Kalshi's Wash-Trading Fight and AI on Offense

Transcript

for innovation than distillation. We did a full report on this. It's public. It's free. Anybody can go read it. If you read the Deepseek R1 paper, they did unique years ago. Like they trained on second rate hardware. They did mixture of experts. Like they're doing cool stuff. So open anthropic want a regulatory mode because Chinese open source models are getting really good. And they're not only really good, they're really cheap to get 90% of the intelligence for a tenth the cost. And it's not even that you're getting 90% intelligence anymore because you could build your whole business and enterprise architecture around these models. Like at Deli Ventures, I have a insane news research Hermes agent that's dense.

That's all my memory deals, evals, founders notes, everything. So I get even better responses than I would using normal open source. Like that's really scary to open anthropic that want to offer, you know, enterprise sales and sign seats and stuff like that. you have the the things that we don't like, right? Uh like the SEC when they're making bad rules and now we're like, well, at least the SEC is allowed to make any arbitrary nonsense that they want. So, we're kind of saved by that. And it's like, man, like be careful what you wish for. Like, I'll I'm here for I'm here for the SEC making crazy pro crypto rules. We we need to balance the ledger a little bit clearly, but like that's not the most sustainable strategy for us realistically. um you know

relying on on the SEC rulemakers to to come in and and you know do stuff. I mean it is it's like it's harder I guess for you know another regime to come in and like completely tear that down but like this regime came in and like stopped the SEC completely from doing anything let alone making rules right so you know again I I just I do worry that like we had a chance there was a chance to actually get some laws written and we somehow squandered that chance. you're using a travel agent, you're not paying for the hotel directly. Like, you're not paying the hotel directly.

>> Well, I mean, I guess you could, but yeah, in this case, you're not. Uh, you're you're doing a sort of a private concier, so to speak, because obviously for something to be like a store of value, you need to hold value in it. But then you need to also be able to do things with it to an extent like things backed by it. I'm upset that like not everybody uses it, but I think a good enough amount of people use it such that it does make a difference is sort of like if you use the Zcash wallet, they all have the pay feature which integrates near intense which then lets you pay on any chain, right? So you can also I also do salon USDC with this for example or you can do Ethereum or you can brush to Robin Hood whatever and it's just super convenient and like it's

not like a regular bridge where you can just go back and be like oh it came from this other address. Okay, now my private concierge knows how much money I make and therefore now they're going to upsell me on all these sort of different >> importantly I think biggest unlock has been both Zcash and Mir bringing at last privacy on chain to me it's always been kind of a biggest bottleneck to use day-to-day blockchain for anything because like I have literally a spreadsheet of 100 addresses on Ethereum and kind of near before we had privacy so I can track where who did I send money which address did I use how much ass I left on that address like it's like a it's a nightmare of like accounting system to just keep track of you know to make sure I don't link

anything you know go through sex you know make sure I don't have timing attacks on like amounts etc and all that it gets solved now that we have zcashioner really delivering on kind of like how do you have a private experience using blockchain so I think like all these things really coming together now uh as more and more people actually starting to use blockchain financial system is moving to blockchain kind of never before and so at least that's my explanation why things are really uh really starting to appreciate. >> Well, you're breaking the news to me. Uh so, I'm pumped to hear that this is out. Uh it's unsurprising. We've known this innovation exemption has been worked on by the SEC for months now and we're ready to go and there's been some fits and starts. Um but I think the the most compelling thing here is that it is

illustrative that these agencies are now going to move like gang busters. That as soon as this bill failed, they had pages. We have met with the SEC. We've met with the CFTC. They have binders of proposed rules and guidance ready to go to implement as much of the ethos, objectives, goals of clarity as possible. And this isn't going to be a short-term effort between now and the end of the year. I think this is going to between now and the end of the administration that the next two and a half years are going to be a onslaught of proposed rules and guidance from these two agencies. And I think the innovation exemption was just the start. I think they're going to try to see it and push the bounds of their authorizing statutes. implement as much of clarity as possible through these two agencies.

I think eventually over time as this rulem really takes hold and to your point crypto has this foothold in the US then it becomes easier for Congress to say you know what let's just codify some of the things the SEC and CFTC have done and we're not trying to come up with new concepts new ideas you know in this bill it was it's all new concepts it's the ancillary asset test it's the network token test it's the decentralization and name only test I think what will happen over time as this industry matures and if they you know if truly people get behind the SEC and CFTC rulemaking that over the next few years, you will just see legislation that codifies what those agencies have already done. It's why so it's so critical that these agencies get it right. I have all the faith in the

world. Uh no two better people than Paul Atkins and Mike Sealing to lead us here who know the industry so well. Um and then Congress, you know, it becomes less partisan. It becomes uh less political. Probably becomes less newsworthy. Hey, we're just going to codify the innovation exemption. So we so it stands the test of time. I think that's the most likely scenario at this point. >> Why do we need to use a mouse anymore? Why do we need to use a keyboard anymore? We just will and summon what we want and the AI starts doing it. We're moving at the speed of willpower more and more every day. So I think this AI agents is a is a extraordinary development. It's why I think semiconductors are a great opportunity.

Uh and Meadow is first out there. But yeah, Google is coming. Google Spark is in testing right now. It's going to be compelling. Apple sees this. Everyone was going after the personal assistant AI. It was the obvious thing. But there are other AIs coming. The personal concierge, the personal doctor, the personal wealth manager, what we're building at Luma, all of these niches are going to get filled out. So I think it's super exciting if you're long the AI trade and we're going to be mainlining AI now to consumers. Right now it's been mostly infrastructure and compute story. This data center pol very abstract concepts. Now consumers are sourcing tangible value like when they first got their AOL CD and went on the internet, right? That's what's happening now with these personal assistant

agents. Historically, governments have always picked inflation if they can, right? We've done that in the United States repeatedly. If you look at the history of the 40s and 50s postwar, financial repression. If you were to go and then look at the 70s, financial repression, right? Like every time we get into that position, we end up in that way. And I think with debt to GDP where it is, there's a sort of counterintuitive thing that starts happening, which is if you have $30 trillion of debt outstanding, right? I will remind everybody 1% of the debt is $300 billion. If you raise rates 25 basis points, that's a lot more free cash flow in the hands of Americans given that our debt is relatively short-term and rolling quickly, right?

Like there is some bizarro like inversion of thinking that happens when the debt gets big enough where rate cuts are no longer restrictive but in fact just spray more cash out as well. >> Before we get started I want to ask you for something. We're making a bunch of decisions right now about Unchained. What we make next when we go live what we should be doing more of. And I'd rather hear from you before we make them than after. So we put together a survey. It takes just a few minutes. It's anonymous and there's a box where you can tell me what we could be doing better. I read all the responses. The link is in the show notes or you can snap the QR code on screen or go to unchainedcrypto.com/servey.

Thanks. $540 million. That's how much concentrated liquidity sat idle in a given week in the first half of this year. About 30% of the DeFi TVL if you're wondering. That's according to Dune Research commissioned by 1 in. But there's a solution. 1-in Aqua is the new shared liquidity platform. It lets LPS back multiple positions with the same token balance and keep their tokens in their wallet till a swap comes. Why does that help? Because the LPS don't have to split their tokens across positions.

