The role of AI in Cybereconomy

Transcript

Hello. All right. So, super happy to be here. Um, my name is Ardam. Waiting for the slides. Just a moment. Oh yep. Yeah. For some reason, they're at the end. So, let me just click through through all of them. All right. So yeah, my name is Ardam. Um, a bit of background about me. I I was a mathematician for about 10 years.

Um, and but around 2016, I I got interested in crypto and since then, yeah, it slowly consumed my life and around three years ago, I joined Cyber Fund as a head of research. Today I'm going to talk about uh cyber economy is a concept that we have internally and it's a concept that we use to understand what's happening with uh with the world with our society how it evolves right um but before I I jump into the our thesis about cyber economy let me uh tell tell you a little bit about our fund so um the first uh point that I is that we're very much thesis driven and uh this is

um you know our founders there they have been operating in crypto since 2014 investing founding building uh towards the cybernetic economy the v this vision that I'm going to talk about uh second point is that we're investors and we funded more than 200 plus 200 startups including genesis investments in ethereum salana coosmos celestia ien layer and many others Um we're also founders and builders. We built Lido and P2P staking solutions with 40 billions in TVL.

Um and yeah, so today what we're doing we're deploying around 40 million per year with a focus on three technologies crypto, AI and robotics. Uh and the reason is because these technologies will drive the next phase of capitalism. Okay. So let me now turn to cyber economy, right? What is it? First of all, the definition is that it is a system of resource allocation and actually any economy is a is a system of resource allocation. But cybernetic economy uh has two characterizations.

One is that it is self-regulating. That's why it's cybernetic. And the second is that it is hyperefficient. That's the second one. Uh and we uh truly believe that that's going to be the next phase after capitalism. So um I would say that this is uh this in part is an answer to many challenges that we have in our current system of capitalism. Okay. And today I'm going to talk uh I'm going to unpack this vision. So let's jump uh into the details first.

What is a uh economy? It's a resource allocation system. How it works is that we have entities on the let's say on the left right and the entities are different. It could be people, could be states, could be firms, right? And what they do is of course they produce and they buy um you know other resources and goods, right? Uh and they decide what to produce, what to buy and the the way they kind of reallocate those resources and goods and and services between each other is through a market, right? and the market is on the right.

Now a very important point uh in economy is that it actually uh self-regulates through uh through a key mechanism uh of supply and demand adjustments according to price and the price you know maybe some a non-trivial observation about the price is that it is an emergent uh emer emerging signal from the market that tells peoples and firms and states what to do right and exactly how they adjust demand and supply is uh is um is basically uh according to this graph that I have there if the demand is larger than supply the price has to uh

has to grow the supply is larger than demand it's vice versa the price uh goes down and the and because it goes down you know the the supply and demand they adjust and that's how capitalism works and that's why we we live in capitalis because it is the most efficient system that we know so far, right? But there is a catch and the catch is um you know that we actually are not all rational agents uh or let's let's uh let me reframe it. We're we're rational but only within the limits of our capacities. In particular, uh, humans lack in discovery capacity. Uh, which you know, basically our s how we search,

how we read, in cognitive capacity, how we analyze and uh and process things in coordination capacity as well, communication and uh and commitment to action. For example, it's it's famous that person cannot closely communicate with more than 150 people and action capacity as well. you know just just the mere action that we can have uh during the day are pretty limited right so that's kind of the catch and that's why a lot of challenges appear in our economy uh be that asset bubbles credit cycles etc etc principal agent problem is another one um and we posit that um there's actually the technologies that we uh that we are

witnessing right now AI crypto they will actually solve this challenge challenges and so in particular let's talk about AI well the point is the point about AI in particular LLMs that they all excel at these capacities they truly are much better at these capacities than us uh and that's why it's it's so let's say they found product market fit you know we all use chat GPT all the companies uh try to integrate this into their workflow and that's why uh AI and organizations they truly are a match made in heaven right so essentially let's think from a let's say academic standpoint what does it

really mean for economy if AI helps us to excel at all these capacities well it means that our limits are uh are further we're almost unlimited and as a result we're more rational right so AI really enhances this cybernetic loop of capitalism from market to price to supply and demand demand adjustment and the upshot is that actually AI supercharges the invisible hand of the market uh in cyber economy uh and Adam Smith would be definitely very proud of us right but the interesting part uh and what I wanted to actually talk more about today

is that this is only one side of the story but there is actually a second side to the story of AI integrating into our our economy and the second side of the story has to do with market and what AI has to do with markets and how it actually improves it. Right? So to unpack this, I want to introduce a new term called instrumented markets. So instrumented markets basically are markets that are programmatically accessible via trading API and uh uh very good examples are stocks, hail rides and and many others uh as well right those that you can access

from your terminal and the you know as you can understand the if if the market is instrumented it's much more efficient because market makers can participate and etc etc but there are also So uh markets that are uninstrumented and actually there's a lot of markets that are uninstrumented such as enterprise procurements for example maybe a state does a tender to build a factory or something like that or [snorts] microtasks like what we do on task rabbit I can go to task rabbit and find a person but it's not uh fully instrumented right I the the specification is not complete there it's very manual and I I want to tell tell

you a story about balloons. For example, I did an experiment and I my daughter had a birthday and I decided to order balloons for that birthday uh solely by Chad GPT. So I told the agent, okay, buy me balloons and I gave him, you know, some directions and turns out that and it did the work actually. It ordered balloons, but they were not as good. Okay. Um so some balloons were not exactly the ones that I wanted. It was pretty expensive, etc., etc. So it it wasn't really good really the best. It was okay but not the best. And I would say that the reason is because the balloon market is not instrumented, right? And what AI does?

