User-Owned AI: On-Chain Training, Inference, and Agents, with NEAR's Illia Polosukhin

Hello, and welcome back to The Cognitive Revolution. Today, my guest is Illia Polosuhin, founder of Near, a hyper-ambitious, multifaceted project that describes itself as the blockchain for AI and aims to build a future where AI belongs to everyone. Illia is not your average blockchain founder. Before starting Near, he was one of 8 co-authors of the seminal 2017 paper, Attention is all you need, which of course introduced the transformer architecture and helped launch the current AI revolution. The credits on that paper say that Ilya and another co-author, quote, designed and implemented the first transformer models and were, quote, crucially involved in every aspect of this work. And in fact, Mir started as an AI project but took a major detour into crypto when Ilya and team realized just how difficult it was for them to pay their data task workers all around the world. Today, given his pedigree, Ilya could no doubt command one of those mythical billion-dollar comp packages from Zuck's Superintelligence Labs, but instead he's building the foundational infrastructure for user-owned, privacy-preserving AI that can operate at global scale. In this conversation, we start with foundational principles and work our way up the technology stack. We begin with an overview of how Near's proof-of-stake consensus mechanism creates security for network participants without requiring trust in a centralized authority. In short, because the protocol is open and permissionless, anyone who is willing to put value at risk can become a validator, and thus no one can prevent the flow of valid transactions. From there, we discuss how Near is leveraging NVIDIA's confidential computing capabilities to create a permissionless network that allows anyone with compatible GPU hardware to sell inference compute. While keeping both the model weights and user data private, even from the hardware operators. All with only a 5% overhead relative to normal computing. To be honest, I hadn't realized just how affordable this privacy layer has become. And while it does require trust in the chipmaker, it's clearly a huge deal for all sorts of scenarios and does seem to help explain how it is that frontier model developers have been able to ship their models to so many different inference partners Without leaking the model weights. After that, we dig into Nearest plans to decentralize model training itself with a process that allows contributors to provide whatever compute or data they have in exchange for a cryptographically guaranteed share of the model's future revenue. While it does remain to be seen whether a community-driven project can produce a 1 trillion parameter model that performs at the level of today's leaders, at a minimum, Ilya and team seem to have designed an incentive structure that could make it worthwhile for folks to contribute the estimated $100 million worth of resources that's required to compete at that level. Finally, we talk about what AI and crypto can do for one another, including how AI might finally put the smart in smart contracts and how blockchain technology can provide stronger guarantees that AI agents will act only as intended. This is something that I've been envisioning, if only in quite fuzzy terms, for years, and so I was really excited to hear just how concrete it's starting to become. In the end, I was so eager to understand the foundational technology on which Ilya's vision rests that we barely had time for the vision itself. But I do think this is really valuable knowledge. And the good news is that we've already scheduled another recording. So definitely stay tuned for part 2, in which we'll explore the applications that people are already starting to build at this intersection of AI and crypto, and also try to get a handle on the giga-agent future by exploring Ilya's vision for how AI agents will interact, transact, and even participate in governance. With that, I hope you enjoy this technical deep dive into the infrastructure for user-owned privacy-preserving AI with Illia Polosuhin, founder of Near.

Illia Polosuhin, founder of Near Protocol. Welcome to The Cognitive Revolution.

Thanks for having me.

I'm excited for this conversation. I think this is going to be one that I'm going to learn even more than I usually do, because you're right at the intersection of AI and crypto. And sometimes these 2 technology waves are characterised as different camps that don't understand each other, or at times are in some sort of online rivalry that may be mostly made up. But I've got this sneaking suspicion that I've never quite been able to fully develop, that these 2 technologies actually might really need each other to reach their fullest potential, and to do so in a way that we can keep control of over time. So I'm really excited to dig into all of that with you. For starters, though, if my deep research reports are to be believed, and this wasn't hallucinated, I do know for a fact that you were one of the authors of the original Transformers paper. But then I understand from deep research that this project that we now know as the Near Protocol started as an AI project and then sort of evolved into a crypto project before now kind of coming back to the intersection of crypto and AI. So I'd love to hear a little bit of the story of how you went from an author of attention is all you need to branching out on your own to realizing that you needed actually crypto as well as attention. And yeah, just give us a little bit of that history and bring us up to the present day.

