0G Labs is growing, now ∅ A community account in the 0G Labs (Zero Gravity) ecosystem

🐼Developer Office Hours #1: Unveiling the 0G Architecture & SDK Demo | Recap
On 26 July 2024, 0G Labs held its inaugural developer office hours, an event dedicated to introducing and exploring the intricacies of their revolutionary "0G" architecture. Hosted by Gathin, a developer role engineer, and Peter, a developer at 0G Labs, along with contributions from Yon, the session provided a comprehensive overview of the 0G architecture and its implications for decentralized data availability and storage. This recap delves into the key points discussed during the session, h...

🐼 Summary of: Testnet V2 AMA
AbstractThe AMA session primarily focused on updates related to the 0G Testnet V2, including the introduction of new features, the validator selection process, and future plans for the 0G network. Key discussions included the improvements made from Testnet V1 to V2, the stability and performance enhancements, and changes to the economic model. Additionally, future goals for the 0G chain, collaborations, and support for developers were highlighted.Highlights:Introduction of new economic models...

🐼 | Research: Theoriq
Review of TheoriqAI and its Partnership with 0GIntroduction to TheoriqAITheoriqAI represents a paradigm shift in the AI landscape, aiming to decentralize and democratize AI technology. Founded by Ron Bodkin, Arnaud Flament, David Mueller, and Ethan Jackson under ChainML Labs in 2022, TheoriqAI focuses on creating a community-governed AI ecosystem. This initiative addresses the critical issue of AI centralization, where a few entities wield significant control over AI advancements. TheoriqAI&a...

🐼Developer Office Hours #1: Unveiling the 0G Architecture & SDK Demo | Recap
On 26 July 2024, 0G Labs held its inaugural developer office hours, an event dedicated to introducing and exploring the intricacies of their revolutionary "0G" architecture. Hosted by Gathin, a developer role engineer, and Peter, a developer at 0G Labs, along with contributions from Yon, the session provided a comprehensive overview of the 0G architecture and its implications for decentralized data availability and storage. This recap delves into the key points discussed during the session, h...

🐼 Summary of: Testnet V2 AMA
AbstractThe AMA session primarily focused on updates related to the 0G Testnet V2, including the introduction of new features, the validator selection process, and future plans for the 0G network. Key discussions included the improvements made from Testnet V1 to V2, the stability and performance enhancements, and changes to the economic model. Additionally, future goals for the 0G chain, collaborations, and support for developers were highlighted.Highlights:Introduction of new economic models...

🐼 | Research: Theoriq
Review of TheoriqAI and its Partnership with 0GIntroduction to TheoriqAITheoriqAI represents a paradigm shift in the AI landscape, aiming to decentralize and democratize AI technology. Founded by Ron Bodkin, Arnaud Flament, David Mueller, and Ethan Jackson under ChainML Labs in 2022, TheoriqAI focuses on creating a community-governed AI ecosystem. This initiative addresses the critical issue of AI centralization, where a few entities wield significant control over AI advancements. TheoriqAI&a...
0G Labs is growing, now ∅ A community account in the 0G Labs (Zero Gravity) ecosystem

Subscribe to 0G Panda

Subscribe to 0G Panda
Share Dialog
Share Dialog


<100 subscribers
<100 subscribers
X Space:
https://x.com/i/spaces/1BdxYrQerWyKX
In this Twitter Spaces discussion, speakers from 0G and Ora Protocol shared insights on decentralized AI and its applications in blockchain and crypto spaces. Key points included the technical aspects of decentralized AI storage and data availability, as well as the future roadmap for both companies. The conversation also highlighted the potential benefits and risks of AI in the industry. Introduction to 0G’s ultrafast decentralized storage solution for AI. * Ora Protocol’s decentralized AI Oracle protocol for smart contracts. * Discussion on the Initial Model Offering (IMO) for tokenizing AI models. * Technical challenges and solutions for on-chain AI model storage and verification. * Future roadmap and expansion plans for both 0G and Ora Protocol.
Halo halo thank you. Sama shopping here on the we have a xer froud. We are waiting for om. Michael also joined. Let me accept Michael. We also have a cotton join us right now. So we have all our speaker being here. Thank you so much for being here.
Good morning. Good morning.
Yeah, hi.
Good morning.
Good morning.
Yep, this ask me anything is a partnership between 0G and Ora. And we have prepared some questions for the founders to discuss. And Alice will help me to be the co host. Why don't we kick it off by sharing some background about 0G and Oracle to begin with. Well, I'm also going to make sure we have more question coming up from disco and other platform. We were going to leave about 10 minutes towards the end for you guys requires Mike and asking open questions. With that, I'm going to give the mic to Michael. Introduce Sergey.
GM everyone. I'm Michael, co, founder and CEO of 0G. We're building the first module AI chain. So we said at the middle of the AI infrastructure stack, we're basically building our ultrafast decentralized storage and data availability solution. Most companies in the space are providing anywhere from 1 to 10 megabytes per second and throughput, but that's still very limited for fully on chain AI use cases for things like fully on chain model training, we need tens of gigabytes per second and throughput. And so that's what we enable through our architecture and super excited to partner with carton and Nora as well over 2 cotton.
Cool. Good morning, everyone. I'm carton, the co founder of Ora Protocol. We're building a decentralized AI Oracle protocol that enable AI influence for smart contracts. Very excited about the partnership.
Fantastic. Thank you so much for the quick background introductions. Alex, would you mind to taking on the first two questions?
Sure. So the first question for you guys is first for carton. So you mentioned that aura is Ethereum stressless AI and it's bringing AI and complex compute on chain. But could you explain why is this actually needed and how this works?
Yeah, yeah, yeah, really good question. So I think like, first of all, like AI is like changing every single industry in our society right now. And I think crypto industry will be one of the first like that gonna be changed by AI permanently. And so like it's really important to have AI that empowers smart contract to do something that we used, we cannot do, right? But we know that smart contract pretty much can't run the AI inference inside a smart contract with when you, especially on Eve. So that's why you need Oracle to do that. And in all in aura, we invent the OPML to run the inference decentralized fee and submit the result into smart contract that enables more contract to perform AI. And so which means like now we are able to use AI in a smart contract. So we can do way more involve innovation. So imagine use like a new AI model to be a market maker to increase the capital efficiency in the AMA. And also use like m model to do a landing and stuff to increase the capital efficiency on the landing protocol. So I think most like in general, like AI will like significantly improve the utility ability of the crypto industry.
I see. Okay. So basically we are using this off chain computation and then bringing this AI verifiably to the smart contracts, which opens up a lot of new use cases.Okay, so and then the other question for you, carton, is about the other bought a product from aura initial model offering. So can you maybe elaborate more on that? What is it and how it benefits to the ecosystem?
