The Ben and Emil Show - BAES 91: A.I. is the Biggest Waste of Money in History

Episode Date: March 13, 2025

This week we've got Ed Zitron, the outspoken AI critic and journalist. He's got a LOT to say, and we have a lot to ask him. We think this episode speaks for itself, and is one of our favorites of all time. Enjoy. Follow Ed here: https://www.wheresyoured.at/ https://bsky.app/profile/edzitron.com Be sure to join us for a very fun very special bonus this week! Sign up for your FIRST MONTH FREE and support the show at https://benandemilshow.com ***LINK TO OUR DISCORD: https://discord.gg/CjujBt8g ***Subscribe to Emil's Substack: https://substack.com/@emilderosa ***Leave a comment! Like this video! Tell a friend about our show! WHY THE MARKET IS CRASHING: https://youtu.be/OSRVh6ns7Z8 Latest MEATBALL SPECIAL HERE: https://youtu.be/uIOdsIn1Tdo We bought suits HERE: https://youtu.be/_cM1XqA9n2U __ RIDGE WALLET: Take advantage of Ridge's one-a-year anniversary sale and get up to 40% off right now by going to https://www.ridge.com/BAES #ridgepod MUD/WTR: Start your new morning ritual and get up to 43% off your @MUDWTR with code BAES at https://mudwtr.com/baes #mudwtrpod __ This episode was edited by Connor Rousseau / @ conrad_roussrad Follow us on instagram! @ benandemilshow @ bencahn @ emilderosa Learn more about your ad choices. Visit podcastchoices.com/adchoices

Sam Orton's been around for decade plus, and he's wormed in with the media, and he's warmed in with everyone. He's a good carnival barker. On top of that, this desperation time in tech meant that when Chat GPT came around, everyone went, finally, something, we don't make things here? What, we have engineers? No, we're only here to screw customers. That guy invented something. Let's do this. Why Sam Wormtman get to fart out whatever? He's like, oh, that's the best thing ever. It's like, what is the best thing ever, Sammy? There is no industry here. This is a PR campaign for one guy and one company, except this company is a complete... It's a time bomb.

We have this software today and you can barely get like a picture of Garfield with boobs successfully. Hell yes. It's just so frustrating because this is what happens when an entirely narrative-based thing takes a market. I would need it to work. I would need to say to Siri, I need you to order dinner for me. I'm feeling like Chinese. I want to spend no more than $35 and I want it quicker than 20 minutes. Sorry, I didn't quite get that. I'd be like, setting a reminder to kick your dog. Like, hey, everybody.

We got a really great episode for you today. If you're joining us for the first time, my name is Ben. With me, as always, is my co-pilot, Emile. We do a weekly. I wish I didn't do this. No, it's okay. I don't know why I did that. I'm out of this kind of guy.

We are the greatest show out there, and you really should. I'm sure you'll enjoy this. Our guest today is Ed Zitron, and we're going to be talking all about the AI bubble that he believes is imminently about to pop. And we're not going to talk much longer, but if you want to support the show, if you want to check out hundreds of hours of bonus content, head over to Beninamil's show.com. We got a lot of great stuff on there. We're offering a free month right now to all new signups. We're a free month, he said. Yeah. So he always does that. Yeah, go sign up. If you like what you see, there's a bunch more fun stuff. Honestly, some of our most

fun zany moments are back there. Get us up to 100,000 subscribers and we're going to release a semi-nude calendar and it's going to be great. Everybody's talking about it. They can't wait to get it. And also everybody at the

higher tier, we're going to be doing our Q&A that we're going to be recording next week. So we're going to drop the phone number on the website for you to call and leave your questions in the form of voicemail that we will play. next week. So without further ado, here's the episode. I'm looking down to town with baby on me. Tell me what's going on. Tell me what's going on.

So listen to love to baby to me. Tell me what's going on. Tell me what's going on. Wow. Hey, folks. Welcome back to another exciting this is going to be a very fun episode uh it's the ben and amel show with you as always is me ben that's a meal and we've got a very special guest for you uh his name is very special guest he's he we've probably seen some of his work on this show actually he he runs where's your ed at uh better off line and i think you have a book coming out next year please remind me the title why everything stopped working why everything stopped working we have ed zitron with us which is Which is going to be great.

He's going to tell us all about what's really going on in the AI space. And spoiler, it's all good stuff. AI is here to revolutionize everything. We're about to have artificial general intelligence. Yeah, baby. Right up our ass. Well, according to some people, that sounds like that could be the case. Yeah, if you're a big fucking idiot.

Big stupid, man. It's past the minute. I can swim. Oh, you know the rules. You know how it goes. So I think the best things to do is to maybe get you to kind of lay out your thesis around the AI bubble because I think

you've probably laid out the best case. You're also one of the most vocal not supporters. Hators. Critics. Critics of the AI thing. Also kind of a I don't know what the word is, but like a rational voice for me sometimes. I think it's hard to not get wrapped up in like

even I'm a pretty skeptical guy. But sometimes you just come across so much hype and like, you just can't help but be like, oh, well, I must be an idiot. I'm missing this. These guys know what they're talking about. It is right around the corner. I better just get ready. And it's the wave of constant hype as well. So you have to, you're kind of like, can this many people be wrong? And the answer is yes. So long and short of it is, you see chat GPT, for example. Open AI as a company, they spent $9 billion last year to lose $5 billion.

no one is making any money on this. Every single one of these companies is unprofitable. The transformer-based models are unprofitable. They cost insane amounts of money to run. They don't really have outcomes. Like there are cool things you can do with chat GPT. You can summarize documents. You can ask it really weird questions and it'll kind of get them right.

But it will also hallucinate in a way that makes it kind of by default not great for financial services. It makes it pretty bad for anything that you need to rely on, such as information. and on top of all of this requires a bunch of stealing to train these models you have to steal from everyone and then cost the world to run requires these distinct GPUs

graphics processing units from Nvidia and then you need a big bastard of a server to run them in big data centers worth all this stuff costs a bunch of money and then on top of all of it like I kind of said no one makes any cash no one's making money after this there's no real profit and everyone is selling

the AI bubble based off of things that it does not do generative AI does not do this. We've been seeing the same shit for a year, two years now. It's the same product. Right. From my perspective, I was just talking about this last week where it seems like there have been two things that it kind of can do. It's like changed the way we think of search models. Like it's changed search engines for sure. And it's kind of coders seem to like it. It seems to be helpful for coders outside of that. And even The search engine thing is like a kind of frustrating thing. It seemed like we had something that worked pretty well for us.

And now they're like, we're going to do it like this now, and you're not going to be able to trust any of the information you have. Yeah. So I get shocked. I used, I've messed around with chat GPT. I bought a subscription when it was first becoming a thing because I wanted to be kind of, you know,

I wanted to stay. No, I pay for chat GPT plus because how would I test the thing that I'm attacking? And then I, you know, I've used things like perplexity and Claude and these other things. I like seeing what they can do. They mostly just disappoint me, though. Especially when I hear people talking about them, I go, that has not been my experience. It's been very hard to get it to do the things I want to do.

What do you want it to do? I want it to be able to be very specific about research. I want it to be able to pinpoint things for me and find it, especially as it's hoovered up all this data, but cannot bring it to me. I'll say, I'll ask it for, because, and everyone talks about, as you pointed to, summarizing documents or whatever, like anything larger than a short internet article, I'll say, I'll say, give me a very detailed, uh, explanation of this or, or summarization of this. I can't read. I need you to read it. And it won't, it's not that detail. Yeah. It's very rudimentary. And I'll say, get more detailed. And it really doesn't. And I'm like, this is horrible. Why can it not pull the important... Because that's all that this is. It's all it is. And even they release deep research, Open AI's deep research.

Yes, but people, they're saying that it's like creating... Deep, it's dog shit. ...scientific documents for them. Yeah, it isn't. So deep research is great because what if, imagine this, you could get 5,000 kind of gobbledooke adjacent words, which also had citations of like Reddit posts and web forums. That's what I call science. That's what I'm looking for in a financial report is a bunch of Reddit posts and fucking hacker news.

And I sound like I'm being sarcastic, but this is literally the example that AI boost the Casey Newton of platformer used. This is his best shot. This is their best shot. It's been two years. Where's the money? Where's the product? Right. And so your point is that like they've been able to get, they've been able to get pretty close.

No. Close to what? Like, they've been able to create these LLMs that people can use. You can go on, you can use chat. Yes, but close to what, though? Close to, uh, close to a, you know, system that can spit out somewhat reliable information. But they're never going to get to a point where this thing is right more, like as often as it needs to be. No, that's mathematically impossible.

Right. And they say, please, just give us more data, bro, and we'll be able to do it. But they, it's the limitations of a probabilistic model. It's a model that is based on probability. It's pretty, it's pretty good at it. But it is still probability. It does not know anything. This is not even intelligence.

We do not understand how intelligence even works in human beings, let alone in models. And on top of this, the thing they're selling it as is some approximation to an autonomous AI that can do all these things. It's never been close, not even close. We have a slightly sexier version of search, which is only broken because Google did that deliberately to juice profits. And now they have made the least profitable software of all time. It just burns capital. And they put hundreds of billions of capex into building. these big data centers, for what?

Really? For what? People, one of the things that really pisses me off, well, okay, there's many of those, is I get people contacting me, going to like, what would it take to convince you that AI was real, that this was the... And I'm just like, why the fuck do I have to do this, but not open AI? Why Sam Wormwood get to fart out,

whatever? He's like, oh my God, it's the best thing ever. It's like, what is the best thing ever, Sammy? Right. Scam artist. It just, it frustrates me because... None of this makes money. None of this really does anything. There's not really business returns. September last year, the information reported that only like 1% of Microsoft 365 enterprise

customers had paid for co-pilot, it's $30 a head. Really expensive. And what does it do? Summarizes documents, generates things. It's always the same shit. None of these companies have found a way to monetize them. None of these companies seem to be able to make any profit. There are no profitable AI companies other than Turing, which is a software consultancy

company that helps them find human beings to train the models. If you look at the fundamentals of these companies, they're bad. They are all bad. Anthropic. They made about $900 million last year. They lost $5.6 billion. Jesus. And every single one of them like this, it's dogs everywhere. And you compared it, I believe, I believe it was you, compared it to the likes of Uber, which no, no, in the sense that they, at the time, it was like, holy shit, they're, they're, they They're losing so much money, but at least they had an actual business that you could use that was something that had actual utility. People used. Yeah.

And they get to it. They now are at a point where they are cash flow positive. And they weren't even losing as much money as this. And even then with Uber. Yeah. Uber's biggest loss, I believe, was $6.2 billion in 2020, which was COVID. And on top of that, Uber is just not functionally even similar to Open AI. to, it would be like if every car, every ride cost the customer $5,000, but cost Uber $50,000.

And it sounds like I'm being dramatic, but Open AI is on course to lose like $11, $12 billion this year. And they play all sorts of accounting wank where they're like, oh, we don't count this bit, which is like billions of dollars. Yeah. It's insane. I feel like I'm going crazy when I read this stuff. Because to your point, everyone's like, AI's here. AI's here. And you're like, where?

And they go, well, there was Matthew McConaughey commercial. That has AI in it. Right. And also that thing is that commercial is bonkers. The one where he's in the rain. Well, that one's insane. Yeah. I'm talking about the Super Bowl one. Uh, yeah.

Super Bowl one. Which one's the one where he's like, it's like, don't be like that guy. He didn't book with like an AI chat bot or whatever. That's all of them. Okay. It's insane. No, the Super Bowl one is Matthew McConaughey's running through an airport. He throw airport, allegedly, even though, obviously.

and British. And I've been to Heathrow and it's like, this is not. And also, he's like running around the airport being like, ah, my, my travel agent didn't use an AI, a futuristic AI from Salesforce. And it's like, first of all, agent force is not a customer facing product. Second of all, it's a chatbot. And the things he describes it doing are not things that chatbots do. And he's like going to miss his flight because an agent didn't tell him that he was going to miss it. It just, it breaks down. And this is one of the largest companies hooking this bullshit during the during the Super Bowl. It just feels insane.

