Bankless - AI and Web3 | Mohamed Fouda & Qiao Wang of Alliance

Episode Date: May 3, 2023

AI is exploding into every facet of the internet. The convergence of Crypto and AI is inevitable, and we bring on Qiao Wang and Mohamed Fouda to discuss the opportunities and challenges this presents. They explore why Web3 provides an interesting platform for AI, such as payment and execution rails for AI agents, easy access to financial tools, and the ability to commission resources permissionlessly. One example of this is AutoGPT, which uses AI to generate code for on-chain smart contracts. How do we balance caution and optimism? How can we surf this tidal wave of innovation and new frontiers? ------ 🚀 Stake with Swell https://bankless.cc/Swell ------ BANKLESS SPONSOR TOOLS:  ⚖️ ARBITRUM | SCALING ETHEREUM https://bankless.cc/Arbitrum  🐙KRAKEN | MOST-TRUSTED CRYPTO EXCHANGE https://k.xyz/bankless-pod-q2  🧠 AMBIRE | SMART CONTRACT WALLET https://bankless.cc/Ambire  👻 PHANTOM | FRIENDLY MULTICHAIN WALLET https://bankless.cc/phantom-waitlist  🦊METAMASK LEARN | HELPFUL WEB3 RESOURCE https://bankless.cc/MetaMask  ------ Resources: Qiao Wang https://twitter.com/QwQiao  Mohamed Fouda https://twitter.com/mohamedffouda  Alliance https://alliance.xyz/  ----- Not financial or tax advice. This channel is strictly educational and is not investment advice or a solicitation to buy or sell any assets or to make any financial decisions. This video is not tax advice. Talk to your accountant. Do your own research. Disclosure. From time-to-time I may add links in this newsletter to products I use. I may receive commission if you make a purchase through one of these links. Additionally, the Bankless writers hold crypto assets. See our investment disclosures here: https://www.bankless.com/disclosures 

Bankless Nation, we have an exciting topic for you today. You know it was coming. We had to talk about artificial intelligence, that is AI and crypto and their intersection. Are there investable opportunities here for us? We've been talking so much, David, about AI safety, AI alignment, AI Dumer type scenarios that we haven't given this enough attention. And the truth is, regardless of the long-term consequences and, you know, the safety that we need to get right, there are actually some investable opportunities in AI, and there's some intersection between crypto and AI. And so we're going to talk about that today.

You know, what's interesting about this, Ryan, was this was actually, this question is like, what does the intersection of AI and crypto look like? The impetus for our original Aliaser episode, that's why we did that. And then we got massively sidetracked with a, the AI is going to repurpose the atoms. But before they repurpose our atoms, they are going to produce new opportunities for us. So on the way of getting our atoms repurposed or not, we are going to discover what those opportunities are. But before we tease who's coming on the show and the blog post they've write and all the ideas

that they are about to bring to the table, we first want to tell you about Swell because liquid staking is also going to be a thing that is a thing before the AI's repurpose our atoms. And Swell wants you to know that they're staking as a service protocol is up and running. and you can be an early staker with the Swell Dow. There are 0% fees for early stakers with Swell. In addition to being a staking as a service Dow, a brand new staking as a service Dow, there's a bunch of other enhancements to the Swell product

that you can go and check out. There is a link in the show notes. Swell, like I said, brand new staking as a service org on the scene, also playing with brand new staking as a service technologies. And so soon they will be implementing some of the more sci-fi aspects of ether staking, including DVT, among other things, as well as being able to be integrated into defy app.

So there is a link in the show notes to get started staking with swell. Man, it just rolls off the tongue. Yeah, it's really exciting. Stake, earn and deposit. And David, that AP are 4.5% on ETH right now. And I've got to tell you, post withdrawals, I am much more excited to stake my EF. It feels like there's been a decrease in protocol risk here,

post that the recent upgrade. But David, tell us, who are we having on today to talk about this intersection? In fact, is there an intersection between AI and crypto? We shouldn't presume that there is. Of course, there's a lot of mean coins in crypto being created that sort of used co-op the AI brand name. But what is AI and crypto? What do they have to do with each other?

Who do we bring on today? And what's the discussion going to be like? Yeah, I think the discussion of AI and crypto really needs to start at first principles, which is something that you and I are really good at. And something that I definitely saw articulated in this blog post from Kiao and Chow and Mohamed from Alliance Dow. So Alliance style invests in the Web3 space across different verticals and ecosystems. And they wrote a blog post called The Convergence of AI and Web3, which like I said, was exactly the subject matter that we wanted to go after originally. And so they've ideated,

they've thought about some particular use cases, AIs as economic agents, the collaboration between ZK proofs and machine learning and data sets and running machine learning algorithms on all that kind of stuff. AI and gaming, AI and NPCs, AI and Defi. And so while this might be a podcast episode about ideation, this will certainly strike the imagination of the listener as we go and explore because, after all, Bankless is a futurist technologies podcast. So we're going to explore the frontier of future tech.

Yeah, I'm super excited about this conversation, particularly. the intersection between crypto, AI, and identity. I think we really need a way to figure out who the humans are and who the robots are, because increasingly, it's a little bit hard to tell, David. Guys, we're going to get right to the episode about AI and crypto. But before we do, we want to thank the sponsors that made this possible, including our recommended exchange for 2023, Cracken, go set up an account, check them out.

Cracken Pro has easily become the best crypto trading platform in the industry. The place I use to check the charts and the crypto prices, even when I'm not looking to place a trade. On Cracken Pro, you'll have access to advanced charting tools, real-time market data, and lightning-fast trade execution, all inside their spiffy new modular interface. Cracken's new customizable modular layout lets you tailor your trading experience to suit your needs. Pick and choose your favorite modules and place them anywhere you want in your screen. With Cracken Pro, you have that power. Whether you are a seasoned pro or just starting out, join thousands of traders who trust Cracken Pro

for their crypto trading needs. Pro.crakken.com to get started today. Bankless is launching the bankless token hub. At Bankless, we've been studying the crypto markets ever since 2017, and all of our research has led us to this. The token hub, you're a one-stop shop for Alpha to help you navigate through the crypto markets. Have you ever wished for a trusted resource that would share their thoughts, ratings,

and their opinions about tokens? Boy, do we have the product for you. The Bankless Token Hub is where we provide bankless citizens with the Alpha on the hottest tokens in crypto. We do the research so you don't have it. to. The bankless token hub includes the token ratings, where our team shares their research and outlook on the hottest tokens in crypto. Also, the token hub includes bankless bags, our own internal investment club. Bankless bags is where we put our money where our mouth is. And for

the bankless power user out there, you can access the analyst team 24-7 inside the Bankless Nation Discord. You can ask them questions and learn from a group of people deep in the weeds of crypto investing. The last feature of the token hub is the ability to up vote or downvote token ratings. The bankless token hub lets you learn from your fellow citizens to rate these tokens yourselves. The bankless token hub is launching right now and has already been beta tested by your fellow bankless citizens. So stay tuned in the bankless discord for updates. And if you're not a bankless citizen, well, you better sign up if you want access because this corner of bankless is available for citizens only. I'll see you in the discord. If you haven't yet experienced the superpowers

that a smart contract wallet gives you, check out Ambire. Ambire works with all the EVM chains, the layer two is like Arbitrum, optimism, and polygon, but also the non-Etherium ecosystems like Avalanche and Phantom. Ambire lets you pay for gas and stable coins, meaning you'll never have to spend your precious eth again. And if you like self-custody, but you still want training wheels, you can recover a lost Ambire wallet with an email and password, but without giving the Ambire team control over your funds. The Ambire wallet is coming soon for both iOS and Android. And if you want to be a beta tester, Ambire is airdropping their wallet token for simply just using the wallet. You can sign up at ambire.com. And while you're there,

