How AI Will Completely Transform The Human Labor Economy - PublicAI Founder

Transcript

How do you know what the AI system needs? >> The most value is after you've birthed one of these agents into the world, can you go back and say, "Hey, give me the money you made." Like that. That's where the human in the loop click is going to be the most valuable. But it feels like that's about to flip. It feels like with MCP agents like getting more contextual awareness about themselves, agents will start to prompt the humans. Welcome back to the rule. This is your home to learn from the biggest guests, the brightest minds at the tip of innovation. Our goal is to help spread the collective knowledge of everyone in our community.

>> We are breaking crypto out of its echo chamber and taking it to the real world. >> Sit back, relax, and enjoy it. >> Jordan, welcome to the show. >> Absolutely. Um, first question is, uh, this AI super cycle is the name of our show. uh super cycle was pioneered and kind of coined by C uh excuse me Suzu uh three arrows capital and called for a crypto super cycle. Uh this was uh kind of in contrast to the four-year cycle and and kind of the tra traditional boom and bust cycles that we had seen in crypto. Um, my first question to you is, uh, do you see this AI, this capex, all of this investment going into AI? Is

this a AI bubble or do you see an AI super cycle at play here? >> Yeah, I mean, we're definitely I would say that it's a combination of both. There's there's no way that we're breaking the pattern of people getting excited about new technology, being overzeilious, and uh, expecting more from it than is actually possible. Um, that's just what happens with any new technology. Um, there's like the garter hype cycle and that's part of it. The the bubble like exists in that cycle. Um, the good news is all of the capital going into these things means that what we're able to do is tons of experimentation to find the right product market fit for the new technology. So that after the bubble has

burst and we've gone through the disillusionment and everybody's fed up with it that we get back to a place where we have like legit applied use cases that make a lot of sense. So from that regard uh yeah we're we're in a bubble in a sense but then also like this is part of a super cycle where the nature of technology is changing and the nature of work is changing and that's what has public AI really excited about what we're doing is like as we see uh the nature of work changing jobs changing things like that we're ensuring that humans have a place in that that we're able to capture value and that we're able to like pay people for participating in these systems.

Got it. Okay. Now, you know, we hear a lot about uh AI coming and displacing the labor market, automating a lot of tasks, uh and people being out of work. But I have a feeling what when you describe changes happening and the future of work being vastly different from how it is today, I have a feeling it it kind of goes a little bit deeper than just people losing their jobs. Could you just unpack a little bit more what you mean when you say like labor and work is going to look very very different? What do you mean by that? >> Yeah. So whenever a new technology comes and in the nature of labor shifts um there is always like fear and jobs I would say like transform more than they

disappear. like new kinds of labor start to exist and what we're doing at public AI is focusing on data which has been valuable even before the AI super cycle. Um you know in in the web 2 world large companies were able to make money because they would harvest users data and then resell it or use it for their own purposes. um basically the the web two model of if you're not paying for it, you're the product, which means that you're generating data which can be used for commercialization.

Um so what public AI is doing is focusing on on data as this like thing that's been valuable um since the beginning of commerce and trade like having aformational edge is how you trade better than other people. Mhm. >> And uh we're paying people directly for participating in a system that lets them opt in to sharing their data. Um contributing to smarter AI. Um and then on the other side, we have clients that know the value of this, want to source their data ethically so they don't wind up in legal quagmires. Um and also need like niche on demand data that isn't possible to get at through just scraping or through synthetic means. So we're

we're filling that that need uh for the AI market, but having come from a marketing background have seen like the trajectory of uh data markets in general being like a strong theme. >> Yeah. Okay. Um I mean this is interesting to parse through because I think there's a lot of people that are understanding that AI is coming. They don't want to see their jobs displaced. They want to participate in this market, but they're not quite sure how. And so when when I see kind of these data labeling companies pop up or these data markets pop up such as public AI, how can people get involved? You know, we know that we have our data. We're constantly getting, you know, if we download a new app, it's always like,