It feels crazier than anything I've ever seen in my career. Real fast, going back to what you were saying about the limits of it, I am a bit of an idiot and I don't fully understand, is it because the way that it functions, it's going up against the data limits? And there's a function of how it's, it needs more data. It's ultimately, it doesn't know anything. it's just, you, you're not, you're not too far off. So the way these models work, and I'm going to really truncate this a bit,

is that you train them by throwing training data at it and then kind of reinforcing things saying this is a good answer, the savanna answer. Training these things costs millions, tens of millions, hundreds of millions of dollars. Now, the limitations are, first of all, they're running out of training data. Right. And that's the only way to make them better. Also, making them better very rarely if, and it doesn't appear to for some time,

have actually made them do more stuff. They're just like, they will pass more benchmarks that they rig. But fundamentally, what we were saying was sound, which was when you train these models, they are still drawing on a corpus of data. And there's a whole thing around training data and tokens and such. But nevertheless, they are still drawing from what they have in the training data and generating an answer as a result. That answer is not like they look at the training data and go, oh, a cat is a cat. No, they say, someone has just typed in, do this. it's most likely they mean this thing, which is cat.

And they said, orange cat. I don't know what orange is, but orange might be based on these things. It's why in the images, you'll see them have like 11 fingers. Because they don't really have an idea of what a human is. They have the most likely image of a human. I get slightly better on that. Now, the problem is with this is multifaceted, but longshore of it is they don't really know anything. You've not taught them anything.

You've taught them how to answer a thing. thing more likely. But in the end, that's all they'll ever do. All they will ever do is guess. All they ever do will roll a bazillion dice to come up with an answer. You can keep feeding it as much data as you want, but you can't teach it to think for itself. Yes, exactly. Ultimately need it to do to lead to this AI promised land that they're... And even if we could do that, and we can't, we do not have enough training data. To train the next models, they need like four times the available data. And the current available data is the entirety of the internet. So they've run out.

Even if they haven't, there is no proof, none, that having enough data would fix this. Because what? They're just going to keep training it until the math gets better and then it wakes up? It becomes sentient? What the hell are you talking about? It's nonsense. But they're already running up against diminishing returns with training data. There's a big story towards the end of the year about how all of them are running into the same problem.

And because the tech industry is completely bereft of any creativity or real, innovation anymore. Everyone's doing the same thing. They're all doing the same thing. So all of them are going, they have very similar models that all pass benchmarks in varying ways do mostly the same thing. And the same thing is not profitable and it's not that useful. Hey gang, it's time to take a quick little break to talk a little bit about your stupid, fat, bulky wallet. It's ugly and it's stupid and you should probably do something about it. What are you 50? What are you over 50? You know, my dad used to carry around a wad, not even kidding about this thick. And it was just loaded with receipts, credit cards, business cards, that you're

never going to use. His license that you need, cash, pathetic, I would always say. We would make fun of them for it. Because it's such a geriatric old man, pathetic thing to do. I was stuck in that mindset, too. I was carrying around my old geriatric wallet, my, you know, my bifold. bifold. And I said, it's time to freaking upgrade. Yeah. So that's why we're using the ridge. It's this for the audio listener. God, I wish you could see it. It's so thin. It's about a quarter of an inch. Is that a quarter of an inch? I think it's about a quarter of an inch wide. And baby, this thing is sleek. It's got a little money clip thing for your dollar bills right there. It opens just like this. A nice little opening to put all your cards in it. I took mine out. this demonstration because I don't need any of you freak screenshoting this. No, that's the last thing we need. Finding out where I live or what my credit card numbers are.

And they also got a handy little cool keychain that I also didn't have my keys on just yet because I don't need you guys screenshotting it and knowing you, you'll be able to like 3D printed or something. But all around, man, sleek packaging, very cool, very neat stuff. They've got over 50 colors and styles. So the Ridge wallet isn't just functional. It's personal. Whether you want sleek black forged carbon fiber like I chose or a wallet that features your favorite NFL team or one that just matches your vibe.

My favorite thing about the Ridge Wallets is they're extremely thin, yet they can securely hold up to 12 cards plus cash without adding unnecessary bulk. That's huge for me. I hate having a big Bokey wallet. And I do. You know, the cards add up. You got to bring all these different cards with you, IDs, whatever. The materials it's made from are so high quality. It's built to last.

One of the best things, though, every Ridge wallet comes with R-F-I-D blocking technology to protect you from digital pickpockets. All right, go out safe knowing your wallet is protecting you when you're out in the world. All right, scammers can now scan your pockets and steal your credit card info wirelessly. Yeah, not.

But Ridge has you covered. You can feel it. It's so thick. Is your wallet R-I-F-D blocking? R-F-I-D blocking? Probably not. Probably not. It's probably leather or something.

Or probably held together with string. One of the best parts? Raise your hand right now if you've ever lost your wallet. Okay, fine. I did it. You did? Yep. That's stupid of you.

The best part, though, the Ridge wallet comes with the Ridge Air Tag attachment. Okay? You'll always know exactly where your wallet is before panic mode kicks in. That's right, Abil. And Ridge isn't just about wallets. They create premium everyday carry essentials like the key cases I show you. I showed you. Suitcases and rings all built with the same.

same sleek, durable design. No matter what you pick, Ridge has free shipping, a 99-day risk-free trial. You got nothing to lose. And a lifetime warranty on all products. Right now, Ridge is having their once-a-year anniversary sale. Get up to 40% off at ridge.com slash bays. Just head to ridge.com slash B-A-E-S to see their biggest sale of the year. After you purchase, they will ask you where you heard about them, so please support our show and tell them we sent you. Can I try to sum up what I believe your position to be, which I believe I agree with? It's that while there are some good practical use cases for these AI models, it's the fact that they aren't profitable, like you said. They are massive, massive resource and capital suckers.

And that's pretty much it. It's that they're not going to get to this point. where they're going to be printing money for these big tech companies who've already thrown away so much money. They're not going to be like fundamentally life-changing for everyday people. Aside from there, yeah, there are utilities in maybe drug discovery and things like that, right? Even then, it's a lot of the stories, and this is one of the more noxious things, you'll see stories about like, oh, they discovered drugs and you'll look and it'll be like, it's not actually large language models.

Like it would be something else. They'll just like plug AI into it. There's another thing as well, which is, now, large language models has an idea there's nothing wrong with them. If they'd been called, like, library models, it would be a $10 billion industry. It would be sold by, like, Microsoft on a page you have to click a few times to get to. It would be a small but useful industry. But there's also a good chance that in a few years we have large language models that just run on a device. Now, you may think, surely, that's good for Open AI.

No, it's not. Every single model is kind of similar. And so what? And then you, okay, strip away all the unprofitability in the stealing and the environmental stuff. and you go, right now, what does it do? And it doesn't do very much. There are some cool things you can do with search with it. But it's always, you get these stories where it's like,

AI's going to change everything. But when you get to the truth, it's, yeah, you can do some cool shit, I guess. Yeah, it summarizes my texts. And even then it doesn't do it very well. Yeah. I'm like, okay, I see the summary. Now I don't have to open it right away. But then I'll open it later and be like,

oh, that's the extent of my AI usage and the Google search. My one thing I found chat GPT useful for is reading this S1 that I've been burying my head in Yes, tell us about that We will get to that And that's a whole separate thing But giving it like a block of text

And saying, explain this in plain English And then having to go in line by line verify that information myself Having a little bit more knowledge It's just like And when you get down to it Wow, that was kind of useful Okay, is this worth a trillion?

Is this a trillion dollar? No, it's not. It never was, it never has been. So what do you say to the people who would assert that we're just kind of in the first innings of these things akin to the internet 20, 25 years ago? Tim Covello from Goldman Sachs fielded this one very well. Yes. In a Gen.

I remember reading. Yes, we talked about it. Not enough return. Too much, Ben, not enough return. I think it was called. Point he made was that when you looked at the original internet boom, yeah, they did use expensive some microsystem service, 64 grand, I think. You didn't use that many of them.

On top of that, he makes a really good point about, like, smartphones. He said, people compare this to smartphones. They say, oh, people didn't believe smartphones to be a big thing. He's just like, false. Jim Covello, head of Equity's Research, I think Goldman, old school semiconductors guy. These people do not mess around. They don't play fun games. He called Intel very early.

Anyway, his point was that he was in meetings in 2000 saying, like he said thousands of presentations, which might be a little much, with people saying, okay, when GPS modules or radars, whatever, get small enough, we will get devices that have GPS in them. When cell phone radios get smaller, we'll eventually have computers. When we have microchips, there was a defined roadmap. No such one exists for AI. On top of that, oh, it's the early days. Okay, but even in the early days of the internet, you could draw a line from where it was to where it would be. The early days of e-commerce was poorly run companies, but you could see the idea of buying stuff online made sense. What's the thing that makes sense with ChatGPT?

Yeah, like Jim Covello said, what trillion-dollar problems does it solve? Exactly. Yeah, it feels, the cell phone thing is very funny, or the smartphone thing is very funny. We had someone on our show recently who was much more optimistic about AI and brought that up. He's older than us too. And he kind of gave us a like, maybe you guys might not remember what a big deal, like what it was like when the iPhone came out. And I remember being like, I was fucking like 17. I fully remember it.

And it was like a massive deal. Uh-huh. And everyone saw the utility immediately. I love this point because I bought an iPhone on the launch day. it was no i i can tell you the literal experience i which every single person i showed went holy shit old young i was in college in penn state at the time i think i picked on my iPhone in state college crazy um every single person who sued the voicemail was like holy shit every fucking no one was like i don't get it every single person who saw it was like damn

that's so expensive like they were right and the difference was they did their uh they did their keynote and they went, here it is, we made this thing, you can pre-order it now. And everyone went, that's incredible, I can pre-order it, I'll have it in a few months. The difference is now, these tech companies do these keynotes, and they go, we hope we can do this someday. And you go, well, what the fuck? Google I am. Open AI does it all the time. They're like, you know, they do a cool little demo, and then they go, I don't know.

You might be able to do that soon. Open AI is pretty honest. What they do, they're honest in the sense that they show you some stuff it might do, but then you look. And it's like, did they cut this weirdly? And you look, what the one that really did that that was so horrible was Google. Google I.O. last year, Sunda Pashai gets up on stage.

He does this whole song and dance about... Generative AI. Agents. Agents are going to change everything. We'll get to agents in a minute. And he talks about agents. He's like, oh, you'd be able to just click here a few things in your email and do a full return for your shoes.

And he goes, and this is a theoretical idea. The fucking nerve. How dare you? You make $200 million a year. Give me it. Give me your money and do a real job, you loser. Isn't he kind of pissed at Open AI now, too? Didn't he kind of...

Oh, wait, no, sorry. I'm sorry. I'm sorry. I'm confusing him with Sundar Pishai. Satchan Nadella, I assume. Or, yeah, Sati Nadella. Fuck. It's just one of the interchangeable MBAs that runs one of these guys. They all have MBAs.

Sundar, the way to tell you, he's McKinsey. He was a former McKinsey guy. Satchin Adela is just a regular MBA. Yeah. And a fucking product manager. Yeah. Wait, so can we talk... Wait, wait, wait.

Where does your disdain for, did you, what's your background, by the way? Oh, I'm a mess. Yeah. Well, we know that now. I was a video games journalist. Ooh. PC games journalist as well. Then I moved into PR in 2008 when I moved to America.