Yeah, so like initial model offering is like for, it's like tokenized AI model. So the reason why I start this thing is like, because I think like open source AI need a new way to monetize their model, right? Cuz like pretty much if you spend most of million dollar to develop a AI model today, the only way you can monetize that is like through API subscription that have people to pay for yours. Oh, close those AI model. And that's why like, I think if we can have a new way, a new paradigm, a new flywheel for open source AI model to monetize their model, that would be like really helpful for the industry, both AI and crypto. And so like in aura, right, do you have like IMO that is like specifically use crypto to empower open source AI? And also we have like a AI Oracle to use AI to empower crypto. And the IMO part is more like tokenize the AI model and then let the AI model become ownable commissionlessly. And also that can help the open source AI creator to monetize their model and create a positive flywheel.
I see. So like both IMO and Antenna Oracle are kind of working together to make this happen.
Right? It's like a close loop system that like, so yeah, when I wanna use like crypto to enable AI to do to grow the ecosystem for open source AIS E community, then we do the IMO. And then once the model on board on crypto with IMO, then we convert that into a on chain AI, Oracle provide services to smart contract. And then once like people use the AI Oracle in their smart contract and create positive ban, positive revenue. And the revenue will share to the model token holder that people bought in IMO. So it's like the closed loop.
I see. Thank you very much. And with that being said, I'm gonna let my co host ask next to questions.
Yes, since we talked about capability and phenomena of aura, I do want to ask Michael, how does this fit in cog data storage and data availability solution? Michael?
Yeah, this is where we're super excited. So if you look at the AI stack, right, you've got the data generation, things like synthetic data or gathering from different sources. You know, this is very broad. And then you've got kind of a whole data pipeline management, like cleaning and various other labeling tasks. And then you've got kind of us in the middle that's providing to storage and ways of engaging with the models through data availability, and then you've got the kind of execution layers and then the agent layers. And so we're at the data infra layer and then aura is essentially working on the execution layer, providing AI solutions through smart contracts, kind of this AI Oracle concept. And so it's a great match. In order to run opml, you first need to store the actual model on chain. So we provide that. And then you have to serve decentralized inference traces through the da stack. And so the larger the da pipeline, the more you can serve through there. And so that's kind of the basics. And we can go, of course, much deeper into each one of those with Carden together.
We would like to have you and Gordon have a dialogue about the things you just mentioned.
Yeah, carton, do you wanna mention, go a level deeper and mention some of the things you mentioned to me as well that make a lot of sense for utilizing our infrastructure for.
Yeah, yeah, yeah, of course. So yeah, we know that like when we have the model on chain AI model on blockchain, the AI model normally is really big and the size of the model is really big. So you gotta have like a new kind of da system that can like publish giant size of the model. Let's say it's not just like few hundred megabyte. We're talking about like fuel hundred gigabyte. And then so that's why like with this partnership, I hope that like zero gravity can be like a really solid infrastructure. Sure. For all the AI crypto project in the future that can have them to storage and publish the AI model in the zero gravity da.
And yeah, just to add to that, we're trying to get to a cost point that's cheaper than a centralized solution. So to the 3X cheaper than that's just putting it into an AWS S3 bucket with the added benefit of having multiple replicas, erasure coding so that there's a security built in and ways of essentially doing disaster recovery because you have multiple copies. So there's a lot of benefits of actually decentralizing the data itself, you know, beyond, of course, knowing where the ownership is coming from and so on. But yeah, super excited about all of those pieces too.
Since we have aura and OG 0 G and many other AI players arriving. And what do you guys see? Like both of you, Michael and cotton, what, how do you see on chain AI shipping out over coming months and years?
Okay, okay, cool. Well, we're still super early in web 3 meets AI. Even in centralized AI solutions, we're still trying to figure out how to get to AGI and maybe at some point the ASI, which is like artificial superintelligence. And we're still somewhat far from that at the moment. So then, you know, why do you need the centralized AI? We can't even do fully on chain model training today with these large Irlen models. Not well, the data bandwidth issues being solved, but then there's other technical issues like how do you do asynchronous and batch training, for example, in a centralized context, you basically just have a, you know, hundreds and thousands of GPUs strung together in the data center that then can run a, you know, model run. And so we don't have that yet. So there's a lot that needs to be figured out. There's one way where essentially you can say like, well, I'm just wanna bring ChatGPT so you can call it through smart contract.Sure. You know, that's one way of saying like, hey, we're bringing AI to blockchains, but that's not really on Chennai. When I when I think about on Chennai, I wanna see us create actual models fully on chain that starts with the data generation to the data cleaning and labeling to, you know, the storage to than actually the execution pieces. So we know exactly what went into these models, who the owner of this particular piece of data is, so that it can benefit, like pardon mentioned about the initial modeling auto offering, and so that we can truly create an open AI, fully on chain AI. So I see that's gonna happen in the next five years. And I'm super excited about the future because for societal use cases, like when you have agents or big models running things like production systems and transportation systems and administrative systems, I would want to know exactly what went into the models for version of the model's running, what the decisions are being made by the model so that we as society know that we're fully aligned or that the models themselves are fully aligned with our interests. And so I'm excited about building that future of abundance. Carton, over to you.
Yeah, cool, cool, cool. So yeah, of course, like it's important to have this like a super AGI there on blockchain. Well, and like all the infrastructure is ready for it. And I think like, so the main thing I really care about right now is like how AI gonna reshape the crypto economy, right? Like, cuz right now, like kudower is not that big, but still quite big. And so like how AI gonna restrict the whole industry? I think it's like gonna be important. And I notice one thing is like that we have so many new asset getting issued every day on crypto, but most of them do not have enough liquidity or exchange or volume. And I think if we have more AI like that can do trading or do landing or something, and then we can like utilize all this like a new asset issue on blockchain. And then so that means like AI can empower the capital efficiency with the blockchain and then like made the society better or like make people have a better portfolio. And so another thing, big thing is like when we head into AGI, we wanna make sure the AGI is not like out of control, right? We wanna like out of control means like either the AGI become like anti human or the AGI become like own by only few people in the world, like probably only own by Microsoft. I guess we don't want that happen. So we want you wanna make sure that like AGI is like own by the people, at least majority of the people. So like ownable, like AGI is like actually what we really want, like the, the a, the API that own by the public is like what we really want. So like, so we need to invent a new mechanism that can have like use crypto to either vote for the influence or like vote for the token ownership or something like it's not clear at the moment yet. But I think we will find out how we gonna have like a ownable like AGI that own by public.
Are you talking about, go ahead, Michael.
No, I was just saying that's cotton. Exactly. Talked about our mission statement, which is to make AI public good. So super aligned. This is also why I'm so excited about the partnership to our kind of value and vision alignment.
We talked about like potential risk and also potential promise of entry AI and down to the day, like how the builders are can utilize and build the entry AI today. I think that would be a really interesting question to discover, like how can the builders starting using and interacting with the launching AI right now, like this month or next couple months? Do you have some thought, Michael and cotton?
