Hype or Bust? How to dominate ai with web3 - Jacob Robert Steeves, co-founder of Bittensor

But so then just living in Peru, because after that I got. I couldn't leave the country and so I got a dog and I got a wife and then you just don't leave. You know, I locked myself in. Yeah. But it turned out to be a really good idea, actually.

Good afternoon, everyone. Welcome to Proof of Talk in Paris. My name is Gareth Jenkinson. I'm the head of multimedia here at Cointelegraph and it's a great pleasure to be live streaming from the event. Next to me, I've got Jacob Steves, who's the co founder of Bittensor. Before we get started with our conversation, I just want to invite you, the viewer out there, to drop your questions in the comment section below, whether you're watching us on x or on YouTube. Any questions you want to know about Bittensor, about Tao, now is your chance to ask Jacob. So please throw your questions in the comments and we look forward to posing some of those questions. Towards the end of this 20 minute conversation, we're going to have to keep it short and sharp because I've got a panel at 2 o' clock sharp. So we're going to have a quick 20 minute conversation about Bittensor and a quick overview of Proof of Talk. Jacob, it's great to meet you. It's great to chat to you in person. Let's start off with a philosophical question. Why did you build Bittensor? What does it solve? What does it prove? And yeah, why does the world need this technology? Especially when we're seeing AI tools develop as quickly as they are.

I think a good way to get into that is with a story. When I was at Google, I was working in AI at Google and I ended up working for a team that is called Google Brain. It's like the highest level, the pinnacle of. It's like the brain of Google. And I was asked to work on a project and the goal of the project was to generate images that would go into the watermarks. So like behind not able to be seen of advertisements to manipulate people's likelihood to click on the ad. And so it's like massive subliminal messaging gray area. Right? It's like, is that legal? Is that not legal? Like, I don't think that our legal structures are even capable of really measuring that. And I kind of think that that was wrong.

Yeah.

And you know, these types of problems stem from really the structure that you have a technology inside. So Google as a corporation is a structure where there are clear incentives for how people move and how those people control commodities and how they Control, compute and how they, they follow up their careers and how they make money. And it makes them ultimately it makes corporations become, you know, at first the heroes and then the villains because they have a fiduciary responsibility and they need to maximize as much money they can make per quarter. And so you get these types of, I'm not saying nefarious, but like questionable ethical decisions that happen in corporations just as a consequence of like, hey, look bro, I'm part of a system and this is how I'm going to feed my kids. It also points to. And so that's why I think it's important that we actually build in the open, right? So blockchain technology is about decentralization. But it's decentralization. What does that actually entail? It means that a lot of people can own some technology. It's sort of like democratically owned. It can, it's permissionless in the sense you can come to own it. Like it's not something that's like above and beyond you. It's not something that's in a walled garden. You can access, control, run, participate in the system. But also the centralist systems have to be transparent. Like the clear structure of that incentive is well understood. Like a legal structure, like out in the open you can investigate the structure of law in a government and see whether or not that is going to have a good outcome for humanity. Not the case with technology. So I think that it's essential for us to build this incredibly important technology out in the open so people can see and understand where the incentives are, so they can participate, so they can own. You know, you can get kind of flowery with your language and say that like we're trying to reach some sort of symbiosis with the machine, right? Where hey, like I, I actually own this thing, I own this technology. Viscerally the machines work for me because I hold this token, right? And I can see how that is the case. I think that that's quite beautiful because it allows people to really viscerally own something that could potentially have a lot of control over them. As you, as my example was showing, like you AIs can manipulate our opinions quite easily and we're building this secondary organism. That's my vision, but it's not the vision that you see coming out of Silicon Valley where their vision is more safety angle, like Claude. You listen to the CEOs of Claude and OpenAI. They want to control the technology as much as they can, right? They're not giving it away, they're Aggregating the control. And they're doing it because it's for your best interest, you know. But is it?

Probably not.

Probably not. But they're going to give something back. And like, if you look at the language that Sam Altman is using about how he wants to fix the potential issues and the social political issues that are arising out of technology that is shaking people's jobs, he wants to institute something like a universal basic income. Yeah. Now we're in the crypto space.

Right.

We're missionaries here. We understand money.

Yes.