Stayed in PR. I started in a newsletter on the side because I was depressed and did the newsletter. I took it from 300 subscribers to 57,700 today. Oh, boy. And then they recruited me for the podcast, Better Offline, end of 2020. three yes well that's right which is now like one of the top

tech podcasts yeah it's getting there it's crazy good for you yeah we're trying to get there too yeah we have we have tens of listeners it's really great yeah you got a decent you got more reviews than I've got

yeah we've got and your reviews are more positive than mine guys he's joking we've got I'm not you have like 900 I have about 300 400 yeah I look at the metrics wait so you just missed Gamergate then you just fucking Christ I did

yeah that would have been if anyone was getting killed it was me What's obnoxious for me is seeing, do you, I'm sure you know the name Mike Cernovich, but watching that guy ascend to where he is has just been mind numbing. I fucking can't stand that guy. It's genuinely depressing to watch. No, I know. It's terrifying, too. It fucking sucks. Yeah. And seeing his power and influence just grow and grow and grow and the confidence with which he talks about just all the bullshit he espouses on the daily. I mean, jokes on me because he has no idea who I am. but you know that's for the best yeah I guess yeah wait so before we move on to anything else I just want to

before we move on I just want to kind of wrap up with this part about this thesis I'm very curious about where you see all this shaking out like the bubble popping yes because I'm

I'm having trouble kind of seeing anything outside of like two things happening like either either they're right all these guys we've said are, you know, very smart, these tech guys, they're right and AGI's around the corner and the world is like about to change in a way we can't even fathom. Or they are wrong, but we still

get this like horrible AI ram down our throats and like implemented in everything and things just work shittier because they've invested too much into it and we just like need to, we just now need to live with this. Yeah, is it a sunken cost fallacy we're witnessing on full? But how do you see it shaking up. Yeah. I will start by saying it's not really possible to say this is the bubble popping. That's the one thing that people can't redefine. Similarly, people can't define success either. There is no success. Like, there really is the second thing. The best they can hope for is the thing you just said about how they just shove it down our throats. However, these companies are slaves to the markets. The markets are the reason that they've been doing this and the markets

will be the reason that they stop. So I, again, don't know how long this will take. I don't know exactly the world of events that happens. But right now, seeing the markets as something that is logical is a mistake, or at least fully logical. Because right now, if there was a logical thing, they'd go, hey, why are none of you companies saying how much money you make from this? Microsoft said they make $12 billion ARR on AI.

Now, if you look real close at that, what that actually means is that's annualized return, meaning that they did 12x what they're currently making which is fucking insane to publish by the way because that means that they're making what like $3, 4 billion

a quarter revenue not profit and that's the only one of these assholes that will actually talk about the money so if the markets were logical they'd be like

what is going on with you freaks what do you mean? What do you mean you can't tell how much you're making show me the money now show me money money now but All right, everybody.

We're going to take one quick final break to round this thing out. And if you notice that I got a little bit of a spring in my step, it's because I got that mud water flowing through my veins, baby. You know, I always used to struggle with that afternoon brain fog and whatnot. But not anymore because I get a little bit of that mud water. It tastes so good. Mix it around with the frother that it comes with. So cool.

So cool. You just be going like that. And, you know, I got a. say, it's March, all right? We got spring right around the corner. We just hit Daylight Saving Center. The rodent saw his shadow and he's like six more weeks. We're not quite there.

But it's the perfect time to refresh your routine. Maybe you're thinking about shaking off the winter blues, focusing on your wellness or whatever the heck you want to do. But it's time for something better than coffee. All right? And that's why we got Mudwater. So every single ingredient in Mudwater's products are 100% USDA certified. organic, non-GMO, gluten-free, vegan, and kosher.

There's also zero sugar and no sweeteners added. Best part about it, each ingredient in mudwater serves a purpose. With organic ingredients for a clean, natural, boost, mudwater's smooth, earthy flavors provide a delicious and natural source of energy, even for a guy like me who doesn't mess with coffee because it makes them all weird. The OG blend contains cacao and chai for a hint of caffeine and hot chocolate-like flavor. Lions main for focus. Focus.

Corticeps to promote natural energy and both Chaga and Rashi to support a healthy immune system. Does your 3 p.m. cup of coffee make you stay awake at night contemplating life choices of the meeting of existence?

I know mine sure does, but not with mud water. Let's not do that again. Switching to mud water can help you say goodbye to sleepless nights. It's got all the energy without the late night existential crisis.

To use mud water, you simply drop the powder in your favorite mug, pour some water on it and give it a mix. Some go wild and cream or honey or even CBD. There's also caffeine-free bloods available. The best part about mud water is it provides sustained energy without the spikes and crash with the traditional coffee.

Ready to make the switch to cleaner energy? Head to mudwater.com and grab your starter kit today. Right now, our listeners get an exclusive deal. Hold on to your seats for this one. Up to 43% off your entire order. Plus, Oh, my God. Free shipping and a free return.

chargeable frother when you use code bays. That's right. Up to 43% off with code bays at M-U-D-W-T-R.com. After your purchase, they'll ask you how you found them. Please show your support and let them know we sent you. Keep your energy

natural and refreshing all year long with mudwater because life's too short for anything less than clean, delicious energy. Keep your energy to up. Guys, we got to keep our energy to up. Bump. right now, I believe we're in the throes of what might

start popping stuff. And I know I'm being vague, but it's like, you know, nothing's ever this simple. It's not like one event happens and then everything stops. Right now we're in this situation where you've got Open AI running out of money. Anthropic running out of money. Anthropic just raised $3 billion, a $61 billion

valuation. You're fucking kidding me. Open AI is right now raising a $40 billion round raised by SoftBank. Now, SoftBank, if we work and Wire card fame? Yep. Now,

let me tell you how insane this is. SoftBank is doing $30 billion of this $40 billion around.

They're doing a syndicate for the next $10 billion, meaning like whoever will join in into the party.

Now you may think, well, SoftBank's big, right? They've got enough money. What if I told you they didn't?

They're taking out a $16 billion loan to cover this. Now you may think that's lower than $30 billion. And you'd be completely right.

I don't know how this works. They've also committed $18 billion to the Stargate project, The Data Center project out in Abilene, Texas. Well, it's the first one.

$500 billion. Up to $100 billion right now. Only one, literally one data center is planned. When you get to the fundamentals, things are very shaky. But then it gets worse. So SoftBank, why do they have to take out $16 billion for, they owe like $46 billion. They also committed $3 billion a year in revenue to open AI.

The math, it's not math. thing. But I think that they close this round and then Open AI becomes fat and happy. The thing that kills people, regular people, companies, isn't spending a lot of money. It's the rate at which they spend it. Open AI, once they get on the SoftBank T, aren't going to stop. They're not pulling back expenses. If this company needs $40 billion a year, no one has enough. Anyone says while you're listening to this, what about a government buy? Go outside. Go outside. To the podcast. You're too silly. The government's not going to bail them out. Why would they bail them out?

But on top of this, SoftBank doesn't have enough money to give Open AI. They could sell their arm holdings, I guess. And wait, what about over here you've got the Stargate project? Well, that's good, right? They're going to build some data centers. They take three to six years to build. They're not going to fucking build those

quickly. Also, Microsoft has been canceling data center projects. D.D.C.C. Analyst came out and they said, and everyone misread this. They said, they've canceled two data centers. I think it was like a couple hundred megawatts. Didn't Microsoft come out and refute that? Microsoft did not.

Microsoft came out and said, we still intend to spend this much money. And we may make some adjustments to infrastructure. Yeah. But worse still, this report said, oh, a couple hundred megawatts off, which is not brilliant, two data centers.

It also said that they walked away from a gigawatt of letters of intent. A gigawatt is like a seventh of their current, their capacity is like seven and a half gigawatts at most right now. Gigawatts is just the power of the data center. It's stupid. But Microsoft is walking away from the commits the data centers, just as open AI is trying, like they're growing even more.

That doesn't math, as you would say. It does not math. But on top of that, Microsoft is the largest owner of an Nvidia GPUs. Yeah. The largest purchaser. Why is the largest purchaser of the things behind generative AI, walking away from building out generative AI. Does that not sound

kind of fucking strange? Yeah, I was going to say, if anything, what I imagine, the thing that's going to start to deflate the bubble, if not pop it, would be, and this is not an original thought, because they are the bellwether, but Invidia earnings.

I mean, once you start to see a slowdown in that, I think the air will just start. I agree. Yeah. He has already, that he's been tumbling. I don't know why this is not... They lost a trillion dollars. in market cap since they're

lower than they were with Deepseek today. I realize this would go out after it. But on top of all this, Microsoft is generative AI. Yeah. They are the company. They funded Open AI there. The reason that Microsoft has such big data centers

is because Open AI needs to make the biggest, hugest models. Do you think Deepseek has anything to do with kind of some of this pullback where people are... That's what people want to believe. People are seeing like, oh, maybe we don't need as much computer power.

We don't need as... Oh, that's a big part of it. But people are saying, saying, well, what if Microsoft, I put out this newsletter about this, and I'm going to record a podcast probably tomorrow about it, about the whole Microsoft data set, pull back. People saying, what if Microsoft has a big secret efficiency thing they haven't told anyone about? It's like, why is it always some secret fucking magic source with you people? What if instead of it's the thing in front of you, it's a thing that's secret? It's a wizard. Microsoft has a

grand vizier. Microsoft has a series of mages. Do you remember the Yahoo guy with the... Chingy? What was his name? Hey, well, Chingy, David Ching. Yeah. And what was going to be a guest on my show? No shit. What was his title? The digital profit. Yeah, I have Shingie coming on better offline. Holy shit. What does he do now? I have no fucking idea. I'm going to find out.

Do you remember this guy? The digital profit? Jim, can you pull up Yahoo Digital Profit? I like David Shing. He looks like something that you would tell a German child about before bed to make sure that they like aren't a bad child. Yeah. Also known as Shingie. Shingy. Yeah. Look at him. Legend.

Oh, God. What a time that was. What is this? Circa 2012? Is that when they had? Because it was when, what's her name was the CEO? Marissa Maya? Yeah.

No, AOL. Sorry. Marissa Mayer wasn't AOL. She was Yahoo. Yeah, but he was at AOL. This is before. Oh, he was at AOL. Anyway.

Good times. Where were we majors? So the thing is, Deep Seek did play into this, and that Deep Seek kind of proved that you could make models that were more efficient. The one thing no one has come up with and I've really been looking is, is Deep Seek profitable to use? I can't find

any information to suggest that it is. It may be that even the super cheap model is still unprofitable, which is so fucking funny. Also, the information reported, I think today, actually, that Microsoft is trying other models to replace Open AI with. Microsoft is AI. This entire thing,

every bit of it, is Microsoft. It is Microsoft's hand up different assholes. And I think when people finally realize that, this whole thing fucking starts unwinding. But on top of that, how the hell does this continue? How the hell does Anthropic and Open AI keep raising? In OpenAI's case, $40 billion a year. And I've had people say, the Uber argument, it's kind of gone away now. They're raising 40 billion. Now it's just taking the piss. But this company, Open AI, they're not going to be profitable ever. They're claiming by 2030. But how the hell do they make it there? And if Open AI

dies, this is all narrative driven. It's not business return driven. It's not output driven. It's not magic driven. There's no experience like using an iPhone that you will ever get with OpenAI or anything with ChatGPT. Yeah, it's kind of just like, oh, this is cool. Well, you clearly haven't used their new agent to...

Oh, you mean operator? Yeah. Wow, to sometimes search TripAdvisor correctly? Jesus Christ. I mean, they have, they tell 300 million weekly users, but are they able to, hey, where's the profit? That's a good question. And actually, here's a fun story about that.