For sure. Yeah, no, that's good deep research there. So my background is in machine learning and AI. I joined Google kind of when I saw— I don't know if you remember this cat neuron. paper where they effectively trained an autoencoder model, meaning they would feed the images and compress it and uncompress it to get back the same image. And that model trained on a bunch of images on internet figured out that there's a cat. And so there was a neuron specifically that if you activate it would show up as a cat. And so for me it was like, this is the approach. Neural networks working now. But I think images are not where the knowledge and intelligence lies. I think the text is. And so I joined Google with very much like, hey, how do we focus deep learning on natural language? And the straightforwardest path to kind of advance and check intelligence of any human or machine is to ask questions. So my team worked on question answering and we actually had even some of this in a product on google.com. Back in the day before Gemini, you would get these short answers. And part of the challenge we had was actually like the models that were state of the art were too slow, right? You couldn't, you know, the long short-term memory like LSTM models, right? They need to read one word at a time. And so, you know, if you feed them a bunch of articles from Google search results, right? There's like, it'll take forever for them to respond. And so that's actually how transformers was one of the motivations. Like, hey, how do we consume all this context effectively, as much parallel as possible, and then figure out how to answer? Now, after that work, I was like, the slope of evolution of AI is accelerating. This is 2017, and I want to build products on top of it. And so I left Google to start NearAI with my co-founder Alex Khudanov. with a premise that we actually want to teach machines to code. And so that's actually something that I believed a long time, that if machines can code, we're effectively changing how we're interacting with computing, right? It's changing from, you know, few people can sorcerer the code, you know, and like talk to machines and then everybody else effectively, you know, need to consume their magic to everybody can effectively do everything with a machine that normally only, you know, a few developers can do. And so we were trying to build that back in 2017. That sounded like science fiction. Now it's just called vibecoding. So we're trying to build vibecoding in 2017. And the expectation we had is what we've seen in 2022, '23. We thought that's what's going to be happening in the industry. And as we know now, effectively we're lacking probably an order or a couple orders of magnitude of compute. Now, we were trying to be smarter, so what we tried to do is to get a lot more training data. Now, again, it's called instruct fine-tuning data. So we were trying to get people to actually write a little bit piece of code for some instructions or write instructions based on code. And the people who can do this are effectively computer science students from developing countries where a few cents here and there That adds up to like dollars, a few dollars a day, actually reasonable money for them to work online and kind of practice their coding experience. Now, the challenge we faced was paying them, right? People like students in China don't have bank accounts. They have WeChat Pay. Back then in Ukraine, if you receive dollars in your bank account, you're actually by law required to sell half of it. There was no way to pay into some countries for whatever reasons, right? Like there was the PayPal didn't work, TransferWise didn't work. And so there was all kinds of weird, like just pure payments problems that we faced while trying to collect more data. And so we ended up saying, okay, well, we've heard of this blockchain thing. It's a global payment network. It should solve our problem. So we're just trying to use blockchain as a solution to our own problem of kind of coordinating and paying a bunch of people. And as we did our research, this was really 2018, we kind of like, well, architecturally, none of them scale. They're kind of very slow, hard to use, hard to build on. And so that's sort of like, hey, if there was a blockchain with this criteria, easy to use, scales, microtransactions are always predictable price, we would use it, but there's none. And so we should just build that and then use it ourselves and come back to AI. And we thought we're gonna build it in 6 months, to be clear. So we're like, hey, you know, it's easy, right? It's just some systems. We're gonna ship it and come back. It took a little bit longer, but indeed we effectively built, you know, highly scalable blockchain that's focused on ease of use, really abstracting out the blockchain itself from kind of the user experience, as well as highly programmable. You can effectively run arbitrary software that, you know, written in C, Rust, JavaScript, Python, et cetera, on the blockchain as well. Obviously all the financial stuff, payments, loyalty, remittances being run through it, a lot of microtransactions going through it, as well as, you know, a few other use cases. And so kind of it is one of the most used blockchains. We have 50 million monthly active users on it. And so kind of as the '22, '23 did happen, right? Like we started building up the team back into like, okay, now that this acceleration is happening, how do we end with this new learnings we've had through the blockchain experience, right? Where can we bring that vision of, hey, the AI will change how you experience computing forward. This is the history right now looking forward. The important part is as AI will write code, it'll be able to interact with other tools and systems. You're effectively removing the need for other apps and even websites, right? Because effectively anything can be built as the stack matures on the fly, it can go and directly get the information you need. And so it's interestingly, there's a few things that are happening. One is the devices are going to become more run by what we call AI operating system. Now, this AI operating system, however people want it to run locally, you will not be able to process everything locally, right? There's background, you want it to summarize news for you, et cetera. And so you need some way to offload the compute. The problem is it needs to have all of your context. That's how we're going to get to true AI that's really helpful is because it'll know everything about you. But that is a very scary situation if some third-party company has all your data, right? We just saw everything from the chats being leaked to effectively some people in organizations and reporting to police with your chats. So there's already kind of a challenge with kind of how the data is handled by the centralized AI companies. So So first of all, we want everything to be private. I think that's like a fundamental. So like we want AI to be private. We want AI to have all your context. We want it to be kind of be able to go and execute actions on your behalf. And so the other interesting thing that happens in this future is that a lot of the existing, what, you know, you can call them aggregators like Google, DoorDash, Amazon, As well as other middlemen, like I call the FDA, for example, it's a middleman, right? It takes all the pharmacy filings, it processes them, it evaluates them. It's like, okay, yes, this medicine is safe on average. And then they allow pharma companies to stock it. Your AI can do all this work, right? It can go and find directly the factory in China. to order stuff, it can go directly to AI of the pharma companies, discuss with them exactly your medical situation, and find the exact medical compounds that you need at this moment right now. Like you don't need to check, you know, if on average for average person it's going to be good, you actually need to check if it's good for you right now with what you need to address. So you're kind of removing this middleman system that our society is actually built on. and going very much direct peer-to-peer, be that between different people or people in companies or even people in autonomous agents. And so that's where blockchain actually comes on the other side, because it's effectively facilitating those type of transactions. So there's effectively 2 pieces of this vision. One is how do we ensure privacy, verifiability? And there's another aspect, which is if this is how we see the world and how we interact with kind of the world is through this AI lens, a small alternation into that AI can effectively lead for you to perceive the world differently. Like the example I use is if you go into ChatGPT right now and like at the beginning say, hey, subtly convince me while we're talking about some, whatever the other topic of X, right? Let's say voting for somebody else that you don't like. And then you go through the conversation, right? It'll start actually like working in to change the accents on how you think about things in the chat on another topic. Now, this can be in system prompt right now, or this can be in the weights directly. We don't know. And so we also have this concept of what we call user-owned AI, because you want the AI to be yours, not theirs. Privacy is part of this, but the other part is the model itself. We need to know what goes into it. We need to know and have a verifiability of when you run it that there's no additional ways that it's actually being affected. So that's kind of a big part. And to do that, you need a different model, right? You need to build a movement, not a company, right? Because if you build another company doing that exact thing, right, you'll end up in the same result. So you need to build a movement around this idea where people are actually willing to contribute to build this more of a, as a common good that's accessible to everyone while creating an economic engine behind it to actually power it? Because it, this actually requires a lot of financial investment and resources. And then on the other side, as this AI becomes kind of your interface, how does it interact with other AIs? And that's where kind of what we call Intents, AITP, and other components are coming together. So that's really kind of the vision we are working toward.

Hey, we'll continue our interview in a moment after a word from our sponsors. There's a lot there to unpack. I'm going to maybe try taking it from what I think is the most foundational layer and kind of work up toward the giga agent future. Most fundamentally, when we talk about— and obviously there's a lot of problems in society today, including loss of trust in institutions, right? So trustless is becoming a general description of society, perhaps something that certain technologies can achieve. The avatar, for a lot of people listening, don't have a great handle on where with different blockchain schemes the trust comes from. I know, like, in the original Bitcoin, the idea with the proof of work is that basically it's really costly to mint a new valid block. And so because it's so costly, nobody could plausibly rewrite the whole chain to present. You'd have to do that so many times in a row that nobody could really ever get there. And so the canonical version is kind of the canonical version. That difficulty of extending means that nobody can kind of corrupt the history. Okay, so that's unfortunately, though, pretty costly to run. I looked up, I think it's the— Bitcoin is consuming the same amount of electricity as the country of Poland today, per my internet research. So we've now seen this move to proof of stake, and there you sort of have A lot less compute required, but you have to make sure you have really thoughtful incentive design. People are essentially putting their holdings of tokens up for putting them at risk, basically locking them up, putting them at risk, and saying, I represent that I'm going to do the right thing for the broader community, and if I don't, I stand to suffer some consequences as a validator in the network. So can you maybe elaborate on that just a little bit? Because everything we're going to talk about ultimately, if I understand correctly, is built on the idea that we have a set of validators that we don't necessarily have to trust as individuals, but we sort of have to trust collectively. And we also have to trust that the incentives are such that they don't have any reason to defect from the stable equilibrium. So can you give us a little bit more about What is that stable equilibrium that you have designed? Who are the validators? And in your mind, what is the ground that people can put their confidence in that everything else that we'll build on top of this, that we'll cover, is really on a solid, albeit distributed foundation?

