Pod Insights Better Offline Radio Better Offline: Brian Koppelman, Cherlynn Low & Mike Drucker About this episode Welcome to Radio Better Offline, a tech talk radio show recorded out of iHeartRadio's studio in New York City.
Ed Zitron is joined in studio by actor, writer and producer Brian Koppelman, Engadget Managing Editor Cherlynn Low and comedian Mike Drucker to talk about the current state of the tech industry, what AI can actually do, and what the tech industry looks like in the next year.
Anthropic and OpenAI Have Begun The Subprime AI Crisis
https://www.wheresyoured.at/anthropic-and-openai-have-begun-the-subprime-ai-crisis/
Anthropic Is Bleeding Out
https://www.wheresyoured.at/anthropic-is-bleeding-out/
Brian Koppelman
https://www.instagram.com/briankoppelman/
Mike Drucker
https://bsky.app/profile/mikedrucker.bsky.social
https://mikedrucker.substack.com/
Good Game, No Rematch: A Life Made of Video Games
https://www.harpercollins.com/products/good-game-no-rematch-mike-drucker?variant=43103018024994
https://www.amazon.com/Good-Game-No-Rematch-Video/dp/1335012699
https://www.barnesandnoble.com/w/good-game-no-rematch-mike-drucker/1145272268 https://www.barnesandnoble.com/w/good-game-no-rematch-mike-drucker/1145272268
Cherlynn Low
https://www.engadget.com/about/editors/cherlynn-low https://www.engadget.com/about/editors/cherlynn-low
https://bsky.app/profile/cherlynn.bsky.social
https://www.instagram.com/cherlynnstagram/
YOU CAN NOW BUY BETTER OFFLINE MERCH! Go to https://cottonbureau.com/people/better-offline and use code FREE99 for free shipping on orders of $99 or more.
---
LINKS: https://www.tinyurl.com/betterofflinelinks
Newsletter: https://www.wheresyoured.at/
Reddit: https://www.reddit.com/r/BetterOffline/
Discord: chat.wheresyoured.at https://discord.com/invite/QUUQUP9szv
Ed's Socials:
https://twitter.com/edzitron
https://www.instagram.com/edzitron
https://bsky.app/profile/edzitron.com
https://www.threads.net/@edzitron
See omnystudio.com/listener https://omnystudio.com/listener for privacy information.
Transcript 17,027 words · 9 speakers
Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.
0:00 Speaker 00 This is an iHeart Podcast. Guaranteed Human. Empower is all about helping you invest well so you can go out and live a little. And their Free Empower Personal Dashboard shows you your full financial life in one place. You'll be able to see your net worth, monitor budgets, analyze your portfolio, set retirement goals and more. Whether you're saving for something big, want to treat yourself, friends or family, or meet long-term goals, Empower helps you see where you stand. You've worked hard for your money and Empower can help you make the most of it. Download the Empower Personal Dashboard or visit empower.com. Not an Empower client paid or sponsored.
0:39 Speaker 01 Every small business owner has that one moment that could have broken them, but remarkably, it didn't. Hi, I'm Kathleen Griffith, co-host of The Unshakeables, along with Stevie Barron, CEO of Chase for Business. We have more incredible stories of overcoming the impossible still to come in season three. Listen to The Unshakeables wherever you get your podcasts and learn more at chase.com/podcast. JPMorgan Chase Bank, N.A., member FDIC. Copyright 2026 JPMorgan Chase and Company.
1:18 Speaker 02 Call Zone Media.
1:21 Ed Zitron? Your scientists have yet to discover how neural networks create self-consciousness, let alone how the human brain processes two-dimensional retinal images into three-dimensional phenomenon known as perception. Yet you somehow brazenly declare seeing is believing? Yes, I do.
1:46 Ed Zitron? I'm Ed Zitron. This is Better Offline. I'm your host. Buy our merchandise, go to my newsletter, where'syoured.at. Anyway, fuck all that. Brian Koppelman is joining us here in the studio. He's the incredible writer, producer, and he's in the bear? Flipping bear, he's a real deal actor. He's also the co-creator, showrunner and executive producer of Showtime's Billions and Super Pump, The Battle for Uber. Brian, thank you so much for joining us.
2:08 Speaker 04 Thrilled to be here with you, man.
2:10 Ed Zitron? We've of course got Mike Drucker, the wonderful comedian, and of course you have a book, don't you, Mike?