So it's $400 million now. And the weekly user is a very strange statistic. Yeah. For a number of reasons. Number one, a weekly user is anyone who uses it once within the seven-day period, which is still $400 million. It's pretty good. They lose money on every single free user.

Every single one loses money. Every single one. They operate in the, they're antithetical to regular software. And usually software, as it scales, gets cheaper. Theirs gets more expensive. It's insane. But on top of that, they're a month. revenue business. Open AI, the majority of their revenue, over 70% of it, comes from subscriptions

to their premium products, chat GPT plus and pro, and then there's like AirDU and government. To be clear, they still lose money on the $200 a month, one, which is so fun. They're so bad at business. So, but here's the thing. In a regular software as a service business, you look at any public company or private company with the software business. You wouldn't do monthly active users if you're selling a monthly product. You would do that. You would explain that. And indeed, monthly active users would be higher. So I estimated in one of my pieces, they used data from similar web,

which is a web analytics company and a sensor tower, which does the monthly active users for apps. I estimate they have about 500, 600 million monthly active users. Now, the reason they probably don't want to put that out there is when you divide their 15.5 million users against that 600 million, that's a piss poor conversion rate.

They're not a good company. They're bad at this. Right. And you're saying so they have a lot of people in there, but a lot of people are not engaging with this. And then they're barely engaging. They're losing the money. Even the people that pay them lose the money.

But even if they didn't, it's still a pathetically small amount. And then you have another problem, which is on top of all this stuff, not making money and all this. Well, you'd think in an industry that exists, you'd have, I don't know, competitors. And these competitors would have a ton of traffic. Right? Wrong. Anthropic. Their website, whatever the Claude website is, this is in the news.

there. 8 million unique monthly visitors, 2 million monthly active users in January. That's fucking nothing. Co-pilot. I think they had 11 million monthly active users on their app. Google Gemini, 18 million. These are bullshit numbers. This is not a real industry. This is not indicative of anything other than the fact that generative AI has been a successful PR campaign for one company, Open AI. Where do you trace it back to? Does it all go back to Sam Altman just being one of the best bullshitters of all time then? A little bit. Yeah. It's a mixture. So what it is is I believe we are, I have this theory of the Rockcom bubble, which is we're at the end of the hypergrowth period for tech. Tech's insane

valuations to be based on one fairly simple thing, which is these companies will grow forever and they'll always grow. And software can proliferate and deliver value to shareholders in a way the physical stuff cannot. They keep giving us new and cool. Exactly. And you find, you find new shit to So here's the thing. Tech is out of hypergrowth. They don't have anything else. They didn't. They haven't for years.

People are going to say, what about crypto? Crypto wasn't like this. Metaverse wasn't like this. It was just like those were people like to mention them. It's a bullshit thing. But they are minute. Even with the 45 billion that the meta burned on the Metaverse compared to this. This is egregious.

What we are seeing is a corporate desperation and the death of innovation, really. And I know it sounds very dramatic. But what else have they got? Delete AI right now. What else do they have to show growth? I mean, just look at your iPhone. It's the same shit. Apple Intelligence is one of the most egregiously awful products ever released.

What about Gen Moji? You can turn... Well, listen, hang on. You can imagine a rainbow ice, what do you call it? Ice cream pop, right? And what if you wanted to mix it? I don't know, with a parrot. There's an entire billboard.

Kill myself. It just says there's exactly that. it just says imagine it gen moji it i'm sure you see it but i actually speaking speaking of the iphone i believe there's a steve jobs quote about how i'm going to butcher this i'm obviously paraphrasing but it's like uh tech companies reach a certain point where they no longer innovate and the bean counters take over and that's kind of what we're experiencing now across the board that's wonderful i'm going to need to i'm going to need to memorize that and say it to people like i found it i mean that's true because it's it's the ever it's you you have to keep the market happy

But you have to, NVIDIA has to show, not only are we meeting our forecast, but we're beating them every single fucking quarter. Invidia's even more fucked up, which I'll get to in a second. But note that Google, Apple, really. Apple's actually had a few innovations. The M-Series

chips are fucking brilliant. Yeah, they really are. They're really good. Like, Apple actually created something very cool with that, and I get shit because, oh, you're an Apple fanboy. No, I rip on Apple. I think the App Store is an egregious shakedown operation that promotes disgusting violative business models like online dating and sports gambling.

I think Tim Cook's a bastard. I think they all are. But Tim Cook could destroy some really noxious. He should destroy Hinge and Tinder and the match group. Like the one company that owns all of them because microtransactions and dating apps are disgusting. I've spent a lot of money. And they still lose a shit ton of money. It's not worth it.

Yeah, I know. But all the beautiful women, all the beautiful women, I need to buy a rose. And this is what a podcast, mate. But in all seriousness. All of these companies, they haven't got anything new. And right now, they've convinced the market they do. So, your question around Sam, Alton. Sam, Womton, he's been around a while.

He is very charming. I've known an alarming amount of reports to say, but he's so nice. It's like, if that's your excuse for covering someone. But in all seriousness, he's been around for a while. He's been a very successful investor. You look in his history, it's a little bit fucking bizarre, though. Sam Morton founded the company called Looped. Looped was dogged

It was like a location-based thing It didn't really have a successful product Emmett Shear CEO of Twitch Literally went on the record and said I'm not sure what Loop does I paraphrase it was like I don't know what Sammoreman does all day It's really weird

Like I took over Open AI for like two days Really when Samormann got kicked out Emmett She was up there So he had this failing GPS start That he sold to Green Dot Because back then you could sell anything to anyone Complete failure of the company Paul Graham

founder of Ycombinator says in the New Yorker or New York magazine. Can't remember which one, that there was no one else who could take over Y Combinator. Why? Why did Paul Graham give Sam Altman

half of his allocation for Stripe, a company that was obviously going to become worth so much money? The root of Sam Altman's wealth comes from Paul Graham's decision to give him half his allocation to stripe. Anyway, Sam Orkman's been around

for decades, decade plus, and he's wormed in with the media and he's warmed in with everyone, and I think he's been looking for his fucking chance. He's a good carnival barker. On top of that, This desperation time in tech meant that when chat GPT came around,

everyone went, shit, finally, something, we can, we don't, we don't, we don't make things here. I, I don't, what, we have engineers, no, we're only here to screw customers. That guy invented something, let's do this. Had Sundar Pishai gone, no, this is stupid, would have died within a year, wouldn't have been a big thing. But because everyone followed Microsoft,

you got the situation where Sam Ortonman became the face of the industry. Every single AI article, chat GPT, chat GPT, chat GPT, every single one. There is no industry here. This is a PR campaign for one guy and one company, except this company is a complete, it's a time bomb. There is no fixing open AI. There is nothing, because if there was, after DeepSeek, they would have gone, we will come out with an efficient model. Instead, they went, no, we're going to make our models big and shittier. Yeah, we're going to give you deep research.

GPT 4.5, their big thing, Orion, finally. comes out. And his thing is like, good news. It's the first time, I'm paraphrasing, but this is alarmingly close to the tweet, which really I should get hobbies. Um, so it's like, good news. This model is the first time it's felt like talking to a thoughtful person, just saying like, fuck my other models, I guess. Bad news. It's very computer intensive and expensive. We're bringing tens of thousands of GPUs online this week and hundreds of thousands soon. That's your response to deep seek. All for a chatbot that's maybe a little bit more thoughtful. That's also insanely expensive.

Right. What did you think of the, oh God, the guy who ran the back rooms experiments where he put two clods in a thing together and they just talked

and they came up with the whole like religion and stuff and, you know, fart coin was born out of it and if I sit there with two dice and I'll keep rolling

them, they're going to come up with all sorts of different numbers, mate. Sure. You get two, you get seven. Yeah. Somebody might even get 10.

Was there any part of you that found it interesting? No. I found it so interesting. I was like these two... No, it's a random number generator, mate. I'm sorry. But just more complex, obviously.

Yeah, the thing is, had these been called library models? And had these been introduced, there's not a thing that has to steal from everyone. It was just a experiment. Kind of cool. Yeah. When chat GPT came out, I'm like, oh, that's cool. That's interesting. Again, it's got utility.

It's just that it doesn't have it... Yeah, but it doesn't have it on the level... I actually don't think you should conflate utility in doing something, though. I know what you mean. I know what you mean, because it's like... But I think that this is part of the big con. They kind of say, oh, it's got utility. And it will, this is, to quote Robert Evans of Cosa, this is the worst that will ever be.

It's been the worst that'll ever be for a long time. And it's just the reason I'm so animated about it is, first of all, a lot of people have been very rude to me. I remember. They've been astronomically rude. But guess what? I'm so much ruder. Casey fucking Newton. This shit bursts, I'm going to be touchdown dancer.

Yeah, for a year. I am going to bathe the Starways in this. No, that's Thanos. But it's, people have been very rude, but also what I am talking about, I know I sound and look and am crazy, but what I'm saying is not illogical. What I've been saying for the whole time is not illogical. Call me a cynic, a pessimist, a pig. They throw water at me when I'm outside my house. They spray me with a hose.

But what I'm saying isn't crazy. What they're saying is fucking insane. What Sam Altman does is insane. Yeah, we've burned $5 billion. But don't worry, we're going to burn $11 billion next year. We have no idea when we'll be profitable. This thing here, this needs 100,000 GPUs. Why?

I don't know. It's very expensive. I don't know if it'll ever get cheaper. I don't know when we'll make money. I need more money than you have right now, and next year I'll need even more than that. What is going on? Right.

I don't think you're crazy. I mean, it doesn't sound crazy to me. I think it's why I found it so comforting when I, you know, found where's your red at and, and heard people like Jim Covello because I was like, okay, other people are seeing this, right? It felt like, it felt like everyone else was trying to make me feel crazy. It felt like, honestly, a lot of the things were the people who would try to scare you about AI and, like, paint these Terminator futures for you. And I'm going, I don't know, I use the fucking thing. I don't think it's going to like, that's the,

these AI safety people are like, well, Dario Emmett. They, what a scumbag. CEO of Anthropic. He loves to go and, like, damply roll around and be like, well, it's going. want to be very scary. Give me a billion. The Miri Foundation, all those guys. Future of life and fucking Jeff. Jeff Hinton, is it Jeff Hinton? He actually, like, got a Nobel Prize and has, like, actual credentials.

But his whole bit is just going, the computer's going to kill us all. I need $20,000 to tell you how. Is it, is it, is this kind of just like how a few years back when crypto really had its big moment? And we were getting the Dow's and, uh, and, and the, All these companies suddenly talking about implementing blockchain technology. Is this just that, but on a much, much, much, much bigger scale? Of just everybody going, oh, shit, in the sense that Microsoft is kind of taking the lead, whatever fucking company it was back then, everybody's just going,

oh, yeah, we're doing blockchain also. It's similar, but so much bigger. Exactly. People mention, like, IBM loved saying blockchain, for example. But the scale of the investment is just crypto, like Microsoft had a blockchain bit of a Zio that they kept in the corner. IBM had a blockchain beer. I don't even think Google played in it. And Mark Zuckerberg with the Metaverse, things like,

ah, you're at Web 3. And he was just like saying whatever Casey Newton would write down. He's just like, oh, sounds completely implausible. See for lunch. But it's different in the sense that there's more desperation around this. Before with crypto, it was like, oh, Metaverse, oh, this might be good. This will get my stock a bump. This is, please, please, please, my, please, AI,

make growth please now, please me? Can we talk about the Ezra Klein show? Yes, I'm sorry. No, no. Let's look at this. I think this kind of lends to a bit of why it's different. And I do want to explain for it, because I don't want to expect that everyone has maybe listened to or read the transcripts of this.

Fucking embarrassment of bitches. So this is from the, this from the Ezra Klein show, which is a New York Times podcast. He had Ben Buchananon, who would be described as the crypto czar. under the Biden administration. The AIs are. Sorry. No, it's about as much. And honestly, anyone should go and listen to it.