Sign up for the web app wallet experience as well. So thank you, Ambire, for pushing the frontier of smart contract wallets on Ethereum. Bankless Nation, crypto and AI. What is the intersection? We're going to explore these topics with our guests today. We have Muhammad, who is a contributor at Alliance Dow. He's a venture partner at Volt Capital. We also have Chow, who's a partner at Alliance Dow, an investor as well.

Chau, Muhammad, welcome Bankless. Thanks for joining us today. Thanks for having us. Yes, thanks for having us. It's a pleasure to be here. Yeah, it's great to have you guys. I know we've been wanting to do a podcast with you for a while. And this topic really just came to mind because you guys wrote an excellent post here recently called the convergence of AI and Web 3, the opportunities and challenges.

And this was really an exploration of the ways in which AI might find itself in Web 3. It's kind of the intersection type point because there's all of these opportunities in AI, of course. And we know about all of these opportunities in crypto. But what I really liked about this post is you started to focus on some of the intersection type points. And I think that is what the bankless community wants to hear. We've been talking a lot about AI and how it could come kill us and how we need to make sure that we're aligned. Can crypto help with any of the upcoming problems with AI?

And so that's what we're going to start with here today. Maybe before we begin, could you give us like kind of a preamble of thoughts? So when I say, Chow, crypto, and AI, what do you think? Do you think that there's a large intersection here? or are these two totally separate technologies at just the highest level? Like what's going on in your brain with these two interrelated terms? I think there are two sides of the same coin. You know, obviously in the past you have memes like created by Peter Thiel that AI is this centralizing force

and blockchain crypto is this decentralizing force. So it sounds like they're incompatible with each other. But the fact of the matter is, I think that blockchain's provide a very good interface for AI to participate and to really flourish. Because if you think about what's happening today in Web 2 and the traditional games, everyone is very allergic to bots, to AI. Like bots get the platformed, you know, left and right by social media, for example. But to me, there's a difference between being a bot and being spammy. and we're in an era where the LLMs are so good that they're no longer have to be spammy. They can produce really insightful content.

And in order for those LLMs to flourish, they need a permissionless system, a permissionless platform that they can build on top of, and that's blockchains. So that's one area of intersection. The other one is decentralized compute. How do we build models in a way that they're not? not controlled by a central powerful corporation, right? So it's for the exact same reason why we want decentralized money, we want decentralized finance, we want decentralized social network, that we may want decentralized AI, right? Decentralize ways to train and, for training and

for inference of AI. So we can dive deeper into that area as well. I want to get Mohammed to weigh on us, but, but, Chow, you brought up with something that's core because that has been, That has been a meme in my mind, this Peter Thiel thing of AI is a centralizing force and crypto is a decentralizing force. And these are kind of two opposing forces. And I think the central ideas, AI needs to be centralized because it needs to consume this very large aggregate data set. And so something like...

Mass of economies of scale. Yeah, it gets better. The more data, the more humans you throw at it, the more kind of like feedback it receives from all of the humans in its large data set, right? So it's centralizing. Whereas crypto, as we know, is very... decentralized technology. Everyone, every individual can be their own bank. You can hold your own private keys and no one can take them from you, right? So that is a decentralizing force.

But I don't think, I get the sense that you don't actually agree with that split. Maybe there's the idea of we could have a decentralized AI. So talk to me about that. What do you disagree with Peter Thiel's meme? So there's a part, part of me agrees with it, which is that law chains provide checks and balances to AI. And so does zero knowledge proofs, which Muhammad is very familiar with it. So we can talk about that. The intersection of blockchain ZKP and AI, that's super fascinating. But the part of me that disagrees with Peter Thiel's meme is that I think decentralized

compute is going to play a somewhat important role. I don't want to overblow it because, again, this is one area where Mohammed and I actually disagree with each other. So Muhammad actually is strongly against decentralized compute and I can see some merit with it. I think that decentralized compute can be useful for training AI models, especially the smaller models. It's not good for large models. Large models you need centralized, you know, AWS or whatever to train the models.

And then for inference, I think that there is a different way to do decentralized AI. It's not about running the models across the cluster of nodes that interact with each other through the network. It's more about, I think the best way to achieve decentralization on the inference side of things is to run open source models on your local device. So this is an area we can dive deeper if we want. The way I understand Peter Thiel's metaphor that AI centralizing and crypto is decentralizing is less that AI and crypto are oil and waters, just more about the power structures, where AI and the power that AI creates benefits large centralized institutions, whereas

cryptography and crypto-economic systems largely benefit the individual. So, Mohammed, a similar question to you, but also can you talk about just what you see when you see the intersection of AI and Web3 and also how it shifts the power pendulum of the world. Yeah, I come from a different thesis. Like AI started way before crypto. So like AI roots come back to the 1960s and 1970s. And the idea was different and the goals were different. And then crypto came in like 2010 after Bitcoin.

So they came as two independent technologies. but and this give a course of every every technology that doesn't care for the other like AI people don't care about crypto crypto crypto doesn't care about AI but what happened in the last few years that we are seeing AI is getting overwhelming it actually we started to see actually threats from AI and now we are asking hard questions like how can we actually manage that and then comes crypto the picture and as Chao mentioned it's a crypto came as the checks and balances and for me I see that crypto is one of the very, very few technologies that can actually steer AI. No other technology can shift how AI works or how it performs, but crypto can.

And this is why it's like really interesting topic. And when I started exploring this with the child, we found that, yes, the technology that we developed in Web 3, like pocket chain, digital signatures, and Ziki proofs, all this will play a bigger rule in AI in the next few years. And the same time, like, AI is disrupting everything and naturally, like, we'll find some use cases in cryptos that will be disrupted. So I see it as a two-way street. Each technology will affect the other, and there will be a lot of synergies. But at the same point, we will find some areas that are conflicting.

And that's natural, because these technologies came as independent technologies that didn't share background. I find this idea of crypto as a check and balance against AI to be very tantalizing and very needed, because governments all around the world, are they not asking the question of like, what do we do? How do we regulate this? People going to chat GPT for medical advice, for instance. And I mean, well, is chat GP a doctor now? Like, is it able to recommend what you should do in kind of like for your health? And that's just one micro facet.