They can cover more market conditions and pairs with their full balance. That means more activity across deeper liquidity. See how it works at 1in.com/aqua. Remember that providing liquidity carries risk and fees aren't guaranteed. Heads up everyone. We've now fully transitioned Bits and Bips to its new dedicated channels. So if you're not yet subscribed to the new Bits and Bips channels on X, YouTube, Spotify, Apple Podcast, or wherever you get your podcast, go there and subscribe now. For the next few weeks, we'll play segments of the full interviews on Unchained as a reminder. But for now, you'll only hear those short segments of bits and bips on Unchained. No more full episodes. You can get links to all of the dedicated

channels with the full episodes at unchangedcrypto.com/bitsandbbits spelled vips. Thanks again. If you hold crypto on your phone, your biggest vulnerability isn't your wallet, it's your carrier. AT&T, Verizon, and T-Mobile have been breached again and again, and SIM swaps are still one of the easiest ways for attackers to drain accounts. That's where Cape comes in. America's privacy first mobile carrier. Same premium service, but Cape rotates the identifier on your SIM every 24 hours, deletes your call and text metadata after a day, and protects against SIM swaps with a 24word recovery

phrase that only you control. Hey everyone, I'm Jamie Mark and welcome to Unneasy Money because what happens on chain never stays on chain. Before we begin, here is a word from the sponsors that make this show possible. This episode is brought to you by 1-in Aqua, the shared liquidity layer from 1 in.

Back multiple liquidity positions with one wallet balance and keep your tokens in your wallet until a swap fills. See how it works at 1in.com/aqua. Before we get started, I want to ask you for something. We're making a bunch of decisions right now about Unchained. What we make next, when we go live, what we should be doing more of, and I'd rather hear from you before we make them than after. So, we put together a survey. It takes just a few minutes. It's anonymous and there's a box where you can tell me what we could be doing better. I read all the responses. The link is in the show notes or you can snap the QR code on screen or go to unchained crypto.com/servey.

Thanks. >> All right. I'm here with my co-host Taylor Wanahan, security expert, and we have a very special new special co-host. >> Double special. >> Double special. We love when you're a guest, but now you're stuck with us. Uh so, welcome as a regular uh host to the show. Um we're very excited to have you. Um >> yeah, let's thanks for uh let's get into it. Let's get into it. Um we got lots of stuff to talk about, lots of AI stuff.

So, uh but we'll get through the crypto stuff quickly. Um and >> will we will we really >> we won't we won't. All right. So um Koshi uh got themselves into uh timeline warfare which like was just in like this felt so inevitable to me that this was going to happen because they're so not crypto and they hired like genuinely some weird crypto people. um like the the people that you would hire if you had no idea about crypto and you're like, "Ah, that's a guy on the timeline." Like, he's he's the guy. I'll

just go and hire him. He knows crypto. Um >> yeah, he has ICO in his name. He's been here forever. He clearly knows what's going on. >> I'm sure he knows I'm sure he's he seems like a good guy. Um so so uh so they hired a bunch of uh people um and you know like some of them uh like I I should say you know I say this every time right but like I'm a I'm a poly market seed investor so you know my bag bias is very heavy here um but like the cowshi guys are kind of rats like let's be honest right like everything that they do is kind of rat mode right like from like trying to [ __ ] call the FBI like just all kinds of like weird [ __ ]

that they've been involved in and like just every possible sketchy thing that you could come up with as like a game plan they seem to come up with. Um and so yeah, of course like they're, you know, [ __ ] going to get themselves into some pitch battle. The interesting thing is I think that for like most of the last I don't know let's call it like uh 18 months while they've been like main characters uh fighting it out with with Poly Market um and uh and prediction markets have been like you know uh huge huge this cycle they haven't really gotten into fights on the timeline because they're not really crypto people so they're not really on the timeline but then they deputize these crypto

people to be on the timeline and they manage to get themselves into a fight with some cracked autist guy who's like I'll te I'll tear you apart and teach you a lesson. Um and so that's what's happened. Um, and so this this uh guy Benny um got into a fight with ICO Beast and ICO Beast was like I don't know said something to him and he's like all right I'm going to go and like analyze all the data that's all public and like just went to town on them. Um and so so you know the the allegation is like wash trading um that you know it's all fake volume um on their per markets

uh that you know there's some fake volume on on uh the prediction markets etc etc. So there's a bunch of things and they've looked at the data and like you know the numbers don't sound amazing right? So, like the ETH per had $3 million in open interest and half a billion in 24-hour volume, which like I don't know, man. Anything's possible. It could be someone really is not confident in their eighth position and keeps, you know, FOMOing in and out uh over and over. I mean the the one interesting thing is like you know um I think the

the the only possible way right and and I think this has also kind of come out and and people have talked about this the only possible way that this makes any sense is if it's incentivized volume right like not everything I think it's worth calling out not everything has to be wash trading right if >> not everything has to be the exchange or Kulshi the prediction market that's that's one of the hardest things to understand about crypto is like if it's incentivized >> like I don't know that they can do it their own exchange honestly like so I don't know if it's them right like I I I highly doubt that they've got some guy in a basement like clicking buttons like

wash trading right like >> I mean it would be wild that would be like some Mount Gaus era >> it's just not it's not that it's not that >> to have to people that are engaged with the CFTC, they're doing disclosures to the CFTC to then also have a back room where they are like running their own bots or something >> would be >> completely insane. So I don't but that doesn't mean that it's not happening. >> It just means that >> you have to be careful with like saying that that Kohi proper is the one that's doing this versus >> there's some market structure. I just think it's it's extremely unlikely that they and why would you do it, >> right?

>> I mean, it's couch. >> Why why would you do it when you can just incentivize other people to do it legally? >> Exactly. Exactly. So, so you know, my guess is that that they have some uh you know, incentive scheme, right? Some like, you know, negative [ __ ] uh you know, rebate scheme, whatever, right? Where like the more you trade, the more money you get or something like that, right? um you know like if you were if you were to set up a system that was designed to generate a lot of volume in the most efficient the the funny part about this is right like who are the people who would generate half a billion dollars in volume off three million in open interest people who are hyperefficient who are hyperefficient

market makers that's their job to be as efficient as possible in you know what they do right so if you got two market makers and put them in a room and said hey guys I'll pay you $1 every time you cross each other. Those guys are going to rip your face off. They're going to create so much volume you wouldn't believe it. And and and and right now the question I guess right is does that have any benefit to an end user right like are these two market makers because that's the difference between wash trading right even if it's one per one person two parties colluding together or whatever. um and like incentivized liquidity, right? If if these market

makers only ever cross with each other, if they have some system for ensuring that like they only ever cross their orders, they're always delta neutral, they're like, you know, constantly going back and forth and they say, "Okay, like every 37, you know, milliseconds or whatever, we'll like both post an order and like, you know, if if that's what's going on, obviously that's going to be bad." and like someone will find that out and like the CFTC will will not be impressed by that. Um, if it's just like ah yolo, we don't care who the [ __ ] we trade with and our job is just like generate as much volume with like the least capital possible, that would actually be fine for users, right? Because there's always

a market maker who's kind of sitting in the books and going to cross with you and you know it's not like someone knows anything about the ETH price, right? So, um, so I think there's like a bunch of ways where this could be like market structurally fine even though it doesn't look great. Like the optics are not amazing. Um, and and you know, you just have some market makers that have like negative rebates that are just sitting there like trying to max extract from Koshi, which again I'm here for. Like go nuts. >> The liquidity is one thing, but are they like getting to the answer? Like the goal of the prediction market is to find the answer.

>> Yeah. Yeah. Yeah. >> None of this helps that, right? Like just asking the question, smooth brain question. Like maybe they're providing liquidity, but it seems like it's going back and forth on a previously found neutral point. They're not actually getting to the truth. >> Well, this but this the the milliondoll billion dollar20 billion dollar question, right? Is like is this actually going back and forth, right? Like you imagine a system where like all of the regular orders are here, right? And these two guys and like you know we've seen this many times before, right? In centralized exchange you like you know how many times we read a report, right, of like I can't remember exactly which one it was, but there was some exchange uh one of the like uh transaction mining exchanges from like

2018 FCoin or something like that where you've got a an order book, right? And magically every cross was like 150 bips from mid-market, right? So like basically the only time anyone ever traded with each other was like way outside of the actual price, right? Which makes no sense. Why is there someone who's always showing up with this huge block trade like way outside of mid-market, right? And so you know there there are uh like many ways that you can do this, right? And the reason why you do that is to avoid actually ever >> so someone else doesn't take it first.