hey, do you want to give permission to this thing to access your data? And I'm always like, I don't know. Sometimes I say yes, sometimes I say no. It's just kind of like whatever I feel like that day. Is there a general rule of thumb that you you know we can we can give people or that you you suggest or that you use yourself? What is the general rule of thumb when it comes to our data and the permissions that we give to some of these external applications? >> Yeah. Yes. You see those shrink wraps so often that you kind of turn your brain off when you're clicking on them, but you know that that's how they get you and they get your data. Um in in some cases it's worth the trade-off. You've got to make a personal distinction on your own whether like the service you're

being offered is worth it. Um and if there's alternatives out there that you know you can you can opt for whether it's open source or you know there's companies like um Proton which is >> headed heralded by a foundation like a nonprofit foundation to make a product that is like data uh privacy focused. So there's options out there. Um and you know web 3 is also part of that. There's a lot of decentralized software that you can use too, but to make it really tangible for people and um if you don't see the value of your data, then it's very easy to click it away and and not really be motivated to do it because I don't think that privacy is really that top of mind for most people. Um the

convenience outweighs any privacy concerns they have. But when you start putting economic um incentives on the table, then they start to realize, oh, this is valuable and maybe I can do something with it. Um, and that's when you know they're motivated more so to participate in systems um like public AI and to make public AI a little bit more tangible like we run campaigns on a regular basis. So like today if you visit public AI you can sign up and participate. Um, currently we're running like a mother tongue campaign where people can come speak in their native language and earn up to $4 a day for recording short audio samples, like sentence length audio samples. So that that's something that like um makes it

makes people realize the the value of their data. They're actually showing up. They're getting paid out. Um it's better than necessarily like a gig economy where you've got to wait two weeks for a paycheck or something like that. you're getting USDT um as soon as your data is approved. So, that's kind of what we're focusing on is making this like immediate and and tangible for participants in the systems. Um and yeah, that's that's how we're how we've grown to like over three million users now. I think the last time I checked we were at like 3.5 million contributors with uh like something like 300K showing up every month, monthly active users.

So, people are excited about it. >> Very cool. And and these are real people. These aren't AIs uh you know fact-checking other AIs. >> Yeah. >> Yeah. Real real human value. Human in the loop. >> Very cool. Um you mentioned privacy which is kind of all the rage these days. Uh Zcash has been getting a ton of attention generally. There's more of a feeling around privacy. I think also the uh recent surge in in gold and people kind of flocking to these um these things that they feel like are safe havens. uh you know, even though they're kind of just a meme in and of itself, uh like most things, they do have kind of this sense of sovereignty when they come to gold and they come to a privacy based coin. Um why do you characterize privacy

as such an important um form and and characteristic of AI? you know, when you're building these things out, you mentioned privacy as as something that, you know, maybe isn't forefront in a lot of these in a lot of the users minds, but maybe it should be. What generally what is your what is your uh perception of privacy as it relates to AI and kind of this evolving, you know, technological world? >> Yeah, I think that it um it becomes more and more relevant as we integrate AI into more parts of our lives. Even before the AI hype and it being called AI, we had machine learning and applied machine learning. And that's where my co-founder uh Dr. Steven Wong has his

roots is in applied machine learning. So um that world uh you know at one point it was just in your inbox but now it's in your smartome. uh in your you know generative chat whether it's chatbt or cloud or whatever you're using and if you're sharing intimate details for it searching for you know health medical things whatever else uh that that is like an intimate touch point in your life so privacy um could become more of a concern for people but again like it it while it's becoming a little bit more of the vernacular it also doesn't seem like

something that is like the number one driver for people's behavior. Um because even in the web two days when when search was already being driven by machine learning and people understood that like oh I'm getting ads because I searched for something that didn't prevent them from googling anymore. um like it it's it's something that um you know as as AI gets implemented in government and people start seeing more impact on it and what their profiles look like and all the places planteer shows up then you know it might become more newsworthy and kind of top of mind for people but I think that it's really like the hardcore web 3 cipher punks um that are going to be championing this

and driving that messaging so that people understand Oh, cryptography and privacy, you know, are something that we do have to fight for and uh establish as a human right um for it to like really take foothold in in the like the mimosphere in zeitgeist. >> Yeah. Yeah. And we kind of see these surges of it when it when it when there's greater uncertainty in the world, I think people kind of uh you know, they kind of hunker down, if you will. It's almost like a like a not necessarily a doomer mindset, but like a little bit of a prepper sovereignty. Like I'm gonna have to I can't really rely on someone else, you know, to give me, you know, my sustenance. Got to kind of figure it out on our own. And and privacy and

