ability to hold multiple GPUs.
It's got like a 64 core amd epic which is
one of the more powerful cpus you can get um i probably should write a blog on the build or
something it took a lot of work to figure out the hardware and like the compatibility issues
and um like at several points we had to upgrade pieces of hardware just because like what we got wasn't going to cut it.
I had to learn about server RAM and like ECC and other different requirements.
So overall, it's been super cool supporting Stanley.
So thanks for the shout out.
Thanks for the shout out.
And I'm like hooked on building servers now.
And I'm like hooked on building servers now.
Yeah.
And I have to say, I don't know.
Can we get a show of hands?
Anyone in the audience love building computers?
That's my grown up Lego time.
But, you know, just like Renee mentions, it ain't easy.
And there's so many little ins and outs.
And it's one of those places
where, you know, I think by having a culture around decentralized compute, and even just like,
as Rene says, sharing blog posts, sharing experience, you know, we really probably can
get to a level where we're building decentralized clusters that can compete cost for cost with the institutional compute.
Yeah, that was going to, that's something that, you know, is at the top of my mind.
I mean, Rene, not to put you on the spot again, but, you know, I mean, having gone through all that,
do you see, like, the value in perhaps, like, using, well, first of all, like now that you have that,
you could definitely contribute to decentralized compute networks
with sounds like quite a big splash.
But also like, is there a value in, you know,
just maybe like picking up some tokens instead
and then just, you know, farming out that work
or maybe it wouldn't make sense in your use case.
I don't know if you can comment on that.
Do you mean like renting out compute power? Yeah, but not only renting it out from like, I mean, something like Amazon, but to, you know, available compute out there and possibly also, you know, making
your models available, you know, through something like a decentralized market or even just,
you know, once your current research is done and you have this server sitting there, perhaps,
server sitting there, perhaps, you know, contributing your compute to others' needs as well.
you know, contributing your compute to others' needs as well.
Oh, I see. Yeah, definitely plan to do all of that as we build out more utility with our models.
I'm sure we're going to make those accessible in some format. I've already been looking into
Akash Network, which is a GPU marketplace, and they have a provider program so you can get paid to provide compute.
So I'll probably do that when it's servers not being in use.
Right now we're supporting an active project. That's pretty important.
So I don't have the server configured for any sort of compute leasing, but definitely interested.
I'm talking to several projects
that want me to operate nodes for their protocol.
So I am doing that type of work.
Likely I'm going to build a second server
dedicated for that though,
just because it is pretty intense
to just like rent out your GPUs all the time.
I am going to share even just an ambition I have for the node Renee's been building.
Let's say it's a number bigger than two, two nodes.
And yeah, really optimistic that the kind of initial projects that we're doing are pretty high level in, you know, for any ecosystem,
and will kind of prove that this kind of compute paradigm can sort of meet state of the art.
But yeah, longer term, the idea that there could be these kind of nodes that
at given times are sort of fluidly contributing to decentralized compute pools, but that through,
you know, some kind of
governance or organization can be activated for specific projects or larger scale training.
It's just very exciting, you know, and, you know, it's one of those moments where it's kind of a
scrappy concept and certainly no one's ready to compete with AWS yet. But I wouldn't be surprised
if we get there, you know, if within four or five years, if you have
a large-scale compute project, if it's a pretty hard decision to make, AWS or Renanet.
I'd also like to add that if you do have that compute capability or those models,
BitTensor, on BitTensor, you can be a miner.
And so effectively, you can think of it like Bitcoin mining.
You know, that was a hardware problem.
This one is instead of proof of work, this is proof of useful work.
So you can DM anyone can DM me if they want and have something that they might want to contribute and start earning money or at least contributing, just to add to that point.
Well, listen, I'm going to reach out and I would absolutely love to learn more.
I have to say, even just as a systems engineer, the topologies of these kind of networks are so novel and fascinating.
So even just getting to hear about what people are doing is so much fun.
So even just getting to hear about what people are doing is so much fun.
I thought I would add to this, you know,
and maybe this is where I can bring in some of the DeFi aspects to it.
And yeah, you know,
I am a big fan of like flywheels and how to add value into the space.
And I just think that sometimes the,
we put in so much effort, you know,
with such precious time and resources,
which is primarily our brain and the time,
the limited time that we have.
And I'm always just thinking about how we can hack
these things, you know, and I just wanted to put it out there
that, you know, things like,
I love that you guys are looking into a cash network,
Renee, and I love that you're building your nodes.
You know, the first thing that comes to mind is that we have all of these communities.
There is a lot of ways to bring value and to generate value such that it can be directed
toward supporting these initiatives.
I was just mentioning to Merrick before we joined on here, how, you know, on the,
more on the DGEN DeFi side, you know, you have,
there's primarily this interesting token inside of,
I believe it's Coinbase in the base ecosystem called DGEN.
And DGEN is a bit of a meme coin at heart,
but what ends up happening is that there are more and more apps using it to
integrate it such that value can be captured and directed towards these objective functions.
So let me just try to exemplify this a little bit. Rather than building or focusing on establishing nodes, you know, as a first step, there might be ways to,
for an example, airdropping or onboarding many communities with a token like DGEN, which can be
a token that many people have access and maybe they transact with. And through the transactions that happens there,
fees are captured, which are then directed towards
buying a cash compute that anybody can then use
and have more access to it.
