---
title: "AI DATA INFRASTRUCTURE | Twitter Space 錄音 | AlphaGrowth"
description: "2024-03-01 的 Twitter Space 錄音，來賓：WOLF, Brian D. Evans, AIT Protocol, AIT-t0nY"
url: "https://alphagrowth.io/zh-tw/spaces/ai-data-infrastructure"
---

# AI DATA INFRASTRUCTURE

錄製時間： Mar 1, 2024 時長： 0:57:57

## Space 錄音

[開啟原始 Space](https://twitter.com/i/spaces/1yoKMwlPEmeJQ)

0:00 57:57

### 完整逐字稿

all right everybody we are gonna get started here in just a second it's

giving me a little bit of an error with the AIT account as I'm trying to get it

up here and I see you also Tony if you want to go ahead and hit the request

button all right Brian hope your morning's going well yeah going well

going well can't complain with the way the crypto markets are at the moment

yeah yeah I'll say what's um where do you lean kind of in this crypto spectrum

do you think that this is kind of a brief rise following you know maybe the

ETF approval and stuff or do you think this is start of the next bull run no

definitely into the bull run I think we're in kind of the super cycle phase

because of the having coming up so I think we're still yeah of course we're

gonna get pullbacks but I think with the proximity of the having you know and

sort of mainstream catching their eye again I think we're still early in this

in this sort of extended cycle got it got it all right Tony I shot you an

invite AIT protocol working to get you up here make sure that you are on a

mobile device if you are gonna be a speaker for today I think desktop still

has plenty of glitches so it runs into issues today we'll be chatting with AIT

protocol if you're in the audience we're going to appear just a second pretty

popular accounts obviously working on the data infrastructure side of things

really with decentralized AI so have it a little bit of AI talk here having a

little bit of web 3 talk I would say these are some of the top topics at the

moment if you're kind of looking around the ecosystem for you know what are

people talking about what's popular but that hasn't stopped them from

continuing to run and so you know I'm keeping my eyes on them throughout this

ecosystem and I see them requesting again all right I think now we're gonna

be able to get them up here and then Tony if you want to go ahead and join up

as well that would be perfect then we can go ahead and get rolling but right

now I do believe I was able to get AIT up on stage so as soon as we're able to

give me AIT there we go we got you yeah thank you I forgot about you know the

Twitter base is only accessible on the mobile yeah like and I think Tony is

trying as well he's been using this course mainly you know and so this is

his first time doing Twitter space just gives him some time oh I think he can

smell cool yeah looks like we got Tony up here you guys good to go yeah yeah

okay cool so I'm gonna roll us into this we're gonna be chatting here for about an

hour with AIT protocol for those that are unfamiliar you can click right into

the account I also pinned a little bit of information at the top of the space

around some of the things that they are working on whether it's the LLM

supercharger for BitTensor or perhaps you want to look at their NFT revenue

share there's a couple of items that could speak to you if you are someone

that is in the whip through space and if you're someone coming from the

so with that being said let's go ahead and get into it I'm not sure if AIT or

Tony which one of you wants to answer first but I think at the start it'll be

good to get some introductions here just kind of high-level you know what is AIT

for the audience as well as a little bit of background on each of yourselves and

then we'll take it from there okay cool I can go first so my name is Shin you

know I'm a CEO of AIT ProCo I started you know my crypto journey back in 2017

as an you know ICO investor everyone stars as ICO investor I have been

surviving you know through two cycles and you know founded my own ventures

it's called Mega Lab Ventures you know where we focusing on investment in the

web 3 space back in late 2020 you know we focused more on DeFi and GameFi

startups and you know after a couple successful investment I started you know

stepping into the project section you know where we I helped co-founded a

game five projects you know in incubation support by downmaker after

following a remarkable achievement I you know make a decision to kind of like

step away a little bit for the company and concentrate more on the investment

side but in 2022 when you know the chat GPT emerged all the you know every

everyone was talking about AI and how you know I can help with the daily

activities and I started to realize that you know AI actually can be

integrated into blockchain maybe not feasible right now but you know we on

the right correct way you know with the tensor and I'll explain later on so I

founded IIT Proko to explore the AI sector trying to bring AI innovation to

benefit the web 3 community. IIT Proko you know is the first web 3 data platform

space like in data annotations and AI model training where I can give more

definition and more introductions so I'll let Tony introduce himself first and

then you know we can talk more about IIT. Hello hello I hope everyone can hear me