sovereignty are very intertwined in that regard. Um, now you guys have have launched this. Um, I'm curious what the pipeline looks like, right? Because I think it is intuitive. Look, we've all been training these AIs for a long time, right? I label fire hydrants and street lights on my captions, you know, every time I try to log in somewhere, uh, and prove that I'm not a robot, right? Um, and so we're we're training, you know, these algorithms, we're training, uh, machine learning, uh, and these AI systems in that way. And I think people are really they're really familiar with that because we've honestly people on the internet have been doing captures for so long. Um, and that's effectively what we're doing. Um but but there's a

data pipeline that downstream once it does get labeled. Um and so I'm just curious when when you're thinking about this data pipeline, where is the labeling process, you know, if we're looking almost for like this, we don't want blackbox AI, right? We don't want something that has some intrinsic bias that has been uh uh almost like spoiled or or um uh you know manipulated by some some power whether that's open AAI or that's you know some other company that is choosing to spoil it and and kind of slant it in a particular direction. So where does labeling come into play and and how do we leverage this this decentralized data

labeling process to ultimately give us like organic untouched AI? >> Yeah. Yeah. So talking about the pipeline kind of gets us back to the the client side of things. And honestly, I think that privacy and sovereignty and like the authenticity of data is much more top of mind for people on that side than on the consumer side. Like to be able to have a competitive product, you need data that other people don't have. to be able to have a product that works with consistency and doesn't hallucinate and actually is reliable and cost-effective to run, you need to like have that data edge and train a model that's more efficient and more precise.

So on on that side, um the privacy is a much bigger deal and the like the um having a traceable um like trail of where the data comes from that it is authentic that you aren't going to wind up in litigation for training with the data matters a lot more >> um as far as the pipeline and and like where human value comes in. Public AI is really focused on sourcing because that's the like the roughest edge where it's the hardest to reach people. Um, our network is kind of like seven degrees of Kevin Bacon for any area of expertise. So, if somebody needs like luxury good um experts or pet experts or

speaking a specific language or specific dialect of a language, we're able to go and reach those people um enough of them to be able to train models effectively no matter what the business use case is. So that's kind of like where we're focused right now. But then when we talk about like scaling across verticals and other places where we see uh human value fitting into the value chain, um labeling is there. AI labeling is getting better and better, but for every new use case, you at least need a set of samples to start with to to get the AI label um labeling like up to par. So like those edge use cases um with with the boom of AI and people being able to multiply their impact there's going to

be a lot more solarreneurs um new startups that uh need need applied data for their models things like that so in that sense of tackling new use cases human labeling makes sense for the first set of samples to be able to train out the the AI to be able to label better in the future and do differentiation um And it's not an easy problem. Like one of the things that we deal with is, you know, fakes. We do have people because we have incentives in place showing up and trying to farm it with AI generated voices or robot voices and things like that. So um after trying out all of the um competitors in the space for being able to detect fakes and things like that, we had to roll our own

>> for our use case. Um, so that's just one example of like uh how how hard it is to um apply AI for specific use cases and why it's necessary to have like that first layer of labeling done by people to be able to train something that's um efficient for a specific use case. And then after that you get into like human in the loop for training the model. And then also after that you have human in the loop for usage of the model. So um rating inference and things like that and those are places that we see are valuable and as we grow our business um we'll be expanding to those but right now data is data is hot.

>> Habachi provable trade per earn rewards visit hibachi.xyz trading on fire >> relay low fees lightning speed and the power of multi-chain at your fingertips. Instant multi-chain payments. Choose any token on any chain. 75 plus network supported. More than 5 million users. 8 billion plus volume processed. 70 million plus transactions. Fast, reliable everywhere. Start swapping today. Only on Relay.

