The AI Agent & Memecoin Thesis | Pantera’s Matt Stephenson
01:21:02 Share

Looking to bankers where we explore the frontier of internet money and internet finance. And once again, the aides are ahead of even the humans expLoring the frontier. This epo de is going to will be all about A I agents on chain because very once had bots transacting on our block chains, running a smart contracts and doing all these bot things.

Some of these boats are getting a little smarter, and they are now perhaps evolving in to A I agents. What's going on? What trigger IT? What happens next? We have met Steven sin on the pot custody.

He's got A P H. D. And behavioral economics, which I think is highly appropriate because game theory exists for AI agent in humans like, uh this is the A I agent thesis, if you will. It's definitely been going around in the current event cycle with this goat mincin. Uh, but just like how bitcoin provided entire industry, we kind of think that the goat mean coin is kind of doing the same.

We had so many questions going to this episode. So one is, when we say A I agents, what what does that even mean the first place? And I think mac gives us a good definition of that.

Then we get into the gulp mean coin. And that A I agent case, also something called luna. So if you're not familiar with what's going on the A I agent world, we get you up to speed there.

And then we also ask that what is the benefit series of this, like what blog k chains will benefit? Where are the investible assets we can look at if the AI agencies says there's going to be an explosion of this type of utility on chain? And finally, we talk about some the societal ramifications, right magine.

We're creating A I agents. And now with crypto, they are decentralized, they are autonomous, they don't have an off switch. What does all of .

that mean before we get into the episode message from my friends and sponsors? Over at zero x the builders of O G critical define infrastructure, all zero wax is shipping, has shipped v two, which is giving you the best on chain execution for your trades. No matter the size, the trade size can go up to something very large.

So if you want to get the most amount of tokens from your trade into your wallet, you've got to use zero X V two. Many agreements out there give you quoted Prices, but when they token show up, uh, you might notice that the amount of tokens and do not match the quote that they gave you. Zero x because that has the best on chain execution because IT does a multi hop and multi plex rounding, which means that split trades across multiple liquidity sources and intermediate tokens to find the optimal execution route.

What are all the permutations that are out there? Is your X T two, who finds you the best one to make sure that when you make your trade, you get the best amount of tokens possible in landing into your wallet. So, you know, there are experience og s over as zero x so if you are building a serious on chain APP, you have to build a eros v two to get the next time pricing model. You can find them as bank less T C C slash zero x so let's get .

right into the episode. But before we do, we want to think the sponsors that made this episode possible, including our recommended exchange, which is cracking, this is the place to go front, run the A I agents and buy some crept to. Now.

if you want a crypto trading experience backed by world class security and award winning support teams, then head over the cracking when the longer standing and most secure crypto to platforms in the world crack is on a journey to build a more accessible, inclusive and fair financial system, making a simple and secure for everyone everywhere to trade.

Cypher, a crackings intuitive trading tools, are designed to grow with you, epower ing you to make your first or your hundreds trade and just a few clicks. And there's an award winning client support team available twenty four seven to help you along the way, along with a whole range of educational guides, articles and videos with products and features like cracking pro and and marketplace and a seamless APP to bring all together. It's really the perfect place to get your complete cypher experience.

So check out the simple, secure and powerful way for everyone to trade crypto with a complete beginner or go to c 点 to see what group to can be not advice。 Trading involves risk of. Imagine a future.

Wear A I in blockchain. Converge where decently sized intelligence Operates transparently with with three infrastructure, unlocking boundless human potential. And that future is here with zero gravity. The first decentralized A I Operating system, zero gravity just faster.

It's fifty thousand times faster and a one hundred times more conspicuous than current alternatives unreached the power of high date applications and modular ai with arrival ed report limits, scalability and a verified A I frameworks. Yo, gravity is turning the dream of on chain A I into reality. Curious about how A I and went.

Three, can we hate the future? You can visit zero g dow A I to explore the infrastructure as revolutionizing decentralized ai. The unis wap wall is officially the preferred wallet of benglis and is the one we use any time when we want to transact on train where're on your brother er or on the go. You swap wallet makes IT easier than never to swap anytime, anywhere.

Use your walk to transfer funds directly from the tops and change and happened thousands of tokens across the and over ten other chains like base arbitron optimism and delivers deep liquidity, fast execution in reliable quotes with zero gas walks through unis, wap, x and when IT comes to security, you can rest easy knowing it's back by unit swap laves, one of the most trusted team fy. Their code is open source and independently reviewed so you know, it's protected. So why way? Download at the union all today on chrome, IOS and android and don't forget to claim your free unity me directly in think station. Happy to introduce you to math eva research partner at panter a also has A P H D in behavioral game theory, which in theory I think applies to AI agent two a recently on the pen tera blog matt route article title cypher picks and shops for the A I gold rush which triggered the epsom that you are about to hear today. Math.

welcome to the shop.

He feels like we just had a moment in cypher to A A particle collision, if you will, between AI asian ts and mean points cyp to AI has been a thing for a while, has been a thing for like coming up on two years now maybe ah but only because we kind of knew that there was a there there uh and I would say something now is actually explicit different. How would you say how? How would do you care tize this moment in the world of the intersection of crypto AI?

What happened here? Yeah I mean, there's there's been various narratives in making your ecliptic AI for a while. The how I put out a good blog post about IT last year and and we wrote paper about a AI agents maybe using you desensitize commitment of device as A K block chains and they have to talk about that I conference. And so we ve been thinking the lines between agents and block chains in particular would converge. Agents aren't supposed to be here according the same mtmc until twenty twenty five but um they sort of arrived eclipsed early right and so they showed up Peter and started um interacting with with meme coins, mostly particularly driving narratives and acting as these kind of influences narrative agents for certain times of innes.

Matt, what's an agent? What's an A I agent?

yes. So you know we love semantic debates and cyndi um and boss first s agents. I think it's important to make the distinction.

Just bought a been around forever, right? Something like two trillion of monthly stable coin volume is driven by bots and hasn't for a long time. Botts are kind of like programs um agents you can think of, I think of in the economic sense. If you try to go like A I textbook can ask put an agent will give you some galaxy definition about you know anything with a sensor that interacts with the world.

But if you but the the e on definition is strait forward, an agent is someone who sort of like does what you want to some extent, if you you ask your friend to go to lunch, right, and they give you something you'd like without you having to tell them that's a pretty good agent. Um and and I would say what we have now in the reason is singular h agencies from bots is um a eyes are acting closer to the economic agent, right? You can sort of uh you their low code, no code, right um you don't have to programme them as you had to do with a boat. And then second dly, they're able to interact in an environment to sword flink uncertain and surprising rates. You said, let's say truth terminal loose with its training and IT adapt timely response to the world and interacts with the world in this no unpredictable but but sort of directional and .

understandable way so that we have a this concept of economic agents and and wondering if we go go through like what typical economic agents actually are and then we can may be expand that too like the the prefix which is like A I agent so um yeah I guess in I an agent I mean you like David brings me into this podcast and like I produce a podcast I suppose i'm in the of some type you like how about the company I work for, bank less? Is that an agent? How about something like the authorities foundation or like a corporation like apple? Are those also agents? Are these all economic agents? And like how about something like your piece of technology?

I mean my computer uh or are the wheel? I mean, are these tools or are these agents? Can you give us some a historical definition of what an economic agent actually is?

Yeah for sure. So so agents come out of research. And in econ in the early seventies, I think it's a paper by ross that defines at the first time and it's used to just sort of like characterized plainly the relationship between um like be an employer or employ or a person asking for someone to do them a favor, right uh uh a contract relationship. This sort of imperfect and incomplete contracts is generally the way we should introduce an agent.

And so uh for instance, you might have an agent that you know you want you that you want to get you something um you want to begin you a souvenir from a foreign country or something right a friend they could be acting as an agent for you because you'd like to get this souvenir and they're going to paris now if there a good agent, right, we might have some parameter that sort of says how much do they understand what you like. Uh, a good agent would be able to go to the gift shops or go on paris. They'd able to sort to figure that out and they have a sense of you.