So I think like with builder, the most important thing is like we explore the new idea. So like in Nora, we normally like host a hyper scholar program that like spend time and resources on like helping developer to develop new idea. Cuz I think like when people use AR Oracle to do something on blockchain, that you usage must be super new, super innovative. That is like disruptive innovation that kind of you we place the old player. For example, if like somebody that like can invent a AI landing protocol, then like all the existing landing protocol might not be used by people anymore because like the AI landing protocol just can support way more asset than traditional landing protocol on blockchain or we have like some on chain AI gaming or like the AIGC and FT. Like once people see the AIGC FT, the inference asset, they might not no longer to go to like speculate the NFT asset. I mean, like most people don't do not speculate entity asset anymore in their market anyway. And so like how we do the developer program is that like we, I spend time and resources with like certain hacker that is like really innovative, really good at execution, then invent something new and like that's like disruptive innovation. You.
I totally agree that we need to think about ways of being very disruptive with this technology because it's completely new rules for from an AI step standpoint, like why do you need to interact with a bridge or a wallet? Then AI agent do something like that for you. So the ways of interacting and the type of application surfaces that builders have now is immense. And it can be across the entire stack.You can build a new labeling company if you have some interesting insight, you can build a new data generation company. Like for example, there's some pieces of data that are really hard to reach from a centralized solution. Why not decentralize paid tokens for it, create a new company around it? Or if you want to do something more at the agent side or even at the app side and utilize aura to do a bunch of the inference and kind of calls against some of these models in a verifiable, decentralized way.There's so many different application spaces that you can use it in. You can look at Fifi use cases, social file use cases, gamify use cases. For example, why not build non playable characters that have their own unique storyline for different types of gamers based on their behavior that's fully on chain, utilizing both the Zero G and the Ora side.I mean, there's so many really cool for applications that can be built utilizing this. I'm super excited about the future of it. And so we are also supporting new builders in this space through our collaboration with Monkey Slabs around our incubator. So if anybody's interested in building in the space, please apply. We're still accepting applications from a June and will be hosting multiple accelerators throughout the year. So excited to support the ecosystem.
Michael, you kind of foreseeing what is the next question I'm going to asking? Basically, we want to get a pic for the roadmap and what ahead of us from both 0G and Oracle Standing Point. I think you already mentioned we have incubator and different program. We're curious what Oracle have in mind in terms of roadmap and what happened next.
For Ora. Like the next thing is that like we gonna have more model to be supported in the on chain AI Oracle system and then like have more chain that onboard and more chain provide more services to people and then provide more customization for the AI Oracle to people. So people can like use it to invent a new kind of protocol. That's number one. And second is that like we're getting ready, like we're preparing for like the second IMO with the OLM token. So that means like people with like Ora's point will be able to have like the right to participate the second IMO with OLM token. And that's like the two major thing like I'm working on and the road map for.
Michael from 0G signing point. Do you want to add anything like in terms of technical roadmap and force in the future.
Yes, so we're moving to Tesla V2. We've done a, in a couple weeks, we've done a refactor of our storage nodes codes. We've also moved away from data availability sampling to a query based da solution to increase performance and actual security as well. So we're super excited about fully releasing that. Then after that, try moving towards may not phase 1, which show keep all of that kind of core functionality. And we're hoping for that sometime late July, maybe early August, depending on what kind of feedback we get during our test and YouTube after that may not.Phase 2, we're also targeting for hopefully end of the year where we'll introduce the infinitely scalable Data Availability functionality, which essentially allows us to horizontally scale consensus layers, not just the nodes on the network. And so we can use an arbitrary number of consensus layers with the same validators so that all of the security properties are maintained. And then we'll also be using restaking from Eigenand and Babylon to bring economic security to each one of those consensus layers as well. So super excited about that, you'd be the first to do that in the space.We're also gonna work on kind of decentralized serving frameworks of somebody as very quick data needs and they needs data or even some type of compute through our partners for specific requests, we can through that serving framework and serve those types of needs. So hopefully we can introduce that as part of Phase 2 of the main map as well.
That's a lot to looking forward. I've been asking question nonstop, so I'm going to give my mic back to Alex to asking couple more questions.
Thanks, Ada. So question for carton. So what is the process for bringing an AI model to aura? And actually for Michael as well, about the zero gravity, like what the process of bringing new AI model.
To bring new Mo AI model on chain, normally like the first step is that you need to convert it, make it become verifiable, right? Cuz like the, the, the, like the whole point for doing that is like when somebody can submit an inference chase to the blockchain smart contract, then you can verify it's true or not.So like making the AI model become verifiable is the NUM, the first step. And then the second step is like delegate the AI model to all the. No, that will do the inference.And the first step is like verify the reserve on blockchain and do the proven fraud proof challenge. And so I think like the first step is normally like Danny Aura that we will like publish an open source, the universal compiler that can convert pretty much OAI model become verifiable with OPML later this year. And also like people can like use OG to publish the AI model to all the to do to all the influence know that will take the job actually and then return the result to the smart contract. And recently I notice a really important usage for OG actually. So like, so for OPML, one of the biggest challenge was the deterministic problem in the AI model. Cuz like once the, because like 0 opmi is based on a fraud pool challenge kind of mechanism.So, but you know, traditionally, like, yeah, if you don't do any modification, it is the AI model inference is pretty much undeterministic. So mean, it means like the proposal and challenger will have different outcome, even they both are honest. So that will be have a problem. That means that you have some problem for the challenge. And, but fortunately, in the newest OPML paper, like we're working on to adjust this problem with a round up layer so that like we you run up the neural network when you done with the computation every layer then but the price to do that is that you will have a kind of long run up chase and then you gotta update the round of chase to the some da layer to make sure another proposal and challenger will do the same round of function that you did. And so with this, it means like actually like when you wanna use OPML on a execution environment, then you actually need like not just the influence trade being published by the da, also the round up trace published by da.
Alright, a pretty technical answer, but thank you very much for that. And Michael, do you maybe have anything to add here.
Adding from kind of a technical perspective or just anything else? I mean, the round of trace, super cool. Definitely need a high performance da for that. So excited about supporting it. Where you looking for anything else specifically or?
Yeah, I think that's fine. Yeah, so the next question for you guys is what kind of AI projects are you using your solution in the industry right now? Like, are you especially excited about some projects?
Yeah, so I think like the actually the most exciting innovation that use AI Oracle right now, actually it's like a project call 7,0,0,7 AI. It's like a protocol for inference asset. So once we make the AI inference become verifiable, and then essentially what we're doing is like turning the inference into a asset class, a new enchant asset class. So like when, so now people can mean a AIGC FT without even know how to draw a art. So they can just like put down prom and enjoy really good art and then sell it on blockchain immediately with like probably like 10 to 5,15 dollar to just issue a new AI and gcnft. So I think that will have like a pretty much like a massive adoption. Cuz I think about what like the most popular the app at the moment in the crypto space is actually pumped up fun, right? So like it's a provide a really convenient way for people to like launch a mean kind by just 2 dollar. And but like we know that like mincon have like a rocking problem, like the price go out really quick and go down really quick. But like if you use like FT to reshape this, like launching new asset class behavior, then it will actually made liquidity work better because like normally like FT is a non fungible. It means like different FT will have like, will lock up the liquidity that like going to that NFT. So it will like provide a better way to for people to issue a AI generated content on blockchain. So I think like influence asset will be like one of the most excited use case for or for launch AI Oracle at the moment.