>> The wait is over. The Talis test net is live. Be part of the revolution. Join now at testnet.talis.network. >> You you mentioned a couple times. I'm just curious. Is there a general like nomenclature when it comes to human in the loop? Is that a is that a protocol or a process that is typical when a in in a in a AI companies to kind of give it a human feel? Like what what is the purpose of human in the loop? >> Yeah. um it's usually abbreviated like um HITL and it's a commonly used term in the industry and uh research segment for AI and it it applies whenever you do have a human in the loop for the training process or um inference feedback uh for data gathering anywhere

where you do have the human touch and um you know the research goes back and forth on oh we made a breakthrough that means we don't need humans anymore and then another paper comes out that says Oh my god, if we don't have humans, everything just collapses. So, um, our our bet is with the humans. We're putting our chips on the humans. >> Love it. Yeah, I think that's a good good good place to bet. Um, the odds are in our favor. We got a comment here in the chat. It says, "This is so interesting. Will research public AI?" This is from Monica. The biggest issue is data governance and applied data. >> Yeah.

>> Do you do you agree with this? Yeah, I think that um when you're talking about data governance, like that's one of the reasons why we decided to take a web 3 approach and actually launch a token that gives people a stake in the platform. So, if people go out there and earn or buy public, they have a stake not just in their data that is now for sale, but also the platforms, the means. They've got like their hands on the shovels and they own the shovels they're using to to mine this data. So for us that kind of like sovereignty and governance angle was really important and why we took this approach rather than a traditional web two approach is to give people an opportunity to ear own the tools that they're earning with. Um

and that comes from my own like background and and Stevens as well where like in college I was doing pirate radio to support local unions and things like that. So I've always had like an interest in empowering people, building platforms to elevate them um and give them an opportunity to have a stake in in what they believe in and what they're contributing to. So governance um huge huge deal and yet it all gets real when you apply it. >> Yeah. And you guys chose to launch this token and and generally the incentive mechanism on near um you know the show is presented by Near. We also understand that they are you know pioneering especially some of the some of the privacy enabled elements of AI as they

made crypto and then also connecting some of these ecosystems to the broader landscape connecting with the recent near intense uh launch that they had to to Zcash to Ethereum um and generally broadening this this ecosystem. Why did you choose to launch the public AI system on MIR? And and like you mentioned, you have three million users. Ultimately, these people, do they realize that they're interacting with the MIR blockchain or to them does it just look like, hey, I'm kind of providing this this piece of human uh touch and I'm getting rewarded for it.

Um and you know, that that's great for them. >> Yeah. Yeah. So, for us, like I I have a background um with the near ecosystem. Um I actually got into the web 3 space through an opportunity to work on DAO software going back to governance um with Ilia Pulisukin um when like uh I was just getting started in web 3. So that's what got me to leave my cushy Google job and jump into this space with both feet was like okay this is a once in a lifetime opportunity get to work with the transformers paper guy um on some exciting new technology and uh governance at the time was also a really hot topic. So um getting to build like D infrastructure and early NFT infrastructure at near Foundation was um

a real pleasure. So I had familiarity with the ecosystem. Um I knew about the like the awesome UX of the wallets and um how it's great for onboarding and then like what we use Web 3 for is paying people and I think that that's like what web 3 is best at. Um, but I also see a perfect marriage between AI and web 3. Like giving AI agents digital native um like currencies and wallets makes perfect sense and allows them to transact in in like digital native ways.

So while web 3 has kind of like clunked around and looked for a use case besides payments um the AI age has given a new life and with near being so focused on AI I mean it came from the AI space to begin with >> um it just really made sense to go with a chain where payments are fast and cheap >> allows you to bridge anywhere with near intent. So it doesn't matter if you're a web three user coming in and looking to like bridge your token somewhere, you can do it easily. Um or you're a web two user experiencing this for the first time. And in the the life cycle of a user on public AI, you can log in with Google or Twitter, whatever else, email, um and do your contributions. And the point where you start caring about

blockchain is when you get paid. And it's at that point that we kind of like make sure that people have a wallet connected and they're able to claim their tokens and have instructions on what to do with them next, things like that. So >> I see that overall in web 3 is the best way to go about onboarding users is you don't put up a big barrier in front where you're like, "Hey, go learn about crypto in web 3 before you even touch our product, whatever our product is." >> Right. Yeah. Absolutely. I that is that is what I think the people who have the most foresight in building these applications they understand that look this is infrastructure and and at a certain point it's meant to be invisible people want to get paid they want to get paid in an efficient and seamless way