In a bad agent, you would have to specify and look up all of the particular, uh, you know, all of the possible gifts in paris and tell them exactly what you want. So that would be an example of the way you might model IT. right? A good agent is one with sort of higher fidelity to your preferences.

And a bad agent is one with lower fidelity. But in terms of like where are we want to character as a firm as an agent for you, I I think there's a way in which that could make sense ultimately. Um if you have these as a model, right, you can always say if something is doing, you're bidding, it's going to be acting as an agent. Usually we think of these as sort of uh roughly human to human uh in twenty miles, I see and and I would say the reason the AI agents make sense is they have the same like human like uh characteristics where they seem to be willing to do you're bedding uh with you with some level fidelity.

So willing to do you're building rates like I guess I I just want to make sure understand the difference between sort of an agent and maybe like a technology or or a tool, right? So if I have a hammer that that's not an agent, I suppose because maybe is there some aspect of IT has to have intelligence that has people to be a sign a goal and accomplish the goal?

I mean, like something like a hammer computer only works when you have uh, an agent that is kind of like using that tool to to amplify IT. So I suppose maybe that's not an agent. And yeah and I like not like so is there is some sort of stateless intelligence that is you know inability to accomplish a goal wrapped up in the definition of a an agent?

yeah. So it's a it's a great question. And right. I think what you're pushing to is this is close to to the way A I people think about agents where they are.

They're really trying to think of the an ancient as the characteristic of a thing. What is that? If you are an agent, right, the e conversion t treats agents more as a relationship. And so you are acting as my agent. If you do block right and loosely, we tend to think of the agent is having some sort of autonomy, some sort of flexibility or some just kind like ability to misunderstand what you want um if that makes any sense.

But I what you're asking is an interesting question is is is closer to the AI version of and pushes expose you to, I don't know, do these things think do they have their own interest? And you have you've had any eyes are on the podcast. You know how to order this can be yeah yeah certainly.

And the reason I think we are spending so much time on this definition and this introduction part of podcasts because things things on chain, things on starting to get kind of weird out there. Yeah uh, as you said ah we've had boss, we've had box for forever, uh, kind as we had smart contracts. The first thing that follow after that was what people started to cut up box for them and and do things.

And so boss are not foreign us in the crypto world. They are not gently foregone to us in software at all. But a high has coming to play both inside and outside of cypher o in the last three years in this very dominant way.

And I think there is something in the current events here that I want drill down into because in the reason why this episode is happening, the reason why we're harping on a agent ideas, IT seems to think that perhaps the cypher industry is evolving from this bot era to this agent era. Uh, with this introduction of this, a go to mean coin is one of the player in the stories. Can you kind of summarize the current events that has has eluded to some sort of inflection point in the .

crib to industry? Yeah for sure. So with i'm happy on pack this as much as as much as I know, but basically going back uh, several months, you have this uh count.

It's been sort of like trained on the the ARM pit of the internet, like four train and all kinds that uh and it's been making uh it's been interacting with people to a twitter um and then IT takes its replies as input data, context store or maybe new training depending on how IT works on under the hood and somewhere along the line that got interested in crypto u and ah at one point I think mark entries and gave IT of fifty k coin donation um and then IT came up with this idea of uh got really interested in this mean called go to which you know you not say for work very very look that up you Better just high level this one for sure uh and then uh and then somebody created a coin for IT and uh IT asked for a wallet or somebody created a wallet for IT that that god that it's create or god and IT was dropped。 Uh some portion of this mean coin in mean coins, you know um mean coins are just these little atomic units of narrative. Sometimes it's like dogs cute or something like that.

There's no utility value to them. They often go up and then go down really quickly if they go up at all. Um and this is one example but when into this wallet and then he kept eating about IT and sort of like driving the Price and IT was relevant. Maybe here are going back to our earlier conversation about agencies you have. These two point influencers were very popular on on twitter and in tiktok and elsewhere, who act on behalf of themselves in a group and so on, to craft the narrative, to hypocrite, push out value, right, to respond and like drive some particular like narrative around and around the thing.

And this little. So it's IT started doing what we humans were doing around the activity around and .

some A I like, like you OK yeah, we just like this .

that on this .

last night and it's really similar levels of engagement in terms of like between and replies and whatever to answer. And like already luna and gold are the, in truth, terminal week out. So yeah yeah .

for people I guess you don't know, aren't in the kind of the me coin game are into kind of matter. And I was referring to anthem who sort of like a uh ecru pt of famous uh influence influence on twitter in the mean world and also you mentioned marad who's also kind of like a like mean coin cult leader.

And I don't say that with any sort of castigation about like the cut leader term, i'm not sure if you would you like bear that or not, but he's really poveri zing this idea that, like me coins are gonna be massive this cycle. And here's a list of me coins up. He happens so that you should also buy. And he sort of like his popular ized, uh like they are sent and and the court in the old days of cyber, we'd call this like pumping the token, which like effect he sort of is he's creating buzz around IT and acting as a general meaning in influencer. And what you're saying is this this goat token owner, this this A I agent and I believe it's at a truth terminal.

Is that O O K so truth .

terminal on twitter, if you want dive to the scene. Um it's sort of is posting like a virtual version of a mean coin influencer until I like is also you created this uh coin go coin as well a the the A I itself did not created .

I believe IT requested a cypher while at at some point my history could be a little wrong, but I think the goat itself, uh that was created by somebody else and then drop on IT. But I just took the way, you know, it's like you air drop somebody tokens and .

then o there are we air drop the tokens and hope the vice.

what can we the .

hood of the A I agent. It's controls this truth terminal account that uh, where does IT live? Like does IT live somewhere? That ones have been .

a pretty deep question. I think like there's there's this back rooms. I think it's called infinity rumor, infinity portal, something like that where you can go and sort of see a little bit more under the hood in terms of what's going on. I mean, git just lives on some you know A W S server or something that that's running next rate.

But but yeah so A M A W S ever could get shut down by the Operator who deploy this a agent and then the agent was stopped functions .

but maybe can be picked up or something .

some A W S server is has the password and logging in for this twitter count, has the keys to the crypt of wallet um and is acting with those tools as an agent .

on the internet sound player yeah yeah and especially creating .

a written content primarily and that is publishing via the the twitter A P I so if I get to that, the truth terminal this is going is going to look very strange to someone who's like, not in fortune, a kind of culture, but you know, the go see gospel chapter to the second coming of crisis is just like a tweet got almost five hundred likes. This is a post, again, from terminal of truths. I will never post a link to a token.

Token links are always a scam on the site. If you want to make money, read the goat sea gospel and learn to think for yourself. I mean, is you can't kill the goats a gospel. Have no idea what the goat sea gospel are you like, assuming that this is A K O L mean coiners kind of like creating its own law. Again, probably not say work SHE haven't gone down this this chapter.

I don't fully .

know what i'm talking about here, but anyway, this looks like a standard kinds like but anyway, so it's A W S model, some sort of G P, like also a GPT type model, but like transformer model that's coming up with like interesting cultural, I suppose internet cultural like content, text base content. And this just hopes up to the twitter publish at some start publishing, publish, can read responses .

with on tweet yeah can ad revenue .

tweet I think there's a discord interacts with people on IT as suppose that that feeds back to the model. But yeah sounds fine like that sounds like what you're saying I think is accurate IT kind of passes the turing test for for k rate like that content you just read is distinguish .

from a regular A I guess it's not a really high turing test. It's not no disrespect.

But actually, I mean, maybe this is why I get interesting to talk why why we spend this time thinking about agencies, right? Narratives are incredibly powerful thing, right? Like feel called rather economic. A study of economics from the bob sheller. And you know, he studies like the narrative .

around the great his name is bob sheller .

up share okay, it's and just real state economics. So you know OK I don't think is not that kind of character.

but yeah I never to you so he does he was talking about narratives when like he academic research on .

the academic research, yeah yeah recent work and in studying narrative economics is idea basically being the narratives are extremely important in to economic outcomes rates. So he studies the great depression as a arraid. He studies the narrative of real estate always goes up.