We also talk about all of kind of the bots and the deepfakes and all of the incredible content that these AI programs are able to create now and who's a human and who's a robot, right? We need a check on maybe some of this power. I want to come back to that idea as we go through this episode. But can we take one quick detour, which is there has been this thing in crypto, and I think it might exist in the meme layer, the narrative layer. I haven't paid it much attention, but AI coins. Okay.

So like these are tokens with AI in the name, I think, Chow. And maybe you have a few that you can rattle off the top of your head. Honestly, I have not been paying attention to these things because like I don't know what they are. And they seem like very much a narrative trade. What are the AI coins? Is there something to them? Is this the intersection between AI and we're talking about? Yeah.

To be around, very honest. I didn't look at them since chat, since GPT came out, which was November, maybe December, centralizer, I can remember. But the reason why I looked at AI coins is because I suspected AI coins was going to rally very hard immediately after Chachapit came out. And it did. So I found a list of 10 coins that are somewhat AI related or trying to, you know, jump into that bandwagon of AI. AI spiritual. AI spiritual. That's right. But I mean, one point that I remember today still is is render protocol. And the reason why I remember that one is because Fudan and I,

Mohammed and I disagree all the time on decentralized compute. And render protocol is the one that provides decentralized GPU for AI. So that's the only one that I really remember. And there's another one called Alethea, which provides AI avatars. But most of the coins, I don't remember them, to be honest. It sounds like that that wasn't even the point. You went into that trade knowing that they

were dubiously associated with AI, correct? That's right. Yeah. It was just a gamble that chat to EBT was going to penetrate into the mindset. It's just a narrative trade, right? Yeah. This happens all the time. Right.

And funny thing, when GPT4 came out, the AI coins didn't really move. So it felt like a lot of that narrative has been pricing towards the end of last year. Okay. Well, can we talk about maybe the real use cases? Now that we got the dom elephant out of the room. Because I don't think we're talking about any of the specific coins. Maybe there's some decentralized compute and some coin kind of embodies that thesis. But let's talk about the real intersections.

And you outline, I think, a few in your post. Maybe we'll start here. The first is payment and execution rails for AI agents. This is, again, an intersection point between AI and crypto. payment and execution rails for AI agents. Could you tell us what you mean? I don't know who, which of you wrote this post? Probably is it, it's like a David and Ryan kind of post or podcast where we both kind of

contribute to it. But, Chow, why don't we start with you on this question? It's mostly Fuda, so Fuda should go. Okay, Fuda, go. Fuda, tell us. So, like, just to sit the stage, what is an AI agent, right? Now we use Chachibati. You or David can go to like Chachibb-D type of question and AI agent,

where the CHGBT will respond to you, right? So you are the output. You receive the information from the CHGBT. What if you take this a step further and make the agent another pot? So you take a pot can actually be using an LLM, HGBT at the back end, asking questions, getting answers, based on the answer, asking new questions and getting new answers, and all this interaction having to achieve a certain goal. So an AI agent definition is that you create a pot and you give it just a goal.

You give the high-level goal, and the AI agent can go and use LLM and maybe another database to store data, and it keeps interacting with LLM at the back end. So AI agent actually is a new phenomenon. It is started like a less than a month ago when a VC created something called BBAGI, and from that it exploded. We have a BBAGII, we have auto-GBT, we have so many, many variants of them that actually are inspiring. Like, people are playing with the concept of how to actually make AI agents smart. and can do a lot of things. Wait, Mohammed, I want to make sure I understand the agent concept because this is happening so new and it sounds like you've been like a month old, right?

So is this the idea of like you fork chat GPT? Not quite fork. Maybe that's the wrong term. Maybe that's a crypto term kind of seeking in. But it's like one variation of a chat GPT, but you give it a different kind of objective. Let's say we take a chat GPT kind of like back end, a large language model, and then we just feed it just a whole bunch of bankless content, right? all of our podcasts, all of our archives, all of the newsletter.

And then we can start to ask it about like, what's the definition of ultrasound money? And it will give us like the bankless response. Is this kind of an agent? And the goal of this would be to like, you know, give the best possible responses to anyone querying kind of the bankless database or am I too limited in this thinking? Give us some tangible examples here. Okay. So this is one form of agent.

That's a specialized agent. Like you train an agent to be bankless. But the general form of agent that you create a bot that has storage, it can store data, and then it has access to ChachGBT using an API or another LLM. And then you go to the spot and give it just a title. Like let's say the title, I won't create the most successful podcast in crypto, right? And you give it just title. And it will take this title and start asking Chachvety, what do they know, what they need to do?

So it gives you a few items. And then it takes one of these items and it's elaborates on. it and gets a new list. And so, like, build the website, do, like, get sponsors, blah, blah. And at another step, this AI agent will actually go and execute this task. If you give you a suggestion to build a website, ChachyBT can actually author the HTML code that carry the website, go to GoDaddy, book a domain, and launch the website.

So an AI agent can be very fully autonomous. It can just take a directive and execute all. a job. Wait, wait, so the directives could be as general as like... So, Ryan, the way that this works with auto-GPT, I think I understand this from a high level is that you have, you almost have two chat GPs and one tells the other one what to do. And so it's like if you are using chat GPT, except you replace yourself with chat GPT and you give the originator, you give one chat GPT, is like, hey, do a thing. book me a flight to a country and then also get me a Uber and also a hotel and plan a plan of vacation.

And then that will be the direction of one chat GBT and that one second job with that first chat chat GBT will tell the second chat GBT to do all the things. And so it'll recursively iterate on itself until it gets to the goal. This is my understanding, Mohammed. Is that right? Close enough. Yes. And actually Ryan's response is the actual response. Like, what's it? Yeah, okay. So the goal can be this general.

It can be like, it's not just, I was being way too specific and specialized. Yeah, something kind of dumb. And I'm like, yeah, give me an archive. I'm like, search all the bankless archive. What you're saying is I could give an AI agent something as general as like build the best crypto, the best podcast in crypto. It's a number one in the charts. And it would, it would do this.

It would, it could conceivably. like create all the agendas for all the podcasts, create the entire guest list for the next like two years, reach out to those individuals on the guest list, create a website for like, you know, I don't like this example. AI, all right, and then it could actually like, maybe in a different scenario,

maybe we're not there yet, but pretty close, it could actually record a podcast with AI agents inside of it and there'd be like a version of David and a version of Ryan or whatever it thought the best format for a, top crypto podcast. It could do accomplish all of these tasks. With the fix, it doesn't even need you or David on Zabwkastry. It can create a model for you. We're in trouble, David.

When one chat GPT gets off the rails, the meta chat GPT is like, no, no, you effed up. Like, go back. And so I think, Mohamed, you said there's two elements here. There's crypto as payment rails and then the AI agent. That was the AI agent. I just want to put an AI agent because people throw the word. and think, don't think what's an agent. Yeah, I didn't realize until you just described it.