And one of the things that we understand is fiat currency and how fiat currency is not a fair trade. Yeah. If I'm playing your game and you can just mint currency, you have all the control and fiat currency isn't that valuable. And that's what we've seen over the last 15 years of Bitcoin's life, is we've seen people's. People run towards truth, run towards a system which cannot be manipulated in the favor of the people that hold the keys. And so if we're talking about a technology of this much importance, we need to make sure that it's not manipulatable for the interest of a small group of people. Now, people that are on the fear side, they want to sell you a vision of you need to watch out because, oh no, the AIs are going to launch the nukes. Oh no, the AIs are going to take over the world and it's going to be an iRobot outcome. I don't think that's going to happen at all. Or at least if it does happen, we're talking a couple hundred years and we're certainly going to go through something before that, that that period arises. And I think that it's something. Have you ever seen the movie Elysium? Yeah. I think that, that we're moving potentially into a world where we have the, the have nots and the haves and they basically have access to this technology. They have the first class access to the best machine learning models as they come out. They control it. At the end of the day, all of the value is accruing to them. And they're giving you basically peanuts for a UBI with an orb. Right. I think that there's a lot of people that don't want that, and me especially, I think that it is essential that we try to use the technology, the most liberating and freedom technology that we've ever invented, which is Bitcoin, and we merge it with the machine. Right. We have affinity with the machines we have unification, we have a form of symbiosis and we're not afraid of them, but actually we viscerally can control them and work with them.

Yeah, look, you mentioned Bitcoin being part of that puzzle. Maybe you can just give us a very brief explanation of what Bitcoin Bittensor is, how it works, why it has a native token like Tao, and why you didn't look to integrate something like Bitcoin for payment. For payments.

Well, I actually tried to build Bittensor off Bitcoin. I was going to try to use the Lightning network originally. But if you look at what Bitcoin is, basically it's a supercomputer that uses hashes to secure the network and secure the state. It's a commodity system that uses that commodity to, to crystallize its own token. And Bitcoin is, if you go a little bit more abstract into what it is, it's a market, it's a perfect market where anybody around the world can plug themselves in and mine Bitcoin to secure the network. So we basically said, what we said is that hey, let's try to think about what that exactly is. Let's take that concept of the digital currency market and let's abstract it. Let's get down to the really nitty gritty of what that means to have a bunch of nodes that do some work and that are validated. So we took that idea and we basically built that's a proof of work, proof of stake mechanism that allows us to co locate and organize in really, really, really, really, really efficient and hyper competitive systems to extract digital commodities that would be valuable for the production of AI. We thought, hey look, we know that there's this grand vision for decentralizing AI, but if your AI sucks, it doesn't matter if it's decentralized. So we knew that part of the puzzle here was actually building something that could compete with centralized organizations in terms of just scale, raw scale and performance. So the reason we needed the token initially was because we needed to bootstrap the system. And we tried to use lightning, but lightning doesn't catch any fire and there's no market that can be created. So yes, of course we're taking advantage of the bootstrapping potential of using a digital currency in order to garner support and garner, garner interest in the currency. And then we turn that whole thing into a commodity that's similar to Bitcoin. It has a specific value and that value is the value of artificial intelligence.

And it has a finite supply right.

It has a finite supply. So we knew, like, hey, let's use. Let's use bitcoin tokenomics because it's most understood in this industry. And I think that's the most fair.

Was it a bit disappointing that you couldn't do it on lightning? I mean, if you're, you know, like you said to me off air, you're a bit of a bitcoin maxi at heart. Not to put you on the spot.

Yeah, I mean, like, I'm a bitcoin maxi because. And I've said this before, I actually tweeted it. It's like, if. If it came down to I had to destroy Bittensor in my way, it's not very possible. But if I had, if I could, you know, to save Bitcoin, I would. I would do that.

Yeah.

Because I. You. What we're doing, that can't be achieved without the shield to the industry that is bitcoin. I think that, like, the initial, like the white paper. White paper is so seminal, it's almost religious. And we exist as like a continent on the inside of the world generated by bitcoin rather than some sort of competitor. And, like, a lot of people that are building cryptocurrencies are trying to go after bitcoin. We're going to try to be better than bitcoin. It's like, no, no, no, no. Like, bitcoin is your friend. Bitcoin is. Bitcoin is the United States. And you're a state inside of the bitcoin universe. So I don't think that it needs to be antagonistic in any way. People obviously will see it. And I just want to say, like, when I first got into crypto in, like, early, early days, the toxic maximalist was something that I really, really liked because there was so much bullshit. There was so much bullshit that you need. It's like a defense mechanism for the industry to help protect people against actual scams. Right. So, you know, we built what we've built. We don't think it's antagonistic to bitcoin. We think that. We think that we actually really are allies with that technology and want to sit there in its ecosystem.