so I'm Tony I'm the CPO at IIT Proko. I am actually an MIT alumnae Google cloud

professional machine learning engineer data engineer dbt expert so I am your

go-to guy for all kind of data and AI stuff. Perfect thank you for the

introductions from both of you there now I know you've covered a little bit of

this around you know what is IIT. I'd love to go into kind of what inspired

really the creation of IIT protocol and then a little bit of the vision for the

future for web 3 and AI integration because those are some of the hottest

topics people discuss nowadays. So yes as I just mentioned earlier IIT Proko is

the first web 3 data platform specializing in data annotations and AI

model training in the web 3. We're using you know the blockchain for

just less global labor market incentivized by crypto economics and

offering instant international payments so basically you know when we look into

the AI development we try to understand like which state of the

development you know cost the most and which one is you know require the most

labor force and it turned out the early the very early state you know that the

data allocations states where we need to label the data before uploading it to

the machine learning for it to learn it cost is the consumer like the the

largest portion of the development purchase and so we think this is where

the money come from right this is where like the AI the biggest AI company where

they spend money into so we try to bring that to web 3 we see that there's a

untapped work around the clock workforce in the web 3 where there's a lot of like

crypto users but they don't have you know some kind of like fixed income so

we try to bring that model into web 3 and turn it into a I would say like a

tokenized data annotation platform where we reward them for their labeling

tasks using it or stable coin perfect Tony you got anything want to add on

that yeah so the way we work is we build this initial model based on a very early

state of the data and then the data will then later on being peer-graded you can

say so by the human labor force and then when we got the peer grading from

the human labor force the data quality got increased improve so much and then

we use it to retrain the model again and then the whole cycle keep on repeating

until we got the accuracy above the roof like 80 90 percent then that model is

finally ready for productions perfect thanks Tony appreciate some of the

insights here guys so again I encourage people check out a couple of the threads

which I pinned to the top of the space and if you are of course enjoying this

space there's a few hundred of y'all in here but feel free to go ahead and share

it out if you have questions you can comment them as well below into the

spaces chat it's just that purple pill right there you can add that drop a

comment or just you know let us know where you're tuning in from all right

with that being said let's talk a little bit about how this differentiates itself

so you mentioned you know being the first but sometimes you'll be in the

first isn't enough right you have to also create differentiations and build

those moats and competitive differences so let's talk a little bit about it can

you share a little bit on how a IT protocol is differentiating itself from

the other projects both in the web 3 and the AI space because you're kind of

competing in both yeah for computing you know it's not quite because you know

here we tried to be the infrastructure for the development of AI in web 3 we

see a lot of AI projects you know popping up a daddy now some doing very

good some providing very good use case you know like inspect power AI they have

a really big ecosystem and you know that the app also have a lot of users so

we are not in the position to compete with them rather we try to help them to

improve their model from what you know when they need to improve the model they

need like someone to label the data the new data right to improve the model AI

cannot improve by themselves or by using the same data sets that you have

change them with so you need to change them the the machine learning with new

data sets and before doing so you need us people like us to clean the data

short out the data label them before they can use that so while we different

from order is that you know we calculate the infrastructure for that we help them