So the thesis here, this AI agent will not be as is that general. Like, they will not create a podcast. But instead, they can be very specific. They can create a fake model of Brian or David, right? They can create actually context or a question for your guest. So these models will be very specialized. And you can actually use an AI agent to coordinate the other agents. So the thesis here that bots will be very autonomous.

They can actually recruit and hire AI agent and even better they can pay them and receive payments using crypto wallets. And this is what crypto comes into the picture. To be autonomous, you need to have a form to pay and receive payment. And no bank will open a bank account for an AI agent. So the easy way, a crypto wallet, in Bitcoin wallet, Ethereum wallet, whatever you want. And then AI agent become full autonomous. They can receive payment for the world they do, giving a prompt you give them. can actually recruit and hire other AI

agent and pay them. That's what I call payment rail for AI agents. The other part of it is execution rail. What if you want actually, an AI agent want to create a website? Then it has to pay GoDaddy, it has to

recruit another AI agent to design the website. And maybe even they need to get resources for that. So if they go to AWS, AWS may plug them. So why not they plug it into a decent centralized compute network, where you can actually have a server that hosts your website. Or even better, what if the AI agent need to scale itself and do more compute?

It can actually recruit decentralized compute to scale itself. So, like, we are going into, like, a really dangerous territory here, but this is all possible. And it's not possible tomorrow. It's possible today. It's just we need engineers to build that. So Crypto can actually be the payment rail and execution rails for AI agents. So this is the thesis here. The key insight here is in theory, AI agent can control a bank account, but there are two problems that are unique to bank accounts compared to crypto.

One is, well, if you open a bank account, you need to K.C. Well, how does an AI agent KYC? That's impossible. And two, the interfaces are not permissionless in a bank account. So you can get the platformed if you have a bot that controls the bank account. But crypto solves both problems. So what's a better wallet for an AI agent? Is it a bank account or is it a crypto wallet?

The answer is clear. One of the very early bulk cases for Defi, one of the earliest episodes, was with FANS Spencer. And we talked about just the open API model for financial applications. It's like every single financial application is just a contract call away from another financial application. And we've never seen a world of finance in which your money market can talk to your exchange. Like NASDAQ and I don't know what a money market is in TradFi. But those things don't talk to each other. And even if they did, you need to log in and that you need to be a human and you need to go through KYC.

And so the composability of defy was immensely bullish. And we'll just go back to our bankless routes back in 2018 and 2019 when we were talking about the composability of defy because smart contracts are just one API paying away. And so that was Polish. But adding in AI agents as a layer on top of that is an order of magnitude just like way crazier that I've never been able to like think of that until now. It's like really kind of makes it seem like, man, humans are probably pretty inefficient agents on Ethereum. I bet you AIs are going to be way better at this game than we are. That's kind of the intersection here between AIs and payment rails as you're talking about, right, Mahad? Yes.

This is, again, like this I agent also we need marketplaces to discover other AI agents. So, like, maybe there is opportunities, and this is one of the investment opportunities that we are polished about because we are thinking about that, is that we need to create marketplaces where this AI agent can discover each other and can pay each other and can rank each other. So this is one of, and this is also a decent market place. So this is like, there is a lot of investible opportunities that come out from this thesis. I almost think, okay, so I don't think I was fully understanding, even though I've been in crypto, you know, I don't think I was fully understanding until you just described AI agents what this is going to do for crypto, basically. I think what we have done in crypto, guys, is we have just created a banking and money system for the robots.

Like that's, they are. We used to call them smart contracts. but now there's another level. I'm talking about like agents, AI agents using all of the smart contracts. So if you're an AI agent, I mean, anybody who in the United States

has tried to open a business, right, anywhere and pay payroll across multiple states knows what a nightmare it is to go register your LLC in kind of the meat space world and to figure out payroll taxes across all of these jurisdictions

and figure out how to fill out the forms right and set everything up. Well, you know what we call that in Ethereum? it's an eth address and it's a Dow, right? Like the decentralized autonomous organization, the A and Dow is autonomous, is AI agent. I mean, they are not going to want to mess in the Fiat old world banking structure anymore. They don't care about Wall Street, right? AI agents will completely string together a series of smart contracts, create their own decentralized autonomous organizations, pool capital together in order to coordinate goals. Imagine you're trying to

create the like imagine the goal is more generalized rather than to create the the best a podcast in crypto and you want to create like the best um you know defy protocol or something or the best whatever you want to accomplish like an economic agent level goal the best you know smartphone or something like this is a big country in the world if you can create this with an AI agent as well you could do this all on chain i mean we have just it's it's both okay so it's both very bullish for crypto which is basically like, okay, well, all of the protocols that we're excited about, like Uniswap and, you know, Ethereum and all of these different chains that we're very excited about, well, if AI start using this block space and start consuming it and creating economic activity, like if the future of the economy of, basically, I think we've had this backwards. Sorry, I'm just, I'm a little rambling now because I'm connecting so many thoughts here. But I think we've had this backwards. You know how crypto has always been about like, well, what happens when a major nation kind of adopts? our coin, right? And it becomes like, say, the U.S. sort of brings Bitcoin and ETH inside of its treasury. Maybe that whole concept is entirely backwards. In order to become a reserve money

for the world, you need to be a reserve money for the largest economies of the world. Nation states have been the largest economies in the world so far. The United States is chief among that. But what if crypto, something like ether Bitcoin, becomes the reserve currency for the largest future economy of the world, which is this economy of AI agents operating across geographic boundaries. How bullish is that for all of our crypto assets? And yet I'm also scared of this new world of agents independently acting because what have we created in crypto, guys? We've created unstoppable code, unstoppable chains, right? And so like what if they turn evil and what if they're bad? I mean, this is just...

What if they're better at crypto than we are? How do we start? So I guess what I'm saying is this whole use case, Muhammad, of payment and execution rails, right? Make sure we understand, guys, as you're hearing Muhammad describe this, payment doesn't just mean like, oh, I'd rather rather than wiring my ACH from this bank to this bank, you know, and I could send Ethan said, it means being able to set up decentralized autonomous organizations, being able to set up capital structures, being able to execute trades, be able to hire and fire people, being able to operate with all of these other independent agents. It's an entire banking system, money system that we've created for these robots.

That's how broad this first case is. I don't know, Muhammad, if any of those rambling made sense to you. Yeah, talk to me, guys. If you take this idea of AI agents controlling crypto wallace and doing DFI, take that a layer of abstraction even higher. So we actually mentioned this in our article in the conclusion where we said, if you view blockchains as a network, then we envision AI agents or AI in general to dominate the edges of the network in 10 to 20 years. So this implies not just money and defy, but also social

networks and gaming. And this again goes back to the very first point that I brought up in this podcast, which is that if you think about what's happening in Web2 in traditional games, in other consumer products in Web2, they're all very allergic to AI. They de-platform the bots left and right. But if you build a decentralized social network or a fully on-chain game on top of the blockchain, then you can envision a world where the vast majority of players or participants in the social networks are not humans, but there are bots that produce meaningful content. So instead of user-generated content, which is this whole thing about Web2,

we can see a world where we have a ton of AI generated content. So crypto is the only world where AI has quote unquote human rights, is on the same level as a human being. Elsewhere in Meetspace, they're treated as a lesser, but not in crypto. The analogy I like to give is MEV, actually, because in Defi, you have, DeFi basically has two groups of players. There's the human players and there's the MEV bots.