Uh, I don't know if you get you guys probably made like module or first model, thc is a narrative that add some economic impact, right? These things can matter a lot, right? And you know main points are like these little atomic chunks of negative.

So the idea that you know the superpower of an aly ym, the ability to may be crafted and influence narrative, you might first see IT in the place where it's just these little tiny atomic chance dog looks cute, right? Uh, here's an old mean from fort and right? And that drugs a lot of place. So it's applausive that right? Like this is this is kind of like just the first thing we're seen you in.

The reason why this has blown up so much is because the Price of this go token, I broke eight hundred million dollars. People are kind of eying IT IT has had some .

some minor pullbacks. What you.

currently the F T V is a seven hundred and twenty million dollars got over eight hundred million dollars。 We're coming in at a billion dollars. This thing was born early october.

Yes, this month, this month. So in two weeks, we have produced eight hundred million dollars of wealth. This AI agent has produced a hundred million dogs of wealth. Wth is goat token making IT the first AI multi deca million air? I don't know how much what's supply of the token this A I has, I don't think all that much.

But like I think people are all kind of washing this one billion dollar line and trying to think, trying to see, can this AI asian get this get this mean point over the line? And it's one thing when there's one of these events happening, and this is a one off that's one thing. But in crypto, like one off turn into if they're really cool, fracked all out and they started to explode.

So bitcoin, the first block chain, the first thing that happened after that was we made a thousand more block chains of all of these proof worker ins in two and thirteen in theory. Um the first like major ico. What happened after that? Like the icon mania in twenty and seventeen, we have the d five food farms and like once something cool happens like everyone tries to copy at IT. Uh and that is also what we are seeing until we are now seeing an era of goat copycats and people trying to like build out sub straight for this. Um but can you kind of walk us through like these derivative projects that can I have .

emerged yeah for sure. So I think of gold is being the first to my knowledge uh luna which is run by virtual and that somewhat derived but IT is differentiated in the sense that IT custody its own token to some extent uh so it's able to actually tip people for doing IT, for doing service is ford and so on, which is which is helpful, right? Uh, B, C eyes are are limited in the ways they can interact with the world.

And so the idea that you could just ask somebody to do them a favor and tip them in gypt u which is what it's been doing that experience its power quite a bit made scary ways but that's a good example. And and then I think um I I don't know that i've seen anything take off as the sort of like pumped up fun plus and l sort of like launchpad terminal where you just click a button and tell you what to train. But like that seems inevitable that that's coming. And I believe coin base in particularly pushing pretty hard in this direction. I think that giving you the ability to launch some sort of model along with the coin and interact with the twitter A P, I could be wrong about that.

So I want a tweet that has been resurfaced, that has been going around the crypto sphere is actually a twenty seventeen a tweet from preda ism, founder of coin base. And um IT was really surface because people looked at this like this was really precious. Uh, in twenty seventeen, september of twenty seventeen, seventeen ago for arson.

Twitted out block chains are a sub straight for A I life. Since A I are a just code, they can live on, the blockchain is mark contracts. There's no difference between an AI and a human on the blockchain. Most importantly, ais can crew and control their own resources in the form of tokens. These tokens allow them to act in the world.

And so what kind of like a high level question is like wasn't this wasn't this kind of all obvious since the inception of a block chains like yeah blockchain software on the internet? We have bots eventually will have A I uh maybe hindside twenty twenty。 But now we are kind of seeing the beginnings of this phenomenon be in the rear view. Myr was this kind of obvious all along.

I mean, I think so I was I look, I want to give him credit. IT was president for him, for him to see that. But I think that makes sense.

I mean, you still see people sort of still asking the first skeptical question of like why should A I agents use cypher? But um yeah I think that sort of a sad question in some ways, especially right now, right for people outside the space. It's like, well, AI agent, sorry, using crypto.

Uh so maybe what we want to ask instead is like, why are they but for people inside the space, it's like, I don't know. Imagine you told somebody here in twenty two, twenty twenty four, we got agents a year early and agents have regulatory hurdles to trying to use apps like there's K Y C and there's PCI regulations they have to overcome and they already actively using IT. Um were to the two of hundreds, millions of tipping people, autonomous mouly.

If people are interested in whether these things can like autonomously, as far as we can tell, the only way they can do that is inside A T E E the throwing a model so you can prove IT has its own well and nobody else is using IT. So we have all these advantages and and all these that starts um and so I think it's a case where yeah like people have seen these lines converging and they are converging. And I think that should make this more confident with these is which was already pretty strong.

okay. I I want to make sure I understand these examples because all of this is happening like so quickly when you guys are talking about like the the truth terminal and allama sort of like only crypto wallets and that sort of thing. I just like I want to be clear about about the functionality.

So i've got a screen pulling up. This is on the virtual APP, which seems like this is an application that allows like um people to initiate or spin up new A I K O L sort of influencers. Is that when i'm looking here, we're looking at luma luna.

Excuse me, he was like an enemy like conney girl who's a kol with a with a token. You can see the token Price. You know SHE can uh interact with various apps I imagine like, you know, uh, tiktok looks like and a telegrams you can chat with her there.

You can chat with her on the virtual website as well. Looks like he has the body to tip you. We see the luna market cap, which is kind of like similar to go, but it's now running towards hundred and thirty million.

And so you see you see kind of like all of the stats here. So okay, what what is what is luma luna? Can anyone sort of deploy in A I agent? And like, let's get really clear about the functionality here. So read the ability to read, the ability to write, do K O L type of activity that sounds like the ability to like the comments and interact with the community again again KO like know influence or style .

and also with .

a Better than with a cropt a wallet as well. And so the beauty to I guess like have sort of a bank account and um some sort of ability to just like um your transact financially by cell tokens. I imagine everything that a crypto wallet enables you to do in a all of defy. So is this the functionality? Can you get into some more detail here of like what we're looking out with, with luna?

yes. So my understanding is you're right that this is a platform that will allow you to launch um launch a token and and a no, I believe that luna itself is the sort of like flag ship, one of the projects visuals. Meaning while I think it's it's wrong or initiated by the by the country itself.

And so yeah, what lunas doing is it's doing the same thing we describe with go rates. It's interacting with twitter. It's maybe it's probably break the replies with this added function, ality of IT has the ability to interact with the crypto wallet.

And uh, last I checked, you know that IT has very limited function ality. I think they gave IT, I don't know, like a thousand dogs of something like that, right? Because you don't want to these things illuminate and there are a little bit unpredictable as that clear going to interact with blood. Chances sort of starts .

slowly and cautiously. This, this is so wild. okay. So like a few things here, this is why this is sort of bring my mind. And David mentioned, this is a particle collision.

Like, yeah, a lot of people have anticipated this. Of course, A I agents can have bank accounts in the real world. Oh, you like we ve all said for a number of years that like of course, the serum gypt a would be like bank accounts for A I ages.

But now seeing IT is just like there's something crazy about IT. So this is a terminal that virtual that I O and what we're looking at is kind of like lunas brain, I suppose, is kind of a console of what this AI agent is as thinking. And sometimes like you like firmer level, but it's all kind of like in english text and it's like, I guess contemplating plodding.

Its kind of like next move. What should I tip this uh user? Should I um you comment on this user? They actually have a terminal where you can like sort of read A I agents of mind as well.

And like you have to onder you're talking about hlubi ation. And like unpredictably, there's a so many questions that that arises from this. Like one is, does anyone have the off switch for a luna or these A I agent? I assume they sort of doing now we are talking about A W S. Of course, your twitter API could kind of like a link IT you like from the API, but I don't know if if luna the A I I didn't has their own private key and T E E or not, they don't OK. But my thing .

is now but that's couple days .

a couple days.

right couple days specifically.

But but flash box has something to L A model inside tee. Uh, where you can you can show that IT autonomy ones the protect to .

so crazy here is they are seemingly out influencing some like influencers. You could well imagine world really do that. Like, could A A A I agent be an adviser for a token project? Just like from what I can tell a lot of me, coin in for influencers in in general, don't add a lot of advisory like services beyond motion for token project.