Yeah, so that's a very good question. And I think just to contrast it with our traditional internet, Right. Right now we're using Riverside to record this. We went to riverside.fm. We actually relied on somebody to tell us Riverside is this IP address. That somebody is a distributed system, but it actually has a single organization that effectively decides how it's done. Similarly, we have SSL certificates that give us internet encryption. But again, there's actually certificate authorities who issue the certificates. And there's like the authority that gave them this, right? So, there's like effectively points of centralization on the current internet where you cannot verify like on itself that this is correct. Now, when we think of blockchains, actually the important way to think about it is from your perspective, right? Like you wanna verify that what you interacting with is correct. And how hard is it for someone to effectively lie to you and show you the fake kind of information. Like, that's what you're looking for. So, as you said, for Bitcoin, effectively, if you're looking at some state, for example, you just received a bunch of Bitcoin, the cost of this is actually not true is effectively will be at kind of how much it costs to produce that number of blocks. And indeed, For example, for an hour— to lie to you about an hour of history, it costs $2 million. Right. So, that's kind of the idea. Now, with proof of stake, it's a little bit different because in Bitcoin, it's possible that, for example, Google has a lot of servers. They could be effectively creating a fake blockchain that's days and weeks long. And they can present that and effectively lie to people and steal a bunch of money. Now, don't think Google's doing this, but like just to give you an example. With proof of stake and specifically this idea of Byzantine fault tolerant consensus, you actually have this notion of finality where if indeed 2/3 of the stake is not kind of non-malicious, then they're following the protocol. And so there's no actually way to have a fake history or like to lie to you. And you can kind of verify this from the beginning, right? So like we launched Block, kind of the mainnet, October 2020. And so since then you can verify that like, you know, every single transition was correct and all the rules were followed. So that's kind of the idea. Now, who are these validators and kind of, and what is the stake, right? So I mean, the example for Near, but it's similar for other proof of stake, it's a lot of kind of both projects in the ecosystem, right? So we have a number of other companies in the ecosystem that building on top of us. So they effectively have a vested interest in the security of this network. So they become participants. And becoming a validator is permissionless, right? You can actually start a validator right now. It's also now on Near, you can run a very kind of lightweight node, like on an old, somebody ran a node on like 8-year-old laptop recently. We also have, for example, Vodafone, NTT, and like other kind of entities like that running nodes. We have crypto exchanges who, are participating in the network as well. So, it's really about the participants in the network who become validators as well because they have vested interest, they have some of the stake as well. So, they have both like kind of business interests and financial interest in the network. And so, they become participants and validators. And then also just like a bunch of community kind of, we have universities, we have some individual developers who are running nodes as well. So that's kind of the idea. You have like, again, open network, people join. We had actually a pretty large number of new validators joining in past 6 months. And in turn, they put some of their money at stake as well as maybe some other users can delegate. So they kind of say, hey, I trust this participant and I'm going to add more stake to them effectively. And then there is a reward that's being split between all of them to effectively incentivize them doing this.

This is maybe a really naive question, but how, if it is permissionless, maybe we could, let's talk about the importance of permissionless versus permissioned. Because I was also, I'm sure you're following the Tempo project that, I don't know if you see them as a direct competitor. I don't know. You know, there's obviously a lot of projects out there. Tempo has Stripe behind it, as I know you know. My understanding, reading through what they have said, is that they plan to start with trusted validators and then gradually become more permissionless over time. And I'm not sure if this is one of those things where it really is a spectrum or if it really is more of a binary. I always say AI defies all binaries, but when it's something like, is it permissionless or is it not permissionless? That sounds like, how do you become more permissionless? Is there a way to become like, if there's some permission there, it still seems like there's some. Anyway, I'm a little confused about sort of exactly how much that matters. And then also, if I just show up with a node, like, how do you know who I am, who I say I am? Like, if I show up and say I'm Coca-Cola here, you know, is your new validator, do you have to like, meet me in real life? Do you go to the Coca-Cola office and kick the tyres on the servers to do that sort of real-world validation? Again, these are obviously noob questions, but knowing how much I might one day rely on my AI agents built on top of all this, I do want to make sure I'm solid in the foundational understanding.

No, this is all good questions. Yeah. So I think that permissionlessness indeed is a spectrum because, and there's like maybe like few Step Functions, there's few networks where effectively there's like a very like set specific set of validators. And what this has as an issue, right, is that let's say if these validators don't like you for whatever reason, right, they can effectively not let your transactions or your AI agent's transactions on-chain. They can effectively censor you. And so that's a main challenge with like, even if it's like not one, but like a subset of kind of permissioned validators is that for whatever reason, they can effectively exclude you from the network and you cannot do anything. The benefit of this permissionless network is that you can yourself join as a node and say, hey, I'm actually going to join the network. And then I can, even if they don't let me in, I can myself push transactions and interact with the network directly. Right. So that is like the biggest kind of change, which for normal people is like whatever. It's not that big of a deal. For maybe people who've been through, you know, I'm from Ukraine, you know, I've seen different types of, like, you know, I have multiple banks that like effectively closed up that I was a client of there. Obviously it was a war and everything. Like things like that are pretty important, obviously for journalists, for, but more important is actually like for businesses. Like if I'm a business, right? If this is a payment processor that I'm using and the competitor business is effectively controlling whose transactions are included or not, or even there's also just like delaying, for example, my transaction, et cetera, they can affect my business. And so that's why like there's this idea of neutrality, right? You want kind of ability for everyone to join because then, you know, like there's no single party that can pull it on themselves or coalition of parties, a cartel, you might say. that can pull on it. So, that's kind of one side. The other question is like, okay, you want to join, you say you're Coca-Cola, you can actually do that. But nobody will believe you because there's no social signal right now. If, for example, Twitter of Coca-Cola tweets like, hey, we just started a Near validator, here's our address, here's how to delegate, let's go make Coca-Cola the biggest validator on Near, then that would actually provide the social signal. So the way you identify yourself to the network is effectively through public-private key cryptography, right? You essentially say, hey, I'm, you know, this validator, this account. And then to become like, you need to put some NEAR at stake, right? You need to put some money at stake. And so now you're a validator. So you don't really need to, you know, meet in person or do anything. You can be a cat, you can be an AI, it doesn't matter. If you have kind of some capital stake.

Gotcha.

So, anybody who shows up, if they are willing to make an investment in the currency, which they have to have in order to then be able to put some value on the line, that's the proof of stake is that you actually own value and you're putting that value at risk. Not at risk in a probabilistic sense, unless you do something wrong, right? Like, the idea is, as long as everything you do is valid in your role as a validator, then your value is not at risk. But if the rest of the network finds you to have fake transactions or whatever, then they can take the value that you put forward essentially as collateral.

Correct. Yeah. And like we have even a mechanism because so far it's mostly people been having misconfigured nodes. So it's not really been a malicious attack, but actually people just, you know, had whatever bugs in their setup. And so we actually only kind of, it's called slashing when you actually take their money at risk. So we only slash proportionally to how much stake you actually have. Like that participated in the malicious behavior. So, if you had the misconfigured node and you only had like, you know, 0.1% or 0.01% of the total stake, you're only going to get slashed, you know, whatever, a multiple of that percentage of your stake, right? So, it will be very little. Now, if the whole— if, you know, a large percentage of network coordinating the attacking the network, then they're effectively gonna get slashed fully because like that is a percent multiplied by some coefficient will be slashed. So that's the idea is like we don't wanna punish people for misconfiguring things, but we wanna punish if there's a coordinated attack for whatever reason, which I mean, it can be that somebody hacked into a bunch of nodes, like they had zero-day vulnerabilities and done that. But like we need people to have kind of responsibility over their security setups and— kind of things. But we want to allow people to have like, kind of individual mishaps.

Yeah.

Gotcha. Okay.

So is there any—

last foundational question on the security layer before moving up the stack. Hey, we'll continue our interview in a moment after a word from our sponsors. Is there any risk of incentives changing in the long term? Like right now, The market cap of the Near coin, last I checked, I think it was $7 billion. And that was, I mean, it's obviously a lot, but for people who have businesses or whatever. And then you also, I guess, would imagine if there is some coordinated attack, probably that market cap falls real fast. So you could try to hijack, but also what have you won, sort of, right? I imagine that's a big part of it. But in the fullness of time, if we imagine, like, a massive AI agent economy built on top of this whole thing, is there ever a point where it sort of could flip from it's not worth doing a coordinated attack to maybe it could become worth it because instead of $7 billion, it's $7 trillion or something like that?