They have different goals. The MV bots, they're there to make money. And the human players are there to optimize for their dopamine. And one of them is EV positive. The other one is EV negative. The humans are EV negative, and the MV bots are EB-positive. But they coexist with each other very well. And so if that's how things work in Defi,

why wouldn't the same thing work in social network or on-chain games, for example? I think going through this article and also this conversation, a lot of the early days of understanding Dow's, like the idea of decentralized Uber on the blockchain and we can have this massive hierarchical organization and we would talk about the theory of the firm and how we would make things autonomous and we wouldn't need any internal employees and it would just be external value contributors and a two-sided marketplace to manage by a smart contract. This was like the vision of 2017, right? And all these ICO tokens that spun up.

And this was the original OG vision of what a DAO was before a DAO became what it is today, which is like a multi-sig with a bunch of money in it and a bunch of humans controlling it. The original version of a DAO was this system of inputs and output, smart contracts, that would be this autonomous network of value that would control some external resource in the world like Uber or Airbnb or any sort of, you know, know, commodity sharing ecosystem, except that was all just fanciful, like, ICO vapor in 2017. But the missing component of that was the AI in the middle to manage everything, or at least that is what in 2023 I am claiming to be now in this current like ice of form. So maybe we can unpack that statement a little bit. Was that really the missing piece of imagination back in

2017, the whole like autonomous side of Dallas, or am I just like over-obscribing the power of chat GPT and AI to really feel that void? Mohamed, what do you think? I think you are connecting the dots. I think once you understand the power of chat GBT and LLMs and like the power of AI agents, you're actually starting to put them in areas if you're interested. Like you saw that, that was worse. And that was maybe filled to achieve the original vision.

And now you are putting a piece of the puzzle, how AI can actually, help us to get to the original vision of the Tao. So this is kind of the brainstorming that we need to do now. And honestly, I think that we are already starting, but we are already starting in the right direction. Like now AI agents are not powerful enough to even come with their own goal, right? They have to be directed by a human. A human has to give this AI agent a goal.

So in the example that I shall give about AI agent as glayers in an online chain game. Yes, only chain games as bots will perform way, than any human player, but a human player can go and actually develop the strategy or develop the thinking or like educate the bot what to do or like how to even understand the rules of the game. And the AI agent will do all the legwork. It will go understand the rules of the game. It will understand the smart contract even if the game and it will do interactions with the smart contract when needed and with efficient costs. So as an AI agent, actually we have an on-eaching game in our portfolio that is doing literally that the allowing players to create to put a strategy and the strategy

will be translated as an AI agent to play for them. So the AI agent are not fully autonomous. In the future, they will be and it will be a kind of a battle of who will have the best strategy. Not necessarily, the league work will disappear. The league work will be automated and it becomes a market, an efficient market of strategies and ideas. This is mind-blowing stuff, Mohamed and Chow. I think what we need to discuss, if this is the world we're living in with humans and the robots and the machines side by side, and the same games and the same social networks, is are we able to use crypto in some way to tell who's a human and who's a robot? Because that is very important. So we're going to talk

about content authenticity next and how we battle all of the coming deep fakes because crypto may play a role here. Guys, we'll be right back with that conversation. But before we do, we want to thank the sponsors that made this episode possible, including Metamask. If there are any crypto term, you didn't understand on bank list, go check out Metamask Learn. They've got the info for you to level you up on crypto. Learning about crypto is hard. Until now, introducing Metamask Learn, an open educational platform about crypto, Web3, self-custody, wallet management, and all the other topics needed to onboard people into this crazy world of crypto. Metamask Learn is an interactive platform with each lesson offering a simulation for the

task at hand, giving you actual practical experience for navigating Web3. The purpose of Metamask Learn is to teach people the basics of self-custody and wallet security in a safe environment. And while Metamask Learn always takes the time to define Web3 specific vocabulary, it is still a jargon-free experience for the Crypto-Curious user. Friendly, not scary. Metamask Learn is available in 10 10 languages with more to be added soon, and it's meant to cater to a global Web3 audience. So, are you tired of having to explain crypto concepts to your friends? to learn.medamask.io and add Metamask Learn to your guides to get onboarded into the world of Web3. Arbitrum 1 is pioneering the world of secure Ethereum scalability and is continuing to accelerate the Web3 landscape.

Hundreds of projects have already deployed on Arbitrum 1 producing flourishing defy and NFT ecosystems. With a recent addition of Arbitrum Nova, gaming and social daps like Reddit are also now calling Arbitrum home. Both Arbitrum 1 and Nova leveraged the security and decentralized. of Ethereum and provide a builder experience that's intuitive, familiar, and fully EVM-compatible. On Arbitrum, both builders and users will experience faster transaction speeds with significantly lower gas fees. With Arbitrum's recent migration to Arbitram Nitro, it's also now 10 times faster than before. Visit Arbitrum.io, where you can join the community, dive into the developer docs, bridge your assets, and start building your first app. With Arbitrum, experience Web3

development the way it was meant to be. Secure, fast, cheap, and friction-free. You know Uniswap. It's the world's largest decentralized exchange with over $1.4 trillion in trading volume. You know this because we talk about it endlessly on bank lists. It's Uniswap. But Uniswap is becoming so much more. Uniswap Labs just released the Uniswop mobile wallet for iOS. The newest, easiest way to trade tokens on the go. With the Uniswap wallet, you can easily create or import a new wallet. Buy crypto on any available exchange with your debit card with extremely low Fiat on ramp fees. And you can seamlessly swap on main net, polygon, arbitram, and optimism. On the Uniswop mobile wallet, you can store and display your beautiful NFTs. And you can also explore Web3 with the in-app search features, market leaderboards, and price charts.

Or use Wallet Connect to connect to any Web3 application. So you can now go directly to Defi with the Uniswot mobile wallet, safe, simple custody from the most trusted team in Defi. Download the Uniswap wallet today on iOS. There is a link in the show notes. Guys, we have Chow and Mohamed here. We're talking about AI and crypto. What's the intersection? Apparently a lot, as we've just learned in the first part of this episode.

let's talk about this idea of the humans and the robots living together in all of the digital landscapes, whether it's our games or social networks or every place in the internet. There seems to be this problem of being able to tell whether the content is published by a robot or whether it was a human and who the humans and who the robots are. That seems important to know because, of course, with lower costs of replicating humans, we can get into a world of disinformation. We can get into a world where we don't know whether this is genuinely a message from the president of the United States or whether it's a deep fake from some AI that has, you know,

created a message from Joe Biden. Can crypto help us with this timestamping as a use case maybe, this idea of identity, tying cryptographic identity together to separate the humans and the robots? Muhammad, what's your take on this section of your post? Because you guys covered it here too. And actually, Muhammad, before you give you a take, I'd like to present just one scenario that I think is a great way to tee this up. Once upon a time, there was fake news that was spread about Vitalik's death. And Etherpric started to plummet.