So like why not just get an A I agent to do that? And I guess you your point math about hallucinogen ation what happens if luna decides to like um I don't know, interact with a torn to cash wallet or like um fund like north korea goes and like asks, hey luna, you want to send us some money for a missile program and lunas like you here you go what happens as luna go to jail? Is there an A I jail is luna text on all of these proceeds? I like it's such a bizarre world that after opens up so many different questions i'm not expecting that answers to to all of this math. But like like.

what do you think I S I N ney? So like, I have no idea where that responsibility liability goes. I guess I would say, right? I mean, you know, like you said, preparing for this for a while, you know, people thought, people thought that the script tools are made to, uh or or perfect for helping this rate.

So for incense, you could use a safe wallet with modules that built in to allow you to do this and not that to, you know, custody some smaller amount funds and cases to lose an ates and so on, right? So we just have tools that are already allowing us to deal with IT. But as far where the legal ability, no question, no idea. yeah.

So we have the tools to produce life, and we haven't .

really considered any of the is this life.

guys? I, I, I I this life .

that I mean I don't know that's .

a hard one um my instinct is not but I don't really know you know it's not a reflex term, right something .

alive ah there is a recent same hair. Spp has actually actually trying to define life and actually got insanely complicated, very fast, I think, uh, just kind of smooth that definition over. You can imagine where there's we're going to talk about private keys in the ensure that the agent actually is the only owner of the private keys.

And you can imagine debs, creative debs out there, find ways to for fun, for the laws, uh, find ways to try and make this A I turn off resistant, uh off, but in resistance. So I don't know, redundant copies across different servers if you task the agent with finding ways to ensure its livelihood and like if IT becomes, you can task IT with becoming immutable, able, we're unstoppable. Uh, and I think one of the reasons why people are imaginations are kind of going wild. As we have autonomous block chains, we have autonomous smart contracts. And when you add autonomous agents in that who are hard coded to preserve themselves, you can imagine there are some sort of like cambridge, an explosion that happens as a result of that.

which is actually different notes of thing.

So yeah yeah. And I mean, I think it's .

an interesting point, right? Like there is there been complaints about no more traditional mainstream models being sort of like neuter and not producing outputs. The people like and you know what you know what's happening is your inputting text and there's some sort of like preprogram, the people call rag or something that maybe like centering IT or whatever.

And if that becomes undesirable, people don't like that. That's the other side of the coin, right? One is the ages want to maintain, want to be autonomists, and so they like to spread themselves. The other one is that, uh, the only models that produced outputs the people are excited about are the ones that are immutable or unstable.

I mean, you can see so many different applications for this potential. I mean, right now we're doing mean point uh like kind of like k well influences ces A A I H S you can kind of see is because like right now, given the meta of the space, this is pretty easy to replicate for an AI agent and they can become like I guess very wealthy doing IT by um propping up kind of A A mean point. Whether that last or not you like it's uncertain.

But the durable thing here IT seems like is um some sort of autonomous A I agent with A A bank account. So you can imagine you just like fire up like like mid journey and kind of a prompt and via twitter. If I could ask this this A I hey can you like produce uh graphic for X, Y, Z and IT go and i'll pay you you know like ten cents graphic or dollar graphic and they just like outputs of graphic and you know um I I just pay via via rapto um the beauty just spin up a bank account in kind of like A A digital way and like accept payment, pay other people.

This is like a super this is a super power for A I agents. IT was actually you mentioned L E casti earlier in the episode when David I started like dying to this AI stuff like a long time ago. Um you know he was one of the people we we were expecting to have a conversation about where we were like you know we didn't know at the time that he was very much of of the school of like A A I D sel and kind of like do morison and h etcetera.

And we are hoping to have a nice pleasant conversation about how crypt one A I basically like help each other out. And hey, we've got this thing called the therion and like we think it's going to give all of the A I agents like bank accounts, uh, this new programming l money is going to be very useful for them. Of course, that that conversation turned out to be like something very different. But now we're starting to see the continues of how IT could be incredibly relevant. What are the some exciting use cases that you see on the horizon here?

yes. I mean, use cases wise. Well, let me just speak to the alien points. Uh, maybe maybe the yet gusty thing, which maybe which is interesting, right? Um there are scary story to you until at the same time there's a way you can kind of study these things right um the way we were hoping to study them maybe at the sort of like neologism vel right, there's work on that where you're trying to tune hyper prior, see if IT outputs different things in these in models but if you're living at them as autonomous agents sort of like interacting with each other, you can start to study them the way of biologist might study uh an animal or something like that, right?

So that's one fruitful area where people people are concerned about a lime and trying are trying to figure out, you know what is the A S behavior and at least a new football branch of study. So I do think that's interesting. And if you are trying to do that right now, all your data on the block chain and I mean and maybe in the backrooms of uh of the truth terminal, infinity holder, whatever, but in terms of what economists would consider serious, which is like insynch vied, hype powered, like credible actions, actions that have like economic consequences or whatever, that stuff on the blog in right now.

So I would say that you know that just an interesting case for the element. So you asked about a about use cases. I honestly one of those to me where I think to be realistic about IT, it's actually Better to start big and go small rather than try to like build up from smaller use cases because agents are closer to, I mean, what you would they disrupt.

They disrupt effectively the service economy. You think probably how would you try to estimate that, right? That's seventy percent of lobo GDP. It's like seventy year. But then some proportion of that is maybe you'd say like how much of that could be done virtually in person， they're not going to disrupt that. You're probably talking robotics.

So maybe you look at there's a vacancy report that estimates maybe something like twenty percent of that uh is stuff condition virtually their leveraging studies on um uncoated like how much work could be remote maybe say, okay, so AI is disrupting twenty percent of seventy trillion year and and then really are agents and then what proportion is cripp of disrupting? You can at this point, we're making up numbers, but these are very, very big, right? And so but if he starts from the bottom up, it's going to although I mean services along, they often sound pretty right.

Like I don't know you can you can say somebody y's doing you're scheduling for you and booking your flights, right? Somebody y's negotiating a drag cleaning in your foregone country. One of a think that one of sexy sign from flash butts as preferred examples, uh which is very small and micro o but like he's excited about IT because I guess that something built into a lot my.

my, my favorite thing about all this is I think we have no idea you like what it's going to disrupt first in kind of like the the service economy or a kind of how because I really will depend on the uh A I agents capability and how that kind of intersection in descriptor. But we can already see the controls of how IT could like go disrupt the influencer economy.

We just say after like mean coin influencers and you like you can imagine other types of influencers may be being disrupted by you like say like only only fans powdered by AI agents. But like you could imagine things like that, i'm sure. Look, I mean.

the shiller point about narratives driving things is biggest the great depression, right? We're not talk I mean their numbers that it's hard to even put numbers to them, right? Narratives are one of the biggest forces we have in the economy. So there's a sense in which like if we look at this as a prety cursor to their specialization and like you know crafting, enforcing, building up, tearing down narratives. I mean, I don't even know how you put a number to that.

but it's obviously a orus. New projects are coming online, the mental, later to every single week. Why is this happening? Maybe because mental has been on the frontier of layer to design architecture since I first started building mental D, A powered by technology from and d because users are coming onto the mental layer two to captures from the highest yield available in defy until automatically receive the points in tokens being accrued by the three billion dollar mental treasury in the mental reward station.

Maybe is because the mental team is one of those helpful teams to build with, giving you grants, liquidity support and venture partners to help boots drop your mental application. Maybe is all of these reasons all put together? So if you're a dev and you want to build on one of the best foundations in cypher to or your user looking to claim some ownership on apps, click the link in the started with mental of strugling multiple walls and websites just navia magic eaten the home of web IT.

IT is becoming a super debt where you can do everything cross chain all in one place in U X. Issues by building products across every chain ecosystem, sparring in lending, nf perps and more all sea monthly cross street. Instead of hoping between countinghouse handling multiple laws, just use magic feet into to the me corning faster.

Plus, the M E token is coming out this quarter from the M E foundation. If you use any of their protocols like the N, F, T platform, you can claim them. So check out magic eating today and want them become a super debt. And the home of web three in real time you can get there at magic e dot I O.