It's a bit, or maybe Ed would disagree with that. But I think the last 15 minutes are the most important when he really kind of grills him about. Yes, and grill him. I mean, this burger is raw. I think, so the last 15 minutes, he really... Let's get to the grilling. Let's actually talk about the grilling. The last 15 minutes, he really explains.

poses, at least for me, this part of what centering labor and workers in the AI discussion means. And it's really wild to see, because he says, even in his question, he says, look, and we've had this forever. When tech is coming, you guys say stuff like, we're going to retrain people. We're going to make sure they have jobs, blah, blah, blah. So like, how can you answer that without giving me the, you know, wrote, we're going to retrain, blah, blah, blah. And he's literally like, yeah, I don't know what to tell you. Like, we don't have a plan. We're going to know why he has no fucking plan. You want to know? Because none of this exists. It's all the smokescreen.

Labor, the people getting replaced by this are freelancers, art directors. Like, people are losing their jobs to generate of AI, but not at the scale that they want it to seem like this. They are, if Ezra Klein was not a dope. And also, look how long this fucking question. I know, the questions are extremely long. I love asking a 600 word question so that they won't be able to answer it. But my point is, whenever they have these conversations, about labor being taken away. And by the way, I don't want that to happen. It's fucking horrible. Generative AI is not doing that. It's doing it on some jobs and it is taken away and it sucks, but it's replacing them with like slop shit that just sucks, but it's bosses that

don't do real work thinking they can approximate humans. And customers hate it. But a point I'm making is, of course Ben Buchanan didn't have an answer. One, he doesn't know a single goddamn thing that he's talking about. And two, how do you answer that question right now? It doesn't do anything that really still, this thing, and this was a point made by Darren Esamoglo of MIT's being on the show twice, legend, the big dog economist. He made this point where it's like, right now this doesn't do any of the things that actually increase productivity. This doesn't increase GDP. This isn't something that the way that these things function is not a, it doesn't do

anything in the real world. It doesn't really approximate human actions. Wendy's has horrible trouble with their genetic AI thing. And surely that should be the easiest one, because think about it, it's burgers. Just taking an order and, yeah, even if you are just taking, I don't love the idea of replacing labor, but if we're thinking on their level, okay, theoretically, generative AI should be able to take a person speaking, multimodal, voice to text, and say, okay, there's a limited amount of menu options. This should be really easy.

Apparently, it fails all the time. And it's just, that is the most basic building block of labor that they surely should be able to, not saying I want this to happen, just like, right? they should be able to do it. And the reason he doesn't have an answer is you would need to know what's replacing labor to answer it. The actual question there would be, hey, do you think this can replace labor right now? I did a bad job actually because I should have said that one of the most important things that Ben Buchanan says is that he believes that AGI or something approximating AGI, he makes note of it very many times that he does not like the term AGI, but something approximating

AGI, he says, is about two to three years away from the things he's seen. And the only thing, the only reason I do think that is somewhat relevant to the conversation is because a lot of times we hear this from people who are heavily invested in tech guys, people who are building these models being like, I swear, bro, we're just, we're so close. If you just give us a little more data, we're going to be, we're going to have AGI. So this is a, but this is a government actor who's like, from what I've seen. I'm so sorry to tell you about people at work in the government. I am so fucking sorry. Yeah. Go and look at Ben Bambi Kennan's LinkedIn. Go and look at it.

Sure, the revolving door is very real. No, revolving door. What does he actually know? What is his expertise? What is the fundamental? Oh, from the things I've seen, what have you seen, Benjamin? Biden showed him some stuff. Well, his, his, he's just got like a fridge.

It's a nice ice cream called. It's a kind of soda pop. Ben Buchanan's sort of working definition of AGI, he said in this interview, is it's basically, it's going to be able to do. what a human could do on a computer which is that's not a definition sure like it's the thing

he's like oh aGI'll be here in two to three years no it won't what are you fucking I'm so sorry I can't look at this with a straight face because it's like oh the things I've seen you haven't seen shit brother if you had you'd been investing in it and not telling people I'm deadly goddamn serious

oh I'm scared of it no you're not you're in the fucking Ezra Klein show wanking off and saying a bunch of fan fiction nonsense an AGI. Oh, it's an AGI that can do what humans can do. Humans can barely do what humans do. On top of that, what? We have this software today and you can barely get like a picture of Garfield with boobs successfully. But this thing in two to three years is going to something. The things I've seen suggest that, what things, man? What have you seen? Why is it always other things I've seen? I couldn't possibly tell you the girlfriend lives in Canada. That's where the AGI is. It's the kidding. How do you, how do you, because I understand and I appreciate what you're getting at. But, What's hard for me is the governments of the world, I mean, like, China's really, really gunning for this to try to, everybody's trying to win. Right. So are they all delusional? What are they winning?

What are they winning? They frame it as this. They bring up JFK in the 60s about the space race and how we need to make sure. How did the space race go, by the way? Well, we made it to the moon. What else? Satellites. Well, okay.

Yeah, we kind of ended there. So did we stop the other's launching satellites? No. Okay. So that, that went great. But on top of that, that's different. That's going to space. That's also very different in the sense that there was a thing we were competing to do. What are we competing to do? The biggest ass that can shit the most, I don't know. Buddy, I think I take the crown there. I think I am a white British man. Yeah. The things I've seen with my ass. Um, Blake runner. But it's, I'll tell you what, it is very, it is very, cathartic because like even as someone who's very skeptical it's I don't know why but it's very easy for them to to scare me for some reason

I read it and I go what you know Ezra Klein he's at the New York Times it's obviously the premier place for sharing ideas and stuff and this is Ben Buchanan he must have done and you go my God we're two to three years away it's genuinely

irresponsible as I know it is this should he should be fucking fired for this it is disgraceful that he did this and it's disgraceful. Casey Newton and Kevin Roos of the New York Times' Hard Fork podcast did the same shit, especially because Casey's boyfriend works Anthropic. He does this close it. Nevertheless, yes, they're trying to scare you.

Yes, Ben Buchanan, what does he do now? Genuinely, what's his job? I actually don't know. But if I had to guess, he's trying to become a consultant. A consultant goes to companies and goes, hey, you should be really scared of the computer. How? I couldn't possibly tell you. It's all the secret of you.

He talks about how the real, the crux of the race, if we're thinking governments between the United States government and the Chinese government, what's going to matter the most is our respective AIs and their ability to sift through massive, massive swaths of data and intelligence to glean from it. Like he uses the satellite imagery as an example. There's tons of satellite imagery that would take days or weeks or months or however long for humans to decipher and close. We already have machine learning that did it already. So I don't know if it's just a matter of that getting more intense. Why does any of this cost MVIDIA GPUs and data centers?

That's the thing. There's a complete disconnect between the narrative and the thing happening. Yeah. Nothing. And yes, by the way, if you're wondering, yeah, every government could be chasing their tail in the same way. Yeah. Macron has any idea. What do you think, Kirstama has any idea how this shit works?

Do you think any of these people do? Microsoft only put the money into GPUs or chat GPT because they wanted chat GPT and Bing. And I am not kidding. This is reported. It is insane how one of the most horrible things, it's a bit of history for you, actually. So when I start better offline February 2024, I remember thinking, okay, I think I've kind of got startups worked out, right? I'll look at big tech because something I don't know, right? These people have strategy and planning and they wouldn't just do a bunch of stupid ship. The more I read, the more I realize nobody has a plan at all. Not a single one of these people has a real plan. They don't have

anything. They don't have shit. They want to sell. They want to sell something, but they have, because they have chased all the people out of the company who build things and replace them with management consultants and product managers and people that go, what will make us grow? They do not know how to build things anymore. And the truth is, they may not be more growth things to build, or they may not be for a while. We may need breakthroughs and hardware don't exist yet. It's possible. There are cool things

happening in batteries alone that could be really cool. But again, not a hypergrowth market. And when I say hypergrowth, I mean, that allows them to get to a four or five

or six trillion dollar market cap and to keep growing 50 to 20% every quarter. They are all heavily invested in this and of course, well, governments are going shit. We want them to keep building stuff here

because that's spending money in our country. And the other countries are doing it. And the other countries are doing it. If we don't do it, we'll look like a fucking idiots, won't we? And if everyone looks like an idiot too,

well, everyone can, Everyone can come to the same conclusion and say no, let's stop this at the same time. And also, man, you've probably read quite a few places about all the countries building data centers, right? You know how long a data center takes to build three to six years? This shit can all be canceled. Microsoft's canceling it. Yeah.

Right. And he's saying in two to three years, he thinks that AGI is. If I was Ezra Klein, I would have a megaphone and I would go right up to his ear and I go, how? How? How is this going to happen, Benjamin? No, Ezra Klein seems to think that he's right. Yes, because Ezra Klein is a fucking moron. He says from everything I've seen, like, you're right, we can't stop this.

This man, what is Ezra Klein seen? I'm so sorry. Again, I'm getting back to the people being mad and being rude to me thing. I will be like, hey, look, this company burns billions. People are like, idiot. But Ezra Klein's like, I have seen things. What are the things? I can't tell. He calls out that he's used, I was shocked. He calls out that he's used deep research.

and it was providing him with scientific-level papers that, like, his staff couldn't produce. And I was like, what? I would kill myself if I worked for him, but... I was shocked. I was shocked. I'm actually not shocked to hear that because he's a moron. But on top of that, I would be, like, if I'm one of his researchers, I'd be like,

fuck you, Esre. I would start writing just completely fake shit in there. I'd be like, yeah, dog poop is the tastiest. Like, just like, see what else wrote, like, Ron Burgundy. But, no, this motherfucker hasn't read an academic paper in his life. If you look at anything on J-Store, just type in any word, you'll get an... It's just so frustrating because this is what happens when an entirely narrative-based thing takes a market. Sure.

And the, I'm not going to say it's apocalyptic. It's not going to be great financial crisis level, I hope. But the entire stock market, like 30%, 35% of it is the magnificent sale. 19% of the magnificent 7 is NVIDIA. NVIDIA is entirely, their entire growth story is insane. They can't, in video, on top of everything, is unsustainable because their entire, the markets are mad at them because they only grew 76% year over year because the market's got used to triple digits. How's that end?

How's that go? How does this keep going? And on top of that, why? Like, really, it's just, all that's going to happen is someone's going to go, why are we doing this? This sucks. I don't like, I feel icky inside now looking at this. we have, end of April, we have earnings for Microsoft, Google, and META, I believe. What happens there?

Meta's a weird one because Mark Zuckerberg can never be fired. Look, that beautiful man. It's just, I've been trying to warn people. Yeah, I don't even know what about, I don't know what will happen, but I do know when this all falls down, there are no hypergrowth markets left. And at that point, I believe there will be a fundamental re-evaluation of what a tech stock is worth. And I think that will be very scary for the markets and for the industry at large. I think it could actually be good long term, specifically for making tech make shit again. But we are in just a farcical era.

And if this AGI bullshit happens in two to three years, I will gladly eat crow because they're going to have to do insane shit. They will have to have something that they have behind the scenes that they've shown not a single sign of. They still can't get it to stop fucking hallucinating. Like, come on. What would it take for you, what do you, how would you define AGI number one? And number two, what would it take for you to go, oh shit, and give me that crow? First question I'm not going to answer because why do what fuck do I have, I, sorry. No, I understand. It's okay.

Why do I have to justify my, I literally, you're the one who sucks. I know, I literally had a DM asked me exactly this the other day. And it's just like, why do I have to? Yeah. The answer is something that's actually autonomous and can actually do shit, that you don't have to like, oh, okay, I prompted it. okay, I prompted it wrong. Okay, I'll prompt it again. Okay, it kind of works.