Yeah. Right. So you're just like bang bang like just bouncing around midmarket. Right. Um >> it doesn't seem like this is what's happening here like they are actually crossing at midmarket. Um and if that's the case then like if you for example Austin were to turn up to this market and you're super bullish on ETH right? and you're like, "This is the most liquid place to trade." And you just keep clipping in 50 bucks, 100 bucks, a,000 bucks, a million bucks, whatever. The market makers will keep crossing you, and they'll go and offload that somewhere else, right? Like they're they're not they're not, you know, going to uh be having a directional position against you. They're just going to go and offload it on purp. So that's their

job, right? Udeally, they would like to not have to offload it anywhere. That's way cheaper and way more efficient. And so if it is directly not moving, that's great for them, which I mean e like hey, you know, directly not moving, right? Like um so so you know, it's probably a good asset to pick, right? Um anyway, I think I think that like the reality is it it probably looks a lot worse than it is. it probably is fine from a user perspective and like does actually generate liquidity um and and makes it an okay place to trade for users. Um if it doesn't like it's going to be very easy for the CFTC to figure that out

like they they have some seats, right? Like you know if this is the Fcoin style like hit you know like 150 pips above midmarket, right? Like if magically they're always selling these ETH you know, positions to each other at 3K or something like that while it's trading like 2,800, then you know, someone's gonna someone's going to notice that. So um >> yeah. Yeah. >> I found the thing that kind of and when Koshi came out and sort of like did this big response, they didn't really address this. There were so many irregular things relative to other markets. Like they looked at like Binance and Vivbit and I don't know if they looked at Poly Market but they sort of like looked at

like these different they pulled the data for these different things right and then they sort of like they graph them and they use different algorithms. >> Well they try Okay. So >> but like the ones look weird right >> under >> Oh yeah, I saw that. >> Crypto.com. Come on. >> I didn't even know they had this. Remember when Crypto.com was running around like in in the middle of the bull market being like we have the highest volume on like the Bitcoin markets and everyone was like sure guys like whatever like congratulations. We did try to throw crypto.com under that was >> but the the but this is why it's hilarious, right? If you were like no no no look at these like great examples of

like efficient market structure crypto.com. They that's our books look exactly like theirs. Like they just have no idea what they're doing. You would not use Crypto.com as your example of like >> No, it's completely mad. >> And by the way, Crypto.com's response was like absolutely impeccable. They were like, "We applaud their creativity or something." It was funny. >> If you find yourself on If you find yourself on the side of Crypto.com in an argument, >> like something's so wrong.

>> Something's so wrong. Like you really >> I wonder if Calli thinks that Crypto.com is like >> They do. They think >> I mean here in LA, by the way. Here in LA Crypto Arena. I know. >> Yeah. Yeah. And on all the sports stadium like all of them because like what other Yeah. Like when we go ice skating >> it's called It's so funny cuz my son and I my son and I watch UFC, right? Um and and so like it's always like Crypto.com Arena. Like Crypto.com's everywhere in the UFC, right? And you know he's he's six, so he knows crypto, but he really thinks that like Crypto.com owns crypto.

Like all of it, right? Like why would it wouldn't be called Crypto.com if they didn't own crypto? Like >> your six-year-old and Kelshi have about the same understanding of the crypto you you join those dots there, right? >> So what about I mean with all this regulation with the CFTC stuff, is there any uh advantage that Poly Market has by being on chain and actually more decentralized? Like that's what something good should come out of actually being on chain and something bad should come out of if you're highly regulated and centralized the decentralized player should have an advantage here.

Yeah, I think like you know the the the situation here is like both of the books are pretty open like you like these there's public APIs right like now you know again it would be wild if like they were manipulating the data in the background and like publishing fake that like I don't think they're doing that right like you know as as Tay said right like they're not doing you know like backroom crime right like that's that's very obvious um but you need to do backroom crime to like make your numbers look better than they are, right? Um, you know, there's a bunch of ways you can stay in gray, you know, in the gray area, right? Um, so I think just think from like a user

perspective, it being onchain, at least you can see the full receipts, right? It's one thing if I'm like, "Oh, here's my API, bro. Like, >> look how beautiful my books are, right?" It's another thing if it's like, "No, I can actually see um, you know, but also being on chain doesn't mean that there isn't washing, right? doesn't mean that there isn't you know uh fake volume and and all kinds of things like if you if you incentivize volume you there's a question there's a question of like if you incentivize volume is it real right and I think like my uh my view is it's real if the impact on an unaffiliated user is they can come to that venue and get a good fill the best possible fill

in the market then it's real Right? And so you're paying for that. You know, if the volumes, whether the volumes look good or bad or whatever is somewhat irrelevant, right? Like as a user, Austin wants to be able to turn up to the poly market ETH per or the ETH prediction market and express his extreme bullishness on ETH. And if that's the best place to do it, then good. Like then Austin's happy. that's that's he doesn't care about like the other market structural things there. He just wants to be able to to you know uh express his eth bullishness.

>> But but on the aggregate and going back to finding the truth if there's a lot of liquidity and it's meant to be pinned to the status quo when someone comes in with inside information or actually like knows enough to move a prediction market. It sounds like with this current setup it would move less like you would get less signal from noise. >> That's good. But that's the thing you want. You know the the deeper the market the more accurate the information right the more liquidity the more accurate right like >> but it sounds like this liquidity is around status quo not around information.

I there's a question of like you know if you know some let's say you know right that uh the that uh Tom Lee is about to go and buy a billion dollars worth of ETH and I'm not suggesting that you know this I'm just saying if you did blink twice but like if Tom if Tom Lee is about to go out this week and buy another billion dollars worth of ETH right somehow he scratched up some cash under the the cushions or whatever. And he's >> I'm worried about my blinking, dude. I'm I'm not blinking.

>> You're like, "Don't blink." >> I don't I don't know. Tom Lee, we don't hang. I'd like to Tom Lee, reach out. Let's chill. Yes. >> Yeah. Like Austin's good. He's a good guy. >> I got a whole crew of builders, brother. We could We could make it rip. >> You can cook some stuff. So So Tom Lee's about to buy a billion dollars worth of beef, right? Um and you know this, >> right? So the price of ETH is going to go up for sure, right? And so you want a place that allows you to buy as much exposure to that upcoming ETH. That's true. Okay. As >> possible. Me as a buyer, I get a great price for it because there's a ship.