They're completely different things as well. Right. Yours is a decentralized machine learning protocol, essentially, and bitcoin is a monetary protocol protocol. I've got a question from the live stream comments. Please keep them coming through. This is your chance to ask Jacob anything you want to ask about Bittensor and Tao. The first one is pretty easy. What coins will be used in the future to buy groceries. Like realistically, is it going to be stable coins? Are we going to be using bitcoin on lightning? How do you see that playing out.

In the future to buy groceries? That's not the vision I really hold for Crypto. One of the best expressions I've heard about bitcoin is like using bitcoin to buy a coffee is like flying a fighter jet to the corner store. Yeah, you're talking about like geopolitical technology. It doesn't need to be used to buy a coffee. Like there will clearly be looser forms of currency that are used that are easier to on telegram, maybe it's ton, who knows? Or probably will remain national currencies in these countries. And that will be a way that, you know, I think that we're actually teaching governments what money is and making them more evolved in the way that they run their local currencies. I think that bitcoin will be the store of value and that's all it needs to be. It doesn't need to be buy a coffee with it. So which token? And definitely not Tao. Definitely don't buy coffee with Tao. It doesn't make any sense.

Yeah, for sure. I mean look, there'd be a lot of bitcoin payments advocates out there that would be like, you're wrong. It's also for payments. And I mean like that's why we have lightning. And that's, that's a discussion for another day. I want to go back to Bittensor and just talk about machine learning, AI, how, how does something like bit Tensor and a decentralized protocol, a decentralized approach to this technology compete and overTake the chat GPTs, the deep seeks the, the GRO. You know, how do you see that playing out? Do you, do you eventually see your technology being the choice for people to use or do they stay with a centralized option?

So recently Musk at xai, they went out and they just bought this massive data center for like a billion dollars and they started pumping out models like Grok and they're reasonably competitive, right? Yeah. The reason they were able to do that is because they had so much money and they were able to garner this, this data center. Right. I think that, I think that in order to produce this like pinnacle element of intelligence, you need to focus on the like sub commodities that make it a reality. So we set out to really break apart that problem. You can't do decentralized training unless the compute that you're running on is decentralized. Or that's actually not true. Sorry, I would actually say that decentralized compute is best a product for decentralized training. And so Bittensor is a set or of interlinked hyper digital commodity systems that interlink and allow us to actually scale horizontally at all those different layers of commodities that you need to get to these higher level order intelligences. So we're at the point where we can create a data center with incentivization in house that can get H1 hundreds or H2 hundreds at a cheaper price than these corporations. So we want to scale that then we want to scale the technology that can use that compute because it's a different type of compute, right? It's actually geographically decentralized compute. So you need to have a market that can get all of those computers together and get them for the cheapest price. And then you need to put on top of that a decentralized training system that can use that compute and that requires decentralized storage and bandwidth. So that's another thing in Bittensor that we're building. So we're really building up the stack and we're getting to this point now where we're really making strides. Like we're competing with these companies in certain verticals and you know, we haven't yet trained the best in class models, but we're scaling our inferences for faster than Google. We're beating Google Automl on gradients, on subnet 56. Right. Our storage pipelines are like Filecoin. They scale infinitely and we can get cheaper prices per gigabyte. So all of that kind of comes together and you can see just by analogy. So how did XAI compete? Well, they just created a data center and they put some machine learning engineers on the top of it and then you can compete in this industry. We're focusing on scraping out all of that inefficiency at each layer of the distributed infrastructure so that we can now focus on these higher order things. And so I actually talked on stage how. But I'm launching the subnet that's going to focus on that. I'm going to focus on doing these very the higher order machine learning problems like reinforcement learning. And I think that's really where like we're unstoppable, right? So we're proving this point that we can scale bigger than you and at a certain point they're going to have to look at us and go, oh shit. Like we're not the teams with the most amount of compute anymore. And so how do we compete? Well, you have to build Bittensor. Basically.

My other question to that then is why does someone like Elon Musk not just go straight into it as well and realize what you're doing is far more valuable and far more powerful than a centralized approach to doing it?

Why doesn't he do that?

Yeah, why don't they just take all the capital that they have and deploy a load of subnets on Bittensor?