2:13 Speaker 05 Yes, sir. It's called Good Game, No Rematch.
2:15 Ed Zitron? Ooh.
2:16 Speaker 05 It's about embarrassing myself with video games throughout my entire life.
2:19 Ed Zitron? Wonderful. I'll be embarrassing myself with tech on this show. Sherilyn Lowe as well, joining us from Man Gadget. How you doing, Sherilyn?
2:25 Speaker 06 I am about to down a protein shake, hopefully, yes.
2:28 Ed Zitron? Yeah, we were just talking about the Ninja Creamy and all our various slops.
2:32 Speaker 04 They're great. Ninja Creamy, I'll tell you, there's no endorsement out there for any of us. They don't need it because people just get these things and they immediately go to TikTok because they want to share how good they are.
2:43 Ed Zitron? Yeah.
2:43 Speaker 04 They're honestly, they're so good. That's all I full endorse.
2:47 Ed Zitron? Full endorsement. This is the Ninja Creamy show now. I'm going to get, what's great about this is literally anytime I mention any product, I will get an email from someone being like, here is a post from 2003 from the CEO, in which case they shot a dog. It's like so insane and he's like, Ed, how dare you bring up? I'm like, I don't know everything, please help me.
3:05 Speaker 06 Well, Fairlife's bottles have a lot of thalates in them, so you know, that's our, that's our thing today.
3:09 Ed Zitron? Yeah, I just eat like too much yoga. So I want to actually start with a really good thing. So Mike, you wrote a really great piece for The Gamer, I think that went out.
3:17 Speaker 05 Yep.
3:17 Ed Zitron? Yeah, very recently. It's called, I'm starting to worry this industry has no respect for the people who work on it. Why don't you walk us through the article? Because I think it's a good subject matter.
3:25 Speaker 05 Sure. Well, as listeners may or may not know, Microsoft recently laid off about 9,000 people, most of which were in their gaming department. Simultaneously, they were claiming that their gaming department's quite profitable and they're making a ton of money and everything's going well. I don't know if that's actually true, but they are saying that. And they've kind of also said that a lot of the money that they were paying these employees isn't, isn't about losses, it's that they want to move this money into AI development.
3:52 Ed Zitron? Did they say that?
3:53 Speaker 05 Yeah, they, yeah.
3:54 Ed Zitron? That's horrifying.
3:56 Speaker 05 Yeah, it's kind of, I mean, they didn't say that word for word, but it was very implied that the budget was, you know, they're more focused. That was kind of the language. The language was like, you know, as we restructure, we're more focused on AI going forward.
4:06 Ed Zitron? It's just so ironic, it's like video games is one of the first real exposures to AI for most people.
4:12 Speaker 05 Yeah.
4:12 Ed Zitron? I remember when Fear came out, anyone remember Fear?
4:15 Speaker 05 Yeah.
4:15 Ed Zitron? It was one of the first, one of the first, it was the first time a guy would like shove his head over and look around before just jumping up and getting shot. Yeah, it's just, the reason this really spoke to me, other than the fact it was very well written, is that I feel like this is the central problem with a lot of tech, entertainment, everything right now, where it's like, the people at top, and you make this point well in the article.
4:34 Speaker 05 Yeah.
4:35 Ed Zitron? That the industry is the problem, because the industry is not being piloted by the people creating.
4:39 Speaker 05 Right, exactly. I mean, you know, and it's a problem that also extends to the, you know, Hollywood, which I also work in, because I'm famous. No, that's not true at all.
4:51 Speaker 05 But it's this problem that like, you know, a lot of the decision makers are not as close to the product as they used to be.
4:57 Ed Zitron? Yeah.
4:57 Speaker 05 I mean, I mean, you could even say the same, same for things like Boeing, where, you know, you don't have engineers running the department anymore. You have business people who've been brought in because they're good at business.
5:05 Ed Zitron? Well, track alliance.
5:06 Speaker 05 Exactly, exactly. And, you know, these people are far away from the product. They're far away from the development of it, so they see these, you know, workers as numbers on a spreadsheet. And for better or worse, video game development is a very long process. It takes a lot of time. It takes a lot of money.
5:21 Ed Zitron? It's getting, it's getting longer.
5:22 Speaker 05 It's getting longer.
5:23 Ed Zitron? The AAA horizon as well.