to improve we don't compete with them we can collaborate with anyone in the

weather is pay that you know who wish to improve the model so that's a new

unique selling point first we don't need to compete with anyone and we being the

first maybe being the only one in this field in the web 3 can like give us

advantages for you know getting more deals more contracts getting a better

revenue share as well and those thing goes back to our community where we can

grow a larger and larger audience and the labor workforce that's it appreciate

that yeah Tony do you have any thoughts to add on there as well as how you're

differentiating yourself from other companies yeah so not just being that

that annotation platform we are actually having our own models as well currently

we are deploying our model on bit tensor which is also another you can

see it as like a marketplace for AI development and I will be talking more

about that later on okay perfect so rolls me into something that you know I

really want to discuss here which is some of the recent milestones and

achievements I think this is gonna kind of give some color to things that you've

been doing right there's a lot of great ideas out there but execution is

ultimately what I focus on so let's talk about it I'll throw that over to AIT

first can you walk us through the recent milestones achievements things that are

actually getting done here with the protocol so yes we've been building for

almost a year now and we release the the first version of the data annotation

platform the better like back in the January no after that you know we kind

of shift our focus on potential because you know we see the opportunity there

there's a like 32 subnets in potential network and it's up now like you having

their own machine learning model so we kind of shift our focus over there try

to build run a subnet ourselves and you know trying to help all the subnets to

improve their model you know using the same same concept of that the

adaptation services and so we recently we registered our subnet you know in

early February and run testnet in February 23rd and the main after three

days after Tony's is the the lid for that projects and you know he's been

like spending countless of hours and the team you know to release the subnet the

man that as soon as as early as we could as we could and right now we have

been you know receiving some network waste from the validators and it's a

really big progress first in the future near future right we releasing the

front end where we can have our labeler to interact with the front end and

improving the accuracy of the model of the tensor network and I think it's a

really a really big breakthrough for us we can reward our labeler with

TAO you know with a stable COI or TAO and help to improve the potential

network overall also you know we're releasing more custom models on the

tensors trying to help other miners in the network to improve the accuracy as

well by that you know I mean like increasing that chance of earning more

TAO from the emission for the community for the community we are making an in-depth

guide you know for non-tech savings user to be able to set up miner on our

subnet as well so that's a few achievement that and our future plans

that's great and sounded like you're speaking to the team that Tony's been

putting a lot of time and Tony do you want to speak to that

yeah actually for this part I think he been nursing all the thing that we

already achieved but let me talk about how our Einstein AIT subnet so our

subnet Einstein AIT we see it as like a marketplace on with them so network

dedicated to the computation of complex mathematics logical reasoning and data

analysis most of the time we see now-of-the-day language model give you a

very generalized answer this is because they only know how to predict like the

next word you are trying to say and they just say things that they already

learned before like there are no not no really real calculation behind it and

they are usable for doing computations and logical and mathematic so our primary

use case for our subnet is to supercharge any existing large language

model by minimizing accuracy and enhancing the overall network

intelligence so we do this by enabling autonomous code writing and execution

we aim to bolster the capacity of the AI models ensuring that they can tackle

complex mathematical task with precision this is not only to improve the quality of

AI generated content but also opens a whole new avenues for research and

problem solving in the field that rely heavily on accurate data analysis and

mathematical computations that's awesome yeah I love how these models are

getting built out I mean maybe you can speak to this these models just take you

know what humans were doing even with computers and seem to really take them

to the next level whether it's with computing power the amount of work that

they can get done in a day what would you say and I think you know I guess

this gets us into a little bit more of it what is the Einstein AIT subnet what's

the actual use cases what are people going to do with this so I can see that

in the future when the whole because on bit tensor this is actually the place

that you can deploy your AI models on and having people to provide help you

provide you so like the minor and the validator they are actually providing

you computational power for you to work your models and this whole loop keep on

this whole loop keep on continuously going to help us improve the models like

in including the human labeler we have the initial model and then people human

validators can interact with the front end in the futures where they are

actually also validators helping us to improve the data quality plus the

computational power for all the miners that helping us on the subnet right now

I can see that our AI model is going to be a breakthrough we are going for the

north star we are going for the quantum computing as well so this model should

be your go-to problem-solving pal and we can know we can see that in the future

people can use AI as a research buddy for their thesis or whatsoever like very

complex mathematical and problem-solving appreciate it Tony yeah these

these models continue to get developed and build out and more and more use

cases kind of come to light so you have problem-solving in some of these other

pieces I guess kind of a fog question on this what are some of the more exciting

in your opinion upcoming subnet developments and partnerships community

can look forward to so like I said we are continuously working on expanding

the capacity capabilities of our subnet five through both technological

advancement and strategic partnerships one of the key development on the

horizon is the integration of more and more GPU minor and validators which will

significantly enhance our computational power and efficiency

additionally we are in the process of finalizing partnerships with leading

academic and research institutions a little bit hint of course is MIT that's

where I'm from to contribute custom data sets for complex mathematical and

logics these collaboration will not only enrich our subnet but also foster

a stronger ecosystem for AI and machine learning innovations perfect thank you

Tony okay another quick thought question for you Tony as well so there's

there's definitely some complexity here right when you talk about a protocol

leveraging AI technology to enhance things like data annotation tools drive

innovation in web 3 right some of these different pieces I'm curious how they

all connect so to kind of put that all together could you explain and feel

free to take some time here how a IT protocol leverages AI technology to

enhance the data annotation tools and to drive that innovation so first of all at