That sounds super fun on our end, it's very infrastructure heavy. So we, on the AI side, I'm particularly excited about some of the, for example, like model training aspects like pond or and chats with flock. And we also announce some fine tuning use cases like Beagle, for example, building on top of infrastructure so that even full models are utilizing our storage in the future to be a problem piece as well. And some of the other use cases outside of the AI world would be, for example, on the L2 side. So you just announce the partnership with 10 Network where they're bringing privacy to serial twos, obviously also integrating with various other leading L twos.Another one I'm particularly excited about is the partnership that we announce with Kumara and Tyco to provide permanent da storage, because we're the only player in the space that's able to do that because of the decentralized storage network that the 0 GDS built around, also integrating a lot of role business service providers. And then on the gaming sides, recently announced partnering with late to make fully on chain games a possibility. So for there, you need, for example, for 5,000 daily active users, you need a da property that's already a bandwidth that's 16 megabytes per second uncompress, and no other da layer can actually serve that. And so we're particularly excited about really high performance use cases.
Awesome. Sounds interesting. And yeah, maybe last one for me is, do you guys have anything to say to like, let's say the traditional AI companies and startups who are listening right now to this Twitter space? Do you have anything to say to them?
You mean the centralized AI companies?
Yes, yes, like centralized data companies or startups, their starting with AI and should they switch to like blockchain or any other like faults.
For them? Yeah, I mean, we'd love them to come over into the ecosystem and build this kind of nascent space of decentralized AI, a truly OpenAI or on Shane AI. You know, there's many words describe the same phenomena. But if we want to create a world that's full of abundance, where we don't have to trust a few companies that basically control all of AI. This is the place to be. This is the place to really create human good, to make AI a public good. So I'd invite all of them to, you know, find product market fit and then come over to the decentralized world to really fields into that vision.
Really cool. Yeah, I think like ownable, like decentralized AI, AGI kind of future is like a right way to bring the is the right size for the civilization especially.
Alright, thank you very much. And Ada, can you continue?
It just really interesting to hear what are the real use cases or what are the like partnership in the work. Looks like both Portugal is expanding the ecosystem rapidly in the pass and also in expecting to bring more traditional wealth to company into decentralized AI era.I was jokingly talking with my friend about in the future, likely we're gonna have multiple AI agency, AI agent fighting against each other with different intention. I just curious how that gonna play out given that have centralized AI that have bad players always they have some banana AI. So yeah, it's just we have building a technology that we open the Pandora box. We have to be prepared whatever it comes. Then with the partnership between 0G and Orang, I think we have a promising technology to our leads.Monitoring and knowing what's going on is in the decentralized world. With that, I do want you asking Michael and cotton, what would be the, which model or type of model do you think will reach AGI first? Like, could you give us kind of like you predictions?
Pardon? You want to go first?
Is that going?
I think LM hundred percent. And like, is the first type of model going to be the hi, right? Cuz like language model is a simply like a world writer like you, they use world writer to represent the world. And so because we cannot expect a diffusion model that generate image become a AGI someday, right? Like it's pretty much impossible. Maybe I'm wrong. And then, but like most likely, I think language model will be like their first type of model will become AGI or receive probably like thing in another way, language model will be the core model for AGI and then like diffusion model and like ostm. I understand anyone aware model will be like a, a, a like a like functional model. It just like how human brand is structure, right? Human brand have different section and different session to do different things. Like some part of your brand is like processing the vision, some are part of your brand is processing the sound, the listening. And some part of your brain controlling all the muscle in your body and like, I think something like that will be in AGI, different type of model, like function on different things. And so it's like we're pretty much in the process to assemble the AGI and how to evaluate this. The quality of the AGI publish, like up to what language model they're using. And cuz like the language model will send out the command to all another model to do different things when they receive a task. So I think the language model will be the core for AGI and the rest of model will be like a functional area for the AGI. And that's the my take.And then another take is that like, I think like decentralized AI actually is like helping us to keep the peace of the world. Cuz I s notice one really bad like trending right now is that people try to convert AI into, to try to weaponize the AI. Cuz like right now, like the whole logic, the fundamental logic to develop a AI system and make it profitable, it's based on that. Like I have this AI system and you don't, and my competitor do not have that. And then I use this like superpower to crash their business. The same logic can be applied on the warfare. Like, so like this is like a kind of unhealthy mentality. I think should be stop. And like, I think the best way to make the world stop people to producing the killer robot is by like build all the AI model open soursely, transparently so that nobody will like build like, cuz once the, you can see in Github, right, nobody is building a killer robot, like open source project. Obviously, nobody will contribute on project like that. And so like the only problem is that once you build open source AI model that you really hard to monetize that. And then if we RES, if we can adjust this problem with crypto, then it means that like we are turning the world to a more peaceful place.
Yeah, I am. I would agree with carton. I think the kind of lrms that use transformers, I think have the potential to be AGI. And my only key question will be how will it evolve? Like, will it happen? Because we, it's embodied, meaning there's an LM and a bunch of other models kind of running in a humanoid robot. And from the sensor data, it could actually learn a lot quicker as a result? Or will it happen more in a completely digital context? So it would be really fun to see how it evolves over time. And also, will it happen with, you know, bigger is better or will it happen because there's a lot more specialized models that are put together through LA LA middleware or something like that? Or maybe it's a combination of both. So it'll be really fun to see how it'll eventually emerge, but I think we're on a very clear path towards it. And I think by EGI, what many mean is that it's ads or beyond kind of human capabilities across most tasks. So it'll be fun to see how how that how the future looks for it.
It's really kind of funny feeling when we, when I hear both of you indicating the future and the promise of it both, I both feel like I'm in the movie and creating a movie. So yeah, thank you so much for sharing.I have a LA last questions before we close this Twitter space. I do want to call out giving an audience a opportunity to request asking open my questions. So get ready for your question. If you want, you can request to speak and we can let you speak asking your questions. And just keep in mind the question have to asking towards CRG and aura, not any other project, we had an instant that happen in the, in another, eh, ask me anything, people ask me random questions. And my last question is there is that anywhere our audience can learn more and get involved with Oracle or CRG ecosystem? Means like other than following traders, is there any way like we can attending events conference with you guys when you show up, like what events you're attending, are you hiring or are like how you engage with broader community, how people get can get involved. Michael, why don't you go first?
Sure. I mean, definitely follow us on X, follow us on discord, follow me personally on X as well and see kind of our adventures. We're usually at a lot of conferences. We're at consensus this week and then next week will be at Super AI. And then after that likely at the Edge City and then at ECC as well in July. So yeah, there's many ways to engage with us in person at different conferences we're hiring for a number of positions. I think we have a link somewhere on our website for that as well. And always looking for working with great zero G ambassadors and to connect on discord. So just reach out to us. We're excited to partner together. And of course, if you're, you know, developer or project builder, happy to engage as well, just use Telegram just for x excited to engage.