5:24 Speaker 05 And so it's, you know, very easy for these companies that are looking for a short term, you know, next financial quarter boost to go, we fired all these people. And this product wasn't going to come out for two years, so it really doesn't matter.
5:36 Ed Zitron? And that's the thing with Microsoft as well, they've, that's that, they've laid off 15,000 people this year as well.
5:41 Speaker 05 It's nuts.
5:41 Ed Zitron? And it's, is the money really even going anywhere? It's just so confusing. But you're seeing it everywhere with Hollywood. There was the Acme versus Coyote thing. Oh, sorry, Coyote versus Acme thing, where they just like mothballed it, they're releasing it now, but they mothballed a movie.
5:55 Speaker 05 Yeah.
5:55 Ed Zitron? To save on taxes. I feel like that shouldn't be a loophole.
5:58 Speaker 05 I, I hate that loophole. I hate the idea of that loophole. I mean, it almost reminds me of that loophole from, and Brian, you could maybe correct me on this, but remember like in the 90s, they made a Fantastic Four film they weren't going to release just so they could maintain rights for it, because the rights for it were, you have to make a movie to keep the rights, you don't have to release it.
6:17 Speaker 04 I don't remember that exactly, though now you say it, it rings a bell, but I don't have particulars on it. But I think what you're talking about, really the answer, I agree completely as you diagnose it with how depressing it is. But to me, these things feel like a systems response, like complexity theory.
6:39 Ed Zitron? Yeah.
6:40 Speaker 04 And well, they're like systems responses to what's happening in the world as opposed to feeling like, because I think it's easier for us to look at the human being and go that motherfucker. And that may be a motherfucker. But another possible answer is that from afar, remove like something's happening, right? And when this thing is happening, it gets into quantum theory, but it is a complexity systems response to this giant change of artificial intelligence, having certain capabilities. And all of this, you can look at all of
7:23 Speaker 04 it. And I mean, we've seen people who had one set of beliefs for so long and on the record in your industry, completely switch. And they may think that's an autonomous decision they're making. But I have found a little bit of solace by reading about complexity theory, because that for me offers a potentially more not hopeful, but sort of a more complete understanding.
7:45 Ed Zitron? So you're saying like a systemic response to conditions, because I don't know about the AI capability side, but the existence of capabilities potentially being there makes sense in that they're all trying to reconfigure for a future. They don't know if it's actually there. The generative AI doesn't seem to do it. But the potential of that, the idea of it is motivating them so much. Microsoft, of all people as well, should know how little money is actually being made from this, considering out of the 13.
8:09 Speaker 04 But that's it. But yes, but you executives, I also, because I've studied these people so much and written about them so much. And I understand how venal they are and how short term they think at times. But they're also.
8:23 Speaker 04 I think looking at data that we don't have very often and they're trying to forecast out a long time.
8:34 Ed Zitron? Why do you think they have data we don't have though? Cause like with Microsoft, for example, and I have probably studied such a Nadel too much at this point. I really, I learned too much about the growth mindset and Carolyn Dweck and all of that nonsense. But the data, if the data is there, they're acting very peculiar about the data because they're only making thirteen billion dollars this year from AI. Ten billion is AR, I was from OpenAI burning giraffes and whatever they put into the machine for chat GPT. But it's, it feels almost like they are reacting to what they hope will happen or what may happen. I'm just trying to work.
9:10 Speaker 04 The data might just be when, when everyone reacts to quarterly, like the problem in our business, the entertainment thing is that these people are responding to quarterly earnings calls that they have to make. And maybe the data isn't about the long-term, maybe the data is literally that we don't know that they're looking at is their little cohort of people and the exact moment they can exercise which kinds of options, some of which they have to declare and maybe some of which they are somehow able to flip on a different market.
9:39 Ed Zitron? Well, I mean, that's what Anthropic basically, uh, Amazon did with Anthropic. They flipped their investment into a certain kind of thing that they could tax deduct. Little fuck.
9:45 Speaker 04 I mean, there's no doubt that there's heartlessness, but I, I just, because Drucker's here, I just want to say, you know, even in the most depressing times or in moments where the technology, the platform seems brutal, people's humanity can transcend it.
10:01 Ed Zitron? Absolutely.
10:02 Speaker 04 And I remember in some dark days of Twitter, um, Drucker was, and it was really amazing, man. And I remember, you know, you and I don't know each other well, but we've known each other a long time.
10:10 Ed Zitron? Like 20 years.