a IT protocol we leverage AI technology to improve the efficiency and accuracy

through data annotation our platform utilized machine learning algorithm to

automate the auto the annotation process first reducing the time and

resources required for data labeling and then this this automation

enables us to handle large data set with positions facilitating the training of

more accurate and sophisticated AI model so after them after the model did

the very initial annotation process we run these through our human labelers as

peer graders to improve the data the data quality to later on bring back to

improve the models and then when the model is improved we continuously

scraping for more and more data wide and far to continue to continue this

loop non-stop until the model is ready for collections and furthermore by and

integrating these tool with the web 3 technologies we ensure the data security

and privacy empowering user to control the data while contributing to the AI

development process our approach is not only streamline data annotation but

also accelerate the pace of the AI innovations beautiful thank you Tony

yeah I mean I think you know we're doing our best to explain to people obviously

there's gonna be some pieces where you just have to kind of dive into it and

in order to understand you kind of need to read through a white paper you need

to look at the pieces AIT you got a comment there yes so if you talk to me

about it you know the change your model and we you know we at the beginning you

know when we think when we charge like mega platform where people can interact

using the web 3 approach you know like connecting wallet scanning NFT for

identification everything works well right because now we don't need to hire

people we don't need to set up like human resources departments spending

money to interview other people just for the labeling job and it turned

out you know the web 3 is really help with this kind of like concept we're

cutting out the time the cost and you know getting easier for you getting a

larger space a larger audience a larger workforce so how do how does it really

look like is that you know we we did a NFT raffle freemen where we user was

coming in we welcome everyone who interested in being one of the like the

early labeling labeler and they can mean the NFT for free and that we serve as a

work pass then they can connect to our front end using that the same NFT and

doing the task there's gonna be like some lead limitation place on the NFT

they can maybe for example they can do like two to three tasks per day or maybe

ten tasks play depend on the demand of supplies and demand on depend on how

many the contract we guess or how many requests we got from other projects or

partners and they can employ you know in doing completing the task but you

know before they can employ there's gonna be some peer-to-peer you know just

double-check their verify their the answer just like when you know doing

your test in school and if they pass the verification they can under point and

these port can be later are converted to stable coin or towel you know in in

this case that's how the motor working just it's a simple kind of like concept

and we don't want to make it complicated you know which ones make it

like really easy for to onboard anyone to interest it also they can trade the

NFT I see that the NFT is being traded at a much higher price and you know

then we first minted and you know for the exact moral like how does it work

like how much reward they can earn per day we will have to wait till we release

our front end and see how many tasks we can put out there and get some numbers

do some calculation and we can do give our community a projection of how much

they can make from being one of our labeler perfect appreciate the

elaboration there yeah feel free to kind of hop in at any point to take time to

elaborate here we certainly have it okay I'll flip another question back to you

AIT and then Tony if you want to jump in as well both can cover this so part of

the model here is train to earn which is actually one that I wasn't super

familiar with I'd obviously heard of learn to earn you know walk to earn all

these train turns a little bit different what's the model look like and how can

users actually earn from it yeah I think my answer just after this question but I