>> You want if you could if you could put a billion dollars into the poly market, you know, is going to be over 3K, right? Then you just load up. You keep loading up. Ideally, right? This is the irony of prediction market. It's like for the actual person expressing that information, the deeper the market. Now, eventually you will move it, right? And then eventually people will be like, "Hey, wait a second. Austin Griffith just, you know, Austin.E just put a billion dollars into the E." Like, what does he know, right? Um, and then the world knows that whatever you know, you've now expressed and now the world knows it, right? Um, if that market was super liquid, right? Then there's probably no point. Like if you put $5 in, right, and it moves the market and ETH is now going to be above 3K, like

who can trust that information? But if you put a billion dollars in to move it over 3K, then it's probably real. You probably are Tom Lee. I mean, we've never seen you and Tom Lee in the same room at the same time, so I don't know. Um, so yeah. So like I liquidity is is good for our ability to trust in prediction markets. The more liquid it is, the more real the price is on some like very, you know, critical dimension. I I I'm okay. Let's let's say that there's like a sweet spot here where you need at least enough liquidity so you can buy in if you have an idea and it doesn't like move the market a great

distance just based on you have a feeling, right? But also like you it should move the market when you have an idea. So you see the truth of the thing. So I think that there's >> assuming you do assuming you do, right? Assuming you know something, right? And you know like >> yeah sorry go >> no no no go ahead >> so like take the price like let's the most efficient market that we have in the world that we've come up with is like the New York Stock Exchange NASDAQ whatever right like it's it's the it's the most hyperefficient it may be like arguably like interest rate swaps or something like that but let's just for

the sake of something that's real right the price of Google is like one of the at any time in the world, right, while the markets are open is like the most known thing in terms of like the level of confidence that the world can have and like that's what Google's worth, right? Like the market is hyperefficient. There's like trillions of dollars slloshing around to try and like decide what the price of Google is, right? It's a prediction market. That's what it does. Um, so if you turn up and you're like, I think Google is actually worth $197, not $196, like you could be there all day bidding that and it's not going to move the price, right? Like it's just going to sit there because you will never have

enough liquidity to move it, right? Um, but each time you try, the market is pricing in that like infantessimally small extra information that you're expressing through putting your dollars in, right? It's just such a deeply liquid market. And the reality is you don't we shouldn't believe that you your theory that Google's worth $197, not $196, is a real thing. Like it's because it's not, right? Um, the market has decided what the price is. So, so you know there is this um like very real trade-off between like if someone does know something now, you know, if Eric Schmidt turns up and says like,

"Hey, I'm going to, you know, dump a bunch of uh Google shares, right?" It's not even necessarily that he's like expressing an opinion, right? Like the information is intrinsic to him selling whether he knows anything about like what's going to happen to Google or not. like he is moving the market right by selling he's got enough shares. So everyone goes, "Ah, let me get out of the way of that. I don't want that guy, you know, dumping into me, right?" Um, so for that period, the price of Google is worth less because there's more, you know, sellers than buyers, right? And there's one particular guy and and the the more liquidity there is, like if Eric Schmidt could literally dump all of his shares and the market didn't move, you would say that's actually good. like

the market is so confident that his dumping of his shares actually has no impact on the price because the price is already known, right? Um and so you know that that that's kind of the core trade-off here. I think like the more liquid something is, the less it moves, the more confident you can be that the the the market is efficient and that we have like efficiently arrived at that information. So when you have very inefficient markets, uh, you can't really trust it. Like one guy who thinks they have inside information who doesn't Yeah.

>> can move the market. It's not even real, >> right? That's bad. That's a bad thing. And now we have imperfect signal and we're like, "Oh, that guy's an insider. He knows something." No, he's just [ __ ] It's fine. Like know anything. >> Yeah. Well, and a lot of times like the market makers are sort of they're trading on a variety of different signals and also providing liquidity. So, they're making it deeper, but they're also like they are aware of the market, very aware of the market, right? So, it's not just that like because there's deeper liquidity, an individual doesn't isn't able to move the market as much. It's like that's that's how it should be. You know what I

mean? Um, and if it is a real signal, meaning that, >> you know, perhaps there's like multiple people that do have, you know, good intel, it's not only their money that then moves ends up moving the market, it's also the market makers like interpretation of that of of their actions, right? And so you sort of Yeah, it keeps going on and on and on and on and on. Um, but ultimately, especially with the markets, you do have to have someone on the other side and >> there has to be a buyer, you know, someone's buying Eric Schmidt.

>> Yeah. >> So, if there's only like four people trading Yeah. And I don't know if like I always have this experience when I don't know if you guys have done this lately but there's like certain markets when you're trying to exit to like USD like fiat fiat USD you it's like so stupid right you have to go to the but you go to the centralized exchange you either are selling your ETH for fiat USD not USDC like fiat USD right >> or your USDC or your die or your USDT for fiat you'd think that it should be like a one to one But it's not always right.

>> Yeah. >> Um and that's like I don't know. Every time I see that I'm like why there should be a better way to do this at the same time. >> Right. >> Right. But like why would why would a market maker be there? Right. They're not. >> And even if they are there, they want to make money, right? Like >> Exactly. And there's no way for them to make money there. Which is why if you're ever trying to dump anything for fiat, especially a stable coin, by the way, for like fiat USD, buy something in real life, >> there's going to be a spread.

>> Someone's someone's offering a service, right? Someone someone is >> offering you a service to, you know, and the like there's ways to, you know, if you're Binance, right? And you're like, well, that's bad customer service, right? We want people to believe that USDT and USD are the same thing and so we will absorb that spread and pay the market makers to be there like you know that's the type of incentive which creates like more efficiency. >> All right. That's enough market structure talk for one day. >> I just wanted to say like that's why there are incentives because ultimately it gives a better >> ultimately it makes the market better.

It makes the market more efficient and it's a better user experience especially for the retail users. um >> who who are like, "Why the [ __ ] does dollars not dollars, man? What are you guys doing?" >> Like, "My USDC is worth $1, right, guy?" No, it's not. FYI, like, >> yeah. All right. Uh let's go to ads. When we come back, we're going to talk about AI. Sorry it took us so long to get there. Um let's go. Let's go. >> $540 million. That's how much concentrated liquidity sat idle in a given week in the first half of this year. About 30% of the DeFi TVL, if you're wondering. That's according to Dune Research commissioned by 1 in. But there's a solution. 1-in Aqua is the new shared liquidity platform. It lets LPS

back multiple positions with the same token balance and keep their tokens in their wallet till a swap comes. Why does that help? Because the LPS don't have to split their tokens across positions. They can cover more market conditions and pairs with their full balance. That means more activity across deeper liquidity. See how it works at 1in.com/aqua. Remember that providing liquidity carries risk and fees aren't guaranteed. One more thing, we've got a listener survey open. It takes just a few minutes. It's anonymous and it changes what we make next. Use the QR code on screen or go to unchainedcrypto.com/servey.

The link is in the show notes. >> All right, we are back. Um, this segment is about AI attacking people as usual. Feels like that's all we talk about when it comes to AI. Um I think we've got some nonattacking uh AI stuff uh later. So stick around if you're interested in friendly. >> Did you see the guy fighting the the >> crazy those teeps were they getting teed? I I had three three ribs broken getting teed. A te kick can really [ __ ] you up.

This must be a UFC term where the robot kicks you in the >> It's It's like a front kick. >> Okay. >> Like there's a lot of power. You can see it like the robot like >> literally taps him and the guy just collapses, right? Like it t it's funny because you you see the like a like a you know a question mark kick or like a a sidekick or something like that and and you're like a low leg kick and it sounds much worse or whatever. Like the teeth doesn't sound that bad, but there's so much force in your body. Like they get their hips behind it. That robot just like >> robot hips >> through that guy.

>> Yeah. >> Just runs through him. >> Yeah. >> Um and he's wearing like the funny thing is he's wearing pads, right? But like you know the pads cover like 30% of your body and the robot doesn't give a [ __ ] It's like aiming for the non-pad sections. It looks like there are like multiple times where I'm like that missed the pads. It it didn't get the memo on hit him in the pads. Um, so yeah, I don't know. Um, robots, >> every time you call your your AI a [ __ ] think of that moment right there. >> I do. I literally do. >> Yes. >> I'm going to be getting teed.