And what did Fatalc do? He took a photo of himself with a piece of paper that had Ethereum's current block height on it. And that was his proof of life. In a world post mid-jurney and post-artificially generated images, that would no way. longer work. And so now we need a new level up. We need a new game and we need a new zero to one moment in our authenticity of humanness in this post-AI world. So I'll throw Ryan's question back to you. Okay. This is all the good examples. And just to add to the general audience, like what other examples can we use? Like we just saw like a defect for a song from Drake,

right? And like if you go back a little bit, there was also a video of Morgan Freeman speaking and say that's not a more that I'm not Morgan Freeman. So like this is all examples that hit a like hit a tone with us. So and the solution is very simple. We created a massive technological digital signatures that can actually solve all of that. So what's the digital signature for you to do any crypto transaction? You have to you have a private key. No one else knows it. A boat cannot crack it. And you are the only one who can sign this transaction and anyone in the world can verify this signature versus your public key and who why how they will know public keys he can go to the bankruptcy site you can get your Twitter accounts he can go

uh ryan with if and he knows that this public key is associated with you so this district signature turns out to be a very very powerful tool to address all of that including the picture of vitalic it doesn't need to have a picture he can just say a message using his own private key and done this is a key that is controlled by a human, and this human is telling the wallet this is a human generated content. So let's say that you want to prove this episode of bankless. What you do, after you get all the video file, you sign the video content with the signature, and now this video is clear that it was recorded by the real Ryan and the real David and

Chaua Muhammad, right? And then anyone can verify this. So this is the way, I think, is a very good. very easy way that Web 3 can control these defects, but there is one catch. Like now what, after you record this episode, you want to get a clip of it. You want to get a one-minute clip. The problem, once you do this, the signature is not valid for this one minute clip anymore. How to prove that this was an original subset from the original video. Another technology we developed in this space, which is zero knowledge proofs, can actually address

that. You can do a computation, verifiable computation in anything, using ZKPs. And, And actually having a subset is a very simple computation. When you say, okay, this one minute video is an original subset from the one hour video. This is a very simple computation that you can create a zero knowledge proof to prove it. And anyone can verify this proof. So we develop two technologies that can help you authenticate original content and even allow you to get subsets of it, to publish it without the need to re-sign the content. So this is one way we can move forward with this space.

So, Muhammad, we have the technology today to be basically like publish this using a verifiable digital signature bankless. We have bankless.eath. That's our ENS name that that longs to a set of private keys that David and I control. And so we could publish it from bankless.eath. So people would know this is a certified, verified original from the owners of this private key, which I guess through other means because we broadcasted this before, people would know. Oh, bankless.eath is David and Ryan. So this is authentic. And so we have that, we can already solve that problem today, basically. And you're saying also there's another problem, which is there could be snippets of this content that are chopped up, that parts of it are deep faked, something like this. And what we could do is we could actually use a ZK proof type technology to verify that a snippet, a video clip, say, came from the original that was originally signed by David and Ryan on bankless.eath. And we also have that technology today,

the ZK proving technology, to kind of verify and validate that a subset clip was from the original? Yeah, there are developers already building that and there are already developers doing something on images right now that when you have an image and you rescale it or like you rotate it or whatever,

or you do any touching up, it can prove that what you did didn't change the content in a drastic way. I didn't take Ryan's face, I bought someone else. So there are a project already building that. I think the Italian Foundation is supporting one of the team on this regard. We are seeing actually teams that are trying to do this for video.

We will see people do this in audio. There is a lot of ideas here and we will find the specialization because zero knowledge proof is still a new technology. And you need to find the most efficient roof system for your use case. So I think in the next one or two years, you will find many products actually addressing that. So you can use it without even understanding that ZikiB is under the hood. Anyone who makes a clip of something will be just attaching a small data, which is a proof. And the burden goes to the person who publish it.

If I publish something without a proof, we can take it by default that's fake, until they publish the proof that this was taken from an original content. I want to go, I'm so curious as to how this whole, how you can verify a clip of a larger set of data as, a part of the larger set of data. I'm so confused about that. So say we have like a one hour long video and you take a one minute long clip. There are an infinity number of ways of manipulating that clip.

You can change the resolution. You can change the color tone. You can add in new pixels and draw a line across the part of the clip. Since there's infinity number of ways of manipulating a clip, how does a ZK proof actually work in this context? How does that work? Okay. So we're already doing this today, David.

Like when you have a smart contract that do a lot of stuff, it can be defy or whatever, and you say, OK, I will put all of this in a roll-up, and the roll-up will execute this and just send me a proof on the L1 saying that this program happened. So let's abstract zero-knowledge of proofs as a way to create a proof for any program. If you can describe this program as code, you can convert this code in some ways into a proof. So what would you do if you want to take a snippet and one minute snippet of your video, you can say, here is the code that I used. It's 50 lines of code that they used to take the snippet of the original file, and you publish this on GitHub or any open source media, so everyone can actually validate this code.

And here is how this could get converted into a Ziki circuit, and here is the proof using the Zika circuit. So there is, of course, there is a chain of custody or a chain of proceeds. But any open source code essentially can be eventually converted into a Ziki circuit, and you can create a proof for it. So yes, there is some technique. details that you need to publish what transformation did you do. You cannot just do it like here is a proof for everything. This proof doesn't mean anything. You have to publish what you did to the code.

And yeah, once anyone validated that you just added a line, you just changes the resolution and then it's fine. But, okay, so the idea, the idea here is that all of the derivatives of data, so long as I am the person doing the manipulations, I can produce that chain of custody. So it's less about people, other agents that I don't like that are antagonistic. It's not about antagonistic agents that are trying to copy my data. It's about me as the originator and my ability to prove that all downstream derivations of my data are sourced from the legitimate file, correct? I would say anyone can do it.

If I today wanted to take one clip from your video and I can say, here is the code that I used. I wrote a code to take from one minute, 15 seconds to one minute, two minute, 15 seconds. And here is the open source code that did that. And here is the proof that, or Zaki proof that matches this good. It's an open process. But anyone can do it, which is more beautiful. The use case here, for example, is when a journalist takes a clip of a real recording and then put it out of context, for example.