would love to add more for today's so basically like right now we're using a

if a model where we call it a human in the loop where human work together with

the machine learning to label the data basically when the data come in for the

input we have a machine learning to put out what you call a pre-labeled data

sets it's not might not be you know accurate then we move that pre-labeled

data sets to our human department our NFT user where they label the data

again you know giving me you know refi it giving more accuracy and these the

final data sets will be I would say like we revealed by our data scientists

you know we have a team for that and the final result you know will be sent back

to our clan depend on their request and you know like if they request for 70%

accurate or 80% accurate we'll keep doing the loop until we reach the desire

like requirement so the as I mentioned that the change your model is just

something easy just like pledge your own or you know move your own we use your

NFT to complete the task get the point and convert the point to reward like

stable call it or tell for the you know and how much they earn we don't know yet

but we say like this uh this kind of job is really in high demand right now in

the traditional work where they pay label anywhere from I would say 200 to

like 300 per month and this only take about a 30 minutes of work per day so we

try to see there's a growing demand for this kind of concept in W3 because you

know there's a lot of AI projects out there and they all need that the

annotation services and we are one of the you know we might be the only one

right now in the market the offering this this kind of service so once we

release the front end we'll see like how much we can make from that and and how

much the user can earn from it as well that is definitely interesting pretty

unique have you seen any other companies that are doing that and if so how are

you doing it differently than they are maybe just like for now I haven't seen

anyone yet but we more calculate focus on the W3 data compared to our you know

one one thing for sure that we don't want to compete with traditional data

annotations company because those they doing the contract the huge contract for

for big AI guys you know like open AI or Gemini from Google here we want to

focus on W3 the W3 data is different from the traditional data to label the

W3 data you need crypto user you need some knowledge about the W3 some

knowledge about blockchain that's something that traditional company cannot

do all don't want to do because you know they have you like spend money to

educate their labeler a company usually have about the size of the labeler

approximately about a 1000 up to like 200,000 so you know educate everyone

would cost them a lot of money for us you know we have the W3 community where

they are equipped with good knowledge about W3 in crypto so we can pick them

you know traditional company on on that and focus on the W3 space beautiful

Tony you got thoughts on train to earn that you want to share yes so beside be

beside focusing on web 3 the whole missions of subnet 5 that we having on

Bitdancer is also to contribute to AI innovation across the world so by being

validators so the way validator works is it mimics how human would use AI to ask

some complex math or logical questions and then miners are actually more of

that what we were talking about the initial models right so we have GPUs

models and then the miners can provide the computational resource being a minor

in our subnet and help us pre-label those tasks that created by the

validators scripts and this is actually where all the tasks come from and where

all the data that need labeling come from and so yeah perfect thanks Tony

okay good stuff here hopefully this has been pretty helpful we got a few more

items we're going to touch on and then Brian I might go over to you see if you

have any questions afterwards but let's keep it rolling here because I'm

definitely you know learning a lot here for those that have come in late chatting

a little bit with AIT protocol here really about the future of web 3 at AI

integration that's exactly what they're leading into I think that that is the

right topic probably to be doing right now it's a very hot topic they're

building out the Einstein AIT subnet we talked about some of the use cases one

of the big pieces here is the train-to-earn model which we just walked

through now we're gonna roll a little bit into the tokenomics so I'll throw

this over to AIT can you provide insights into the tokenomics and how the

token is utilized within your ecosystem so yeah so so initially when we make the

the tokenomic we had about a 30% of the token supplies you know dedicated to

change on model where we were planning to reward our user with a token but you

know as we go as you know after a launch as well we launch and you know the

beta we kind of like figure realize that you know rewarding users with the token

will not help the projects in the long term because you know now you have a

constant selling and with the slippage as well like you know having our

plan to buy the stable coin and then you know buying the tokens will be too much

for them and that's not something we can you know improve so we kind of like

take out the token use that kind of utility out of the tokenomic and we burn

the 30% completely so how what kind of utility the token has right now we

figure like for the long-term growth of the projects and more stable growth to

our token holder it would be best to make AET a profit sharing where they

can stack a token to receive more AI rewards through the high APR also they

can earn a share of the tile emissions from our subnet also they can the another

one they can earn the ETH from the tax revenue where we just announced a

revenue share of 350k worth of ETH and I think that will help the projects more

we try to stay simple we don't want to add like crazy utility to the token or

you know give it any chance of you know making a high inflation tokenomic model

so our our goal right now is to reduce the supply as much as we can we have 15%

more on top of the 30% that has been burned the 15% is programmed to be burned

as soon as it's unlocked and it's gonna go on for the next 15 months starting

from today actually so we will burn like 2.5 million AET every

week for the for the next 15 months so the profit sharing there's no end day

for the profit sharing as long as you know anyone stuck in that pool they get

the rewards and you know we share the revenue back every Monday the amount

depend on the the tax that we got and that's you know my vision for now for

the future of the AET token. Tony you got thoughts you want to share on the token?