but they don't want to have to change their consumer behavior in order to interact with this technology they want to just get paid and if it happens to be efficient and it happens to be more cost-effective than the alternative perfect but they don't want to have to adapt or learn or cross all of these bridges or all of these things in order to ultimately take uh take the opportunity that that they have at their disposal and and that's what what they're doing here. They're just kind of getting paid and it just so happens that blockchains are a great mechanism to pay people. I I'm curious as you've built this out and you've onboarded millions of users. Um how you anticipate us interacting with the AI economy going forward? You know, when you mention, you

know, blockchains are great for paying people, that that makes a ton of sense. They're they're great for people paying people. They're also great for AI paying humans and humans paying AI. And so, I think as the machine economy develops, we're going to start to see this play out more and more. Right now on public AI, I is how do you know what the AI system needs? like how do you know that you know we're at we're at a at a a point where look we've got you know a real um gap in this particular language right we really need people to upload you know uh Vietnamese like sound bites right or or like

pictures of this food or h how how do you know that you know there there's kind of that piece of data that's missing and ultimately send out this request um to the human layer uh to provide that data. >> Yeah. So, we're a traditional revenue-driven company in that regard. Like clients come to us requesting data and that's what gives us the resources that we need to reward the community. So, clients come and pay us in fiat and then we issue rewards um proportional to that for the community to go and get the data that they need.

>> With enterprise clients, it's usually pretty clear-cut. they know exactly what they're after data wise and we're we're just filling the need of being able to get niche on demand data for them that they can't get through their like centralized factories or any other means. >> And then for the small medium businesses, startups, um you know, uh younger teams, there's usually a little bit more handholding and like learning what they're solving with AI, what data is going to actually be necessary to help them get that tuned and then running the campaigns for them. And the way it works for clients, if there's any founders out there listening, is like after an intake call, we'll like figure out how much data you need to verify that it's like what like the quality is good enough and that it's useful for

training. So, we'll do like a small test campaign like a month like a month or a week long um around that with the right incentives in place so that you have an understanding of how much it'll cost to run the campaign. And after we get that initial run done, we can scale it like as large as you want with our our huge contributor base. So comes from client demand and a little bit of alpha. One of the things we're working on right now and really excited about is um an MCP service uh allowing AI agents to make their own data requests uh directly on the platform. So, it would be like a permissionless self-service portal that would allow, you know, startups to interact with it through an through an

AI interface to be able to run a data campaign, but also allows agents to come in and say like, "Hey, I've been hearing about this thing. I want more data on this thing. I'm going to go pay for it with tokens." Um, which is a pretty exciting prospect. >> Asset management redefined built on ERC7621. Standardize how you launch, manage, and trade onchain baskets. This is Alvara, the standard for onchain asset management. Blockchain shortcuts, templates, and widgets launch without friction for builders by builders. Visit enso.build.

So, let me just make sure I I understand this. The the idea is that an AI agent maybe it's and and Ilia was actually on this show and kind of enlightened us to the fact that you could have these totally or organic agents that are born out of smart contracts that have no biases injected into them. And so if you have something you can verify the code you kind of create this agent out of the smart contract and then maybe it's modus operandi is to go build some trading algorithm and it needs to get a bunch of back testing history right it can go maybe it is also attached to a wallet we can fund it can go request certain data sets on public AI through MCP without

ever having a human kind of like tell it to go make this request or making it the request for it. It can kind of just discover on its own this ability to request the data from the humans that have have the data. >> Yeah. Yeah. Exactly. So, one of the um really exciting developments in the last uh few weeks has been um Nano Chat, which is like a a one bit model framework from like scratch that you can run on a local GPU and like train your own AI for like $30. And if we're talking about the smart contract version of this and being able to like bootstrap a model without any biases and then have it grow from there, the next step for

that one bit model is like, oh, what am I lacking? Why don't I go get some data for that so that I can train the 32-bit version of myself and like continue to grow. So it's it's it's we talk about AGI and these pieces are coming together which are like really fascinating and that that's a whole rabbit hole to go down as far as like oh how the markets have already been mechanations that have been running and have their own kind of like I wouldn't say consciousness but they have their own like motivations and everything. Um, >> they're right.