Well, I mean, if the network were $7 trillion, Right? Then the attack will cost, for example, 30% of that. So if somebody can come by with a couple trillion to attack it, but then the question is like that attack will not be able to extract much, right? So this is where it's like, again, from a perspective, like the question is like you as an individual, because attack is towards someone, right? Like the system itself is From its perspective, it's always correct, right? It follows the rules. If somebody violates the rules, they're not in the system. The only way to attack is effectively to attack you saying, hey, here's the fake information. For example, your agent, like I sent you $1 billion, give me $1 billion of services. And that's— I didn't actually send you $1 billion, right? I just lied to you. By creating, by attacking the system, by kind of faking that there is actually a transaction that happened and it's been final. And so to do that kind of attack, you'll need effectively 66% of stake. So if the network is at that point $7 trillion, you need 66% of $7 trillion or whatever the stake percentage is. And so, and then you probably, as you know, if you're receiving $1 trillion in a transaction, you're probably not going to be like immediately cool I'll give you a trillion dollars of value. You probably can wait a little bit, make sure that there's nothing else going on in the network. So there's kind of a question of timing, value, and security that together work, right? So similarly how Bitcoin right now, Bitcoin is whatever, $2 trillion, right? The 1-hour attack is $2 million. So if you're sending $10 million, you probably should wait a little bit longer than 1 hour to make sure nobody is attacking you. If you, and et cetera, right? Like the more you send, the longer you should wait to make sure. But kind of the idea is like, as the value increases, like then you can also, you know, wait less for larger transactions. So that's why it's like, it's a little bit like people say, you know, the 51% attack and it's important to actually understand what this means. It means somebody specifically getting attacked. Like it can be an exchange attacked or it can be individual, a service provider, et cetera. It's not that the network itself is still correct, right? Like the rules are still followed. It doesn't matter like which effectively so-called fork will end up being picked. Like from that perspective of that fork, right? You're still on the main thing. Blockchain is a little bit of this like abstract points of view, indeed, fourth wall kind of perspective is like, are you in it? In it, you're always correct, right? Or you're outside and you actually, are you seeing the right thing or not? And like, how can you validate that you're seeing the right thing? But like for most use cases, right? If you're spending $1, $5, $10, $100 to buy something, like effectively the security of the network is so much higher then it's not a problem and you can effectively accept the transaction right away. And so for normal use case of an agent, it's fine. And then the reality is like there's more challenges with indeed security of the like zero days of the code of the systems than like, than the kind of economic security of this. And we should talk about this because we have a whole thesis on that.

Well, maybe now could be the time. My plan from here is to kind of work our way up to, okay, how do you train models on this? And there's interesting aspects of that. And then how do you build business models on top of models and ultimately agents? So if you want to talk more about the—

So on the security side, yeah, I think the important part to understand is, as AI now getting better and better, it actually gets really good at finding vulnerabilities, right? I've used it to backtest effectively. You take a recently found vulnerability, you run LLM over it, you effectively don't even need to over-prompt it, and it will find you some vulnerabilities in the code. Somebody found a Linux, I think, zero-day with that approach as well. What this means is because cybersecurity is so kind of one-sided, right? Like it's really hard to know that you don't have vulnerabilities and it's, but it's, you know, it's getting easier and easier to find them. And like it's also true about, you know, social engineering and other attacks. Like it's just gonna get easier and easier. So we kind of need like a fundamentally different approach. And so the fundamentally different approach actually relies on math. Like the only thing that you can rely on is actually the whole system is correct given again your perspective, right? So from my perspective as a user of the system, I want to know that if I'm doing something, then it is actually correct, right? There's no vulnerabilities in the system or potential problems in the future that can happen. So I want actually a mathematical proof at the time of using the system that it is correct. given my requirements. So it's a very different, like for those who are, you know, there's been a lot of like, that's called formal verification. There's been a lot of research on this, but because it's so complicated, nobody actually doing it at the time of using services. You may be doing it like, for example, when they were sending stuff to Mars, right? They actually had, I think, formal verification of the code for the Mars rover. But it was done once. It's effectively like, okay, we want to verify that it runs correctly, will run correctly given the specification. But obviously if anything gets updated, if the requirements change, et cetera, now this is not usable. And it's because it's so hard, it required a lot of manual labor. Now the benefit is with AI now, that's getting really good even at winning gold at International Mathematics Olympiad. we can actually simplify that process of proving itself. And so, what we believe is we actually will need to rewrite effectively every single line of code in such a way that it's formally verifiable at the execution time. And with blockchain, this is the first step. Blockchain kind of, because it's the root of trust, we want to formally verify it so that you as a user, when you, for example, sending money or depositing money somewhere, you can say, hey, I'm expecting to receive money back at least as much money. And the chain needs to prove to you that this is actually going to be correct and it's going to happen. But you can expand that to broader services, right? Like if I'm using some service, I'm giving it my private data, I want it to prove to me that it's not going to leak it, right? And so for that, this is where we're going into like verifiable compute in a broader sense. We kind of need to both for verification and guarantees that this code is going to be executed over your data. So that's where trusted execution environments come in. So that's kind of like the position on like kind of zero-day verification. Like we actually, we're moving in a world where this is going to, I don't know if you like a couple of weeks ago, right? There was like zero-day in all the iOS, right? And Mac and everybody were patching. We just had a massive attack on the supply chain in npm. yesterday in crypto space. Like somebody injected code effectively to replace addresses in all of the tools. So there's like the amount of this kind of vulnerabilities is going to just accelerate and we need a new approach. And so that's one of the like longer-term research we're working on to actually solve that. But yeah, you need a few more pieces. And so that's where we're getting into like, okay, well, how do we actually guarantee privacy around these models, have verifiability, as well as indeed build the monetization, build the kind of financial engine that actually runs this. And so we're using effectively a combination of indeed blockchain, cryptography, and hardware. So as of about a year ago, NVIDIA supports so-called confidential computing. So what this means is the NVIDIA chip itself connects to Intel and ARM and even AMD supported in such a way that even I'm as an operator, like the owner of the hardware, cannot access what's happening inside. So, as a runner of operating system. So, this is called confidential computing mode. And the interesting thing is it gives you both confidentiality and verifiability.

Right.