Okay, now 17 minutes into it, I've got something that kind of sucks. I want something you press a button and it just does a bunch of shit and all. I mean, from my experience, they barely have memory of the last thing. It can barely reference the last thing it's spat out for you.

And it's the most frustrating thing about these chatbots is because I go, if I just keep tweaking, we'll get there. But you'll say, okay, that's close. What if we did it like this? And then it shows you something completely different. And you go, what the fuck is the matter with you? That's because it doesn't know anything.

Yeah. And they have, they have, as context windows where you can dump more stuff into them, sure, whatever. But I would need it to work is actually the answer. I would need it to be like, I need you to go. I need, I need to say to Siri, I need you to order dinner for me. I'm feeling like Chinese. I want to spend no more than $35 and I want it quick. quicker than 20 minutes.

Sorry, I didn't quite get that. I'd be like, setting a reminder to kick your dog. Like, I, but actually, no, I would like, I want all the coverage of this client from the last two years, and that's all I want to say. And now I want the spreadsheet that comes out to be fucking perfect. I know that sounds like a lot, but it actually isn't. It's actually in line with what everyone's claiming you can do already, and it's not even slightly close.

I am in a document heavy business. I am an early adopter. I love my do-dad's my gizmos. There was actually a really nice Reddit post from a guy who probably came out in the last month. He's, I forget what he does in tech, but he was like, I just want to show you guys how incapable of these things are of doing the things you think it can do. And basically he set out to do research with it and asked it to do all these things. And he was like, you know, at first it looks impressive. It starts talking to you.

And you're like, wow, look at that. And then as soon as it starts trying to do anything somewhat technical. And not even technical. I mean, he's like, as soon as it was like, great, now I'll just start using an online database. Like, let me open Google Sheets and it can't sign in and it can't do anything. And you're like, you're just watching the thing just shit the bed. Yeah. And it doesn't know how to actually, you know, form data in a way that's going to make sense to a human.

And it's just really unimpressive when you get up close to it. I've got one. Remember? And this one impressed me because I, this is coming from someone. I've used chat GPT and AI shit a handful of times. I don't find any practical use for it. It doesn't affect me. I should use it as a stock trader for calling, I don't know, X1s.

Why should you? I don't know. To pull pertinent summaries from earnings calls. So that right there, by the way, I know you're making a point. That is the narrative being used on you. That is the AI gaslighting. It's making you feel bad for the thing not really doing. You're like, I should make room in.

my life for this thing that I would have to pay for. That's not... Oh, I mean, the last year, it's all Twitter threads being like, here's how I'm using AI to like make myself a ton of passive income and you're fucking up by not doing it. You're a stupid fucking moron. Can I just be clear? Any industry that attracts those people should make you very suspicious.

That's very true. You were saying what impressed you. What impressed me was there was that famous story of the AI who was tasked with something and it came across a capture that it had to, you know, prove you're not a robot. And it went around it by hiring a task rabbit and lying to the task rabbit person and saying, I'm a handicapped person. This doesn't sound real. I'm sorry. You don't, do you not believe this? This sounds completely made up. It's, do you, do you remember this one? Bringing it up. I don't, I'm going to, if I'm, if I'm, if this is true, by the way,

I'm still unimpressed. Yeah. This sounds made up just because of the task rabbit part. Yeah. Taskrabbit is not a well-designed website. It was, yeah, it was a person or however. Yeah. Open AI created operator, which is allegedly made made meant to do just that which is meant to run websites and it can't do it yeah how did it order the task grab it was it connected to the task grab at API I have no idea that's the thing all of this stuff is just this weird narrative it's like my mate yeah my mate uh it's like the bit where ralph wiggum walks in and the simpsons and goes yeah the baby looked at me except it's holding up her entire economy yeah it's insane it's it turns my stomach because we could have

the panic that's kind of happening right now. I realize some of it's Donald Trump driven, obviously, but it's also the Mag 7 way, and the Mag 7's getting crushed. The socks, it looks like shit. Well, they had to come up with something. They did. I believe it was you, Ed, who said,

our economy is dominated by people who don't do real work and don't solve real problems. So you have companies... Excuse me, I'm not doing the accent. So you have companies selling services that are cheap enough to sell to a boss who isn't going to use them. These people do not participate in reality.

In other words, it's the bean counters. These companies have reached, so it seems, just their absolute limit in terms of what they're able to innovate and iterate on and continue to build. And they're just like, I don't know, let's make this fucking thing that we can get our sales people to sell to this other company that their fucking vice, vice president of vice sales, something, something is going to be just impressed enough by it to buy a subscription that's going to, like, make their boss look happy and it's going to make us look good

that we met our sales quota for the quarter and everybody's happy and it's just kind of this all, this big coat. Did I... You have successfully described how heck has made money for years, nailed it.

And then there's the other layer of they can't even sell it like that. They're not selling it. They're having real trouble because people keep in like, yeah, we were using it. It sucks. And they're like,

and the companies go, wow, what if it didn't? Right, when? And they're like, we'd, Oh, I was just making conversation. This is going to stay pretty bad for a while. It's got reasoning now. Great.

How does that work? I can't tell you it's more expensive too. Sorry. Yeah. It's just, it's crazy. How do you feel about the argument that, that, you know, maybe that AGI is not coming, but it's more going to be these very much more limited in scope. Agents.

Not even agents, LLMs that focus on a very specific thing. That's cool. Those can be very, like, I can see... But will they be profitable. Good question. That is a great question. They also will not scale. They will, you'll get spec.

They'll be useful for some people. And there are startups that use specialized models connecting to distinct databases and those exist and they will exist. In fact, the biggest problem is that all these models are very similar and thus they're easily commoditized. Deep Seek was trained on outputs from Open AI. It's really funny because they were like, should we sue them for copyright? It's like, yeah, damn, someone stole from my plagiarism machine.

Yeah, like, shit. Dang. Good luck. Suing a company in China. Oh, yeah. You're going to, I'm sure they're going to help you out there. I think I have the perfect thing that kind of summarizes this entire thing, which is a fun bit of utility, but absolutely no real way to profit off of it. And it was someone who took a clip of our show and ran us.

It's really, I think it's so cool. It's so fun. Here, let's just play this clip for it. I think you'll appreciate it. One ran us through an AI that not only changes your language, but makes it look like you're speaking it. The language itself is really impressive. Damn, that's so natural. It's weird to watch.

I can't wrap my head around it. I like this one, too. How about tweet, Diet Coke, roll quote, say that? I forgot about that. The lip thing, it gets pushed over. It's so funny, because all of that has some utility. Yeah, it's fun, but it doesn't make money. It doesn't make money and also, how does that scale into something?

Because, again, you get back to the library model idea that this could be like a $10 billion industry. Sure, there's probably like a $300 million startup that did just 11 labs, I guess. They're worth billions now. Yeah. They shouldn't be. It's also very funny. It's always the problems they solve are always things that are kind of sad to me. You know, the great promise of tech was like, you're not going to have to do annoying things anymore.

But instead, like, the only thing I could think of for that, scaling it on to a bigger model is like, okay, now media companies could somehow now make sure their content is available in all across the world and different things. Redid's doing that. They're trying to. But I'm like, that used to be a thing they would get like different actors to do and there would be a different thing. And now it's just like, all we've gotten rid of is like artists and voice over people and like with SORA. They're like, you'll never have to hire a center. Saur looks like shit. Sora looks like a dog's dinner.

It's also insanely expensive. It's so funny how bad sore it is. And some fuck nut on Twitter is like, check this out. And it will be a guy whose face changes three times as he turns. Like a caveman using a laptop. They did it in the Michael Jackson black or white video fucking 30 years ago. Wait, did they? Yeah, you remember the face is morphing?

Yeah, but they did that deliberately. It's white. Oh, right. Yeah. It's not a hallucination. Yeah, yeah. It's great. I really do like it as well.

because it's very childish, how they, it's like, check this out, and it looks like shit. Yeah. But it's a full-grown adult who makes $100 million. You're like, I, do I, do I, do it? But the tech media's like, God fucking, holy shit, holy piss, piss in my mouth. Wow, incredible. Yeah, again, if we're to apply this to the real world, sure, I can see visual effects companies and studios using this stuff to, why are they not using it now?

Why are they not using it? That's actually my, that's the very basic thing. They could be using it. Why are they not using it now? It's not good enough. When will it get better? It will. When we fork over more money.

It will need more data. But what's crazy, though, is the... Yes. But also, they need so much more data for SORA than they will ever have. There's not enough video stuff. If you think that they need the entire internet's worth of text to train chat, GPT, how much video do you think they need to train SORA? They're literally paying people to take videos of people walking around because they do not have enough.

it is an insane idea that they have no proof we'll ever get there. You can't even, I think runway lets you slightly get some reliability between shots, but like, I don't know, if I'm making a movie, I want to make sure the guy looks the same when the camera changes. And it's also not quicker or cheaper. It's just none of this makes any sense and it looks bad and it sucks, but because everyone's kind of agreed it's important. Everyone sits there and claps for whatever they do.

It also kind of hurts that they've built this all on the back of our data, right? Like, all of that is, if you've engaged online or made any kind of content or anything, like, that's just being fed into, into all these machines so they can now put a bunch of people at a work who they're like, well, just... But they're not even going to... They've managed to put, like, SEO writers and copy editors and art directors out, which is really fucking sucks.

Yeah. But they want to do more and they can't. They can't even, like, they've just done it to, like, people who are, like, making contract work. they've made people who are already kind of struggling struggle more. And they've done so for the low, low price of tens of billions of dollars

burned for no reason for this horrendous, ugly machine. And people want it to be true so badly. And I think some of them want it to be true because they realize that they've said it was true too many times and that they can't back off. Actually, I have a counter-argument. If you said this was great and important

and then go, actually, I no longer believe that. I actually think that that's a, That's a brave thing to do. I think there's something courageous about saying I was wrong. If I... And I really am not. If AGI pops up in three years, I'll eat shit. Like, I absolutely will.

I don't care. If I'm wrong, I'll tell people while I'm wrong. There's nothing wrong with doing that. I'm not because unless there's just... If the only argument is magic, which, by the way, is how Sam Oldman described GPD 4.5, it's a certain kind of magic. Motherfucker.

This magic? Fuck you. Fucking... Anyway, the difference is not that stark either, which is the most, like, I'm a pretty ordinary user, want to use it for some, like, easy research stuff. But when Deep Seat came out, I got it before all the hype. I was just, someone had tweeted about it before the meltdown and everything. I was like, sure, I'll check out another thing.

I was using it. I was like, feels just like any other one doesn't feel any different to me. And for them to be like, that's our magic one. I'm like, well, I don't know where. Yeah, that's what's confounding to me is they went, just like you said, oh, so the other one sucked then? Yeah. Hey, we got, now we got the good one. Hey, that one that we said was the best three months ago, it actually sucks ass.

Yeah. And this one's thoughtful. Yeah, this one's good. But yeah, I just, every time I see, it seems like, it truly feels like every other week, Sam Altman is tweeting. And then 4.40 and then 4.5 and plus and stuff. And I'm like, what the fuck does it do, dude? Yeah, I do think they've done a terrible job on themselves with branding, even as someone who was like kind of following from the beginning and subscribing, I was like,

what is what? And, oh, 3, and mini, and... I genuinely think what happened with Altman was he thought this is gonna just exponentially go like this. And when it didn't, he's like, fuck, I have no idea what to do. Like, oh, God. Because the branding is insane. It's confusing. Just as a user who's like, not completely... There's 03, there's 01, there's 01, there's...