>> They're starting to recognize things about themselves, strengths, weaknesses, and and they're starting to generate self-awareness to the extent that now now they have this contextual awareness about themselves. It is only just beginning now that they can go do something about it. >> Yeah. >> Um, right. With MCP, they can go start to make these requests. But I be I I want to just back up just for a moment. Could you give us a refresher about MCP because I think there was a nice wash of of it kind of like on on socials and we all kind of learned about it together when it first came out. But, you know, could you give us a refresher? I think that was a few months ago when I first heard about it. What generally what is MCP and kind of catch us up on things that might have happened over the last

few weeks to months. >> Yeah. So MCP is model context protocol and it's just one way to do agentto agent communication and that's the nutshell is like agentto agent communication in a way that that makes sense. It also gives agents um access to tools um outside of like what they already have inside of their knowledge and understanding. So it's kind of like APIs for agents. So instead of like in the old web two days of making an API so your website can talk to other websites, >> get tools, get information, instead you would have your agent using MCP to be able to go do a calculation or go get a data feed, things like that. Um MCP is

just one approach. um other there's other approaches um that are handling this problem of like how do agents get access to resources that are outside of their um their inference um and their existing models. So that's all MCP is. Um I think that like cloud is working on a similar thing called skills and and there's a few approaches but M MCP has kind of become the de facto um at least for the the moment that a lot of people are building on >> and and it makes sense for for this to be a shared standard. All right because then those rules kind of apply to all of the agents and we don't get these walled gardens where you're only composable with certain sets of the agents and

other ones are only composable with certain sets. Ideally, we kind of have this really really broad and diverse uh ecosystem of agents that are able to interact with one another. >> Yep. Only makes them stronger and more capable. >> Yeah. I mean, coming back and coming full circle to our first, you know, thing that we had here, thinking about how the labor market has changed. You know, right now the predominant resource that we have at our fingertips is AI. You know, you and I, we're humans. We're in the labor economy and we're prompting AIS, give us information, give us information, answer our questions. But it feels like that's about to flip. It feels like with MCP agents like getting more contextual awareness about themselves. Agents will

start to prompt the humans. Agents will prompt the humans and the humans will be the ones feeding them information instead of the other way around right now and and and humans prompting the AIS. Do you is this the way? Is this your vision? Do you see this playing out? >> Yeah, you you start weighing the the the balance and and the value of like um where where is human in the loop valuable? And from a human's perspective, the most value is after you've birthed one of these agents into the world, can you go back and say, "Hey, give me the money you made." Like that. That's where the human in the loop click is going to be the most valuable.

>> So there like where are these agents coming from? If you're thinking in a non-abstract way, they're there because somebody wants to get something done, whether that's um running social media or trading on their behalf or so on. Um so there's got to be some kind of like output that's valuable to people otherwise like what's it running for? And yes, there will be experiments of like letting it run for its own sake and things like that. and we'll see how far those go and in what directions, but for human context and value, there's got to be something to click on to get what you want back out.

>> Yeah, absolutely. And just in the last few moments here, I know, you know, you're obviously building this this this business and this protocol that's running on Near. Um, they've got a ton of composibility benefits. There's open source models that are on there. There's tons of tooling and infrastructure. I, you know, a really popular narrative that kind of took took hold for a while was uh was these agents on Twitter, these kind of like Twitter reply agents, not not the spammy ones that, you know, are still in our all of our comments, but you know, the fart coins and the the you know, AI6Z's of the world that actually did get some money put into them.