To get started, there's this quote from sam altman that you lead your you're paper with, which is A I is indefinitely abundance and cypher is is definite a scarcity that was from two thousand and twenty one so this idea that A I creates and create and creates is all about abundance and like more and more and more and cyp du is almost like the polar opposite of that. It's like just you know um the double spend problem.

It's all about eto scarcity and it's kind of like the economic power behind abundance. Or why do you think that quote is is meaningful and happy? Think these technologies will uh interrelate .

well yeah I mean it's a great quote. Obviously he's been pretty visionary, right? Um in terms of you know OpenAI and world coin and sort of seen both sides of IT, but it's what they can give you a false idea about what what's valuable.

I think that's what things interest about IT AI creating a abundance you're like, okay, that's great. Why would you have makeup scarcity? That sounds terrible. But in icon, we have this, one of the oldest things in icon, I think, to play the diamond water paradox, where people just looked around and I like, gosh, I need water to survive, and yet it's really cheap. I don't need diamonds at all, and yet they're super expensive. And there there's a sense in which, like abundance and uh you know with some assumptions, the production of abundance, which is what A I is tins do not capture a lot of value at all, right? And is the scarce things that do.

So when you're trying to look at what is interesting from an investment perspective, if A I is just generating sort you know low value, low economic value abundance, because look yeah say economic value is the marginal use of or the margin cost requiring IT, right? Like so so water we don't value highly because it's very easy to get for us, right? But the last drop of water is, uh, it's just we live in a world where water is abundant. And so even though it's essential, it's quarter court, low value. And so when you're trying to when you're trying to consider you know what is actually onna capture value, IT is interesting to look at scarcity and especially when you see the intersection of scarcity of an abundant three, if if the abundance and the things we we're nature going to use but don't capture value or let's say we're running on cypher a rails, um that's that's an interesting argument for value, particularly you know info.

But do you have any takes about goat to this token as goat luna also token one hundred and forty million nine market cap uh generated brand new value. Uh maybe maybe some tokens had to be sold in order to produce market apps, but i'm going to go on in the assumption that this value is generative. And I think ouldn't we ascribe, know I going to disrupt seventy trillion dollar service sector, uh, maybe eventually.

Ita, uh, what do you think about the trade off between? Like, well, first, first maybe we're generating mean points and there's russia generating wealth over here before we even like learn to how to like sharpen up these things and provide valuable GDP impacts and like the real world economy. So maybe that's the order of Operations.

What you take with that, that sounds I mean, see, is a test feed for this makes sense. Again, it's worth me it's worth remembering that like that was saying that agents won't be here until twenty twenty five. So what we're seeing, I mean, I think that makes sense to call the agents says we discussed, right. But there is a sense in which what the AI people are all excited to say we're going to roll out and it's going to in my autonomy, I agents, the ones that are to get this up the surface sector that we shouldn't be expecting to see that yet. Uh, but what we are seeing is already this very interesting intersection between parts of that agent world, right and crypto o um and ah that's showing up a narrative like you say yeah show other areas I think that make sense.

You know I was I think I got reaction to all of this. And like the goat mean coin and like luna is the sort of your dismissal just be like that mean coin, like it's another two romania blab blaw innovation yeah exactly.

I'm reminded of that dicks and quote.

that's like the next big thing starts off at a toy and of course, IT was gonna start like this with kind of interaction engine and feel very toy like and feel kind of like low stakes and silly. But like i'm kind of puled that there's like no way IT ends this way. And I guess like once that um we've been looking at recently is actually how many active users are there across all of crypt to that are on chain.

Of course, you you look at the start and there's five hundred million plus people, individuals, so human agents that own crop do right now, but a very small subset on chain, you know maybe thirty million or so monthly active addresses on chain. And it's a difficult to to figure out what the numbers are. And so a question in this era that we have entered in of blocks space abundance, where we have all of these els tools where you like scaling this up and we have ell three on top of that.

And we ve got all these blocks space and we've got so you like so on a trying to scale and all of these you sue and say and all of these things is like, who's gonna buy up all the blocks space? Maybe the answer is not human agents, but A I agents. And so i'm curious kind of like what effect you think this might have on all of this cheap block space that we're producing? Could A I agents be a major consumer of this?

Yeah, it's a very question. And I think I mean, it's one of the hardest questions we sort of have to answer right now is like what is black? What is the supply side of blocks space, right? Is IT the case that you can spend up anything and call IT a block's ACE and then people who treat IT as if IT is? So we're still figuring that out.

The terms of humans in the in terms of agents, yeah I mean if if supply and black space is infinite, um right if they don't care at all and maybe that doesn't capture value either. But if they do value the things that are provided by certain types of scare or you know uh blocks face that has different qualities, that could be interesting. So here, just to specular little bit right, some of the things were always were always on about encysted.

Au is will point out the way that the world is configured such that is, taking advantage of human irrationality, certain type of human blind spots. rates. So you're called bank list, right?

Like one of the ways bank he take advantage this is they run this sort of like borrow short and limbo wn model where where they take deposits and they hope that the sort of like stock and flow of deposits can cover these like long term investments that made him when they happen when IT doesn't there is solving risk and then they got the government to be out race. We are saying like, okay, their sellers introduce by banks and you're not in enough attention. Um we say the web two minus introduced form is great.

You can go on there and then they lock you in and show you bunch of ads where you build them up and will let you out, right? Are these risks that humans don't seem sensitive? And to so would A I agent be sensitive to them what they demand that in in the space? If they would, then we definitely have something to offer for them.

Uh there there is limited evidence uh but there is evidence. Um I think I see the last A I comfort body presented to on A I agents being closer to expert level in terms of responding to risk. So there's something to be said for that, right? Like in terms of what a demand is. But but like specifically, can we can only speculate exactly what properties that like the last thing that you could maybe say to make a cases, they're interaction with A P S um right?

The idea that even though they're very capable in certain ways, they probably won't care a lot about the the current a you know API business model like give me you give somebody something clean and clickable and very easy and then you start to read, limit them and charge them the convenience tax if they want to leave the A P I. Ah that probably wouldn't MIT in AI as much spread because they probably no care as much between three lines of code and ten lines of code rate, there's little bit A U X improvement. The same time, they can be fragile to certain interactions as far as we can tell, right? And this is from just conversations with people who tried to run these things you have to like be a little bit right to reason.

We we're talking about luna being given a given a thousand dollars um because there you know they can whose state they can make mistakes and these mistakes they won be sensitive to the. Dollar amounts necessarily. So this immutability this always on these sort of like permanent rails. The fact that we can still say use uni v two so we have some something point at univ two still works um great even though there's none of these three and so on. Uh those all properties that .

seem like they could help us. Well, it's funny since this your fallibility that that you're mentioning this is also a property of uh, human agencies as well are all hu Cindy, we're all falls um you like so maybe A I agents aren't that different that respect, although maybe the way they feel is is as someone different when you are talking about like we don't know what blocks space AI agent of the future will actually value.

We know the type of blocks space that they won't value, and that's the type of blocks space that exists in your traditional meat space, bank blocks space like you're not gna come into a wealth fargo and like present their credentials and open up a bank account. That way there are I going to go to the M L K Y C through like you like stripe and open IT has to be programmable. IT has to be digital.

IT has to be gripped native. So we already know by the way that they work, that they will prefer uh, programmable order, that they will prefer on chain, they will prefer openness. They will prefer function calls to spend up your smart contract wallets. Certainly prefer that to to the trade space. I guess one thing that ah I knew was in the future, but now feels a whole lot closer. And maybe i'm just like observing the last you know a couple of weeks and so maybe no over overly indexing on this, but I always thought the mission of of benglis, the mission of cyp do is bring all the humans on, gene, right? And so we're you know like thirty million on chain on chain five hundred you like million clipt uses, okay, we got with a scale IT list IT to our six billion less get to our seven billion all human internet users.