10:11 Speaker 04 And yeah, it's true. And, um, I remember that there were many nights you were like not sleeping that well, uh, sleeping at the wrong times.
10:19 Ed Zitron? Yeah.
10:19 Speaker 04 And like sleeping during the day, up at night.
10:22 Ed Zitron? Half a dozen once.
10:22 Speaker 04 I remember him really like, like, um, very vulnerably talking about sadness and depression and getting. People to entrust sharing and him actively trying to like save people in a place where like where people were so callous almost by profession on there by like everything they wanted to do that right well everyone was at home playing Grand Theft Auto which could you guys fix that and get the next one out everyone's at home playing Grand Theft Auto and being callous and he was like literally going let me explain depression to you and what you can do and what the resources are and why you shouldn't kill yourself and which I think is great of course you would lead with empathy on this too and look at these
11:04 Speaker 04 assholes and think about all the engineers and it's the right way to process it I just think often it's not a binary.
11:11 Ed Zitron? No, and that's the thing I've been saying like a lot about the show is yeah I talk about the pigs and the assholes and the scumbags and all different names and the voices I do for them but it's also about the fact that there is like I'm pissed off Casey Kagawa friend of the show said this to him a lot it's I'm brokenhearted romantic because things like what Mike did actually one of the reasons I read Endgame as well is it's like yeah there are a lot of these financial horrors in there but there's fun dorky shit online still there's still one like most of my friends are from the internet I know it's like I'm like the drill quiet crying like being online it's like like everything I got is through emails but I still think there is a lot of joy in this and I think the central
11:54 Ed Zitron? thing about your article now is just like beneath this capitalism crush is actually some really like wonderful things being built there are still wonderful games being built yeah there's like an entire economy on Minecraft for better or for worse about selling mods and stuff there are people on Roblox other problems with Roblox obviously who are like building games on there and selling them there's still some cool shit happening in tech yes it's just being interfered with because every three months someone gets upset at them.
12:21 Speaker 05 Or finds a new way to make money off it which you have to you know shove into a new category.
12:25 Speaker 06 When you're saying the chase right like that's what it is right now I think that what you're talking about the like macro picture of everyone sort of having that reaction and I think we will course correct in time I think we're right now at that stage in history where It's not as cotton dried to me as the NFT crypto sort of bubble where like it was clearly
12:45 Speaker 04 That was criminals.
12:46 Speaker 06 Yeah, exactly.
12:48 Speaker 04 Fucking people over knowing it, con people.
12:50 Speaker 06 Right, and I argue that on some level, these are criminals. There are criminals at play in this scenario right now. I will say that with what Mike was writing about in your article, the Microsoft thing was all the more like, I think when my team saw the news last week, my instant reaction was, didn't they just ratify a union contract with like their first ever in the US too? Like, to be saying that on one hand, we really care about workers' rights and really want to protect workers, and the other, you're like, and those are their quality people, right? And now we're like, oh, game developers, nah, 9,000 of you can go because AI NPCs are all the rage right now. And they really make a lot of sense, AI NPCs. They've never said a bad thing, never said anything that didn't make sense.
13:27 Ed Zitron? How was that Nvidia demo from last year?
13:29 Speaker 06 Or was it this year? I can't remember.
13:31 Ed Zitron? I think they've done two. Like, every year Nvidia wheels out a demo, it'll be like, the generative AI NPCs are here. And within one day, someone has made it Sersla. Or like, it's just said something insane.
13:41 Speaker 06 Or Tay, Microsoft.
13:42 Ed Zitron? Oh, Tay was the 2016 AI bot that learned from the internet.
13:45 Speaker 06 To be racist.
13:46 Ed Zitron? Like, kind of a proto Mecha Hitler situation.
13:49 Speaker 06 Oh, that was Grok, too.
13:51 Ed Zitron? Yeah, Tay walked so.
13:53 Speaker 06 Yeah, yeah, yeah.
13:53 Speaker 04 Was that the Microsoft bot from like those years and years ago?
13:56 Speaker 06 2012 or something.
13:57 Speaker 04 Oh my god, I forgot about that.
13:58 Speaker 06 It's been, but the thing is, back to Brian's point is that this keeps happening, and then we keep course correcting back. Like, it feels like there is a systems response. And Ed, I think you're looking for the joy. I wouldn't use the word intervening or interfering. I would say it's noise. And the way for companies and people existing in this industry to deal with that is to focus on what you think is good. I find the irony in saying that, which is I think these tech bros that you're referring to, Brian, they're also trying to focus on what they think is good. So they're cutting out noise from their perspective. And I don't know, it's that movie that was just released featuring those four dudes in the mountain lodge.