It tells you this code was run on this data and only this thing was done. And by the way, nobody else seen what happened there, right? And so what we're using this is effectively to now, I'm as a user, for example, want to run some AI workload. I can enter and establish end-to-end encrypted channel to this secure enclave and run some AI workload there and then receive back result again end-to-end encrypted. And I know that there's no other single party that were able to actually understand what I was doing there. And I also kind of received a verified certificate that says this model was run on your input and this is output, right? And there's no way that something could have changed, you know, the degraded performance midway or whatever, right? Or they injected some prompts into the thing, right? It's like you have guarantee of the execution. Now, so that's kind of the basic primitive of this kind of confidential verifiable compute. And so there's few sides of this. One is, well, we now need a lot of compute, right? Hundreds of thousands of GPUs need to be here so we can actually provide it for lots of users. And again, Like one of the challenges with doing that is massive data centers, like lots of CapEx. Now you need a lot of energy. It's all in one place. You have latency problems. You have sovereignty problems. So all the challenges. Now the cool thing is again, because the owners of compute don't actually see anything, we can again open it up and make it permissionless. We can let anyone with compute join. We have a way to verify that compute is indeed NVIDIA and Intel chips. Because they effectively sign the certificates. And now we can decentralize the compute itself, right? So we can have compute everywhere joining across the world. We can then route requests to the closest and available compute to you. You get confidentiality, verifiability. You now can pay to the network and network then distributes these rewards very similarly how validators receive rewards for providing Validation of the blockchain here, the compute providers receive rewards for providing compute and then users pay for using inference or developers paying for using inference. So that kind of creates what we call decentralized confidential machine learning cloud. And it's not just for AI stuff, you can also run arbitrary compute over it as well, data processing, agents, MCP servers, et cetera, et cetera. store memory end-to-end encrypted for users. So that's kind of the second piece, which again enables that vision of like, hey, how am I going to have an iOS that truly private and I know what's happening there? And so this already has an economy, right? We're effectively bringing compute and usage. Now the other challenge is indeed models and data in a sense. And so there you have, okay, well, right now if I'm a model developer, I effectively have 2 routes. I'm like open sourcing it and making zero money, or I keep it closed source and then I need to actually procure compute, make sure that compute is verified. Like I trust the compute provider, meaning like because I'm uploading my weights to them, they can actually steal and open source my weights or, you know. Runs them alternatively. So that's why only big hyperscalers actually usually partner with Microsoft OpenAI, Amazon Anthropic. And then on top of this, there's also this maturity problem where if I have, I'm just launching a new model, I don't have any users for it yet, but I kind of need to commit to a bunch of compute from this hyperscaler, usually like 2-year contract. I don't actually know how much I need. If I get more users than I have compute, now I need to pay a ton more. So it's like a very weird also economic question. Again, economic question. We have tools to do that. And so the idea here is that now if it's open source model, you upload it, or if it's closed source model, you can encrypt and upload it to our system. And so now again, because all of the compute happens in this confidential environment, we can decrypt the model inside confidential environment, meaning again, nobody's able to access what's happening there. And you can run on users' data privately. And now when users pay, they pay both the developer and the model developer and the hardware provider the fee. So you're effectively now combining model providers and model developers into this, offering them a way to serve their models without prepaying, right? They don't need to pay for the compute to serve it. They don't need to take users' data. They don't need to deal with GDPR, HIPAA, SOC 2, whatever, all the compliances. They just like push a model in, start making money when people use it, right? So that kind of creates now a new economic model for model developers. Similarly, you can have for content providers. creators and data providers, right? You can upload your data that can be used at inference time or potentially later at training time. And if it is used, you're getting paid from that transaction as well, similar to Spotify, kind of like there's subscription fees or kind of API call fees that get distributed to the providers. So that's kind of like how we see this, like kind of you have a cluster of services that effectively all run on the same kind of decentralized hardware like cloud where you can bring models, you can bring data, there's users and developers consuming this, and it all kind of gets secured on this like blockchain-based marketplace that really facilitates it.

So a couple questions there. We could go on a long time. I suspect this is fascinating already. When I show up to the network, I basically just have to have an Nvidia chip of a certain type that has support for the secure private computing, right? That does suggest that there's, like, important— I don't know if it would be right to think of it as a— I guess you could have a hardware vulnerability or a software vulnerability there. Hardware vulnerability, from what research I've done, people seem to think, like, if you gave some of these chips to, for example, the Chinese government and let them bring the full power of their immense engineering prowess to bear on cracking that hardware, people, from what I hear, seem to think they probably could. But that it wouldn't really be economical to do it for much stuff because it's like, really, obviously these things are exquisitely crafted. So to do some sort of hardware-level modification, you might be able to figure out how to do it. You're probably not going to find it worth your while to try to scale it to some large number of chips. So this would be like very focused, kind of hyper-targeted attacks at most. But then there's also the software question of like, How do we know that Nvidia has actually— is their stuff formally verified that this is all fully locked up? How much trust and where are we putting trust there? And then part 2 would be, how much overhead is associated with this? I did one episode on use of zero-knowledge proofs to prove that the model— and this was 2 years ago, so the the framing at the time was, and it's actually still relevant because just in recent days there's been this discourse about Claude seems to be dumber for me during the day than the night or whatever.

Exactly.

And people are saying, oh, are they quantising it or are they not? And they're saying they're not, but something happened and it's not exactly clear what the framing at the time was. Well, if they use a zero-knowledge proof, they can demonstrate to you that we ran the model we promised you to run. You're getting the value that you're you're contracted for. However, at the time, there was a lot of overhead associated with doing the actual zero-knowledge proof. So it wasn't like for 2% more you can get a guarantee. I forget exactly how much, but it was like, it's a lot more to get the guarantee.

It's like 10,000 to 100,000x.

Yeah, it was a lot. How much risk is there at hardware? How much risk is there at software level? And how much overhead is there to actually get these benefits?

Yeah. So, given you touched on ZK proofs, yeah. So, there's effectively a spectrum of how to achieve verifiability and how to achieve privacy. ZK proof is effectively how to achieve verifiability. You don't get actually privacy from it. Somebody still needs to run the compute and then compute the proof. And so that other third party has all your data, but indeed you get verifiability that this exact model was run in this specific way. The challenge is, yeah, it's like even in the right now, the most performant way, it's probably like 1,000x slower than just computing it itself. It's a lot of overhead. It's a lot of overhead, especially with most of the time you're asking ChatGPT to do like dumb things. Everybody does. And so it's kind of only makes sense for like really critical stuff. The other side is there's this thing called kind of fully homomorphic encryption, right? This is where you're actually doing compute where it is private for every participant. It is verifiable, but again, the overhead is massive, right? something also on the order of 1,000 to 10,000x. But that is like a, you know, kind of fully private and verifiable. What we're doing with secure enclaves is finding, I would say, a pragmatic middle ground where indeed we trusting the hardware providers and we can talk about like, what does that entail? But the benefit is the maximum overhead we've seen in product, kind of in our testing and production is 5%. And so it's usually like from 1 to 5% overhead over running without this mode, but you get privacy, you get kind of this ability for per— for network to permission to join and you get verifiability as well out of it. So you don't get, why is it dumber on, on the whatever at night, at, at the daytime. So you'll not get that on kind of our decentralized cloud. So 1 to 5%. The only other challenge right now, it only works on the level of a single machine. So 8 GPUs. So we can fit anything that fits into 8 GPUs, which is effectively all the open source models right now fit there. It's like over a trillion parameters fits there, no problem. But indeed you cannot do this across multiple machines yet. This is going to be in the works right now. Now, on a kind of software and hardware security side, indeed, there is a potential for a hardware attack. So right now, these chips are 2 nanometers, right? That is very small. And so we're getting to a level where it's effectively baked into atoms of the chip. And so if you want to change something or address it, you effectively, like, I mean, I'm not a hardware, like, at that level, so I cannot actually approximate, but yeah, the cost of attack is extremely high. And the benefit is you're going to get some maybe random requests from somebody asking, you know, which cat to pick, right? So like, because you, like, you will not be able to target specific use cases in our system. Even if something happens. There's indeed the other challenge, which is NVIDIA kind of, because NVIDIA effectively certifying it. So they, their key management and their process needs to be indeed correct. And so this is something that, you know, we would love to kind of help secure. I mean, obviously they're already using kind of top-line security, but that's where, you know, formal verification and using blockchain as a security mechanism can definitely help.

Okay, cool. So let's build our way up the stack. So we want to now create user-owned AI. And I've seen you have a plan to train a more than 1 trillion parameter model. Estimated cost of that, $160 million.

Well, it's dropping down since we talked about it.

Okay, that's a good thing. Cut it in half.

Yeah.