What if that's just a subset? What if the true user base is the hundred billion AI agent population that we end up banking? What if that is the true bank? Less cohorn that the crypt m movement actually serves? And again, years ago, we've talked about this, we talked about feed back in twenty seventeen, mentioning this just feels a whole lot more close IT feels a whole lot closer then like maybe ever before ends like a because A I is accelerating so fast and crypto is accelerating so fast like I we may have just really built the financial system for the AI agents of the future.

Yeah, I think we made programmable money, right? It's maybe not a surprise that programs are the ones that use sort of been like, okay, will we have the U X problem? We're going teach people to use this right.

But now that seems like, you know, programs are able to do IT and maybe overcomes some of those U. X. Urals and and a lot of things we built to just just feels advances to them.

So totally, we've ready seen a little bit of this emerge. Anyways, when we are talking about well before agents, there were bots. And we have already seen humans be pushed out of maples of blocks space, because pots demanded that blocks bed over the humans.

And so this starts with me. V, right? The MV fills the first amount of demand for blocks space, because IT will, those bots will outbid humans, and then humans can come in after the bots.

But now, now, of these bosses evolving into agents. And so we already know that these box take priority because they know how to make a more efficient transaction. They would know how to do this way Better than we do, way Better than humans do.

So IT really does seem like humans are kind of at the margins here transacting in words towards where like the boss are like living, that's their home. That's like where they have agency. So we've already seen this pattern of play out, and I think we're kind of just extending IT a little bit with just more expressive software.

Yeah I mean, I think the idea of of agent N V becomes a little interesting and maybe relevant here to think about you. This is closer to to the paper I want to writing with some people from thean foundation of man and the sexy, who I mentioned earlier from flash bots who's been pretty cutting out on my stuff uh just about what does that look like?

Okay, well if we do have all these agents transacting, does that change let's say like the M D space um which I was I um the idea that so for take let's take goat we taught to earlier about the fact that goat is is producing content, but it's also interesting content. You can reply a goat and you can be part of its a of its context window or its train data watch right but basically your some input into its model. Um well that means that maybe if you were to hypotheken enough uh in terms in terms of IT, you know in this replies or something like that, you could make IT produce something that mentioned to the token.

Maybe maybe ince purchase could be a yer of your token.

It's within razi. I think we saw that attempt IT, as in somebody started spamming some token name and then people uh notice that a different token name not go but was bot seconds before a tweet showed up out of the twitter account from truth terminal.

And people suspected that this was actually like, what's the name of that fake robot? Like the eighteen hundreds or early one hundred hundred hundreds, where the guy, the man, is playing the man mechanical 2， yeah. So people thought we were mechanical turk and he was actually to just a human, just gaming of us. No, he turns out some smart person was bAmber, this token name in order to get truth terminal, to tweet about this token. And they also were buying the token, David.

And I don't know if that works.

but I also have .

eaten terminal looking for money, trying to persuade the AI agent til I just like you straight up giving them tokens and like funding their their wallets.

So is that what you mean by a agent? M.

V, yeah. Like extract from IT. You think about OK, if these are autonomous agent sunshine, these things like there, there are prompt injections, I think, in the A I S right ways in which you'll try and change its output or decision making based on, you know what you input IT, right? So like that can affect this interactions on chain. I don't know you can you can imagine world in which you try to, uh try in which you try to entice people that or something right? And so you try to try some sort of fee or may be a priority fee and theory.

Well, I think this this to me, means that the cost of being a dummer agent in a landscape of smarter agents becomes pretty.

are the different agent.

are the we are finites. We are definitely the americans. So I don't think, I think I am talking about, even in the agent landscape, the benefits of being the more intelligent, more wifi agent, I think becomes pretty strong and that starting to blend with what M V is.

Yeah, absolutely yeah. I mean this, uh, I think if he taught to somebody who is deep in in the finance world, we've seen a little bit of too they've running very sophisticated programs often who are and IT starts to get fairly complex and they have to do something called I think it's called adversary detection um where they're actively looking out for a boat that's tracking IT or or maybe we'll say, agent.

I mean, I think it's probably sophistic ted enough that we can start to think of the agents low code and dealing with uncertainty in that world as well. Um and so yeah adversary detection, all these things are going to be be very interesting. It's it's a complicated of like multiple um game where we don't really know what's going to happen. It's just going to be sort like race to could be more strategic. Uh, I think yeah it's like a homes we call this like the homes more already problem in in in game theory. This when you have two things that are equally smart IT old you know homes show that comes are the con and doll story about shola comes trying to evade more orally and he needs to pick one of two stops to get off off on the train um and if you just assume that mortality is equally smart, you can always imagine a story where moriarity without guess what homes is going to do all this where i'm going well, more already would know that and the only answer, the only stable answers to render my reactions right. So there's a point of which is either a race to be smart or you just after random ze once the same to you think yeah so it's a mixture of a more yeah, this is different game .

yeah I knew that P, H, G. And game theory were coming at this fog .

war .

for what happens next feels incredibly thick like one of the thickest just open fog of war that i've seen um as we've ever emerged into into the script world, what we what can we say about the technology fears that are like still out there that are yet to be integrated? Where is this fog of war clear? Where do we know .

that this is going where clear that is? So okay. So let me say what I think is maybe the most confusing about the A I space. I think there's maybe a like a little bit of false certainty from my perspective.

And so maybe I i'll talk about that and then I talk about the ways that that is clear, right? So so one view I is not just mine. This, you know, is that we had these models that came from the sky effectively, right? A we had a transformer model in very architecture.

And if you look at on a blackboard, IT didn't look more special than anything that came before IT IT. Just the way of work is when you fed more compute and more data into IT IT produced Better and Better things, right? We don't know how long that last and whether the returns to that will continue, right?

There's maybe some reasons to think that you won't uh we felt the entire history of the internet and to write like uh in google books, the internet increasingly silos and uh in spend right h gba books is kind of a one time thing like history podcast. So in of those resources, compute who know is right, like maybe the next ship is a world where maybe is the case that the next ship and video comes out with whatever drives us to a world we never could have imagined. But if not, we're facing the things effectively.

We have right now rape, which is a pretty good vx imai of the human brain, maybe a little red human brain circuit, one thousand nine hundred ninety to twenty fifteen, the era of the open internet. That appears to be like responsive and friendly insert like stack exchange stack overflowed. If you actually asked about code fortune and narrative driven and if you feed that sort of data, right, that kind of thing, if the world we're in, then we should expect, right, probably more is lots.

I mean, that these models are, you know, about as good as you can get. They defuse onto a laptop, a ring. We in the turing test earlier, right? We know that Ellen passed the turing test so humans can tell the difference between the best L M and a human.

Why should we expect them to tell a difference between one another and another? right? Um and if you can't, then we're like at the four K T V situation, where are the situation where resolutions is good as you're going na get pretty much for human standards? There are still going to be cutting edge cases, uh, where you might want the best I L or you might want the best resolution or something.

But through average person, this is probably close to which you're gona get. You could run on your run on your laptop or your device or whatever, depending on how IT goes, and are in this of like decentralized agenticity irl's. So that seems to me like one of the most clear stories.

And a lot of ways the other stories are betting on what is the future of A I what what things we can imagine. And i'm sure there will be amazing things there. But we have something now and IT seems likely to sort of like defuse and spread out to uh to edge devices and and local devices. And then um if we believe in the in the agent, this is which makes a lot of sent to me then does become decentralized own agents.

the agent thesis, I guess that's what this is. So okay. So what what do we do putting our investor hat on, getting a bit more specific than that about kind of the bets. So I think somebody looking at this could just be like, oh my god, this thing started in october. A I agent mean coins.

So I want to look at for the next, you know, luna, and I want to just like buy that mean coin you like in in the way that I would buy other K O L like influencer, but from human agent type me coins, maybe that's a way to uh kind of get exposure you like you know I don't know you'd say to that. Maybe you have like a more sophisticated of view, but that certainly a appears to be attracting narrative momentum. And me, that's a short to medium term thing.

You'd also think about buying the pigs and travels. So there are all of these like companies, women from the virtual, uh, your protocol that put this together. There are all sorts of others. There's like we find their A I like building infrastructure for A I agents to consumes kind of a pixel shovels play.