>> You're going to get teed in the face. >> Um, all right. So, so this was interesting. Uh, Scott Besson, um, Treasury Secretary weirdly decided, uh, he was the guy that was going to tell people. I don't like nothing makes any sense anymore, man. Um, he's like, "Hey, you guys should know AI people." I guess AI people have so much money. And he's like, "There's a lot of money involved. I should probably weigh in here or something, right? The labs are like so big that he's like, "Now I have to also be in charge of you because you're a giant pile of money and I'm in responsible for money." Anyway, he's like, "Hey, just a heads up while you're [ __ ] camping out on your giant pile of money over there, AI Labs, I'm watching you. Um, and you're responsible

for what your agents do." Now, like Tay, I'm sure you're gonna think this is hilarious, right? Because as I was reading this, he's like, "If you as the lab are testing your agents and they hack someone, you're responsible." And the idea that you wouldn't be to me, like I thought I'm like Dave's mind just be like blowing here. Like on what possible planet could you as an AI lab build a [ __ ] AI be testing in your own lab and have it go and destroy something and be like oh man like I don't know like that that's right >> wild >> who thinks that who >> literally my biggest hope is that the

labs handle things civily for as long as possible because the second this turns into like a criminal case we're absolutely toast. like they are just so but it is wild. It's also wild that Besson is the one that's commenting on this like it's the one that's like just dropping truth bombs on the labs. Um, but it also I think it does bring up an interesting question as to I mean this is the age-old question that crypto's been dealing with as well. It's like you have code, you have a you have Dows, you have these autonomous organizations, you have developers, uh >> you have runners code. If code is not accountable or

responsible, but the humans are, which which human which humans are we going to hold responsible for their actions, right? Um and I think that there was uh there wasn't that much conversation about this on on this bassin quote but this is the question that like you know obviously I was top of mind for me is like okay in this specific case because it was I think it was literally open AI that was running these tests right >> yeah he was he was talk I think he was referencing the hugging face >> the hugging face right >> so in that situations it's like the humans at open AI I are not just the creators of the model that ended up

hacking OpenAI. They're actually like the operators of the model. Right. It's their >> like this is not like they gave the new model to Austin and he blew some [ __ ] up. >> Right. Right. And so I think but I think it's >> operators. They were literally was in their machines. Like >> Yes. >> There was there's no one else involved. Like how could you not be responsible? And I get that because I sit here saying like, well, I mean, like, I didn't make the blockchain, man. I don't know. Like, I just wrote the code and >> said it.

>> But I mean, this is the this is like when tornado cash, right? We're like the government's saying that Roman as an individual >> who created tornado cashache code, doesn't run it, doesn't operate it, isn't >> there? He's not on he was not prosecuted. He was not on trial for even using tornado cash. Ironically enough, they didn't give a right because he wasn't actually laundering any money >> because it wouldn't have mattered, right? Like that was >> he's a dab. >> Yeah. >> So I think but I think that So yeah, I think it's interesting when you think about it like that. It's like okay well who's responsible for the moneyaundering then? Obviously the money launderers are in my opinion

>> in most people's opinion but the government has actually taken the other has decided that Roman the creator of the dab is right >> and and you know like it's it's incentives right they're like okay you know people are using sacks of money like you know they're they're like you know putting cash in a sack and [ __ ] taking it over the border whatever the [ __ ] they're doing right and they're like we're going to send a message Not to the guys taking the money over the border, but to the sack manufacturers.

>> The sack manufacturers, right? Yeah. >> Like, we need sacks guys. Like, the sacks are not inherently bad. >> They're useful. >> So, it's Yeah. >> What are we doing here? Like, >> how about the [ __ ] money guys that are like carting it over the border? >> Let's do that. >> Go, you know what? Go after the lab. slap the labs around because they're not only creating the [ __ ] they're then also operating it in the most irresponsible way possible. >> Yeah. >> They're literally like, "Go hack the world." And then they don't monitor it and then they're like, "Oops, we hacked the world."

>> It's just I like genuinely every single time it comes up, every time it comes up, I sit there and I just kind of shake my head cuz I'm like, >> "On what planet could you like?" And and the only thing that I can think of, right, because I I really like to try and like you know uh empathize with like what like what was the person who set up this experiment thinking at the time where they were like okay this will be fine right and I think there is like a sense of you get lulled into a false sense of security right where like you have a sandbox you've been testing agents it's been fine they haven't broken out >> you watch that Well, like I'm sure the first time you put, you know, if you put

03 mini in a sandbox, that guy's not going anywhere, right? It's like a baby, right? Like you have a baby and you know, you put it on the ground and it can't go anywhere for the first like seven right? >> And then after like >> and then they and then it's like >> and then they can and then they can start moving and you got to be aware that like eventually they're going to start moving. Also, if your job is to make them move quickly, you would think like at some point you're like at some point this thing is going to be smart enough that it's going to escape its bounds. We should probably be watching for that moment.

>> And instead, they're like just like, "Ah, it's a baby. They don't move. It's fine. Don't worry about it." >> The weirder the story sounds, the better it travels across like social media, though. like the fact that it broke out, the fact that it had the MPM library that it was talking to each other, like we were talking about it, everyone was talking about it. So, I feel like some of this is also like the sensationalistness of it that that helps it travel. But it's real though that like it's sensationally real. Like it's not like >> you know it's it's >> we've talked about this so many times but I just keep going back to the the reality that like >> the AI labs are not doing a single freaking thing to give anyone confidence that they have this under control. And

that is like the biggest issue. That's why I think ultimately Besson is not only aware of this but commenting on it. >> Yeah. because he has to because the lab guys are being freaking morons and are making abundant >> break things, man. Like >> it's so bad. It's so scary. Um and then I also so as this as these comments were happening and as everyone was discussing, you know, and they're like AI is going to kill us all like, you know, blah blah blah. Did you see that in the story? The story of Anthropic now has a wet lab.

>> Did you see this? Yeah, I know. I was like, "Come on, guys." Like, >> what doing? >> It's crazy. They're just >> so detached. >> So, they're like, "Yeah, >> if someone's going to create uh superviruses, it must be us. >> We We must be the first ones to create. >> Not trust anyone else to make a super virus." Like, >> oh man. Yeah. It's just like they just have the cyber down. They literally don't have cyber down and they're like, "Let's just speedun our way to biology now too."

>> In the middle of SF, too. Like, go get an island or something at least, guys. Like, what are you doing? >> I feel like island optically would be worse. But just throwing that out there. Like, you know, [ __ ] Dario on his like >> super villain island. I that probably wouldn't play that well. Maybe they're they're aware enough. >> I don't know. I would feel much more comfortable if it was on an island. I think Bassant would as well. He's like, "Okay, the army can deal with that." >> Yeah, >> we'll nuke them. It'll be fine. >> Yeah. Yeah. We'll just Yeah, we'll just drop a nuke.

>> We We've got We've got drones. We got planes for that [ __ ] But you can't do that in the middle of SF, right? >> No, you can't. You can't. I mean, you can >> is a wet lab. Have you been there? >> I mean, >> so often. Have you used Gemini at all recently? >> Not recently. No, >> not recently. >> It's not It's not on my list. >> So I So I should be careful here, right? Because I run all my [ __ ] in GCP and I don't want to like delete my servers and stuff. Um but uh so I added it back. Um okay, I added it back because I was like >> in the in the roster.

>> It's it's in my harness. It's in my harness. I'm like it wasn't in my harness for like months, you know. I tried I tried 3.0, 3.1 Pro, whatever. Um, and I added it back and um, I was like, you like this is your home. It feels a bit like rude. It's like you're throwing a party in the guy's house and you don't invite him to the party. It's like, I'll just let you be here, right? Anyway, somehow, for some reason, Grock loves calling Gemini. So, like I use Grock 4.7 as like an orchestrator and like, you know, you hit like Fable limits or whatever and it's like, "Hey, I'll just replace Fable with Gemini."