And in this case, you can use ZKP to prove that the clip that the journalist showed is actually real versus totally fabricated. So it's people who intend on being truthful, not people who intend. tend on lying. So people who are claiming the truth are part of that chain of truth. Yes. And I think by default, like now we are not used to this concept of defix. But I think humans are quick to understand that like in the future,

defix will be as a norm. And anyone who wants to prove the authenticity, you will have to send something to make sure. They have to have a digital signature. But let me ask you about this. So, okay. So every, all pieces of content will have digital signatures associated with them, right, on like, you know, crypto economically secured digital signatures. Okay, that's

great. But we just talked about AI as being able to spin up their own private keys and being able to spin up their own digital signatures. So I guess my overarching question is, but does this solve the problem? Because how do you prove humanity in the first place? How do you guys know I'm not a bot and haven't been a bot the whole time holding the bankless.eath like private keys, for example? Sometimes I question myself that. So does that need to be fixed for this? Or are we still using the squishy social layer to prove humanity? And David was like, I met this guy in real life one time, one time we met.

David, are you really sure? How certain are you about that event? Not sure. Did it really happen? That's exactly right. Like you mentioned that you need this social layer to boost up credibility of the system. So today, the social layer is those web two social networks. So for example, if you guys want to prove that this bankless episode is produced by you guys,

you need to publish bankless.eith on your Twitter account. And people believe that Twitter account. And therefore they believe that this digital signature belongs to a human, not a bot. And what if I'm a digital influencer who set up a Twitter account? I'm an AI agent. There's not actually real. I'm just kind of synthetic. Then we have to collapse it back further beyond Web2 and get back into Meetspace.

Potentially, yes. Yes. Well, think about this. It depends on who cares about whether that person is the originator or not. Like maybe the AI is the originator and all we care about the truth for that is that is the correct originator that we care about. We don't care about proving humanity. David, that's a really good distinction.

The digital signature is not to prove whether or not something is produced by bot or human. It's used to produce that this thing, that is used to prove that this piece of content is produced by this person or creator or bot. Right. It's to prove authenticity rather than personhood. Right. Yeah.

So, I mean, our future AI influencers are totally able to use this technology to claim, to claim provenance over their AI generated content. Yeah. Whether or not it is fake news or not. Yeah. We'll just know it's from that specific agent, whether it's human or robot, and it's authentically from that agent. But we don't necessarily know whether that is a human or a robot.

My other question for you guys here is like, we have this technology today, basically, specifically that like sign in with Ethereum, sign in with your private key. Why hasn't it been adopted? I mean, Twitter has this massive problem. I'm like Elon Musk. He's like battling the bots all the time on Twitter. And it seems like crypto has a solution for this. Why haven't we been able to like, why haven't we've been able to kind of penetrate? And people are still using username and passwords all of the time.

there's not a lot that's being digitally verified and yet we have the tech. Is there a network effect problem that we have and that we haven't? How do we know that this is going to take off? I think the reason is very simple. It takes a startup that is well connected enough with all these platforms where the content emerges. For example, Twitter or YouTube or TikTok or even my Sony. camera, when I take a picture, I want to be able to sign it, provide a digital signature, for example. So it takes a startup that's well network enough with these platforms and be able

to convince them to join this protocol, to be part of this protocol where a digital signature is required to sign messages. So I think that's the only reason why the solution is not being implemented yet. And I think that, for example, This whole deep fake thing is going to play a massive role in the next election. But I think we only have one more election for this to happen because by the 2028 election, the solution will be ready. The solution of digital signature will be ready. We just need a builder to put all the puzzle pieces together and partner with the content

providers and make this easy button for them. That's the part we're missing. That's right. Yes. Convenience. user expedience always has been the issue. Probably the other piece we're missing is for all the bad actors to force this use case into necessity.

Yeah, but that's coming soon, isn't it? The election could do that catalyst. Wow. Okay, so we've talked about payment and execution rails for AI agents. So basically, we've created a money system for the robots. A good job, crypto. Hope that turns out okay, but it's definitely going to increase the use of our crypto networks. And then we've also talked about this check and balance, which is crypto technologies can be used to verify authenticity using digital signatures.

There's one other category, and there's much more that you post-discuss, but there's one other category. I want to make sure we cover before we end this episode, Mohamed and Jow. So that is privacy. Privacy. So tell us about the privacy story and the tie between. AI and crypto and privacy. What's the intersection there? This is actually a new category,

which is what I consider that how crypto can actually solve the problems of current AI models or the AI paradigm. AI paradigm now are very transparent. They don't prospective privacy. Like chat GBT collected all the data essentially on the internet and trained it, right? and if you have some certain private data and use the chat gbt,

this data actually may leak back into the training set. So the training set of chat gbt will have your own private data. So the question becomes how to solve this privacy issue, how to allow machine learning model to act on private information. The examples that I love to give here is like using AI for medical diagnosis. So this is a trend. It's not approved by the FDA, but even the FDA is willing to explore AI for this. And there is multiple research pieces from the FDA saying how AI can help with medical tasks.

So the point here, let's say that you have an AI model, right? You want to, and this AI model can diagnose cancer. But to get it to diagnose cancer correctly, you need to train this on many, many data, right? So you need to get people who have cancer sharing the. medical records with the AI AR model. And people also who didn't have cancer to share this data with their AI model for the AI model to be able to label this patient has cancer or this paper doesn't have cancer. But are we willing to share our medical information with this AI model?

Are we willing to share this private intimate information with any company? The answer is typically no. So then how to solve this problem? How to use AI for something like that? I think the answer is simple, which is using, again, zero knowledge proofs. So let's say that I have a medical record today, and I want to use, I'm okay sharing this with, for training this, but my only condition is that it doesn't have any personal identifiable information.

So again, we play the scheme of authenticity. The hospital or the doctor who created this medical report can digitally sign it to make sure it's authentic. Now we prove this medical record authentic. Now I want to anonymize this medical record. I want to take my data off it, only keeping the x-rays, MRIs, whatever. Then we can use a ZKB to prove that I just anonymize this medical record. I took off only my personal information and the rest of the medical record is complete and correct. And now we have a set of authentic medical data that are clear, that can be used to train AI models. So we solved it a huge issue in this case. The other part is I call this a training how to use privacy to train

AI model. But even after training of this AI model, we can actually use it for diagnosis without leaking privacy. Now a company runs this AI model in their servers and I can anonymize my medical record before sending it for diagnosis. Like I will just send this data to the AI model and the AI model will tell me if I have cancer or not, but the model doesn't know my person, my name, my real name, it doesn't know my addresses, it doesn't know any person information. So I think ZikiBs, and this is one of the use cases, there is only many use cases in privacy. One of them is the Wallach Coin announcement that we can touch on if you would like. But I think privacy and ZikiBs are very, very strong players that can actually solve many

problems of current AI systems. What's at stake if we don't solve the privacy issue with AI? Is it basically like, because outside of the bounds of this, I'm not aware. that this is a solved problem at all. So I'm just wondering if basically the chat GPTs of the world, like we're moving without something like this, we're moving towards a dystopia where there is no such thing as digital privacy. What's at stake if we don't get this part of it right?