Yeah, like to reach out to me, please reach out to Michael 1st and then.
Reach out.
Information for my Telegram. And I'm joking. So like you can follow my X and also like aura protocol on X and we will present on some event, especially if CC this year and we got two speaking slot on if CC and then, oh, our team will be there. And also we always co host event with 0G and he come to this side event with us and meet both team.
Fantastic. I'm just going give the audience couple minutes to thinking about the question. If not, we can give a give back to give back some time to the people back in the real life. I know for sure in our disco community, we have a many community member join the call in like 3 a. M, 5 a. M. Some of the people have the whole family listen to the ask me anything super push diff. Hopefully in the future we can have more ask me anything in different time zone so you guys can don't need stay late and can enjoy the compensation.I'm just gonna give another minute. If not, we can actually raise.
Your hand, please.
They just need a request on mic. But I think we covered a lot of today, including backgrounds of background of our companies and our partnership, what that means, what it unlocks, what are the promise or potential risk of the decentralized AI. And so it's a lot being covered. Cool. I see people reacting with heart, have no questions. With that, I think I'm going to end the space earlier. So give some time back to people's life and have a wonderful day, everybody.
Yeah, have every yeah, have a have wonderful day.
Have an awesome day. Super fun to be on this x spaces together with carton. Thank you.
Thank you. Have a good one, everybody. Yeah.
0G Twitter:
My Twitter:
Warpcast:
X Space:
https://x.com/i/spaces/1BdxYrQerWyKX
In this Twitter Spaces discussion, speakers from 0G and Ora Protocol shared insights on decentralized AI and its applications in blockchain and crypto spaces. Key points included the technical aspects of decentralized AI storage and data availability, as well as the future roadmap for both companies. The conversation also highlighted the potential benefits and risks of AI in the industry. Introduction to 0G’s ultrafast decentralized storage solution for AI. * Ora Protocol’s decentralized AI Oracle protocol for smart contracts. * Discussion on the Initial Model Offering (IMO) for tokenizing AI models. * Technical challenges and solutions for on-chain AI model storage and verification. * Future roadmap and expansion plans for both 0G and Ora Protocol.
Halo halo thank you. Sama shopping here on the we have a xer froud. We are waiting for om. Michael also joined. Let me accept Michael. We also have a cotton join us right now. So we have all our speaker being here. Thank you so much for being here.
Good morning. Good morning.
Yeah, hi.
Good morning.
Good morning.
Yep, this ask me anything is a partnership between 0G and Ora. And we have prepared some questions for the founders to discuss. And Alice will help me to be the co host. Why don't we kick it off by sharing some background about 0G and Oracle to begin with. Well, I'm also going to make sure we have more question coming up from disco and other platform. We were going to leave about 10 minutes towards the end for you guys requires Mike and asking open questions. With that, I'm going to give the mic to Michael. Introduce Sergey.
GM everyone. I'm Michael, co, founder and CEO of 0G. We're building the first module AI chain. So we said at the middle of the AI infrastructure stack, we're basically building our ultrafast decentralized storage and data availability solution. Most companies in the space are providing anywhere from 1 to 10 megabytes per second and throughput, but that's still very limited for fully on chain AI use cases for things like fully on chain model training, we need tens of gigabytes per second and throughput. And so that's what we enable through our architecture and super excited to partner with carton and Nora as well over 2 cotton.
Cool. Good morning, everyone. I'm carton, the co founder of Ora Protocol. We're building a decentralized AI Oracle protocol that enable AI influence for smart contracts. Very excited about the partnership.
Fantastic. Thank you so much for the quick background introductions. Alex, would you mind to taking on the first two questions?
Sure. So the first question for you guys is first for carton. So you mentioned that aura is Ethereum stressless AI and it's bringing AI and complex compute on chain. But could you explain why is this actually needed and how this works?
Yeah, yeah, yeah, really good question. So I think like, first of all, like AI is like changing every single industry in our society right now. And I think crypto industry will be one of the first like that gonna be changed by AI permanently. And so like it's really important to have AI that empowers smart contract to do something that we used, we cannot do, right? But we know that smart contract pretty much can't run the AI inference inside a smart contract with when you, especially on Eve. So that's why you need Oracle to do that. And in all in aura, we invent the OPML to run the inference decentralized fee and submit the result into smart contract that enables more contract to perform AI. And so which means like now we are able to use AI in a smart contract. So we can do way more involve innovation. So imagine use like a new AI model to be a market maker to increase the capital efficiency in the AMA. And also use like m model to do a landing and stuff to increase the capital efficiency on the landing protocol. So I think most like in general, like AI will like significantly improve the utility ability of the crypto industry.
I see. Okay. So basically we are using this off chain computation and then bringing this AI verifiably to the smart contracts, which opens up a lot of new use cases.Okay, so and then the other question for you, carton, is about the other bought a product from aura initial model offering. So can you maybe elaborate more on that? What is it and how it benefits to the ecosystem?
Yeah, so like initial model offering is like for, it's like tokenized AI model. So the reason why I start this thing is like, because I think like open source AI need a new way to monetize their model, right? Cuz like pretty much if you spend most of million dollar to develop a AI model today, the only way you can monetize that is like through API subscription that have people to pay for yours. Oh, close those AI model. And that's why like, I think if we can have a new way, a new paradigm, a new flywheel for open source AI model to monetize their model, that would be like really helpful for the industry, both AI and crypto. And so like in aura, right, do you have like IMO that is like specifically use crypto to empower open source AI? And also we have like a AI Oracle to use AI to empower crypto. And the IMO part is more like tokenize the AI model and then let the AI model become ownable commissionlessly. And also that can help the open source AI creator to monetize their model and create a positive flywheel.
I see. So like both IMO and Antenna Oracle are kind of working together to make this happen.
Right? It's like a close loop system that like, so yeah, when I wanna use like crypto to enable AI to do to grow the ecosystem for open source AIS E community, then we do the IMO. And then once the model on board on crypto with IMO, then we convert that into a on chain AI, Oracle provide services to smart contract. And then once like people use the AI Oracle in their smart contract and create positive ban, positive revenue. And the revenue will share to the model token holder that people bought in IMO. So it's like the closed loop.
I see. Thank you very much. And with that being said, I'm gonna let my co host ask next to questions.
Yes, since we talked about capability and phenomena of aura, I do want to ask Michael, how does this fit in cog data storage and data availability solution? Michael?