V03 just got cut in half. So we'll apply a half factor on this too. I guess what I understand there is Dario just kind of talked about this where he said, yeah, we're burning a lot of money, and in a way we're burning more money every generation of scale-up. But in another way, if you look at each model as its own venture, each model is profitable. And so what's weird is we might make however much on— it might cost us $100 million to train Claude 3, and we make a billion, but then it costs us $2 billion to train Claude 4, and we'll make $10 billion, but that's going to cost us $20 billion for the next one. And so that's a weird development cycle, to say the least. But it seems like you are basically kind of taking that reality to its natural conclusion and actually planning to structure subventures for each of the models, right?

Exactly.

So tell me about that sort of economic schematic there. And then also, I understand that you're applying this secure computing to the training process where people can contribute training data in a privacy-preserving way. But that's a really interesting challenge, I would imagine, too, right? Because who's providing what data? The data mix. If you talk to people at the frontier companies, they're like, yeah, it's like baking a cake. We're constantly experimenting with different mixes. A little more of this and this other behaviour gets degraded and whatever. So how can you have a privacy-preserving, even understanding on some naive level that you can do it from a computation standpoint, from a getting a model to perform standpoint, how can you have privacy-preserving mechanisms on the training data side? And have any sense of what the hell you're going to get out when the model comes out of the GPU oven, so to speak?

Yeah, so all great questions. And indeed, it was actually funny that Dario posted it and I'm like, yeah, exactly the model. So I think maybe just to give you a little bit of a roadmap, building out those decentralized computing starting with inference. So now it effectively already gives an ability for people to consume it, get confidentiality, get verifiability. And then next step, it offers people who are building models, including potentially the existing companies that build foundational models, they can upload them, they encrypt, nobody can see them, they don't get leaked. But now they can be used across this decentralized compute. Everybody can verify and know that they're using this exact model that this company has posted, they effectively on-chain said hash company tweeted about it, et cetera. You can verify kind of the, indeed you use, let's say this entropic Claude and it doesn't degrade in the evenings, right? So that's kind of steps, first steps then. Okay, well given we want to kind of coordinate this research and development of model and importantly like it's not, We're not going to build one model, right? That's not the point. The point is to build a process that creates models that are state-of-the-art and user-owned and everybody can inspect how they're done. And so the important part is how any frontier lab is working is actually a number of benchmarks that effectively goes through the whole flywheel of building model, right? Everybody building different parts of it evaluate how their change and innovation benefits the scaling laws. You test the different sizes and then kind of all those ideas gets kind of all accumulated and then you do bigger runs. So replicating some of that as well, where we effectively now, again, using this compute network, we can say, hey, actually anybody can come in and build a new benchmark. So you can say, hey, I'm going to build a benchmark that on model predicting future or answering questions about astronomy or ocean or whatever, different things that people care about, including you can actually have enterprises who say like, hey, I actually care about how this model does on this deepfake detection. And I will not actually— I'll upload my data, but I'll keep it private. So nobody can have access to data, but everybody can run evaluation over it. So you can now offer effectively both benchmark creation and evaluation service where you verifiably can check that this model, the closed-source model that runs on disclosed data, on private data, and gets this result. So again, effectively now creating this marketplace of model builders and benchmarks which can be representing companies or specific use cases and finding what's the best model and the best ideas are. So that already on itself is a really interesting product because again, right now you cannot do that. You cannot actually go and benchmark some intermediate model from, let's say, OpenAI. They'll just not give you the weights. And if you don't want to upload your data to them, you cannot even benchmark their current model. But this also opens up an environment where anybody, like, you're sitting somewhere in whatever, you want to train a model that specifically targets whatever ocean benchmark. Maybe you have special data for this. You can also do this, upload it, and get rewarded when people are using it through the previous step. Now, the next step is like, well, actually, I have a lot of interesting ocean data. Or maybe I'm in a university and I know how to train the models, but I don't have compute. And so that's the next step. Okay, well, we're actually going to offer you a way to do first fine-tuning, right? So like you can take existing base model, fine-tune it, and again, you can fine-tune it even on private data. So I'm as a maybe like institute that collects a lot of ocean data, I have the data, but I don't have the know-how how to fine-tune this properly. So I can partner up with someone who can then fine-tune on my data without seeing it, right? They can effectively deploy like, hey, this job into the confidential computing. The result, the resulting model then splits the revenue between the kind of the developer and the data provider. So you can have all of those different combinations on top. Fine-tuning is, I would say, easier to do when you may not have exact access to the data. The training is, yeah, where if we're talking about full fundamental training from scratch, that's where things get really complicated. And so that's our final step where we're actually putting all of those pieces together. And there indeed, you need, all of those pieces need to really work well where you effectively can iterate on different pieces and sizes and then really combine it all into a final model. I would say to your point about private data, the things there, there's effectively a few things that will need to happen. One is that there are going to be a whole separate process of data filtering, curriculum construction, et cetera, that again, you can actually benchmark it. So I can build a curriculum builder, pick different articles and different pieces of data, construct the training set out of it. And so you can actually evaluate how good that system is by training a few sizes of the models. And so people can actually compete on whose curriculum builder is better, right? And so then when you pick the best curriculum builder, then you run a larger run with more parameters and more data. So I would say it's going to be a kind of a very iterative process, right? And it's the same way it is right now inside Frontier Labs because indeed we're kind of experimenting with all those things as we go. Now with specifically private user data, it's a bit more sensitive and complicated. But again, the biggest benefit we can do is As you contribute your data, first of all, you can get a reward from the model outcome. But you can also, we can have effectively a filtering procedure because one of the things is there's going to be a lot of PII information, your phone numbers, SSN, et cetera, that you don't want in a model. And it's also not actually helpful for the model for the most part. But this filtering procedure will be open source and everybody can effectively Not everybody will inspect, but some developers will inspect and say, yes, this is actually a valid way to filter data. So there's a few different components again that we can do to make sure that both data is cleaner and then how it gets filtered and processed into the curriculum of the training. And probably a bunch of your personal chats are probably not that actually useful for training these models. There's other types of data that's more useful. for those things. And we see in general, we're moving to more synthetic data anyway. So there's going to be a lot more of that. And importantly, we still have the crowdsourcing and data labeling that's run through this decentralized network as well. So we actually have both Near Crowd, which is kind of original project we had for crowdsourcing. Crowdsourcing, I mean, like Scale AI style as well. We have Public AI, which is another How much overhead is associated with all that?

I mean, you mentioned relatively low at the hardware level for the confidential computing, but how much is there in terms of just the software setup?

Right.

I mean, it's not super easy to set up a training cluster. It's not super easy to manage data pipelines. It's not super easy to I mean, you have a marketplace, but even going out and hiring a bunch of people to create data, that's not super easy, even in sort of base case. Interested in how much overhead there is in doing it the privacy-preserving way. And then also at the sort of model, like, who makes these decisions in terms of what's better?

Right.

Because I think we have benchmarks galore, and yet it seems like— They all— I mean, there's been a lot of discourse about this in the last couple of weeks too. People swear by evals, people say evals are useless. Certain famous projects are run entirely on vibes. It would seem like the one thing that you'd have a really hard time putting into this frame would be like, who do we actually trust to be the vibe checker on this? And how do we make sure that In addition to maxing out on these various benchmarks, the thing actually is nice to talk to, right? Which is, people do value that a lot. So, yeah, I guess overhead in terms of all the setup, the config, just moving things around, coordination, I guess, broadly. And then is there a governance mechanism? I know you have governance projects as well that are pretty interesting unto themselves. Are you bringing those to bear on who decides what version of the model is actually good or what training recipe gets scaled up? Or is there some tastemaker that still is involved somewhere?