I'd love it if they did. They should. But I mean, like, there's a lot of stigma towards the crypto industry in general. Right. There's been so much bullshit that, you know, pump fun, you know, I'm sorry. Like, our industry has gone through. I think it really has a crisis in terms of our mission and what we really stand for. And, you know, Bittensor is trying at every stage to change that by actually focusing on real technology, by research, releasing research papers, by showcasing real value margins on decentralized technology. And yeah, they're going to start looking. Right. But we're miles ahead. I don't care if Google puts their compute on Bittensor. That's fantastic. I don't care if Google makes it makes a subnet. Fantastic. Why would I consider them to be an enemy? They're not. When we align ourselves with these protocols and by necessity, we have to do it in a transparent and fair way, we are actually reforming these industries to make them better. And so, yeah, I guess Elon could make his own decentralized technology. I think that he would have to hire a lot of my engineers, but maybe he can. But I think it would be better just to join forces and work together right now. Also in the industry, I think we've really stimulated the creation of decentralized artificial intelligence as an industry. We're starting to see a lot of other projects come online and my message to them is let's, let's co mine, let's work together. The only way that we can really achieve this aim is if we do it together. And we don't see ourselves as an industry of antagonists.

Right.

You said, like, how do the bitcoiners think about you launching another token? It's like, well, we're all just helping each other. Let's see it that way.

Yeah, I mean, to me, the two are very, very much different from each other. I've got another question from the live stream from AI today. How are you going to solve the security issue when AI start paying other AIs with Bitcoin? And how will storage be kept under control?

Storage? Oh, like privacy preserving storage?

I think so.

I think, I think that's what it.

Is that, that, that is how the question came to me.

So, you know, this is something that we're actually at the forefront of. We, we have. I just want to say it's like one of the things that Bittensor has that's very cool is we're no longer a single company upgrading our protocol. We built a mechanism so that the people that are upgrading the protocol are actually highly incentivized to do it and are competing with each other to make progress. And so it's because of that, for instance, that we have, as soon as it was possible to do trusted execution environments, we had subnets doing it. And so what a trusted execution environment allows you to do is, is write a program that nobody can backdoor. There's no man in the middle. And so you can run a machine learning model that you don't have to just trust, like, oh, well, I don't think they're. Basically, what I'm saying is that we're going to have privacy that is actually more private than the corporations that are claiming privacy. Because we can actually say that cryptographically. There's a whole Cambrian explosion of mechanisms on Bittensor that are using that all the way from storage to model inferencing. So we're seeing that on shoots, we're seeing that on Target Tarragon, I think that they have targon on Summit 4, has the largest like co located set of GPUs that are doing trusted computing. Fantastic, right? These things are being really commoditized and the technology is working with us to make it possible for us to compete at an application layer and say like, oh, well, you know, it was before that enterprise would only use another corporation because they can trust a corporation. Why would they trust the miners on Bittensor? Which is a really good point. But now we can say, hey, well, why would you trust a corporation? Why don't you just trust cryptography? Don't trust verify, you know. So I think that's really what we're going to see in the next one.

I've got 10 minutes left before my next panel, so I would love to keep going with this conversation. It's been really great to chat to you, Jacob. I think what people like you are doing for the industry is really great. It seems like a very selfless approach and something that's very dedicated to building important technologies. If there was any parting thought that you could give our viewers out there in terms of decentralized AI, why they should explore using something like Bittensor or a number of other deepins out there that have really great sort of approaches to this technology. Why should they be doing that instead of pledging their compute to a centralized server or service provider?

That's a good question. You know, I think that, you know, we started doing this a very long time ago for the right reasons, and we made a point of not owning the entire thing. And I think that that's turned out to be a really important aspect for sure. You know, it sets you apart from everyone else. Bittensor. Yeah, don't come work for Bittensor. Come own Bittensor. Come and be Bittensor. I'm like, the foundation is being pulled out, you know, So I mean, look, look, there's, there's, there's brand and maybe, maybe you don't want to be associated with, with that brand. And that's, that's on you, right? Totally fine. But I would say come and, and actually build and take ownership. A good example would be shoots on Bittensor, which has a subnet that's done very well. John, he showed up like a year ago and started building shoots and his token value on his subnet is like 25% of Tao. That means that he is going to come to own a massive share of this network. And when we launched this, like, this sub. Sub tokens. The sub tokens on Bittensor, we did it. The first sentence was like, how do we give more of this control away while maintaining property rights, but ensuring that new people can come and really just like, take my job and I'm also going to be in there with you. Right. So I just launched a subnet, so I'm down in the trenches and, you know, yeah, don't come competing because I want to beat you.

There we go. Jacob, thanks so much for your time. We're going to wrap it up there. If you want more information about Bittensor, go find them online. You guys are on X. You're on all the social media platforms on their website. We will continue live streaming from Proof of Talk today and tomorrow. So please keep an eye out on Cointelegraph's x account, our YouTube. Thank you so much for watching. Send your questions through if you have any burning questions for our guests and panelists coming up. I won't tell you who we've got coming up later today because there's some really exciting names and tomorrow we also have a great lineup of speakers joining us, so we'll catch you later. Thanks so much for watching and see you soon. It.