Yeah, this is where we're super excited. So if you look at the AI stack, right, you've got the data generation, things like synthetic data or gathering from different sources. You know, this is very broad. And then you've got kind of a whole data pipeline management, like cleaning and various other labeling tasks. And then you've got kind of us in the middle that's providing to storage and ways of engaging with the models through data availability, and then you've got the kind of execution layers and then the agent layers. And so we're at the data infra layer and then aura is essentially working on the execution layer, providing AI solutions through smart contracts, kind of this AI Oracle concept. And so it's a great match. In order to run opml, you first need to store the actual model on chain. So we provide that. And then you have to serve decentralized inference traces through the da stack. And so the larger the da pipeline, the more you can serve through there. And so that's kind of the basics. And we can go, of course, much deeper into each one of those with Carden together.
We would like to have you and Gordon have a dialogue about the things you just mentioned.
Yeah, carton, do you wanna mention, go a level deeper and mention some of the things you mentioned to me as well that make a lot of sense for utilizing our infrastructure for.
Yeah, yeah, yeah, of course. So yeah, we know that like when we have the model on chain AI model on blockchain, the AI model normally is really big and the size of the model is really big. So you gotta have like a new kind of da system that can like publish giant size of the model. Let's say it's not just like few hundred megabyte. We're talking about like fuel hundred gigabyte. And then so that's why like with this partnership, I hope that like zero gravity can be like a really solid infrastructure. Sure. For all the AI crypto project in the future that can have them to storage and publish the AI model in the zero gravity da.
And yeah, just to add to that, we're trying to get to a cost point that's cheaper than a centralized solution. So to the 3X cheaper than that's just putting it into an AWS S3 bucket with the added benefit of having multiple replicas, erasure coding so that there's a security built in and ways of essentially doing disaster recovery because you have multiple copies. So there's a lot of benefits of actually decentralizing the data itself, you know, beyond, of course, knowing where the ownership is coming from and so on. But yeah, super excited about all of those pieces too.
Since we have aura and OG 0 G and many other AI players arriving. And what do you guys see? Like both of you, Michael and cotton, what, how do you see on chain AI shipping out over coming months and years?
Okay, okay, cool. Well, we're still super early in web 3 meets AI. Even in centralized AI solutions, we're still trying to figure out how to get to AGI and maybe at some point the ASI, which is like artificial superintelligence. And we're still somewhat far from that at the moment. So then, you know, why do you need the centralized AI? We can't even do fully on chain model training today with these large Irlen models. Not well, the data bandwidth issues being solved, but then there's other technical issues like how do you do asynchronous and batch training, for example, in a centralized context, you basically just have a, you know, hundreds and thousands of GPUs strung together in the data center that then can run a, you know, model run. And so we don't have that yet. So there's a lot that needs to be figured out. There's one way where essentially you can say like, well, I'm just wanna bring ChatGPT so you can call it through smart contract.Sure. You know, that's one way of saying like, hey, we're bringing AI to blockchains, but that's not really on Chennai. When I when I think about on Chennai, I wanna see us create actual models fully on chain that starts with the data generation to the data cleaning and labeling to, you know, the storage to than actually the execution pieces. So we know exactly what went into these models, who the owner of this particular piece of data is, so that it can benefit, like pardon mentioned about the initial modeling auto offering, and so that we can truly create an open AI, fully on chain AI. So I see that's gonna happen in the next five years. And I'm super excited about the future because for societal use cases, like when you have agents or big models running things like production systems and transportation systems and administrative systems, I would want to know exactly what went into the models for version of the model's running, what the decisions are being made by the model so that we as society know that we're fully aligned or that the models themselves are fully aligned with our interests. And so I'm excited about building that future of abundance. Carton, over to you.
Yeah, cool, cool, cool. So yeah, of course, like it's important to have this like a super AGI there on blockchain. Well, and like all the infrastructure is ready for it. And I think like, so the main thing I really care about right now is like how AI gonna reshape the crypto economy, right? Like, cuz right now, like kudower is not that big, but still quite big. And so like how AI gonna restrict the whole industry? I think it's like gonna be important. And I notice one thing is like that we have so many new asset getting issued every day on crypto, but most of them do not have enough liquidity or exchange or volume. And I think if we have more AI like that can do trading or do landing or something, and then we can like utilize all this like a new asset issue on blockchain. And then so that means like AI can empower the capital efficiency with the blockchain and then like made the society better or like make people have a better portfolio. And so another thing, big thing is like when we head into AGI, we wanna make sure the AGI is not like out of control, right? We wanna like out of control means like either the AGI become like anti human or the AGI become like own by only few people in the world, like probably only own by Microsoft. I guess we don't want that happen. So we want you wanna make sure that like AGI is like own by the people, at least majority of the people. So like ownable, like AGI is like actually what we really want, like the, the a, the API that own by the public is like what we really want. So like, so we need to invent a new mechanism that can have like use crypto to either vote for the influence or like vote for the token ownership or something like it's not clear at the moment yet. But I think we will find out how we gonna have like a ownable like AGI that own by public.
Are you talking about, go ahead, Michael.
No, I was just saying that's cotton. Exactly. Talked about our mission statement, which is to make AI public good. So super aligned. This is also why I'm so excited about the partnership to our kind of value and vision alignment.
We talked about like potential risk and also potential promise of entry AI and down to the day, like how the builders are can utilize and build the entry AI today. I think that would be a really interesting question to discover, like how can the builders starting using and interacting with the launching AI right now, like this month or next couple months? Do you have some thought, Michael and cotton?
So I think like with builder, the most important thing is like we explore the new idea. So like in Nora, we normally like host a hyper scholar program that like spend time and resources on like helping developer to develop new idea. Cuz I think like when people use AR Oracle to do something on blockchain, that you usage must be super new, super innovative. That is like disruptive innovation that kind of you we place the old player. For example, if like somebody that like can invent a AI landing protocol, then like all the existing landing protocol might not be used by people anymore because like the AI landing protocol just can support way more asset than traditional landing protocol on blockchain or we have like some on chain AI gaming or like the AIGC and FT. Like once people see the AIGC FT, the inference asset, they might not no longer to go to like speculate the NFT asset. I mean, like most people don't do not speculate entity asset anymore in their market anyway. And so like how we do the developer program is that like we, I spend time and resources with like certain hacker that is like really innovative, really good at execution, then invent something new and like that's like disruptive innovation. You.
I totally agree that we need to think about ways of being very disruptive with this technology because it's completely new rules for from an AI step standpoint, like why do you need to interact with a bridge or a wallet? Then AI agent do something like that for you. So the ways of interacting and the type of application surfaces that builders have now is immense. And it can be across the entire stack.You can build a new labeling company if you have some interesting insight, you can build a new data generation company. Like for example, there's some pieces of data that are really hard to reach from a centralized solution. Why not decentralize paid tokens for it, create a new company around it? Or if you want to do something more at the agent side or even at the app side and utilize aura to do a bunch of the inference and kind of calls against some of these models in a verifiable, decentralized way.There's so many different application spaces that you can use it in. You can look at Fifi use cases, social file use cases, gamify use cases. For example, why not build non playable characters that have their own unique storyline for different types of gamers based on their behavior that's fully on chain, utilizing both the Zero G and the Ora side.I mean, there's so many really cool for applications that can be built utilizing this. I'm super excited about the future of it. And so we are also supporting new builders in this space through our collaboration with Monkey Slabs around our incubator. So if anybody's interested in building in the space, please apply. We're still accepting applications from a June and will be hosting multiple accelerators throughout the year. So excited to support the ecosystem.