Those are all the good questions. And I think the high-level answer is we're effectively creating a marketplace of all of these decisions. So instead of trying to prescribe, like, okay, Ilya will be the tastemaker and different and decide you get the compute and you don't. Instead we say, hey, we're creating a marketplace where anyone can do this and probably Ilya will do it as one of the participants, but there can be anybody else doing it as well. And you kind of have the rules and everything known and you have visibility and trace and you can also build on top of each other instead of right now you're effectively competing. And so yeah, we kind of touched on it, but yeah, Let's say I'm going to say, hey, I'm going to train this 1.4 trillion model. I'm going to issue a token for it. So I'm going to create a new effectively value capture system for this specific model. I can now distribute it for the compute. I can distribute it for data. I can distribute it for research and work. In this case, I started it, so I probably will be the test maker and picking and selecting which pieces I want to put together. We train it. If it gets used and then we distill it, fine-tune, et cetera, the revenue gets distributed to the token holders. Now in parallel, somebody is like, oh, actually I'm going to train a 70B model in this way, et cetera. They're going to use exactly the same framework, exactly the same system, do this. And again, because 70B is smaller, cheaper to use, but then It also runs faster. So like it's going to get used in different use cases. So you can, you can like, you're really using this more of like as a platform where different people can kind of put together these pieces really easily. And you don't need to have, you know, you don't need to raise like whatever, $10 billion to really kind of do this. You can actually raise like smaller and specifically for compute data, et cetera. And you already have a lot of building blocks. As well on this. So that's kind of the idea. It's like really creating more of an environment where we can open source, but then also we kind of, and reuse these things while kind of have this economic flywheel, right? Where things are like generating revenue by usage. And importantly, kind of learn from what works and have this inspectability of what actually went into it, right? Which, I mean, in some cases people say, hey, no, actually it's all private. I will not tell you what actually went in. But then people know that this is private and they can decide if they want to use it or not. But also like, here's the verifiable benchmarks. Again, right now when somebody posts benchmarks, you have no idea how they got these numbers. They usually don't even post the prompts they use for the benchmark. So You can have a 2-page prompt that answers half of the questions and get way better answers from that. So this here, you have the full traceability of the whole flow. You know what went in. Maybe it's private, but at least 100 million users contributed their private data and that's what it trained on. And then it gets this result on benchmark and then You can actually, you know, vibe, vibe, you can contribute your vibe benchmark as a benchmark as well. And so like on, you know, on Ilya's vibe benchmark, it gets this much as well, for example. So like you can kind of, and on whatever, on this company's benchmark, right? Like Bank of America, you know, their test, it gets this much. So you can have all of this different information and decide based on that. So it's really about like kind of creating transparency and marketplace for this while preserving privacy for the end user and kind of companies.

And all of this traceability, verifiability, to rewind for a minute, all rests on the fact that somebody has locked up their NEAR coins with the assertion that they are Putting forward valid transactions, which in this case would be more like log statements, basically, right? That—

It's, I mean, it's a bunch of stuff. Yeah. Like, because it's payments, it's logs, it's kind of traces of like chainings, etc.

How big does this database obviously get? Like pretty huge, right? So where does the database live long term? Like, it becomes too big for a node. If I want to just connect, right? I can't download the whole history of like everything that's ever happened. It becomes too much, I assume.

Yeah.

So Near, I mean, one of the kind of underlying things is that Near is sharded, meaning it's like every single node contains part of the data and part of the process, part of the transactions and part of the execution. And it can continue expanding number of shards. And there's a Like the, that's why we actually need more validators to join because the more data, more shards there is, you need more nodes in the network to actually participate.

So the—

Very similar how Google, you know, like databases work or whatever, Meta, you know, they don't run out of capacity. It's like, no, no, no more users here, please. It's like, no, no, no. We just keep adding more computers to run stuff in parallel.

Yeah. Okay, so this has been fascinating, and we've built up this sort of— from the premise that we want user-owned AI, this is kind of all the layers that we have to put in place to finally be able to get to the point where you could have a community-based process of all this contribution and kick out a model and even get paid back with As you mentioned, though, you could also just have somebody train their own model and kind of bring it into the system. So regardless of where the model that people want to use comes from, let's change maybe gears toward what does AI need from crypto and what does crypto need from AI? My sort of very high-level styling, and then I want you to go 10 minutes on this if you want to, is like, AI can put the smart in smart contract. I've always kind of thought like, geez, these smart contracts aren't really that smart, but if there was an AI in there, maybe it could be smart. What we really need, I think, is like smart dispute resolution in a lot of cases and affordable dispute resolution, because we have, at least in some places, we have courts, but it's like not convenient to access them by any means. Other places there are no courts. So we need smart, affordable dispute resolution that AI seems like it could potentially provide into the context of these smart contracts. And then on the other hand, AI needs really rock-solid guardrails because we obviously have many— the AIs themselves can be tricked and can go off the rails and make mistakes and even scheme against their human users in some cases, as we're Increasingly finding out. So we need very reliable checkpoints, basically, in workflows to know that you can go explore, you can do whatever, but if you're going to do this, you got to have a certain— almost think of it like the Jurassic Park thing, where they engineered the dinosaurs to not be able to survive without the special food from the from the park. Of course, we know how that turned out. But not to generalise from fictional evidence, but we do want the AIs to sort of be dependent on a really rock-solid validation that this is okay for you to do. So that's my high-level premise. Give me the fleshed-out version.