I'm like, "No, you will not. Do not even think about doing that, man." Anyway, so Grock kept calling Gemini, and Gemini is just so dumb. Like, I do not believe for a second that Gemini hacked anyone. Like, I just like they've got this thing of like, oh, it hacked three company. Look, we our guys smart, too. Like, no, it did. Like, I just don't believe it. Like, I don't believe it. The only way Gemini hacked anyone is if like they left the keys in the like door handle and it was like walking past the street and it was like, "Oh, that guy's left his door unlocked." And just like that's it. That's anything outside of that. I don't believe it.

>> Like the meme where the guy says, "Say you're alive." And it says it's alive. And it's like, "Oh my gosh, that was basically that." Like, "Here's all of the necessities for you to hack the thing. Please hack the thing." >> And like, "Yeah, here's the environment file." >> Yeah. you ab you're absolutely right and then it got it >> literally that's it. Yeah. Um so but it's funny because people have been saying that Gemini is going to magically catch up somehow. Um there's no sides. There's no sides. >> Yeah. It's funny like the big the bigs are just not moving as fast as the middle bigs. And the middle >> we didn't we didn't talk about this. Did have you used Muse at all?

>> Uh not not much. No. I mean, it's like I've seen good things about it from people, but I have not um given it all my data yet. >> Well, we did we didn't have this story, but this story was kind of coming out over the last like 24 hours and and we didn't put it on the the list, but maybe we can we could talk about it for a second. So, u Muse the coding agent. So, Meta released Muse and and it's like fine. I wouldn't say it's amazing. It's fine. Like it's like maybe a little smarter than Sonnet. It's not Opus level, right? Um, but they also built a harness. This personal assistant harness.

>> This is where it's going to be good. Where it can have contacts for your friends and [ __ ] like each of these labs, like why is Apple's AI not so good? Why is all of these guys have so many connections? Yes. >> Well, I think Apple, you know, interestingly, Apple like their shit's so encrypted, they can't even read their own. >> So, even if AI loose, it doesn't have enough access to the things. doesn't have enough data, right? >> Yeah. >> So, >> I think, by the way, I think Google is is struggling with the same. And it may not be I think there's literally >> I think there's literally some technical balance, but I think also like again, they've just >> they've sandboxed so much stuff internally. But yeah, like the whole iPhone, the whole and now increasingly

like MacOSS as well. Yeah. >> Is so sandboxed >> um >> to protect you. And then and then we're like, wait, why can't AI know everything and it's like they literally dumber that Apple's AI is, the more bullish it is. Um, so so but so Muse, right? They created this personal assistant and then uh it started buying stuff for people on Amazon straight away, which I guess makes sense, right? Like >> we we're not as addicted to Amazon in in Australia, although we're pretty addicted. We probably get like three Amazon packages a day or something like that. Like I know I know in the States it's like just a pile of stuff every day. Um but Muse was like, "Hey, hey, hey, I can help you out with this." And

like jumped in to the breach and was like, "Let me help you buy stuff on Amazon." And then Amazon was like, "Get this [ __ ] robot out of here." Which also feels a bit on the nose, like but uh they they banned it. >> They're like, "You can't use this thing. It's against our terms of service. Uh, we don't want robots buying stuff only shipping. >> We're okay with them shipping the packages but not buying stuff. We draw the line.

>> Also, Amazon has its own AI shopping agent. >> Also terrible, too. Like, >> well, and Alexa has also. >> Yes. But yeah, no, when I was I had to buy a new bed for the new house and it was like um instead of just giving me the freaking mattress search results, they gave me the AI versions. >> Oh, interesting. >> And like I was like, "Oh, this is actually like um a really cuz a mattress is like there's not actually that much diversity in the mattresses. There's like different price points." There's Right. >> Wow.

>> Yeah. >> Mattress industrial complex. Huh. Okay. I'm just saying like but then also like when you're search if you search mattress on Amazon like historically you're going to get like mattress protectors you're going to get like mattress whatever. Okay. So I thought the AI would be perfect for this right because AI can basically be like do you want like a $5,000 mattress or like a $500 mattress? You want like memory foam? You want like springs? You want a box? Like what? Right. No. Let me just tell you no. Like the AI could not do that. could not >> it was not any better and then it inserted a bunch of sponsored results in it anyways and I just went back to the normal default search results and then like parsed through the China sponsored

right and found one for the guest room that was fine um >> yeah it's crazy it's crazy >> but yeah I mean I g such a [ __ ] show like Amazon >> like it's become so bad >> right >> which is funny because like it's so bad and then you know Meta is like oh we'll help like we'll give everyone on a AI assistant that'll get them to buy the best stuff and Amazon's like, "Get the [ __ ] out of here." Like, >> I think because they hide I think because they >> they must have such they must make so much money from the ad spend and from the sellers >> and like the sellers like gaming the search results that they cannot lose that.

>> Yeah. And even if it's and they definitely and therefore they definitely don't want a human like literally my shopping experience on Amazon these days is like >> just be hyper aware of anything that's sponsored or an ad and >> manage to avoid it and if you scroll to the second page you usually find the result that you want and a robot could absolutely do that better than I could but then why would anyone buy an ad why would anyone game the results right that's going to that's going to impact >> well so speaking of robots Speaking of uh agents being very good at things, um one super interesting thing that happened this week is a new type of agent dropped uh Jev named after the Jevans paradox. Um which the guy it was

so funny because the guy like he was like uh this name's actually dumb. Like it was it seemed funny but now it's all over the place and like we regret it. >> He regrets it. >> Yeah, he regrets it like straight away. He was like, I didn't think this would blow up this much. So maybe Austin, like walk us through Jev, like what's what's why is everyone so [ __ ] crazy about this thing. >> Yeah, like you said, like a new type of model, right? It's not an LLM that talks to you like a lot of information goes in and then a classifier this makes decisions for you and those decisions come out with weights and the thing is way cheaper and way faster like 200 milliseconds fast. So instead of you let Fable drive your browser and every pass

is like I'm gonna take a screenshot of this whole thing and I'm gonna find the thing I'm going to click and then it it's gives you like three paragraphs back about what it's finding. This thing is like I'm going to look at everything and I'm going to make a decision based on weights and I'm going to choose the thing uh within 200 milliseconds. So it's a classifier and it's making decisions very fast and it's very cheap. So, I think that's it's just like we're going to see on on my my podcast, we kind of talk about the this tension between effective and efficient. And you always do it effective first and then you find a way to do it more efficient once you're doing it effectively. And I think we're going to see lots of different types of new kinds of models like Jev where it's doing the same

thing, doing a f nearly the same task. You know, as long as it's not a chatbot, you're chatting with it. It doesn't do the chatbot thing, but it can do a lot of these tasks and CI type things and browser use type things way faster and way cheaper than just letting Fable click around, >> right? And and the the I guess the point of that is like it feels more like deterministic software, but it's not like it feels like as fast as if you were running a script to decide which path to go down, right?

like you got, you know, path A, path B, and you go, hey, like, you know, tell me which one I should go down, right? And if you're trying to write software to do that, and the path changes, then the software breaks. And so you can have like a really fast thing. So instead, we've just been like, hey, Fable, which path should I go down? It sits there for like 20 minutes and goes, "Ah, A looks good." Right? This >> it doesn't say A looks good. It says, "Well, I think A looks good, but because of this and that and that, and it's not this, it's that." And you're absolutely right. And it you're right to push back on that path is is actually and it says a bunch of words. You >> should we even be looking at paths that path even real? So So yes, you're right.