Michael, you kind of foreseeing what is the next question I'm going to asking? Basically, we want to get a pic for the roadmap and what ahead of us from both 0G and Oracle Standing Point. I think you already mentioned we have incubator and different program. We're curious what Oracle have in mind in terms of roadmap and what happened next.
For Ora. Like the next thing is that like we gonna have more model to be supported in the on chain AI Oracle system and then like have more chain that onboard and more chain provide more services to people and then provide more customization for the AI Oracle to people. So people can like use it to invent a new kind of protocol. That's number one. And second is that like we're getting ready, like we're preparing for like the second IMO with the OLM token. So that means like people with like Ora's point will be able to have like the right to participate the second IMO with OLM token. And that's like the two major thing like I'm working on and the road map for.
Michael from 0G signing point. Do you want to add anything like in terms of technical roadmap and force in the future.
Yes, so we're moving to Tesla V2. We've done a, in a couple weeks, we've done a refactor of our storage nodes codes. We've also moved away from data availability sampling to a query based da solution to increase performance and actual security as well. So we're super excited about fully releasing that. Then after that, try moving towards may not phase 1, which show keep all of that kind of core functionality. And we're hoping for that sometime late July, maybe early August, depending on what kind of feedback we get during our test and YouTube after that may not.Phase 2, we're also targeting for hopefully end of the year where we'll introduce the infinitely scalable Data Availability functionality, which essentially allows us to horizontally scale consensus layers, not just the nodes on the network. And so we can use an arbitrary number of consensus layers with the same validators so that all of the security properties are maintained. And then we'll also be using restaking from Eigenand and Babylon to bring economic security to each one of those consensus layers as well. So super excited about that, you'd be the first to do that in the space.We're also gonna work on kind of decentralized serving frameworks of somebody as very quick data needs and they needs data or even some type of compute through our partners for specific requests, we can through that serving framework and serve those types of needs. So hopefully we can introduce that as part of Phase 2 of the main map as well.
That's a lot to looking forward. I've been asking question nonstop, so I'm going to give my mic back to Alex to asking couple more questions.
Thanks, Ada. So question for carton. So what is the process for bringing an AI model to aura? And actually for Michael as well, about the zero gravity, like what the process of bringing new AI model.
To bring new Mo AI model on chain, normally like the first step is that you need to convert it, make it become verifiable, right? Cuz like the, the, the, like the whole point for doing that is like when somebody can submit an inference chase to the blockchain smart contract, then you can verify it's true or not.So like making the AI model become verifiable is the NUM, the first step. And then the second step is like delegate the AI model to all the. No, that will do the inference.And the first step is like verify the reserve on blockchain and do the proven fraud proof challenge. And so I think like the first step is normally like Danny Aura that we will like publish an open source, the universal compiler that can convert pretty much OAI model become verifiable with OPML later this year. And also like people can like use OG to publish the AI model to all the to do to all the influence know that will take the job actually and then return the result to the smart contract. And recently I notice a really important usage for OG actually. So like, so for OPML, one of the biggest challenge was the deterministic problem in the AI model. Cuz like once the, because like 0 opmi is based on a fraud pool challenge kind of mechanism.So, but you know, traditionally, like, yeah, if you don't do any modification, it is the AI model inference is pretty much undeterministic. So mean, it means like the proposal and challenger will have different outcome, even they both are honest. So that will be have a problem. That means that you have some problem for the challenge. And, but fortunately, in the newest OPML paper, like we're working on to adjust this problem with a round up layer so that like we you run up the neural network when you done with the computation every layer then but the price to do that is that you will have a kind of long run up chase and then you gotta update the round of chase to the some da layer to make sure another proposal and challenger will do the same round of function that you did. And so with this, it means like actually like when you wanna use OPML on a execution environment, then you actually need like not just the influence trade being published by the da, also the round up trace published by da.
Alright, a pretty technical answer, but thank you very much for that. And Michael, do you maybe have anything to add here.
Adding from kind of a technical perspective or just anything else? I mean, the round of trace, super cool. Definitely need a high performance da for that. So excited about supporting it. Where you looking for anything else specifically or?
Yeah, I think that's fine. Yeah, so the next question for you guys is what kind of AI projects are you using your solution in the industry right now? Like, are you especially excited about some projects?
Yeah, so I think like the actually the most exciting innovation that use AI Oracle right now, actually it's like a project call 7,0,0,7 AI. It's like a protocol for inference asset. So once we make the AI inference become verifiable, and then essentially what we're doing is like turning the inference into a asset class, a new enchant asset class. So like when, so now people can mean a AIGC FT without even know how to draw a art. So they can just like put down prom and enjoy really good art and then sell it on blockchain immediately with like probably like 10 to 5,15 dollar to just issue a new AI and gcnft. So I think that will have like a pretty much like a massive adoption. Cuz I think about what like the most popular the app at the moment in the crypto space is actually pumped up fun, right? So like it's a provide a really convenient way for people to like launch a mean kind by just 2 dollar. And but like we know that like mincon have like a rocking problem, like the price go out really quick and go down really quick. But like if you use like FT to reshape this, like launching new asset class behavior, then it will actually made liquidity work better because like normally like FT is a non fungible. It means like different FT will have like, will lock up the liquidity that like going to that NFT. So it will like provide a better way to for people to issue a AI generated content on blockchain. So I think like influence asset will be like one of the most excited use case for or for launch AI Oracle at the moment.
That sounds super fun on our end, it's very infrastructure heavy. So we, on the AI side, I'm particularly excited about some of the, for example, like model training aspects like pond or and chats with flock. And we also announce some fine tuning use cases like Beagle, for example, building on top of infrastructure so that even full models are utilizing our storage in the future to be a problem piece as well. And some of the other use cases outside of the AI world would be, for example, on the L2 side. So you just announce the partnership with 10 Network where they're bringing privacy to serial twos, obviously also integrating with various other leading L twos.Another one I'm particularly excited about is the partnership that we announce with Kumara and Tyco to provide permanent da storage, because we're the only player in the space that's able to do that because of the decentralized storage network that the 0 GDS built around, also integrating a lot of role business service providers. And then on the gaming sides, recently announced partnering with late to make fully on chain games a possibility. So for there, you need, for example, for 5,000 daily active users, you need a da property that's already a bandwidth that's 16 megabytes per second uncompress, and no other da layer can actually serve that. And so we're particularly excited about really high performance use cases.
Awesome. Sounds interesting. And yeah, maybe last one for me is, do you guys have anything to say to like, let's say the traditional AI companies and startups who are listening right now to this Twitter space? Do you have anything to say to them?
You mean the centralized AI companies?
Yes, yes, like centralized data companies or startups, their starting with AI and should they switch to like blockchain or any other like faults.