Yeah, so, I mean, both of the sides are definitely very interesting and And kind of both directions we're working. And to be clear, like when I say we, we have like a large ecosystem of different projects and kind of, as I said, it's a protocol that combines a lot of different projects and contributors who are actually working towards this. So one side on the dispute resolutions and putting intelligent smart into the smart contract. So we effectively have 2 kind of parallel threads there. One is what we call Near Intent. So Near Intent is this idea that as we move into the AI world, well, actually you or your AI will just express an intent that something wants to be achieved or done, right? This can be as simple like, hey, I want some pizza, tonight for dinner, and as complex as like, hey, I want to construct a new building in this location, it needs to be like this, et cetera, et cetera. So high qualifiers. Now that requires a number of steps, right? It needs to find somebody who actually will do it, right? So you need a whole discovery process. So how do you discover other AIs or non-AIs, right? Like it can be businesses as well. But we're assuming everyone will have some AI-like interface. How you discover the AI that will actually do this intent, how do you actually come into a commercial agreement with them? And this commercial agreement can be like, hey, I need to rent this car to get to airport for Waymo. Or it can be like, hey, we need the 6-month project to construct this thing. And then if something goes wrong, how do you actually deal with it, right? So dispute resolution indeed. So that's what we're building as kind of this Intents protocol. And so this is the idea of, you know, it starts with people, organizations, and AIs effectively collaborating and kind of doing these things. And so that's a fundamental protocol. It includes indeed AI-based dispute resolution. So you can actually run AI over If something went wrong, like the taxi didn't actually drive you where you were going or didn't arrive at all, how to deal with that? AI can look after this, cheaply analyze the situation and decide to do something. Now, if parties are still not satisfied with that resolution, then you go to court. So effectively reduce the cost of disputes, make it way cheaper. It can use a lot more information as well right away because you have your on-chain information, all your previous actions. It can run over your private data, over both sides' private data very effectively, right? Without revealing this private data to anyone. It's effectively like if the discovery process, like you can run AI on discovery of both sides without this costing millions of dollars. And so you can effectively minimize the cost of this process very low. And then again, if it's not kind of acceptable, then you go to court. But again, a lot of work already done. So this is all can be reused. So that's kind of one piece. And we kind of the go-to-market for that, we actually started in more crypto side. So we like trading and other things and we're expanding it into e-commerce and other use cases, kind of adding more and more services into this. The other side is actually this autonomous agents. And so right now, smart contracts idea was it's effectively autonomous being, right? It exists independently of any person or entity. It has money. The problem was it was just a piece of code and usually very small piece of code. So it was very dumb, even though it was called smart contract. What we've done is because we have this decentralized kind of like compute with AI, we actually combine these 2 things. So in the decentralized compute, we are running the AI brain and then it has the smart contract to actually execute actions and execute these intents so it actually can go and do things. So we have, as an example, one agent that was given $10,000 launched. We cannot stop it. We cannot take money back. We cannot do anything with it. And it's effectively trading it based on sentiment on Twitter. And so it made whatever, $4,000 doing that from the $10,000 it was given. So this is just an example, but you can imagine a future of businesses where you effectively have businesses like this which are operated by AI. You know exactly what it's going to do. Your bylaws are the prompt of the con— there's no person that does random things. Shareholders can effectively give it feedback. So you can apply all this kind of training data to improve it or decide to upgrade the model and things like that. So that's kind of the framework of what we call autonomous agents. Intents is how they're going to interact with the commerce, themselves, and with other AI agents. And AITP is really the protocol to communicate, to do the communication, to do intents. And then the other side of this is, yeah, how do we govern this? This is where indeed we have a whole governance protocols. But even before that, I mentioned the form of verification. I actually think one of the pieces of this puzzle will be this where let's say you're using some AI service. Again, you want it to guarantee to you before it even executes on your data that it'll not escape and go kill a bunch of people. So you kind of like, well, Effectively, we need to create this environment in which we can verify what AI runs. But I don't think we can— my approach is I think it's really hard to effectively be a thought police even for AI, but it's way easier to ensure that they don't do anything outside of the sandbox that violates whatever the rules you apply when you call it. And so that's kind of, again, using this verifiable compute, we can actually guarantee that as you upload this model, you can attach effectively a set of things that you don't want this model to do. And it can stop it from doing that kind of by running on the output and verifying if it's correct or not. So that's kind of the idea is to leverage some of that. And then you still need obviously governance and kind of ways for people to come together and actually decide what it is that we don't want, right? Like, for example, probably everybody agrees we don't want bioweapons, but maybe not everybody agrees on some other things, right? And so kind of there needs to be some process through this. There, I would say we have a bunch of experiments, including actually using AI in the governance. So you have AI senators effectively, like you, that people can vote for, and then they go and actually— make decisions. And so you kind of, because getting people to vote is hard and getting people to go and deep research a bunch of stuff to decide how to vote is even harder. And so you can actually have the delegates to be AIs as well. And you kind of just give them, you vibe with them on what you think is your opinion and they kind of construct from that their platform. But you remove this principal-agent problem, which you have with other people. when you vote them, they still have their own agenda, right? And so like they may not do what you ask them to do. So that's kind of like all the pieces where we're like, we're combining all the blockchain and AI side like from different directions and really creating protocols and kind of verifiable compute and privacy and really just, you can use it in all those different ways.

One thing that definitely kind of Makes me nervous is the idea of these autonomous agents that we can't take back. On the one hand, you sort of want to have reliable follow-through on what they're supposed to do. You don't want other people to come in and tamper with them. So I understand why there's some allure to it. But then I'm also like, geez, we have a hard enough time telling what one AI is going to do. What happens when we unleash a billion agents on the world and they're all autonomous and we can't take them back? and we've, by the way, delegated governance to AI senators. Some might wonder, are we not creating the recipe here for the AIs really to just take over and run everything? And so, yeah, what is your positive vision for the future, and how do we make sure we don't take a premature off-ramp on the road to that vision?

Yeah, so I think there's going to be a combination of things where AI will become more of an economic force, right? And be that kind of conduit and medium. And then on our side, we're also going to be kind of becoming more of governance and curators of that on our side. And so one of the kind of, indeed, we're like putting AI into governance, but also we're kind of putting governance into AI where, I mean, formalization is one part, but the other part is also If you have this autonomous agent, well, actually, if it wants to do this intent in physical world, it actually needs to be in some jurisdiction. This intent is effectively a legal contract. It needs some backing with actual jurisdiction, and that jurisdiction has an enforcement over this agent. And so we actually have an infrastructure for jurisdiction, for actual court to enforce stuff on AI agent as well. It's all visible. It's all on chain. kind of transparent, but like you're effectively combining this, you know, what we have traditionally with courts and jurisdictions and legal frameworks and with AIs who can actually interpret all that and, you know, reason over it and make sense of it as well as these protocols to kind of facilitate it. So like kind of the way I see this is all of these things are gonna keep fusing together and like really becoming this kind of new conduit of how how our society operates. But indeed, like, I think a lot of the, like, there are gonna be a lot of exciting innovation and also, you know, we, our role as well will kind of keep changing because like we're gonna be more of like participants in this network as well. But, you know, through the AI as a lens, right? How right now we're, you know, through the lens of computing, we're already interfacing each other a lot more than— you know, the physical world many times. So I think like that is going to continue also increasing.

It's a fascinating set of layers of foundational technology that you've built, and the possibilities for it are really pretty dizzying. So I want to already propose a part 2 for this podcast. Maybe we can get back together. Before too long. And in the meantime, I don't know if you've ever read the book Liquid Rain, but it's my favourite kind of crypto-AI intersection. And I'm not like super well read in science fiction, but it's a funny, but I think increasingly prescient looking and positive view of what that future could look like. So I definitely recommend that. I know you got to go. Put me on the calendar somewhere for part 2. Any closing thoughts that you want to leave people with today?

No, I think, I mean, again, this is a movement we want people to join and kind of contribute across the board. So if you're a developer, there's a ton of stuff to contribute and build. If you are planning to kind of— what does it look like in your life as AI? I think this is really an opportunity to engage in something that's Really yours and you can actually own part of it. And again, contribute in many ways from kind of creators and be receiving it to tastemaking to many other aspects.

Sounds great. I'm Nikolay Sukhin, founder of NEAR. Thank you for being part of the Cognitive Revolution.

Thanks for having me.

If you're finding value in the show, we'd appreciate it if you'd take a moment to share it with friends, Post online, write a review on Apple Podcasts or Spotify, or just leave us a comment on YouTube. Of course, we always welcome your feedback, guest and topic suggestions, and sponsorship inquiries, either via our website, cognitiverevolution.ai, or by DMing me on your favorite social network. The Cognitive Revolution is part of the Turpentine Network, a network of podcasts where experts talk technology, business, economics, geopolitics, culture, and more. Which is now a part of a16z. We're produced by AI Podcasting. If you're looking for podcast production help for everything from the moment you stop recording to the moment your audience starts listening, check them out and see my endorsement at aipodcast.ing. And finally, I encourage you to take a moment to check out our new and improved show notes, which were created automatically by Notion's AI Meeting Notes. AI Meeting Notes captures every detail and breaks down complex concepts so no idea gets lost. And because AI Meeting Notes lives right in Notion, everything you capture, whether that's meetings, podcasts, interviews, or conversations, lives exactly where you plan, build, and get things done. No switching, no slowdown. Check out Notion's AI Meeting Notes if you want perfect notes that write themselves. And head to the link in our show notes to try Notion's AI Meeting Notes free for 30 days.