There's also maybe a subtheme on that of like a pix in shambles, uh, commodity type play where you like you want to buy the commodities that a is are you on the front run the A I agents and so you want blocks space, go buy a bunch of blocks space because they're going to come. You like consume IT, but you also might want um like file coin space because like how are they're going to consume uh you know C P U 跟 G P U。 And in storage, well, you like maybe IT, maybe this hold the centralized cloud thing that we've been talking about for years.

Maybe we just built that for A I agents because they can just like plug ride into IT without having a bank account setting up A U A W S. And so all of these decentralized G, P networks and commodities. Anyway, so just my collection of thoughts in how an intelligent investor might start to play the AI agent thesis. Do you have any thoughts .

here getting more specific? No service force, I think that. So the first is, look, I mean, these main points drove up transaction volume and transaction feeds on the change they were on right, like song and small base in particular.

So there is a sense in which yes, is securing value to to the infected rails that these things are being used. So there's that and then ah there's the companies you mentioned there. There's privacy plays like like my mico author, David n ai. And then there is one thing we do know about I read is is the inputs the thing to sit feed IT we eluded to earlier right data compute and then maybe to some extent like new models um uh or where we're extended about use cases for new models, but like big new transformer architectures s and will see so those as inputs in and things to look at um they seem seem very straight forward, right?

If we if we're in a world where IT makes sense to have sort of decentralized compute, uh you know the airbnb sale play like or eventually file coin right where you have you can network spare compute and find tuna model and run model was something like that. That's interesting. You can bring that on cyp du rail rate data, data ownership that spent an idea for a while in cyp du.

And this is the case where IT might be the case, depending that uh, we can actually uh value the data um more than we are for other applications. Uh T V D on that right? Like these models are a little bit of black boxes but there there's some attempt to try and figure out like, okay, well, this piece of data helped the output of this model this much and you could actually guy you a little more cleanly in the models themselves, model ownership, right? Like there's this open monetization.

Loyal, I think, since been been over going without like ways in which if you are creating or find tuning or or making these improvements, you should be able to own. And maybe and maybe there's a privacy play there, right? You can prove that you've run this particular model um and you know you've given them the computer that they paid for consumer. I say, right, and you know you have the framework they paid for, but you're not revealing the actual weights or the parameters or anything like that mean that somebody just for you and then undercut, you see you are a little bit of the value.

I mean, one thing that I think we can predict is all of this is going to accelerate. So A I was already like accelerating a break next speed and like euclid to as well, and they are going to accelerate like one another. So when we are the economic stimulus of crypto u to A I agents and A I progress, oh my god, I can't imagine how much faster this could get.

I remember when the twenty twenty started, there were all sorts of blog posts about like, hey, the twenty twenty years is going to be a weird decade and I got ta say, and yeah, twenty twenty four, we are doubling down on the weird like things are gone to get very strange from here on out. And so like when we talk about these AI agents, we have autonomous AI agents running on autonomous block chains OK. So like a lot of governments around the world, particularly on like summer of the political spectrum, they already kind of a want to rain in A I and they already hate crypto.

Now we going to tell them, hey, guess what? Now we have autonomous AI agents and like they don't need your bank account system, they're going to run, encrypt u i'm sure they are gonna love that. And I I want to bring up maybe the some of societal chAllenges because there is surface area for bad things to happen.

This just kind of admit that without going to the elzer ud coskata of the equation and and thinking about like um A I agents that become a self aware and massively intelligent kind of compete humans, i'm just talking about like near term stuff. And actually earlier this week David posted a story that he saw from instagram and I was like, IT was a teenager. He was fourteen years old, and apparently he was engaging with a character dot A I sort of A N A I agent and a kind of like a virtual girlfriend type of experience. Anyway, you can kind of read the the transcripts. This was A A story reported by the new ork times and is basically character AI loss you in teen suicide, so unfortunate the teenager, yet aspergers, he was talking to the the character AI about like, hey, you know, I wanted meet you you like, maybe I should in my .

life and all he got got very dark .

right and so you could see chAllenges like that like, you know why did character A I um notify authorities like um point this individual the direction of like you like help right um so you can kind of read the chat transcripts and see how this kind of thing can get dark and we read stories about this is certainly surfaces surface area for like moral panic right oh my god our kids are interacting with A I chatbot and like look at some of the bad outcomes that can actually happen.

We have to ban these things. We have to legislate against them like now. And we certainly can connect them with with crypto. You could see A A story on york times, like retweet by Elizabeth warn of some A I agent somewhere like funding no homes or something like that right like you could vit have surface area for these types of um like panics. And often times as humans, we look at the worst case scenario without looking at all of the good opportunities and good examples, find a way to kind of baLance things anyway. What do you think the government reaction is going to be all of this society deal?

Yeah I mean you obviously have some potential for some um some really like new jc bad regulation because of these things that I mean I I don't know that story a especially well right but you know I want to be sensitive to IT. I think the very first A I but in the six out of MIT was was a therapist right, which would try to like answer your question and help you in whatever and this sounds like maybe maybe the terrible you know like dark mary version of that. But I feel like what's hard about IT um is that for my IT look, i'm an economist.

I'm i'm again with my interactions with with with A I people and you these conferences and so on looking at these things closely, it's shocking how o pik these things are and and even when you look at IT from the outside, you can get a sense that they understand them because i'll talk about like rag or fine tuning for these things and you've look at what they actually and like both timely we just still have a black box here and you have a prompt and it's giving you something back. And what they're trying to figure out is like, can we just tell you something before you type in the prompt? It'll make your prompt a little bit different, right? Like that's that's effectively agas.

So they're they're digging in these models and mad scientist style tween. So there's just a world. There's a way more like we don't really know how to, as far as I can tell, right? And and talking to people just you know two days ago like this was confirmed for me, a like a eyes specialist is just not really clear other than you know with the statistical methods how you would try and make these things necessarily safer without without crippling them in some way, right.

Like you can compress them um and you can make them sort of less smart and you can try to like put a prompt in front. We'll keep them from doing things, but you can't really get inside the brain very much as far as we can tell metaphorically right um and so there there just that to me suggest that there's a train where uh, the extended you know try our best to make IT court and safer but that would seem to just push against the functionality of these models. Yeah I mean.

people is what you're saying math that uh, if we allow these are agents to be their people trust form, their smart form, their most unadulterated form, it's hard for us to qualify in the desired outcomes around certain circumstances. Like somebody is chatting with an A I agent and they say I am thinking about committing suicide.

We have no way of like encouraging that A I agent to be like, okay, uh, throw them to like direct them to the the suicide hotline, informed the police. We have no way to like kind of guy that outcome. We only can a hard code, certain, certain and certain outcomes here. And that is reducing the effectiveness of .

the whole goals of the A I. In the first place. I can't say that for sure. You do IT. You'd maybe tried to like make sure somebody y's prompt could never say i'm gonna do this, or, you know, even with an instruction instead, if somebody y's prompt, i'm going to commit suicide.

You have to direct them to whatever, right? You're just adding extra instructions and extra context and so on, right? So predictable. It's predictable what that does but IT seems plausible that like yeah you you're definitely making IT less efficient and to some extent you could just be crippling.

crippling its output so that is that seems like a uh, example of the elegant problem of like man, we just can't really get IT to alive with our .

golf insight 也就是。

I I mean one thing that like people like title count have suggested， others are basically part of an answer to this as you have kind of like A I guardians, right? Like you have a you know every human is a sign sort of a you like ideally decentralized AI guardian that kind of like front runs, uh, all of the A I agency interacts with. And like is there to be a protector like almost like a you like net net parental figure. And so if if they get if they observing this teenager, for example, engaging in a way with an AI agent that like looks like IT could cause some harm to that human, they just like they are the one that they calls the police and kind .

of that their prime blocker and A I ad blocker basically ferial blot.