It actually instruments basically any market. That's the crazy part about AI. And the question is okay well but how how does it instrument? How exactly does it work from the uh let's say uh from a hands-on perspective, right? And the answer is intense. Okay? Uh some of you are familiar with intense u because uh because intents are ubiquitous in crypto but actually they um this concept is very relevant to AI and we call internally this an intent economy and the simple workflow is like this. Let's say uh you know an entity has a spec right and AI is able to read

and understand this specification and then from this specification they extract intents what exactly we want to do right after that with with constraints by the way of course you know it's not just I want to do something there's a lot of constraints etc etc and then uh a really important step is generating a local auction or maybe some other mechanism M okay and this is what uh has been what is not there right now like it doesn't exist okay we call it generative markets and you can think of it as for example on the example of balloon an AI not would not just go to some websites and order balloons for me but they would say oh okay Ardum wants to order these balloons

with this and these constraints and we're accepting applications we're accepting bits for this that's what they they would do okay and then they would discover counter counterparties, maybe on other websites, maybe other agents would discover this local auction, right? And then once they discover counterparties, they would negotiate and then they would commit and they would commit to some contract, right? And uh you know in parenthesis is smart because of course a for agents it's much easier to commit to a smart contract rather than uh legal contract and then they execute right and uh as a as a last step they obtained the verifiable outcome and you can see that in this intent economy it's it's um it's very efficient it's

outcome based you know it's very focused on the delivery of of what the user wants. Um, and I'll I'll touch on this a little bit later, but be before but now I want to talk about what are the core components that we uh need to build u in order for this to actually work because uh some steps there are are still sci-fi. They're they're not they don't exist right now. And the first is well for this to work the AI needs to actually be aware of what they're doing.

They need to be aware that they're coordinating uh econom you know economically with other agents and they have different smart contracts and different let's say uh utility functions. They need to understand that they they need to negotiate. They need to understand that they have secrets something they they don't want to disclose or and something that they do want to disclose. And we internally call this mechanism aware AI. And as far as I know there's there's no existing you know there's uh there's no mechanism aware AI. I think it's an open direction both for research but also for um for startups as well. So that's the first one be a neutral environment. And

with this we're all very familiar because um you know the the more neutral the more permissionless is the environment the better and the more efficient is the is the econ you know economics economic system that is built on top of it for and why do why does AI need it? Well, they need to express intents in this environment, transfer value on this substrate and coordinate between each other and uh blockchains is is a perfect uh has a product I would say product market fit actually here uh because they can they can transfer value well by transfers and coordinate between each other using smart contracts. see generative markets. They need to be able

to generate smart contracts on the fly to have auctions to sell puts and futures etc etc to engage for AI agents to engage between each other uh with economic contracts and of course they need identity and reputation something that has been talked a lot on this conference and privacy as well. So these are kind of core components. Uh and uh let me uh do an upshot on this slide is the upshot is essentially you know um if I would characterize the economy that we have right now would call it attention economy. It's a it's a popular term. Why is because it's in in the incentive of uh all the apps and

websites is to keep the user uh on the platform for as long as they can to show them ads. Essentially, that's the attention economy. And the point is that well, with the uh with the introduction of agents, um there's less and less eyes on websites and apps. There's less and less um uh you know, demand for these banners who who we cannot really show banners to agents. And that's why uh the payments protocols like X42 uh have a have a big promise. And that's why we think that you know the economy is transforming actually from attention to intention economy or we call it intent economy.

Um so yeah and then there there will be generative markets. So not only markets that exist already but constant generation of local submarkets and local resolutions uh of intents and this is by the way exists in crypto of course every time you know on cow swap you you make a you make an order this this is what this is what happens but the same must happen with AI agents and the and the and the third point is of course we need rails to build all to build all this and blockchains provide this perfect rails And I make a question mark here because because actually this comes uh you know if you think from first principles the

AI agents really do need some rails right they they will operate in some uh on some rails and it's not a given I would say that uh blockchains will be the rails it's uh right now uh because there's there's a lot of uh friction there's a lot of things that don't work exactly like they should um and this is our call I would say uh in in this from uh let's say from cyber funds perspective is that we we think we need to um shape actively cyber economy and what does it mean well let's take back step back it's actually inevitable that we come to cybernetic economy it's

inevitable that this will happen but the question is it will it happen in 100 years or will it happen in 20 years and what we will what we will have to go through until we get there. So the path is up to us whereas the the goal or so to speak the the destination is cyber economy is inevitable. Okay. And and when I when I talk about the path um it's it's like u you know one might think well okay does it really matter what path we take and the answer is actually it does because we can make a really painful detour into into some economic equilibriums where we give up our data

we give up uh for example control for AGI to some states etc etc and become really cyber slaves and cyber slaves is also this image or meme that we have internally that that you know in some sense it doesn't exist right now of course but in some sense it does so if you if you know if you if you look at the data of how many of how much people uh are looking at Tik Toks and really are on this dopamine loop and it's hard for them to escape it really does start to look a little bit like cyber slavery but I would say I mean the the potential for cyber slavery is even worse. So for us it's important to be conscious about

it and to accelerate towards open equilibria where uh you know [snorts] we prosper and uh we benefit humanity as much as possible. And so yeah that's that's the the main message you know we choose to actively shape cyber economy and we accelerate towards stable economic equilibriums those where there is no um >> let's say disbalance of power. Yeah. It's it shows that I have two two and a half minutes for some reason. All right. So yeah that's it. All right. Thank you so much guys.

It's been a pleasure.