For them? Yeah, I mean, we'd love them to come over into the ecosystem and build this kind of nascent space of decentralized AI, a truly OpenAI or on Shane AI. You know, there's many words describe the same phenomena. But if we want to create a world that's full of abundance, where we don't have to trust a few companies that basically control all of AI. This is the place to be. This is the place to really create human good, to make AI a public good. So I'd invite all of them to, you know, find product market fit and then come over to the decentralized world to really fields into that vision.
Really cool. Yeah, I think like ownable, like decentralized AI, AGI kind of future is like a right way to bring the is the right size for the civilization especially.
Alright, thank you very much. And Ada, can you continue?
It just really interesting to hear what are the real use cases or what are the like partnership in the work. Looks like both Portugal is expanding the ecosystem rapidly in the pass and also in expecting to bring more traditional wealth to company into decentralized AI era.I was jokingly talking with my friend about in the future, likely we're gonna have multiple AI agency, AI agent fighting against each other with different intention. I just curious how that gonna play out given that have centralized AI that have bad players always they have some banana AI. So yeah, it's just we have building a technology that we open the Pandora box. We have to be prepared whatever it comes. Then with the partnership between 0G and Orang, I think we have a promising technology to our leads.Monitoring and knowing what's going on is in the decentralized world. With that, I do want you asking Michael and cotton, what would be the, which model or type of model do you think will reach AGI first? Like, could you give us kind of like you predictions?
Pardon? You want to go first?
Is that going?
I think LM hundred percent. And like, is the first type of model going to be the hi, right? Cuz like language model is a simply like a world writer like you, they use world writer to represent the world. And so because we cannot expect a diffusion model that generate image become a AGI someday, right? Like it's pretty much impossible. Maybe I'm wrong. And then, but like most likely, I think language model will be like their first type of model will become AGI or receive probably like thing in another way, language model will be the core model for AGI and then like diffusion model and like ostm. I understand anyone aware model will be like a, a, a like a like functional model. It just like how human brand is structure, right? Human brand have different section and different session to do different things. Like some part of your brand is like processing the vision, some are part of your brand is processing the sound, the listening. And some part of your brain controlling all the muscle in your body and like, I think something like that will be in AGI, different type of model, like function on different things. And so it's like we're pretty much in the process to assemble the AGI and how to evaluate this. The quality of the AGI publish, like up to what language model they're using. And cuz like the language model will send out the command to all another model to do different things when they receive a task. So I think the language model will be the core for AGI and the rest of model will be like a functional area for the AGI. And that's the my take.And then another take is that like, I think like decentralized AI actually is like helping us to keep the peace of the world. Cuz I s notice one really bad like trending right now is that people try to convert AI into, to try to weaponize the AI. Cuz like right now, like the whole logic, the fundamental logic to develop a AI system and make it profitable, it's based on that. Like I have this AI system and you don't, and my competitor do not have that. And then I use this like superpower to crash their business. The same logic can be applied on the warfare. Like, so like this is like a kind of unhealthy mentality. I think should be stop. And like, I think the best way to make the world stop people to producing the killer robot is by like build all the AI model open soursely, transparently so that nobody will like build like, cuz once the, you can see in Github, right, nobody is building a killer robot, like open source project. Obviously, nobody will contribute on project like that. And so like the only problem is that once you build open source AI model that you really hard to monetize that. And then if we RES, if we can adjust this problem with crypto, then it means that like we are turning the world to a more peaceful place.
Yeah, I am. I would agree with carton. I think the kind of lrms that use transformers, I think have the potential to be AGI. And my only key question will be how will it evolve? Like, will it happen? Because we, it's embodied, meaning there's an LM and a bunch of other models kind of running in a humanoid robot. And from the sensor data, it could actually learn a lot quicker as a result? Or will it happen more in a completely digital context? So it would be really fun to see how it evolves over time. And also, will it happen with, you know, bigger is better or will it happen because there's a lot more specialized models that are put together through LA LA middleware or something like that? Or maybe it's a combination of both. So it'll be really fun to see how it'll eventually emerge, but I think we're on a very clear path towards it. And I think by EGI, what many mean is that it's ads or beyond kind of human capabilities across most tasks. So it'll be fun to see how how that how the future looks for it.
It's really kind of funny feeling when we, when I hear both of you indicating the future and the promise of it both, I both feel like I'm in the movie and creating a movie. So yeah, thank you so much for sharing.I have a LA last questions before we close this Twitter space. I do want to call out giving an audience a opportunity to request asking open my questions. So get ready for your question. If you want, you can request to speak and we can let you speak asking your questions. And just keep in mind the question have to asking towards CRG and aura, not any other project, we had an instant that happen in the, in another, eh, ask me anything, people ask me random questions. And my last question is there is that anywhere our audience can learn more and get involved with Oracle or CRG ecosystem? Means like other than following traders, is there any way like we can attending events conference with you guys when you show up, like what events you're attending, are you hiring or are like how you engage with broader community, how people get can get involved. Michael, why don't you go first?
Sure. I mean, definitely follow us on X, follow us on discord, follow me personally on X as well and see kind of our adventures. We're usually at a lot of conferences. We're at consensus this week and then next week will be at Super AI. And then after that likely at the Edge City and then at ECC as well in July. So yeah, there's many ways to engage with us in person at different conferences we're hiring for a number of positions. I think we have a link somewhere on our website for that as well. And always looking for working with great zero G ambassadors and to connect on discord. So just reach out to us. We're excited to partner together. And of course, if you're, you know, developer or project builder, happy to engage as well, just use Telegram just for x excited to engage.
Yeah, like to reach out to me, please reach out to Michael 1st and then.
Reach out.
Information for my Telegram. And I'm joking. So like you can follow my X and also like aura protocol on X and we will present on some event, especially if CC this year and we got two speaking slot on if CC and then, oh, our team will be there. And also we always co host event with 0G and he come to this side event with us and meet both team.
Fantastic. I'm just going give the audience couple minutes to thinking about the question. If not, we can give a give back to give back some time to the people back in the real life. I know for sure in our disco community, we have a many community member join the call in like 3 a. M, 5 a. M. Some of the people have the whole family listen to the ask me anything super push diff. Hopefully in the future we can have more ask me anything in different time zone so you guys can don't need stay late and can enjoy the compensation.I'm just gonna give another minute. If not, we can actually raise.
Your hand, please.
They just need a request on mic. But I think we covered a lot of today, including backgrounds of background of our companies and our partnership, what that means, what it unlocks, what are the promise or potential risk of the decentralized AI. And so it's a lot being covered. Cool. I see people reacting with heart, have no questions. With that, I think I'm going to end the space earlier. So give some time back to people's life and have a wonderful day, everybody.
Yeah, have every yeah, have a have wonderful day.
Have an awesome day. Super fun to be on this x spaces together with carton. Thank you.
Thank you. Have a good one, everybody. Yeah.
0G Twitter:
My Twitter:
Warpcast:
No activity yet