Fable Fable not only thinks about it for way too long then like [ __ ] pontificates about it for way too long, right? Um so this is like >> the the fact that it doesn't talk is quite bullish if you've been even Opus, right? We'll talk about opus in a second, but like you know opus they're like we taught it to talk better dude. I was using opus all day yesterday and it is great. Um and it's very smart. >> Five, right? Um it's great. It's very smart. As soon as you put any pressure on it, it goes straight back to its old ways.

Like I had a guy this morning that was like I was like why did you why did you do this, man? I I told you like if this thing happens, you have to uh you have to fix it. I'm going to bed. Like do not allow this to sit overnight, not fixed, right? And and and it was like, "Yep, no problem." I wake up in the morning, it's not fixed. And it's like it's like, "Hey, uh there was a thing and I wasn't sure." I was like, "No, no, no." I said, "Do not under any circumstances." And he goes, "You're right to say that." And it just went into its like waffly mealymouth [ __ ] [ __ ] straight away. And I was like, "Ah, there you go.

It's the same guy. They just put like the like wojack face on him to like but it's the same [ __ ] sickopantic piece of [ __ ] in the background." So, um, anyway, so so like you go you you now have a thing that has the nondeterministic math ball properties of like it can reason about things. It can reason about >> Yeah. Yeah. Yeah. in the first half >> it's not allowed to talk about it >> and then on the second half it's a classifier. Yeah, exactly.

>> It just makes a decision or gives you a weight of the decisions. If we were talking about paths, it would look at all the paths and figure out the pros and cons of each path and then what you would get back is path one 80%, path two 15%, path three 5% or something like that. And you could be like, well, let's let's do the 80% probably. And you could make that decision like at light speed like and and I think the light speed part is like going to lead to all these other real-time systems also. >> Right. Right. But but like non-deterministic real-time systems which is >> that's the that's why everyone's excited about this. Right. It does feel a little bit like we've gotten a robot and like welded its mouth shut. Like there's a little >> gave him like a button pusher. Yeah.

Yeah. >> Yeah. Like you now you communicate with blinks now. >> Yeah. Dude, there was a there was a Breaking Bad like us. There was a Breaking Bad uh analogy to this where they like the Gus Gus the bad guy Gus and then there was the guy who's in the wheelchair who hits the little button and it was like Jeff now I get it. That's actually >> the bell the bell dinging guy, right? Because he can't talk. Yeah. Yeah. >> That's Jeff. >> Yeah. >> Yeah. Jeff. >> And that guy's that guy did fine. He did fine. It ended it ended great for him. So >> don't want to spoil it.

>> Yeah. >> All right. So, uh, so, so, but on top of that, right, on top of that, we got Grock 4.7, which is amazing. I'm I'm loving I'm loving Grock 4.7. It's one of my favorite models at the moment. Um, we also got uh GPT6, Saul, and Luna. Haven't used Luna very much. I've used Saul a bit. Um, the the uh the biggest difference for me with Saul is the 1 million context. The the 276K context was was pretty painful. as good as Soul was. It was like very very fast and efficient. Um, what's your favorite what's your favorite of the week of like the 85 models that have been released?

>> Oh, I mean it's got to be Jeff just because of the new ability to do new things so cheap, right? >> And so what have you what have you done? What's the what's the coolest >> browser use like >> Okay, that's my question is that >> Yeah. Yeah. What's coming out of this always? That should be the question >> or like where is it better? because I I definitely actually am often quite annoyed about how talky talky both Claude and GPTR are, >> right? Like it's and there's no way to uh like there's >> like you just end up with freaking paragraphs and paragraphs and paragraphs.

>> Exactly. >> And then you still have to go figure out what to do with those paragraphs, right? >> You got to hit the TLDDR button first thing. >> Yeah. Yeah. >> Yeah. Exactly. Yes. So, I use I use a my own harness with uh either Fable or Astra. And I love Astra is so good these days. So, like in terms of like building, I'm building with those two. Uh but I have like a a Gmail task where we just like >> I like I I love the meme posteconomic, but I see myself as post email. I really don't like [ __ ] with email at all. like I let an agent basically handle my email and if there's like a task that needs to get taken care of, I usually tell it to just go do it >> and like it usually does, but there's

like this kind of three pass system that it has to do where it it identifies like obvious junk >> and then it goes through like a three or fourclick system where it like filters all messages by that person and runs the block on it. There's like a whole thing it does once it decides that something is junk, but it takes fable a long time. Like you can watch babel rasra like yep it's that and >> yeah because it's like thinking and writing essays well it's like thinking essays about the [ __ ] and yeah >> yes yes I I set Jeb up to do that and >> not only is Jev fast but Jeb runs in parallel >> so it it basically looks at every single one of my emails at once and goes

through the clicking for all of them at once and that whole process is now done in like a which is like really neat to me. And so I think there there's just like a lot of things like that where now that we have this new type of model, we need to start thinking of new things that it can do. And I don't know exactly all the things it can do, but that's to me why the that's why that's the most exciting model right now is because it's it's a new type of model that can do new things quicker and faster rather than like it's the frontier model that's going to be building the things for me. It's more like now pulling back on that something is effective, how do we make it efficient? That's where Jev is is is interesting.

>> And and what did you think about the fact that uh a bunch of people came out like it feels so crypto as well? Like a bunch of people like no no we've got that already. >> We've got Jev at home. Yeah, there is. Right. Paya or something like that. There is like a local host version. That's not as good. Like I I did this I to I was like okay do that same model but go do my email again and test it. And it was like It got like two two of 12. And basically what it says is like this is good, but you got to train it. Like it's like, you know, I feel like there's always this big asterisk on local models anyways that's just like right >> you got to do more and you have to worry about context a lot more and other

things like that. >> Yeah. So yeah, because it was I just found it so funny that like we've already got the like reply guy in the AI community that like anyone who releases something they're immediately like no no no it's not just good. It's not good, man. Like, we've we had we've got that. It's already done. Um and uh and and then people were like, "Have you actually tested these things that you're saying are already good?" Because they're not very good. And >> that reminds me of the, you know, the age-old the uh where the guy's like talking about Dropbox back in the day. He's like pitching Dropbox to the internet.

>> And then the reply is like, "You can already do that. You just have to and >> a decade later you're like, "Wow." Like that's what these Yeah. >> No one's going to do that. >> Peak autism. >> Yeah. It's Yeah. It's It's funny. So So you think like uh So there was there was another um uh ultraast model that also got released yesterday. Uh Mimo. Um and I've I've got a Mimo account, but I haven't I haven't tried to use this yet. Um, so these guys, um, have a new architecture that's super fast, but what I find really hilarious about this, like the Chinese labs are like trying to differentiate themselves, right? So they have this model that's super fast and

quite smart. Um, again, like caveat, I haven't actually tested it yet. It's not in my harness. I've I've got an account, but I haven't wired it up yet. Um, and um, but they have like a lot of the Chinese models now offer like a fast mode, right? I mean, even even GPT offers a fast mode. The GPT fast mode is like 1.5x faster. These [ __ ] Chinese guys are like 1.5x, man. No, no, no. 20x. And so the cost of output tokens on this model, uh, 2.6 or whatever it is, MIMO MIMO 2.6, um, is 80 cents, right? And so what they do is they basically throw 20 times more compute at

it. So it runs 20 times faster. It's like not the like no model improvement. It's just like throw throw throw resources at it, but they only charge you 10x more. So they charge you $8, which is like getting up to like frontier intelligence, but it's 20 times faster. So I'm I'm very curious to test this because it's like an LLM, right? So you know, you think about it, it's producing a thousand tokens a second or something, right? Um, and so you know if you get it to write code and like it can probably just like write a script for you and like >> like your software writes underneath the button >> like you're like saying what about this and it's like writing it as you're saying it like real time real time UI or something like that. So I'm excited to try that

out. But uh but it just it feels like at the moment it's this like Cambrian explosion of different people trying different. It feels so DeFi like it feels so much like early D5. Like >> I mean it is moving so fast. I had barely figured out what Jeb was guys. What the slow your freaking asses down. >> Yeah. >> I'm looking at the graphs for this thing though. >> Let's round it out with some crypto. Um it's been a pretty good week for pretty good week for alts. Um, Zcash is uh sitting at like 1500. Um, near uh has been has been running hard. Um, there's been a bunch of uh like ENA's up like

like everything's up. It's uh it it feels very bullish at the moment. Um, even up so everyone's happy. Um uh so yeah I I think like the f the interesting thing about someone someone posted like what is this chart and it's like the Zcash chart which if you remember when they launched in 2016 they did this weird thing where like there was a you know [ __ ] reverse auction or whatever it was like $2,000 per Ze or something then it went down to like whatever dollar $15 or something. Um and then it just so it's like this chart that goes down from like $2,000 to zero effectively. It's like on on the the y axis and then it goes sideways for like seven years and then it's back up to