And I'm like, which one do I fucking need for? Yeah. And that's the funny thing. You go to them, you like, what do I need? And they're like, well are you trying to put tits on garfield or you just do that with the gpd4 are reasoning when it was my favorite thing where the tech media really shut their pants was when oh one came out their reasoning model and you watched every single right i'd be like yeah it's really good it thinks for longer not one of them seemed to be able to come up with a use case because there wasn't one and there still isn't this is the main thing the reason the reason deep

see chat everyone up is because they made a much cheaper reasoning model and that was oh one open AI's like moon magic. There's like no one could possibly do reasoning now everyone's doing reasoning. And what's reasoning? Well, it thinks about thinks, it's not thinking. It basically runs prompts and then goes hey, is this bullshit? Is this bullshit?

Is this bullshit? Which helps for a much larger price. And then you think, wow, reasoning sounds amazing. So now what can it do? And the answer is science, science. And you'd think open AI would I don't know, run a PR firm.

If I'm launching a product, I might want to say what it does. But the thing is, is that Sam Altman has actually done something kind of incredible, which is he's kind of revealed a big failure in the markets and in the media. And that the media does not know what they're talking about when it comes to this stuff in many cases. Yeah, I think a lot of them don't understand what's going on. But on top of that, he's used the fact that for about 15 years, they'll just print whatever the fuck the tech companies want them to. So they've just almost just like, I can say anything and I'll get news.

And for a while he has, except even he has run out of news. Right. And it's just, it's pathetic because if things proceed as they're proceeding, people's pensions are going to be fucked. People's, like, the economy is going to take a meaningful hit from this. This is going to be, and then there will be a dark age in tech. It's scary. Which lends itself to why you're kind of sounding the alarm, because I'm sure some people are

going to be like, oh, this guy's just a fucking hater. It's like, no, there's real world. There's absolutely people's lives. jobs, tens of thousands of layoffs from the tech companies, guaranteed. Billed trillions of dollars on the line in retirement phones. What about those guys who are making 800K as like AI

director or whatever? That guy... Well, they're for their jobs, which is very funny. But all the fake McKinseyites doing AI bullshit, those guys are being there sent to the glue factory, like Boxer and Animal Farm. I do think the dark age of tech is a very real thing. Like, I mean...

We've reached a limit. Even... You could tell. The internet has changed. for the worst. It's a, it's a much more difficult thing to use. Even just Googling was the most like recognizable thing on the obvious product. Yes, it was the most recognizable thing on the internet. People often so many times their homepage, like if you went, if you started a new tab, it would go right to Google because you're like, that's where the internet starts. I start right there. I type in what I want. And I never thought in my life that that would become a cumbersome experience where I'm like, oh, it's spitting me AI slop.

I did a whole story about this. It was my big break last year. It's called the man who killed Google search. There's a guy called Prabagar Ragavam, who hates me due to the fact that his SEO is all me now. Ah, so proud. I couldn't be happier. So Google search in 2020, there was something called a Code Yellow. Code Yellow happened, which is said, hey, queries are down across the board. Now, queries are when someone searches for something on Google.

It's a useful metric in the sense that if queries are down, something has changed. However, it is not a good barometer of success of Google. And indeed, there was an argument between the head of search, Ben Goams, and the money people, Jerry Dishler and a guy called Prabagovang, who's the head of ads at the time. Long story short, you can go and read it. Prabagal Ragavam pushed and pushed and push, so they made changes to increase the amount of times that people were searching on Google. You may think, surely changing things to make more searches happen means they made it worse.

Completely fucking correct. Prabagal Ragavand took over Google search in June 2020. Ever since then, well, things in 2020 went bad for a lot of people, but Google search specifically really started to take a turn. He basically changed it so that you had to make more searches per thing you were looking for. When I say he changed it, Prabagal was the head of it. He was the head of ads and the revenue side barking to reduce the code yo, to have more queries, whatever.

So that they could sell more ads. So that they could show more ads. And though there is no definitive answer, and I found all of this because of the emails as part of the antitrust trial against Google, they're all out there. They're all in the article. And what it looks like they did, and I compared with like the various companies that cover search, so outlets the cover search. If you look, it looks like they rolled back updates that suppress spam. So you've got worse results. They also changed it so it was harder to see ads in Google search. And if you look at Google search from 2020 on, it just goes shittier and shittier to the point that now it's just you'll be lucky if you find a result on the first page and it'll probably be a Reddit post and now they have a deal with Reddit so they can feed Reddit into it it's just so fucking sad and it's what I call the rot economy it's that all these companies are growth focused I mean it really is sad it's it's like just what you're described I mean a race to the bottom just to get more eyeballs on

ads race to the bottom suggests that they care all of these companies make their experiences worse to make more money look at Facebook Facebook's borderline unusable Oh, my, I know. Look at Yelp. Every time I've, on occasion, I'll use Yelp. I used it the other day to find a dog trainer. And it just showed like 10 sponsored results first. And I, as I was making my way through, I realized, oh, shit, I'm not even in the actual ranked, you know, the number one or the mafia. Yeah. The mafia put them on the top. I do, I do real fast. I want to share with you guys one. And there, there's many out there. But it speaks to, again, it's on a much more. serious level. It's like the AI doing our face and the language. It's not necessarily something that's going to make enough money over the long term to justify or make up for the tens of billions of dollars in infrastructure and research and all this shit that they put into it. There's the Commonwealth Bank of Australia. Just a few months ago outlined how artificial

intelligence is helping to transform the experiences. They had a 50% reduction in customer scam losses aided by the implementation and use of safety and security features which use AI, including something called name check, collar check, and customer check. A 30% drop in customer reported frauds due to measures

like generative AI powered suspicious transactions alerts and AI powered app messaging, helping to reduce call center wait times by 40%. But how much... Yeah, it's great for this company. Only one of those things was generative AI by the way. Yeah, Gen. You'll notice that I powered suspicious transaction

one thing and all the others were just AI. What that means is none of this was generative AI. It means one thing and also suspicious transactions. Bullshit. I'm sorry. You would never hand off a finite. Regardless, even in that massaged press release

or whatever that is, AI, AI, this is the actual real, this thing we're talking about. How much did this bank pay for that? Also, they put it out so that their fucking shareholders are happy to think they're futuristic. They actually only have generated. Notice how they love to conflate these AI things. It's so cynical. Yeah. It's so

cynical. I mean, it's a Super Bowl commercial thing. Just a few years ago, it was blockchain, blockchain, blockchain. And now it's we're using the power of AI to fucking AI, your AI. Super Bowl commercial from 2024 Microsoft Copilot. It was just people typing prompts into Microsoft Copilot.

One of them was, draw me, sorry, write the code for my 3D open world game. And you know what happens when you type that in? It gives you a rundown of how you build an open world game. It does not do the thing it fucking said in the fucking commercial. These people have been lying for years. And it's sickening. I have an annoying one for

co-pilot. I mean, I'm still probably one of the only people who likes the Microsoft Office suite. I think... What the fuck? We've talked about this on the show. Don't act like you're surprised. I can't remember... Well, my brain sucks, so. I use Google Drive and for anything that's going to be... Use Google Drive. That's a sick thing.

I use Google Drive. We use it for the show. It's just easy for document. It's not easy. Google Drive was like an Desherpaint, Dropbox. Dropbox for documents. Oh, no, no, no. Oh, okay. Just for docs. Oh, okay.

That's saying Google drivers a cloud storage thing is. No, no, no, no. Just purely for docs sharing. And then any kind of, no, no, I'll get. Any kind of like, any kind of thing that I'm not going to have to, like, share easily. If I'm, like, doing personal taxes stuff, sure.

I like Excel on my thing. I like using a word doc. All of a sudden it updated and co-pilot was just all over a, a word document and I was like I don't need this like and I was trying to be quick I was trying to add something and I just could not get into the document and I was like

why are you fucking me like clipy on crack or something basically Gemini's doing the same thing in Google box. You cannot get copilot off I just want a fucking blank word document and you cannot have it you have to have copilot but what if you want to create a parrot like yeah what if you need to generate the ice cream come? What if you need to generate

of the paragraph? Yeah what if you want to put tits on the paragraph? Yeah well if you need a big book paragraph But that's the thing, though. What you're experiencing there is why I think this is calamitous as well because consumers fucking hate this shit too. You don't get anyone who's walking around

being like, God damn, like, you get perverts on Twitter being like, I love AI. But like regular people, like, I don't know what this is. I used it to write a limerick or I actively despise this intrusion into my life. Which is insane.

It's just contempt from the tech industry for the customer. And I think the tech industry has turned on the customer. completely, because they've been able to make so much money, making things worse. And eventually, the tide will turn. It feels like a, a digital let them eat cake kind of thing, where they've become so out of touch with what regular people want and need and are actually, I mean, I understand

there's a, there's a bit of like, people don't know what they want. We're going to give it to them. Like, they wouldn't have known to ask for the- They're probably Steve Jobs like 18 years ago. Sure. But at this point, it's like, I don't, I don't need the thing. Nobody I know I know at least five people And none of them

Use Gen Moji Like it's such a fucking I feel like if a friend Not even my mom uses it If I got a Gen Moji from someone It would be like There sent me a picture of their asshole I'd be more open to the asshole

Yeah It would definitely be a oh you're one of those guys Yeah exactly like Like learning someone's a fan of disturbed My worry is that the internet is Oh We're not We're not going

back like the you know no one seems to care right for example like even things that we kind of used instagram for example they've been making it worse and worse and consumers complain and we just continue to log on every Twitter great example they make it worse and worse yeah we log on every day and we log on every day and we go this product's fucking worse and they go we don't care I mean my mind was blown when we we talked about shrimp Jesus way long ago we're like what the Yeah, we were the first. Surely, and then it was like months later. What's going on over at Facebook?

And then I was shocked to see Mark Zuckerberg come on video and say, I love it. I love that we're like... It looks like a clown without his makeup. I think it's pretty cool that the computers are talking to themselves. I actually have a pushback on this. So when I started the show February 2024, and I was saying these things, people are like,

you're just complaining. It's only fairly recent that people have really... People have realized they're the frog in the park. They've realized, wow, everything has got worse. Why is this like this? And the answer is growth. It is the desperation for growth. It's the fact that these companies all want growth all the time.

And it manifests as everything being shittier and more broken and more. But I think people are waking up to it. I think people see it. And I think what they have to realize is, and ironically, this is a prediction from Shingy. I think people are just going to stop using these things. I think that the more likely thing is people are just going to go, wow, Twitter is just like every response is other AI generated or someone writing the 14 words and it's

it's just okay I'll go on blue sky if they don't like blue sky what they're gonna do use Facebook great I'm gonna find my friends between the 18 sponsored posts and the different shell games

that Meta is playing to get me to have one more advertising impression yeah have a fun fact about Ryan Barwick over at Marketing row wrote a piece last year about how Facebook advertising managers the people that put money

into Facebook ads for their clients are losing money and that Facebook you'll be like I'll put $1,000 and it'll go across a week and spends it in a day. And Facebook's just like, yeah, that'll fucking happen. What are you going to do? Advertise on the other map.

It is a mob-like operation. At some point, this has to break. There are so many things that are so shaky about all of this stuff and you can dismiss me as a cynic, miss me as a critic, whatever. But if you look at what's happening, you look at it, you've got all of the beloved apps becoming worse, all these companies making more money than ever, and then they're taking all that money

and putting it into this unsustainable, unprofitable, ecologically destructive thing that no one wants. How does this end well? How does this turn into AGI? How does any of this work? Do you even think that if they made AGI, which they will not do, that they'd make it useful?