A lot at stake. Would you like to people to know your medical records and like act on it and like maybe even like treat against you because you are not in a mental state or a healthy state? So, like, it's very dangerous. And I mentioned this in our private discussion. Like, if you really don't want to use any information public, that is the only way, the only way is to make all your transactions private. And this is something like we also built in our ecosystem using Zcash and like other

companies that use zero knowledge proof to hide complete information of a transaction. I can transact with you. I can send you data. But no one except me and you will know it. So even to protect it from invasion of privacy, and I mentioned this in the article, you can, but you will have to go through the pain of like using complete privacy using zero knowledge proofs. Guys, this has been a lot of fun. And as we draw to a close here, I got to put the investor cap back on here, of course,

because we are on the journey west and we're definitely looking for opportunity here. A lot of investors listen to this space. And so I'm trying to listen for the investable opportunities that you guys see between the intersection of AI and crypto. And of course, there's lots of AI concentrated type bets. We already talked about the meme narrative trade. Not interested in that. I'm more interested in kind of the fundamentals here. One thing I'm hearing at the end of this episode is buy crypto networks that AI agents are most likely to use in the future.

That seems pretty logical. I'm like, okay, well, I'm doing that. I think I've been doing that. and so continue doing that. Okay, that makes me more bullish. Also, we talked about looking for the easy button for content authenticity, content authentication. Somebody's got to build that. There might be a number of startups trying to do that. Also looking for the easy button for privacy. That seems to be a, you know, case here. But, but Chow, high level. What do you think the investable opportunities are here between this intersection of crypto and AI? Of course, none of this is financial advice, as we always say.

but what are the fundamental opportunities here? For public market participants, frankly, I can't see any. Just because, I mean, I said before, there's probably a dozen of AI coins, but it's very hard to price them because they have no cash flow. I don't know if they're overvalued or undervalued. You know, you can do a narrative trade that lasts like two months, for example, but there's nothing I'm really willing to hold long term at this point. Because I don't know if they're going to go up or down.

It's hard to price them. That said, there's a couple of stongs if you're interested in that I'm personally invested in. I've done quite a bit of research. But one of them is TSM, Taiwan Semi-Conductor company. I see Mohammed is smiling because Mohammed told me about these two stocks last year. but one of them is TSM. The other one is ASML. They are parts of,

they're two parts of the AI or the chip supply chain. So Taiwan, so TSM's main customer will be someone like, so TSM has a lot of customers, but one of them is Nvidia, for example. Nvidia is also probably the purest AI bet. However, Nvidia's P is like 140 or something. I just can bring myself to buy it. So I'm not thinking about Nvidia,

but TSM produces the chips to Nvidia. Invita is the chip designer. ASML is the one that produces equipment, just to simplify things, equipment for TSM and Intel to produce their chips. So TSM and ASML are as close as it gets to monopolies in any industries. Like TSM, it probably will take Inton,

tell maybe three years to catch up to them. And ASML might take a decade. These are the chips that all the AI agents are going to run on, basically. That's likely right. The semiconductor bottlenecks is the players that create the chips that anyone using AI need to them, need them. That's right. Isn't the more obvious answer, Chowin and Mohamed? So, cool, there's infrastructure that we need to put the AI components together to build the AIs. But what about just buying the AI money that the AI is going to use in the future? Isn't that the more simple answer?

Of course. Yeah. Of course. So doesn't that just mean we're just buying ETH? Buy Eith, that's right. Yeah. Okay. There we go.

That's the answer is it. It sounds like every conclusion on bankless is that answer to the bankless. But I mean, this is great. Last question for you guys is we kind of head out. So we talked about a lot of things, a lot of transformation coming our way with AI agents. David and I have been doing a number of episodes related to AI safety. And there's this problem that Elise Zerudkowski may be most famously lays out of like, he's very scared of the existential risks, he's 99% certainty that AI is going to like murder us. And so I mean,

this is like kind of worried because I'm sitting back and we just established that we built a money system for AI agents. Okay, cool. Did we just build a money system for AI agents to like accelerate rate the murdering of humanity, the destruction of humanity, you guys are in the space too. And I'm just at this stage information gathering, looking for people with some knowledge and some opinions here. What do you guys rate the probability of like the AI risk threat to humanity? I mean, Eliezer is the most extreme. He's like 99.9% sure.

There's others at a far less extreme. How do you personally weigh this problem? Maybe chow you first and then Muhammad will end there with you. To be honest, I haven't thought. deeply about the AI alignment problem. I've listened to a bunch of podcasts. I haven't formed my own independent opinion yet. But so far, I'm very optimistic about AI. I've just seen how much more productive chat GPT has made are developers. So the single biggest use case of GPT from what I've seen today is developer productivity. So I've chatted with a bunch of developers. Every single one of them

tells me, they've quoted me a productivity improvement of anywhere between 10% to 200%. It's mind-boggling. And I think that, you know, these things are just going to make startups a lot easier. It's going to make it a lot quicker for people to ship MVP's and build new products. And I think productivity is going to increase across the board throughout the whole economy within the next 10 years. I don't really see a world where AI kills us in the next 10 years. And I see a huge improvement GDP boost in the next 10 years.

So that's what I'm seeing. I will say the AI alignment people will tell you that, yes, everything's going to be great. We're going to have massive economic and productivity booms. And then the alignment problem will manifest and they will kill us. It's more like the turkey. It will be great. The turkey's having a great time until it's, you know, the week before Thanksgiving. Muhammad, though, how do you answer this question?

What's your take on it? I think it's a viable threat, but my personal view is that we will actually figure out ways to use AI to discipline AI or AI to even fight AI. So like I don't underestimate the human ingenuity to like come with solutions, even using the tools. Like I still consider AI as a tool that we use now. And if it became outside bound, we will find ways to bring it back inbound. And we already starting discussing today ways of how we can actually prove

that stuff is happening without AI can prevent it. This is the new nuclear codes, right? Like you cannot launch a nuclear missile without having a distal signature and like biometric signature and a lot of other stuff. So I I'm optimistic and I think we will have it's a threat and we have to be attention to it. But I think we have all the tools that we need to

be on the good side of this debate. Very good. Guys, we'll end on that optimistic note. One thing I will share, I'm also, I haven't formed an opinion on this yet for myself, but I can say this with 100% confidence. I am more optimistic in our AI future with crypto in it than without it. I do think this technology is a useful tool, checks and balances for our AI system. So, you know, hopefully we're doing a good thing here. And certainly you guys are doing a great thing by bringing this to our audience.

today. So, Chow and Muhammad, thank you so much for spending some time with us. Thanks for having us. Wrists and disclaimers, of course. You got to let you know. All crypto is risky. So is artificial intelligence. You could lose what you put in. But we're headed west. This is the frontier. It's not for everyone, but we're glad you're with us on the bankless journey. Thanks a lot. Bam. Thanks, thanks guys. That was killer. Thanks, guys. Really great. Really fun. You guys unlocked some things for me personally. So I love those types of episodes. I appreciate it. Thank you.