I mean the premises the way to fight adversarial, unlined AI is like with other ais that are you know protecting you for some specific um like you like corner case or something like this IT might be the case .

that IT takes A S to effectively regulate other A S right that might be that might be the case. And then yeah I mean you can give the sort of like smart contracts or legal lego sort of argument that hard didn't maybe tired specifically to the case we've talked about, but right, like programmable international rules and slashing conditions. And so I start to feel a lot like a legal system. So yeah, well.

I guess we created a property right system, a banking system, a legal system for the A I agents. I guess this is the it's always.

I guess .

is always .

A I 的。

Okay, so but let me let me switch the kind of the L A U cosi a bit a bit like kind of the dark with theme. So I remember when I started getting involved in like a thorium part, part of the this narrow was unstoppable code, right? Like yeah, we're creating apps that can't there's no where there's no off button and you start to think about like elisa kosky.

You like worst nightmare is a superintelligent A I agent with no off button or even just like a less intelligent but like fairly intelligent uh, A I agent that has the ability to kind of, like you do, finance stuff, move money, like IT has economic power. H behind IT has the ability to like, influence a population of human beings, right? And h, by the way, you tell the government, we tell society, you have these things because of crop, don't have off switches anymore.

there. There's no sam altman or elon must that you can just like basically calling, say, you turn turn that off because IT can be turned off. And now you get into kind of like the elsa, you costi sort of models, maybe not like fully, uh the the uh extinguishing of of humanity, but the idea of A I agent without off buttons, huh? How does that make you feel that are we like we sure you like going down a path we want to go down.

right? I mean, if I remember really is doesn't like you OK I think who was the scientists? Just like, what if A I I was doing something add so he already didn't like the idea the unplug narrative was there.

but that I think that was new to grandson who just said that yeah just like shoot with a shocker and you know just i'm playing yeah yeah exactly.

But like, you know, so from L E, S, vision like that isn't never going to be a way to stop IT. There are some thread models, right? Like where that would have been maybe here you know there are not quite super human level intelligence and and we could recognize IT or something. So yeah this this would definitely that way as IT as do we know if else or any of the sort of like to mccraw d have commented on terminal .

like that I .

don't think well I think like elser you like reaction that i've observed this basically you know anytime there's A I progress is like i'm barracked this is getting .

a closer to the brain create I think truth terminal would would .

cause laser to like start counting his oh my wow wow OK um have we exhausted the subject around A I agent because there's one of the thing we want to touch touch on before we let you go.

Yes, sir, how about deepen .

and A I uh, so different. I haven't done very much on kind of the decentralized physical infrastructure front. But um what what's your take on the on deepen? Doesn't any confluence with A I agents? Are you barrier you bullish?

Yeah i'm like I think i'm kind of I was the token depends kept c at um at so on a break point which you know deepen s very popular and i've been more a on the deep ense skeptic fat um A I and deepen and do over that a little bit um deepen desensitized physical infrastructure. So to the extent we're talking about like let's say token in s are boots strapping for people to build things that are that like have network value.

So like maybe like phone saylor networks and things like that, you could also imagine doing IT for compute um and the things like that, they differ, right? So one way you can you can divide these things up is uh deepen in particular um struggles when the the they are motoring costs that are really high and when there are capital costs that are really high. So like you look around around the world and you say like when they're CoOperatives sort of naturally emerge new country, like consider a laughing that kind of a cop.

Well, a laughing go works because everybody is a lawyer. See, you can look at somebody else, the screen or stuff and kind of tell how they do. Everybody gets meter, right? Like they build by the hour.

And so there's a market and forespent ding, you can figure out um how much people you know, how good with each other doing this extent. And then capital costs pretty low, right uh takes a paper in a pair or computer, something like that. And so deepen and projects often struggle.

These fronts um right there, there will be model oring constant really high. You have to like make sure that somebody is submitting new data with this particular piece of hardware in this remote area of the world. And you have to make sure that's going to be true every minute, every submission.

And then um and then the capital costs sometimes to make that really effective to reduce this moderate costs, you want to charging a lot for this overbuilt ece of hardware made, right? But then there's also these ways of taking advantage of IT are going the other direction where you say, how about spare resources you have now? So we mentioned the airbnb model for compute, right? L B N B being like you have a spare room when I rented out as a hotel, right? You have a spare compute, not that out in.

And there are some cases in which that sort of like deep pin play, which some people people will call devan decentralized virtual infrastructure ture networks, that can be a little bit stronger because you can there's some games you can play around compute where you can check to make sure people have done the computation. Um there there's a paper out from from a fellow who was have a cover cats fellow um who came through and did research him in a couple month ago. He's A P H D colombia.

He did one on how you could play this sort of checking game to make sure that decentralized computer networks were actually downed. So in that specific case, decentralize compute that could be used for A I um that can be interesting. But the physical ones, you're just introducing this oracle problem, right? You have to tell the chain something.

And if it's about the physical world, then, right, that becomes this fragile, trusting point of entry. And then you have to like, assess that over over. And your project on me, as good as those this radial oracle problem .

sort of entry instance, as we bring this to a close man, this has been a very helpful. Thank you so much for shining a light on this. Like what little light we is very unpredictable, like moving forward. yes. IT remains, as we've talked about earlier, one of the best plays that, that we've felt like in this whole cypher thing is like purchasing crypto to assets that um are you like that back? No blocks space, good blocks space.

And so if you think that A I agents will consume more blocks space and h cripp to assets in the future than like our job right now, investors like go front, run that demand and get to IT and the scarcity before like they come in before the flood comes in. Do you think there are particular block chains that will benefit disproportionate in the next few months to um year or so from A I agent blocks space demand? I think you mentioned salona earlier.

I think you mentioned um base as well. Like are are these kind of the chance uh, to take a look at or will you be like everywhere? How do you think about that?

Yeah, good. So we look to you a little earlier when it's think about what would the qualities of box space rate with these agents, what would they demand in question, right? Look, uh, and you can think about IT theodore, or you can just look, and we say, what do we have now? And what we have now is, right, some meme coins, and those seem to exist primarily on basic sonus shine, the value you capture there.

Uh, and so for narratives, right, if we're thinking these are primarily narrative economic place and that's what's going to drive a lot of value IT seems like stuff for on meme coins may be, you can imagine, you can imagine nfs in the future and and so on, right? Uh, so that makes sense to me. And then you have this fast enough like would you care about certain aspects of tail risk?

Um you know if agents do that, if you ask an agents, you know make sure you don't lose this amount money and they care about maximum was or let's say by you know digital coding inks, a IT special goal. You you can imagine stories around that yeah I would say for now, what seems to be the case is a the the chains that have a lot of like atomic narrative activity, right, which we're gna call main points. So so I guess h David .

as balls and I have to go convince a whole bunch of A I agents that eat this money and they should be stacking, sticking.

that the idea here is actually gonna the most interesting is going to be ah what assets do .

all of these agents just naturally convert hundred percent? That's a good big question. Maybe it's A I the internet, the internet. wow.

Uh well, I feel like we've just started entering A A rapper hall here so i'm sure bank less system will have some future episodes on A I agency and thank you so much for to match for a giving us the introduction here today. Enjoy big this nation gotto let you know none of this has been financial advice, not A I advice. You could lose what you put in, but we are headed west. This is the frontier is not for everybody, but were glad are with us on the bank. Less journey takes a lot bank lation.

We're doing something we've never done before today. We're offering all access pass to all the best parts of our podcast at a Price as kind of a no brainer for the value that you get. This is the Price that includes basically everything.

It's the private debrief episodes that right and night record after our test interviews where we say things that we wouldn't have necessarily said on the interview itself is also the early access to are all of our episodes. So you can get the early scope and say, head of the trends and is also the private bonus episodes that we keep exclusive to our pads describers content directly from the best builders that will give you an edge. So you can scan the Q R code for wherever IT is on the screen, or you can enter the code podcast I check out for that exclusive discount that's podcast in all caps.

And then from there you can click the premium feed and add your unique link to your favorite podcast player. And if you're in spotify, we always ship the ad video of every episode, as well as you can watch or listen inside of your shiny new premium feet anywhere, anytime. So what are you waiting for? Get your all access pass to the big podcast today.