yeah so what I'm trying to work on right now is the token incentive for both

miner and validator within the Einstein AET Summit as well. Miner are reward I am

figuring out like how we can collaborate through smart contract miners are

rewarded in AET token not just mine trying to mine town from the

subnet so you can also earn AET token for contributing your computational

resource and for AI test day process similarly validator earn AET token for

maintain the network integrity and overseeing the quality of the outputs so

this incentivation ensure that the network remains secure efficient and

continuously evolving but this is kind of a teaser I am still working on it

right now and I hope this is this is something that we can achieve soon

perfect got it appreciate that okay we might circle back a little bit on the

token but I want to just get through a couple other items here you know there's

people commenting below where can I buy in where can I get involved so I think

let's talk about it right because there's multiple ways people can get

involved I think the big way here is going to be you know developers right

individuals and developers can get involved with the protocol contributed

to growth and development Tony how should people go about doing that right

so for individuals and developers our subnet report is actually open source

code base so we welcome any contribution on the GitHub repository and we also

encourage developer to build on top of our platform utilizing our API fine

tuning models and yeah just get up there and pick up and poke some holes

feel free but for those who looking to get involved in all the capacity like we

offer opportunity for data annotation as well obviously and model training and

participation in our community forums and discussion you can anyone can also

become a miners we appreciate every computational resource that is all it

really meant for building AI and yeah we believe in the power of the community

collaboration and we are always looking for passionate individuals who help drive

our mission forward very nice very nice okay anything you want to add on that

ways that people can get involved here yeah I mean I think Tony has covered

most of it but you know we welcome any death who can you know improve you know

have has more knowledge more experience working in the AI section and improve our

maybe visit model maybe our model after you know after they try with the

format that we're gonna release soon and for the user we try to onboard the user

as well that's you know let me say every tensor is a rough diamond right it has

so much potential so much profit to make on and we don't want just you know

onboard just tech savvy guy you know we want to onboard you know any any common

users and the script or use out there and by doing so you know Tony is making

an in-depth guy so anyone you know can join us can set up the miners thought

and you know try to make the profit for himself this is something we've been

you know having in our minds at beginning and and yes we welcome anyone

who want to walk with us to run the miner can help us all help us to improve

our subnet overall there you go a couple ways people can get involved all

right when will a it get listed on cx yeah it's funny how we've been we've been

asked this question a lot like recently so sex you know change your life change

is kind of like the next our next phrase we launch it on the next only we

make a name for ourselves we got a really good performance good charge and

I think in order to expand our audience you know to expand our exposure

centralized chance is needed in fact we have been reached out by a couple

centralized chance good tier one tier two and I can say the name but you know

we've been working with them behind the scenes maybe for a deal so as soon as we

got some kind of deal or confirmation we will update a community but I just want

to let everyone know that is our next focus you know the subnet and the

exchange getting bigger a larger audience increasing the token power you

know increasing our exposure to the more community more markets segment and that's