Do you think that they'd, I don't know, release it to the public? Fuck no, they'd use it internally. They don't have shit. They can't make Google search work anymore. They couldn't find a way to make Google search profitable and good. They couldn't make Facebook usable and profitable. They can't even run their companies well. What the fuck are we meant to do with this supposed magical intelligence? Well, I think that's

kind of the problem too, right? So like all these companies, at least for a while, they were good at getting everyone's attention. But then once they had that attention, they were like, well, how do we make money off of that? And they were like, I guess we just ruined the thing that people like with ads. They did really well for a while, though. There was a period, like, I'd say up until about 2014, 2015, when things were fine. They weren't, like, nothing was ever put. They weren't, like, nothing was ever perfect. They were problems with Facebook, but it was before they really turned the screw in the butthole. Back when people were putting filters

on their photos on Instagram, and it wasn't cringy. It was like, look, I made it look like an old picture. Isn't that cool? And they were making bank, they were making billions of dollars, but these motherfuckers needed to grow. Not good enough. It's wild to look back at Facebook. I think about

this every once in a while, how I believe it was 2013. The Facebook, when Facebook IPOed, the big controversy as you, I'm sure remember, was we haven't quite figured out how to monetize our app, which in hindsight is fucking bananas. It was

mobile. It was they hadn't worked out how to monetize mobile. It was zero. They had zero dollars in mobile ad revenue. They didn't have zero dollars even. That was the thing. I remember at the time, not knowing anything. I was a dib shit back then. I don't know why I'm now. But if you read back then, it was people saying there's not enough revenue

in mobile. They haven't worked this out. It's insane that, like, you can say that Facebook got treated unfairly, but they did. and then they turned on the spigot and it was off well no

they proved themselves to the market and Cheryl Samberg she was the yeah she was the COO she's a McKinsey

she's McKinsey they're everywhere these NBA pricks the growth team the original growth team at Facebook have you olivann

Naomi Gly Alex Schultz and have you olivand don't know I've said him yet

all those people were picked by Cheryl Sandberg and Schmath and to lead the growth team those people run Facebook

oh you want Here's a fun story about Facebook for you. You should tell you everything. Now, 2017, a guy called Andrew Bosworth puts out a letter internally at Facebook called The Ugly. And one of the things he said, and I'm paraphrasing very lightly, it's quite close to this. He says that a terror attack could be carried out on Facebook or someone could be bullied. And it's all just, all things are justified undergrowth.

You would think saying something like this and it got out to the press, people like Alex Kantrowitz over at BuzzFeed News reported it, you'd think a guy like this would be fired or disciplined somehow. He is currently the chief technology officer at Meta. that's what this company rewards that's what these companies are built on these are the people running the tech industry these are the people who are making the decisions that shape our lives that shape the culture

that shape the direction that we're heading in and yeah and horrifying right wing machines growing meta is responsible for what we see today everything there are other participants sure but meta was the driving force of the conservative movement and it goes back years and it's disgraceful that Mark Zuckerberg's got on away with it

fucking scumbag. He's going to remasculentify the workplace. He is the least masculine man I've ever seen. I don't know, man. He does BJJ now. Yeah. And he wears looser shirts and a chain. Kevin Fedeline. Wow. That's

I wonder what happened to him. Well, maybe we leave it off on a positive note because we were talking about... Kevin Fedeline. Kevin Vedeline. I do like your point that it does seem like people are pushing back against this. And there is a real

there is a real desire for something else. I think I don't know where that comes from, but do you have any more like optimism for that? Like is there someplace where tech could go in the right direction for people to There's a shit ton of cool stuff happening

in batteries and silicon. It's, I mean, I like Anchor. Anker's a cool company. They make cool little battery packs and gallium nitrous. Oh, I had a couple of Gallium nitride, I forget what it is. I always mentioned this one because I love it so much. It's like they found a way to fit more power

in a smaller place. The batteries can power your phone super fast and they're tiny and they've got cable in it. It's not a novic. Have you heard of a Novix? No. I think they're probably a scam. GAN is like a very real thing though. That's cool.

There's a shit ton of stuff going on in batteries that could really change things. Eventually we're going, 5G is going to do the thing it would say, I believe that. Yeah. I don't know as far as social networks go. I have faith in blue sky.

I really like blue sky. And also remember, despite the fact that Instagram is the only game in town, TikTok came along and they had a shit ton of money plunged into them, but they took a meaningful market share away. I also think that at some point someone's going to go, wouldn't it be a good business if Instagram was 80% less shitty? And someone will do that and put a shit ton of money into it and just fucking brute force it.

You would think that one of the big tech companies with all the money that they have, that they can clearly put into data centers, would just go, What if we made a good Instagram and we used our marketing arm to push this on people? And people went, wow, this is exactly what I used to like about Instagram, except I no longer get interrupted by the most annoying advertisements possible. And I can see all the people's shit I want to see in chronological order. Wow, wouldn't that be good? At some point, this will happen. At some point, I actually think after the reckoning happens, this is more likely than people realize.

Blue Sky is genuinely cutting into Twitter is. dying. I have to eat some crow about blue sky and I don't mind owning up. I said, I went over there and I did not like what I saw. I was like, oh, the vibes are bad over here and I said it on the show. I was like, yeah, the vibes were off. What? Yeah. Recently. But what I realized was, what I realized was, I barely gave it a chance. And without it knowing, I think I was having a reaction to a thing not having an algorithm for me, just feeding me stuff. And I was like, what is going on over here? I stuck with it because I saw some commenters. being like, it's great. I don't know why you had the experience. I followed a bunch of people

followed you on there. I was like, and then all of a sudden, I started using it more and it knew me more. And I was like, wow, this is actually a pretty enjoyable experience. It felt like Twitter before, where it was a curated feed that I liked. Twitter, I mean, I've become so, uh, I used to really like it. And I thought it was a great tool. I thought it was great. It now, my whole growth came from that like yeah it now gives me just an anxiety yeah like 14 words a it shows me stuff i have no interesting it feels like they made a part of the internet that feels like a bad part of town yeah damn yeah wow yeah you could be safe you should be fine maybe there's a shop you want to go to you don't want to hang around yeah and it's somehow still so addictive of just like what's gonna piss me

off now or make me scared. I do want to own up and say that blue sky has been a nice respite from that and because it's a tool to me. You know what I mean? We do the show. It's a great way to call information and see what people are talking about. It's been more useful. People are like, hey, I wrote this article and I'm like, great, I can go right into there instead of people being like, ah. So, I mean, there is hope that someone can, someone can go like. Blue sky is relatively amateur as far as when it started. Like, it's way more respectable now, but kind of thrown together in some case. Like, Jay's very good

and the team's very good. I wish it had bookmarks. I wish it did too and that feels like something they'll do. Like group DMs as well. What I'm saying is if Blue Sky can happen, there can be another Instagram. Yeah, it is a real hopeful thing.

And consumers are getting pissed off. Like I said, when I started the show, I got emails from people saying, you're just mad at everything. Now people are like, I thought I was the only one, which is super cool. I love receiving that.

It also sounds like something I'd make up, but I swear I'm not. But it's, It also feels like society is more open to this. So I'm allegedly going to be in New York next week. It's going to be a huge thing. Like, the shit I am saying is being quoted in publications that a year ago would have been like tech industry is actually full of beautiful boys that we love.

Yes. How dare you speak of our damp beautiful boys and their beautiful computers. And now people are saying, huh, maybe everything doesn't. Yeah. And what I think it is is everything kind of sucks. Like every app fucking sticks. Every app's messing with you. You need consumers who are going,

hey we fucking deserve more than this but it's one other thing sorry it's okay but just it is I have tinnitus my ears ring all the time because they ring all the time I don't think about it like I hear it now because I think about it I believe that

tech has become digital tinnitus that everything kind of sucks every app's manipulating us everything is growth hacks so we use it more in the way they want us to to the point that we're all kind of numb to him but I think that they've pushed too hard I think they've finally made people go, wait, what the fuck?

And I think as that happens, users are going to stop using him. That is what's going to scare the shit out of them. And then you'll think, well, they'll fix their services. Now, they fired all the people that made good stuff. They only have growth hogs now. Face app is still good. I love FaceUp. FaceUp, one of the original AI scares.

I love that. Man, dude. FaceUp was so funny when it came out. It's so fun. Everyone's like, sending data to Russia. I don't care. They can have my face. Well, no, my favorite thing with that, though, was they were like, yeah, it's sharing our analytics about each user with the holding companies. Like, yeah, we only do that with American companies. I mean, that's always the big, when Deep Sea came out, it was like, oh, have fun with the Chinese Communist Party having all your time.

Yeah, ask it about Tiananmen Square. Oh, that was the fair. Gotcha. Checkmate. But, yeah, I wish that Face App still did the race swapping. I think that that was a shame. That's race app. Yeah, was it race app? Oh, damn. By far my favorite tweet about like the China thing was there was just some medicine that they discovered or whatever.

And it was the headline of like, you know, Chinese discover some cure for whatever. And someone just retweeted it and was like, yeah, but can this medicine tell you what about 10 minutes square? That's great. QED. Well, Ed, what, what do you want to plug? Say whatever you want. Better Offline.com. Just go there. It's got all my shit.

Follow me on Blue Sky at Ed Zitron.com. I have like 140,000 followers on there now. I know you're crushing on there. And I can't quote people anymore if I slightly disagree with them. Because like, 1,100 pigs will be like, you piece of shit. I'll fucking kill. I will get like, I will murder. And it's like, okay.

Like I made a comment earlier about. You got the Ed Army. Yeah, it's great, but also scary. I made a comment earlier, I quote tweeted this journalist called Karen the Tiller, I think her name is, and she's like, was talking about how men didn't like the fact that she's muscular. And I quote saying, yeah, there are men who are probably like, they're scared of a woman who could crush them up into a cube, like a car in a garbage dump. I didn't get like, 30 perverts. And I was not thinking sexually. I was not, and there's like, yeah,

I'd actually love that. I'm like, fucking can't. I can't say anything without you hogs. You can't discount how many perverts are on the internet. And using the point emoji, which is the best emoji. Just use, if you have to see. Oh, I love the point. Yes. No, no. No. No. I can't. It's my friends, Caleb Wilson and Ari Fassan, who I do a football podcast with 60-minute drill, but not doing seasons over. I do an NFL podcast. It's so horrible. But it's... Wait, football.

NFL. Oh, okay. Do you know Smith, baby, Raider? So, how British are you? I'm extremely British. I mean, you can't tell by... Well, I'm just like... Oh, I'm very fucking annoying.

I know baseball. I know football. I do not like soccer. Oh, this is incredible. I moved away. He's got a shirt from Kentucky. on there. Yeah, yeah. I've got one of my favorite defunct baseball teams. I love baseball

too much. I love baseball too. But yeah, there's anyway, very long answer is better offline.com. Okay. If we, if we, when this comes out should we tag you on Blue Sky or something? Absolutely. I'll post it on Blue Sky. Oh yeah, my boy, I got to get on Blue Sky then. I guess I did. But thank you

so much for coming on. Thank you. I feel much more calm about you know, Ezra Klein and Ben Buchanan trying to freak me out. I feel less calm than ever about it. I'll be able to sleep better tonight so thank you ed for coming on this being great yeah thank you coming up on this week's episode of ben and amel show i guess but i really got it yeah mr white yes sign it now that is a great use of a i do you think so yeah i always loved that in movie trailers when they do that

where they just totally chop up different sentences to make and they just act like it's totally normal and we accept it like it's normal we have to get out of the white house no it's just uh it's funny it's one huge photo of a guy like in the middle of a tango dance pose with his wife and he's clearly like in his 60s and it says like thanks for letting me dance the tango again and there's one wow some pitcher for the uh Detroit tigers has his gave him a whole like his uniform and like a thing and was thinking i couldn't throw a single pitch before I went to the doctor Because my heart was full of orbies. There's some other doctor in L.A. who's just putting orbies.

Yeah, you know, I'm going to prescribe you 50 milligrams of orbies. It's a racket. They're in cahoots. And then this guy's like, I can get the orbies out of your heart.