So there are many ways that you can create value flows
by bringing all of these ecosystems together through a token that generates
value, which then can be utilized and leveraged towards unlocking some of these very difficult,
challenging things that we have, right? Which is like, how can we raise capital and start generating this value such that it can scale
and it can be used in a variety of ways
that the more integrations,
like there's a cash fee that buys CPU that others can use,
there are additional features or additional use cases
that can continue to generate more value,
which then can again be used towards improving the whole design space
access to whether it's compute or anything else. So I just wanted to just touch on that briefly,
and I'm being very hand wavy at the moment. But the whole concept there, right, is that
I believe infinite,
the concept of infinite gardens has many different flavors.
And one of the flavors that I continue seeing
are communities just creating amazing value flywheels
out of thin air, just by really thinking about the tokenomics
and how we can direct them towards these objective outputs.
So I just thought I would bring this up.
Yeah, interesting.
You know, we had a great conversation last week about, you know,
DeFi incentives, you know, with respect to DeSci.
So that's a very timely commentary.
I think that that's going to be an ongoing thread in these discussions and spaces going forward.
So thanks for that insight.
We've got one of our audience come up.
Pamela, did you want to tell us about your interest in decentralized AI or DSi or if you have a question for the panel?
Yeah. Hello, everyone. Thanks for your comments and having me here.
So I want to tell that part of this iMexico work is to disseminate about the bans in science using technology to improve diagnosis, for example.
And speaking from reproductive women health
problems with artificial intelligence, we could have faster results.
And I remember a few days ago, I was reading about ETH spin-up called DA NOSE, and they
are developing an algorithm that can be used to analyze ultrasound images of the womb.
So I think this should enable doctors to diagnose endometriosis more quickly in the future.
Because, I don't know if you know, but normally it takes up to 10 years to diagnose endometriosis.
Using AA can help avoid triggering other conditions.
can help avoid triggering other conditions.
So I think we should make some plays where this artificial intelligence
are exploited and how the knowledge of experts in areas of medicine or science
participate and we improve the use of this technology.
Well, I think it's all. Thank you.
So, I don't think we've got like a ton of time left.
I have time for one more question from the audience.
I know, Aaron, not to put you on the spot,
but I know you kind of had an announcement of an announcement to make.
Yeah, always have to go meta with all of the different conversations.
conversations. So kind of leading off of some of the different just, I guess, branding
considerations that Bankless, Bankless DAO has gone through over the past handful of months.
Bankless DCI will be going through a rebrand as well and really working to align strongly with
really the values and intentions for this account and for kind of the movement in the community that this space is creating and working towards.
So starting next week, be on the lookout for that rebrand and kind of what the new public positioning for these different spaces will be.
It's still heavily in alignment, but a little something to look out for in the upcoming week.
I got to say, I'm so happy to see Bankless dipping its toe into the Desai space.
And just couldn't be more excited to see
what's next and and be a part of it um you know i feel like in the past couple years all of us
web3 denizens have seen um so many different entities that kind of come into the space
professing values like going in a different direction and i i sort of feel like bankless's
core mission to increase accessibility and fairness to banking services, like is is one that they've kind of kept to.
So, yeah, really excited to see what you guys do over here in Desai.
OK, OK. Thanks for those kind words, Science Stanley.
Yeah, we'll be back.
Like, you know, I mean, this is, you know, one of my favorite topics.
I always look forward to this space every week.
So definitely we'll be back.
So, you know, I'm not seeing any audience members, you know, requesting to speak or ask questions. So I think probably we'll wrap it up there. What do you think, Aaron?
Yeah, just want to give space for any of the people up here on stage.
If there are any other points you'd like to emphasize, just keep in people's brains kind of top of mind when they think of AI and VSI or any other points that we didn't quite get to that you'd like to take a moment to touch on now?
interested in getting involved in D-Side from a profit standpoint. It's relatively easy still
to become a miner on BitTensor and get paid in their native token Tau. So feel free to
DM me or anything like that and I can give you some more information.
I thought I would just thank everybody for having me on here.
And similarly, you know, if you are a data scientist, Ocean Protocol runs data challenges
quite often.
They help to onboard and showcase how you can bring work that you're familiar with,
such as working with Python and just doing data science,
and then deploying your models
and building analysis on top of the ocean stack.
So we often run data challenges,
and we're always looking to integrate
with other networks as well.
So it's been very awesome to play with the cache recently.
And I've also been following BitTensor for quite a while.
And it's an amazing space to be, the Web3 space.
And in particular, DCI is a critical component, I think, of the future for how we do things.
So it's just great to be here.
Thanks for having me.
And if anybody needs to ask any questions, please feel free to reach out.
I really love how there are so many incredible ways to get involved at this really exciting intersection point that can fit each person's skill sets and interest. And that's very much true across the entire DSI ecosystem as well.
I just came off of the DSI London conference this past weekend,
and there's some really exciting progress being made across so many different
orgs in the DSI space.
And we'll continue to have more of those on every Wednesday on this Twitter
space. That's not going away, even if we're if we're going to be changing the name a little bit.
So if you're also interested in a specific topic and would like to see speakers and leaders in that space discuss that topic as it pertains to BSI, let us know.
let us know and we can reach out and get some of those people on.
And we can reach out and get some of those people on.
Or if you are kind of leading efforts in a particular domain,
also let us know and we'd love to have a conversation on the space that you're in particular building out.
So really appreciate everyone coming in today and definitely follow all the speakers up here.
Take a look through the audience,
some really great people, a part of this community, and we become stronger through coordination and just collaborating across our networks. And that's one of the intentions of
just bringing together people in this type of way as well. So please take advantage of it.
Heck yeah. I hope everyone has a wonderful rest of their day as well.
Thank you. Thank you everyone. Thanks everyone. Thank you.