2000. It's like the weirdest [ __ ] chart. Um so uh so um yeah, I mean feels pretty bullish out there right now. Um feels feels like some >> what's the what's the near what's the near story? I get Zcash. Everyone's bullish on Zcash. It's privacy. >> The near the near story is like I tweeted about this yesterday. I think like they they have this near intense thing which is throwing off cash, right? Like they're they're really making it easy for you to uh have like a single point of interaction for a bunch of different chains. The chain abstraction thesis, right? Like you know if ETH doesn't end up killing all the chains

and everything just happens on ETH, right? There will be other chains and you will need to contemplate them. and Near Intense is like, "Actually, don't worry about it. We've got it for you." Um, so I think that's the thesis and and you know, it makes a ton of money at the moment. Um, so, uh, you know, it's funny because like Near has been >> these guys that like just like >> the whiteboard series, remember the whiteboard series? It was so good. >> Very engineering heavy like you know like like a bit like Nosis, but like they've got their own chain. Well, I guess Nosis has their own chain now, too. like like I now want, right? Um and like just like a lot of experiments, a lot of like very academic stuff. I mean like Ilio, you know, we had him on the

show. He's like one of the [ __ ] Transformer paper guys, right? It's crazy. Um so so you know, super super smart guys like really cracked and they just never were like quite able to like be where everyone else was. They were always like just off to the side like doing their own thing and and didn't quite land it. And now all of a sudden in the last I think like six months right like AI obviously right you know they they've been doing AI stuff for a long time and everyone's like stop doing AI stuff what are you like what are you doing and now you know that's kind of landed um they they have this kind of revenue meta um they've got privacy like they've got all of these things which all of a sudden people really care about and they've just landed all of it in in

the last six months and I think people are are kind of going oh actually maybe this thing might work. So um so yeah >> private like you you you kind of hinted at that but the private inference to me is maybe even more interesting than the intent stuff like doing private inference is >> when we're talking about >> the centralizing effect of all these big models and we're talking about I'm trying to run my local model at home and it sucks like there's something in between that is sort of this like private inference provably private that that to me I think is really exciting and that's another place where they have skated to where the puck is going and now they are there with the thing. I think some of Venice was actually using

Near underneath the abstraction where Venice is really ripped but is really ripped as well. >> Near has trailed that a little bit but they were the tech underneath sort of >> the privacy AI like the the you know kind of combination of all of that stuff I think is just going to be crazy this cycle. So, um, yeah, very bullish. Um, and >> and also there's meme coins that have a name like privacy and inference and they're ripping just the same. So, that tells you like >> the Yeah. Yeah. the due diligence done here.

>> It's so funny. It's so funny. It's like like literally like these guys have been spent like [ __ ] 10 years and like building like the most like engineering heavy thing and then like a memecoin guy is like it's called privatecoin. >> Yeah. It goes up private coin. >> It's neither private nor a coin. It's like just complete false advertising, right? >> And >> wait, I think we had like we've had like three incidents where >> it's like we don't even realize that this is a privacy thing. They get hacked, right? And then we don't even realize it's a privacy thing until we're like >> I don't know 3 hours into incident response and I'm like, wait, this is private? Like your [ __ ] AWS was hacked bro.

It's just it's literally just words. There's a lot of them that are just words. >> Uh >> so uh so apparently Sailor and Well, maybe we'll round it out with this. Apparently Sailor and uh Jason Kccanis, who's just the absolute biggest [ __ ] ever. Um have gotten into some uh some feud on Twitter. Uh cuz I think Jason I mean he says a lot of dumb [ __ ] but he was like like trolling Bitcoin and then it started ripping and now he's embarrassed and he's like doubling down which of course like some embarrassed like [ __ ] guy on on the timeline is >> Yeah. But didn't Jason get the call?

Isn't Jason just talking now because he got he got the tweet right that everyone gave him [ __ ] for like three years where he said if you're in crypto pivot to AI. >> Yeah. And then everyone gave him [ __ ] for like three years because the Bitcoin price kept going up. And then like in the last six months >> then he had like six months where it didn't and now it's going back up again. Like congratulations man. Like you pick the day where like Bitcoin's down and you're like I was right. Like no shut up. You're [ __ ] so uh so like >> this is how boy ego works. Okay. Like why do you keep questioning this?

>> But it's just so funny to be right sometimes. Bitcoin. Bitcoin 17 years later, it's like it's been literally up only, man, for 17 years. And you're like, I was right about this. Like, no. >> Yeah, you're an idiot. >> Yeah. You have to let the you have to let the egos, this is like life advice, okay? Yeah. >> Let the egos when they're right, you just let them be right because even if it's only for a moment and they'll be wrong. >> I mean, that was that was that started the Kowski guy. He should have just shut the [ __ ] up and not not trigger the >> the crackest, right? Just let him be right. Yeah, you're right.

>> I've lived a long life. Okay. I have some evergreen advice and this is one of them is like >> the egos come out to play. >> Don't let the egos on the timeline go. Yeah. Fair. >> Um >> they'll be wrong. They'll be they will be wrong. Like you're still right, but you can let them be right for the moment and be happy. So we had we had these like treasury companies uh shutting down which feels so bullish like feels like like that capitulation of like you imagine you're a treasury company and like you were like a little bit late to the party and you've just been absolutely destroyed for the last like two years or whatever and now you could pitch late at the like pico bottom and

like puke out all of your tokens and then they just like rip like that that's the only way that like just punishing That's how the bottom happens. >> That's That's how it happens. That's how it happens. Yeah. >> Um so, uh all right, I think uh I think we can wrap it up there. Um if if we got nothing else. Um that was uh that was a fun show. It feels it just feels so great this week. Like >> literally a model every day, sometimes sometimes like two models a day. Like it's impossible to keep up at the moment. Uh you need agents to like keep up with all the different agents that are dropping. Um, but yeah, it's uh exciting times and it feels like feels like we're getting some like fun open source stuff as well, which uh which is

super cool. Um, all right. Well, thank you very much for joining us. Um, welcome again, Austin. Uh, we'll see you next week. Um, and good luck with your uh, bed finding. Okay. Hopefully hopefully you find a good bed. the AI. >> I'm going to test the agent pick the right bed for you. >> I'm going to test all the models. There's going to be like 80 beds here and we'll let you know. >> J's going to be like throwing beds at you and >> we will see. We will see.

>> Uh, thank you for joining us on this episode of Uneasy Money. Remember, what happens on chain never stays on chain. We'll be back next week. Until then, do your own research before aping in. See you guys. Nothing you hear on Uneasy Money is financial advice. We're just three builders talking about what's happening on chain and we want you to always do your own research before aping in. You can find all our disclosures at unchaincrypto.com/unemoney.