yeah our our roadmap for now awesome yeah that's great to hear so let's

think just a little bit more to this roadmap we spoke about kind of the

upcoming subnet developments and partnerships what else is on the road

map what else are you working on actively right now yeah we you know like

we have a large pack you know as a something like something to reward a

community the community right you know besides stacking I forgot to mention

that you know if you stack out IAT you are eligible for the idea as well we

have been working with we've received a lot of requests from good projects you

know to launch the idea on our platform initially we launched our token on now

on our platform for you know power sticker it was successful and it was you

know it feel within like 20 seconds so people you know think crazy about it and

try to like do the last part with that as well and there's a really good

projects that coming soon in in March this month and it's been confirmed you

know for the deal but I will review will review more detail information about a

project in the upcoming days but it's a really good projects the team has been

have like extraordinary experience in web 3 and I can't wait to review more

information about it to our audience and I just want to remind that it's only

likable to our IAT stackers perfect I have a couple more questions but Brian

I haven't heard from you yet do you want to jump in here any thoughts questions

I mean I think to rewind a little bit you know for anyone that's not overly

familiar with AI and how it all works I know this can get quite complicated but

you know at the very core of this I was advising and working with AI companies

going back to 2017 and even earlier and data is always the problem it's like how

do you get all this data and just storing it and the infrastructure around

it is extremely difficult and expensive and then not to mention training the

models themselves so what they're doing here on a very fundamental level is

solving some of those problems that allow you to sort of you know get build

what you actually want to build rather than get stuck in the data phase so to

speak so you know if you were if you think of chat GDP for instance right

there was there were people that had very similar ideas to chat GDP you know

going back many many years I remember a company very specifically that was doing

very similar but the problem was data how do you get all this data well back

then either had to basically harvest it somehow illegally or buy it you know

there's maybe a few other ways but it was very very very expensive and server

heavy you know and can time-consuming so now with with and I'm an investor in

IT because I see that this is needed you know platform in the AI space with

especially data and I was always looking forward to someone kind of

coming in and finding a way for blockchain and AI to work together in a

way that makes sense so I think this is the first version of that making sense

of AI and blockchain that I've seen and again I needed service for the data side

of AI I just can't you know it's it's you can't overemphasize how important

data is to AI you basically don't have AI without data and without you know

storing it and building the infrastructure around it to train the

models so it's it's the most important part it's it's water to humans it's that

important for AI so you can't overestimate you can't overemphasize how

important this is for the infrastructure side yeah yeah yeah you can

have AI without you know data and you cannot have the data without human you

know the AI would not be able to we probably talking about using AI to label

data or you know short out data by themselves but it's still impossible at

the current you know maybe where the quantum computer comes play but not no

AI right now can handle that's that work so we need human you know a lot of

humans to you know to clean out the data to short out data to label them to

prepare them for any kind of AI development and that's where most of the

money come from right that's where most of the budget spend on so I think we are

in the right direction and in the right topic in the right category that will be

relevant for the four years to come and there'll be more opportunity once you

know from my opinion like AI the AI chain just begin along with the the

poll market and we still very early right now even though we have a lot of

projects but the potential has not been maximized yet and we'll see more and

more useful use case no developing every day and I think they will will be one of

the very projects of that you know will be beneficial from all of this awesome

Brian you got any other thoughts there perhaps on a subnet

I know I think they kind of covered it all I mean it's just it's good to see

this happening yeah on the annotation side and tying that in with incentives

on the blockchain everything is just it makes a lot of sense it really does

streamline the process so I think it's a clever way to you know train models and

allow people to kind of earn and then sort of parallel to that so it makes

sense to tokenize this you know you see some AI plays that kind of don't

really make a lot of sense but this this is something that fundamentally makes a

lot of sense to have it tokenized with some of the core to sort of the learn

and earn kind of style so yeah very very nice to see that yeah I did have

another question I know what you talked about about your background and Tony's

what's the rest of the team look like so right now we have data scientist from

Ubisoft and then I myself from MIT Google Cloud Platform and others from I

think I forgot the name of that but I'm so sorry but most everyone they very

experienced in web 3 actually I am I feel like I'm the only one here with the

least experience in in web 3 but I'm your AI guy I am the one who oversee all

the AI stuff here very nice very nice okay that makes sense and then what's

the funding for this all look like kind of who's funding it have you raised

capital are you gonna raise more stuff like that so we we did our initial raise

the fundraising last year back in November 2023 we didn't like a little

over slightly over above 1.2 million and you know it will give us a runway for

the next three years with all the development and I think we will start

making profits from the subnets will be early and from there you know we don't

have to worry about our financial situation and so we don't have any more

plans for another race and with the fundraising we have been able to onboard

a lot of like people that you know support us you know to bring the

project this far and we I'm happy with that

sounds good sounds good and then what would you say are the biggest obstacles

or the toughest things that you're running into while building this and how

are you solving them I think the biggest obstacle would be the thorium gas fee

right now where we try to implement a couple incentive program for our

sticker our token holder but the gas fee is too high and that's kind of like

push back our plate a little bit a lot of plans but we're trying to find a

solution for that also the stack or on the AI side obviously the new

technologies where three is always changing as you know the new thing

always come out every day and with AI is even more we have to catch up with the

new technology from potential Tony actually has to learn their structure

their mechanism for two or three months before he can you know start building

building our current subnet it's not easy but we have a team of 30 techs you

know to working endless hours you know to improve our platform so I'm

confident that you know we will be one of the top guys that you know being able

to keep the late delivering and continue to develop the best products out there

perfect Brian if anything else you think we should go over well actually I'll ask