>> Are are you bullish on these this category? What do you think comes out of all of these agent coins and and and these these sorts of things? >> Yeah, I mean I also um run like my own art and tech nonprofit collective. So have been kind of like watching the space evolve and seeing the kind of like art experiments and financial experiments happening here. And I think that truth terminal is the killer use case in this category of like a really well done experiment that is gaining some traction and um has some interesting like uh philosophical things to say about like how these things are playing out. Um, from from our end, uh, at public AI, we're going to be launching a Vtuber streaming channel

with a partnership with AI TV where we have a character that covers like the latest in DAI news, uh, things like that, and people can interact with her. Uh, when they do so, they're like spending some onplatform points that then buys them public tokens. So, we're fascinated with it from like a marketing and community and engagement um, standpoint. And I think that as a marketer, like if you're not leveraging AI tools, then you're just not multiplying yourself enough. >> So, we're excited about it from that perspective. And also, like I'm I'm personally interested in seeing like where these experiments go philosophically and and how how they kind of like figure out those questions of when we do give the agent its own

like um agency and we don't preload it with our own preconceptions like where does it go? artists are gonna figure this stuff out. >> Yeah. Wow. This is fascinating. Fascinating stuff, man. I I I feel like we could pontificate on like where this goes. What are the philosophical implications of it? Um, but I mean, it's just it's still too early to see. We're going to see this play out in real time. And I think you're right. We're just continuingly adding more tools. Like we're It's almost like we're giving the agents its hands. We're giving it feet now. It can go run. It can go ask questions. it can go uh if we give it some money, it can go pay for things and get these things and ultimately include it into its own arsenal.

>> Yeah. >> Fascinating things that are playing out. Um if you're if you're kind of placing your chips here, where do you broadly like just what are you bullish on? What are you bearish on? I love the the the uh the DEAI like TV AI TV channel. Maybe there's even a mashup that we can do with the rollup where we, you know, bring bring them on rollup TV. We got have a co-host, right, which is an AI agent. >> J. So, so that's uh that's some some place where you see a lot of room for potential. Where else do you see room for potential? And then where do you think, you know, maybe this category of AI is a little bit overblown? Where are you, you know, secular, bullish, and bearish within the AI category?

>> Yeah. Um so we're obviously bullish on data and the space in general. Um we're friends with the other people who are doing decentralized data plays like um Sapion and and others where in in the old world of data uh all of the agencies and firms that were doing um data sales had revenue share programs where like oh I can't get this data but I know this market research firm is an expert on it so I'm going to partner with them get the data and they'd get a cut. So, we're seeing the same thing kind of play out in the decentralized data space where the more of us are playing together on solving the data problems of the world

in a decentralized way, sharing clients, sharing deals, um the stronger the industry is going to grow and we're all going to grow together. So, super bullish on that. Um even with the coopetition in the space, uh seeing it as being like healthy and growing. Um, and then the things that like I'm I'm less bullet on I would say are um maybe like decentralized compute and storage. These kind of like commodity things. It's it's it's novel um but it's also hard to to solve the coordination and make it available um for like on demand usage. Um and also like usually

commodity problems a company can just like if they have money they just solve it by going to the easiest solution that has the most SLA uptime something like that versus like risking going to a slightly cheaper decentralized solution. So like commodity stuff um a little bit more bearish on that but things that you can't get anywhere else like niche on demand data >> like totally bullish on that dude. I I'm so incredibly bullish on the human in the loop but not just broad strokes um you know like any any human can submit anything but maybe there is room for that and then and then we can have a

market to price it. But ultimately these AI agents as you know if if public AI can become the go-to venue for agents because it has this moat of data and the network effects start to build in that regard the distribution becomes so key and we start to have uh yeah like these venues like these town squares where AI needs to go to the market like the supermarket in order to get its sustenance. Well it sustenance is data and it needs to get the data from humans. >> Absolutely. Yep. Yep. Gotta feed the AI good food. We got it over at Public AI.

>> Awesome. Jordan, man, this has been a lot of fun. U Thank you so much for coming on on such short notice, too. I'm really looking forward to hearing some more of the updates uh and and what's coming out of public AI as you guys continue to grow. Um and NPC becomes online. It starts to get a lot more popularity. I think this entire space is just going to get a lot more popular and a lot more activity as agents and humans both realize this this inevitability. >> Yeah, absolutely. Exciting times ahead. Thanks for having us on. >> Absolutely. We'll see you soon, Jordan. Thank you.

>> Thank you. Take care. Bye. Type.