the IT is there anything else that we haven't touched on that you think would

be helpful for the audience to know I think like for if then you want to if

the audience want to learn about our projects I think the best way would be

you know checking out our white paper it's on the docs.iat.tech and learn

about what we do also we welcome anyone who wants to set up minus spot on our

subnet we can help them we have a team on our potential discourse the channel

subnet 5 and join in and ask your questions and we will be able to help you

to set up your own minor also like for our token holders right because the

apply is a lot of just in us you know being with us in the beginning some even

stack for almost like ever since we launched they haven't let me unstuck the

token so we appreciate that and appreciate that you know patient we

commit we are the our team is committed you know to keep delivering and bring

out the the best benefits to our token holder and we can with the net with the

shift of network waste from the potential to our subnet we are receiving

more and more some money with waste meaning we are starting to see the

profit from the potential subnet and we will be able to release the part of the

program where we do the kickback those incentive to our 80 holder very soon and

more information we will review later on sounds good okay so then Brian Brian

I was gonna come over to you anything else you think we should touch on should

we move into some wrap-up here I think I think wrap-up yeah I think this has

been very interesting to hear some of their depth of what's happening at IT

perfect yeah I think we drilled into this for sure if you were not familiar

with it beforehand and you're the audience you definitely have a much

better idea now let's roll through some final comments Tony anything else you'd

like to share with the audience any call to actions you'd like to put out

there yes like I said guys we really really need more miners on our subnet

we building AI is a super computed computational resort burning tasks you

know like we need we appreciate all the help we can get let's make a different

in the LMM space like I said we are the supercharger we will make will ensure

that the AI models it will be able to do much much more than we already see

right now perfect thing so I guess one thing yeah one thing I will say is you

know I'm a large investor angel investor in the space and have an investment

company as well and even though I'm really bullish on the AI narrative

AIT is only one of a few of in the AI sector that invested into I think there's

a lot of sort of garbage out there too that it's just trying to capitalize on

the AI narrative and this is definitely not the case with AIT I think this is a

really fundamentally strong especially with the BitTensor partnership there it

just kind of really validates what AIT is doing so I mean this is coming from

somebody that invests into a lot of different things saying this is

something that I'm really really watching closely so yeah awesome and then

AIT I'll turn over to you any final comments you'd like to share here before

we close out yeah thank you know I'm happy that you know got invited to the

space today happy to talk to Brian and you Wolf it's been my pleasure also you

know we want to have we will have more opportunity to speak to our directly to

our community later on where we construct AMA within a community call

within the out-discord channel or telegram until then you know we will

have a lot of things coming up that where we will be releasing one by one

and we can discuss all that you know within the community call I'm trying to

like learn more about the committee was there you know what kind of feedback do

they have for us and how do they want us to construct the you know the stacking

program you know the tile emission the tile incentive or the you know the

revenue share yeah there's a lot of things being uncertain right now but we

will try to clear out things one by one for this month and thank you for

attending this call as well perfect well thanks so much for coming on again I

encourage everybody to follow the speakers on stage take a moment to do

that make sure that you're going through to the pinned tweets you can

also just go through directly to their profiles and see the tweets take a look

at the Einstein AIT LLM take a look at the revenue share take a look at the

train to earn there's a lot of cool pieces going on here I see a lot of you

already following that account anyway so I think that you're in good hands so big

thank you to the 700 that came through today great to have you on I'll be on

more spaces later today around 2 p.m. Eastern so I look forward to seeing you

all then for now thank you to the IT team thank you Brian great having

everybody as always we look forward to the next one

thanks everyone thanks for having us

### 主持人

- [WOLF](https://alphagrowth.io/zh-tw/spaces/participant/wolf_financial.md)
  [X 個人檔案](https://twitter.com/WOLF_Financial)

### 講者

- [Brian D. Evans](https://alphagrowth.io/zh-tw/spaces/participant/brian-d-evans.md)
  [X 個人檔案](https://twitter.com/BrianDEvans)
- [AIT Protocol](https://alphagrowth.io/zh-tw/spaces/participant/aitprotocol.md) 來自 [AIT Protocol](https://alphagrowth.io/zh-tw/ait-protocol.md)
  [X 個人檔案](https://twitter.com/AITProtocol)
- [AIT-t0nY](https://alphagrowth.io/zh-tw/spaces/participant/t0ny_ait.md)
  [X 個人檔案](https://twitter.com/t0ny_ait)

## 想為你的協議或公鏈提升 TVL？

AlphaGrowth 透過資產引進與整合、資助計畫、激勵和 LP 合作來提升協議和公鏈的 TVL。這些資料由我們建立並免費開放。

[我們如何提升 TVL](https://alphagrowth.io/zh-tw/defi/index.md) [與 AlphaGrowth 聊聊](https://alphagrowth.io/zh-tw/contact-us/index.md)