14:36 Speaker 04 Oh, uh, mountain head, mountain head.
14:39 Speaker 06 Mountain head, yes. So it feels like that. It feels like everyone's got their little bubble, which is also at the same time created maybe and supported by tech. And if we continue to operate in silos, I don't know, I feel like I'm making a.
14:50 Speaker 04 If we only think of those, and often they are men in those roles, though they're not all, but often they are. If you think of some of those, you said bros, but if we only reduce them to those guys running around trying to kill someone in a sauna.
15:07 Speaker 04 Where I think in a way it allows, in a way it allows us to like, um, not worry about them because we could just mock.
15:14 Speaker 06 Right.
15:15 Ed Zitron? That's fair.
15:15 Speaker 04 Some of them are smarter, like just raw synthesizing power.
15:18 Speaker 06 Yeah.
15:19 Speaker 04 Some of those people, not all of them, they're smarter than everyone in this building combined.
15:23 Speaker 06 Yes.
15:23 Speaker 04 A couple of them.
15:24 Ed Zitron? I.
15:24 Speaker 04 And I'm not saying that that makes them.
15:26 Speaker 06 Good or anything.
15:26 Speaker 04 In any way good.
15:27 Speaker 06 Yeah, yeah, yeah.
15:27 Speaker 04 Or in any way good for now, because even if you say like, I agree with you that some of them may believe.
15:33 Speaker 06 Mm hmm. That they're doing this.
15:34 Speaker 04 That they're doing good.
15:35 Speaker 06 Mm hmm.
15:35 Speaker 04 Right? But if they're thinking in thousand year chunks, that's really bad for people. Yuval said this, and Yuval, I think, is amazing. And he said, well, the problem is historians like me, he goes, you may look at it now. He was on a podcast, he said, you may look at it now and say, but those things worked out okay. Look where we are. But as a historian, I have to look at the cost of human life.
15:53 Speaker 06 Yes.
15:54 Speaker 04 That happened for all the intervals, all the little intervals to get from there to here. And that's really what you guys are just talking about, rightly so, is all the, you know, the devastation in the wake. But, but can I just ask, because you might all might be so jaded, but do you not think AI is like mind bogglingly great at times?
16:13 Ed Zitron? No.
16:13 Speaker 06 I, I, I can see the balance. I think Ed's like far off.
16:16 Speaker 04 You don't, wait, you really, I think it's like the single, I think it's.
16:19 Ed Zitron? Why?
16:20 Speaker 04 The single greatest invention in my life.
16:23 Speaker 06 What is, what is your favorite thing that AI has done?
16:25 Speaker 04 I think the ability to have super high level conversations about really esoteric, about systems theory, right?
16:32 Speaker 06 Where does that info come from?
16:33 Ed Zitron? How do you know it's correct in them?
16:34 Speaker 04 Well, you can, well, you have to do a bit of work, right?
16:37 Ed Zitron? I don't want to talk to something that I have to verify constantly. I talk to people.
16:42 Speaker 06 Well, you have to do that with people too, Ed, to some extent. You can't trust everything I say.
16:45 Speaker 04 How do you feel about the automobile? Or is the horse drawn carriage still your thing?
16:49 Speaker 06 Dang.
16:50 Speaker 04 No, I'm just saying.
16:51 Ed Zitron? No, I'm wondering what the point is.
16:52 Speaker 06 Mike looks very uneasy right now.
16:54 Speaker 04 I mean, the, the point.
16:55 Speaker 05 No, I just have jokes in my head.
16:56 Ed Zitron? No, no, I'm sure you got rest.
16:58 Speaker 04 No, because I think you can decry the industry and the way people are using it. And I, again, I'm pretty sure I was reading Eliezer Yudkowsky before anyone.
17:06 Ed Zitron? Yeah.
17:06 Speaker 04 In this room, I, I mean, how to actually change your mind is a book that I was obsessed with, gave out as Christmas gift.
17:10 Ed Zitron? I can't take Yudkowsky serious. I just, it's like.
17:13 Speaker 04 You've got to take him serious. He didn't, I mean, you have to take, one has to take that guy's brain seriously. He, everything you guys are worried about, he called out 15 years ago. In detail.
17:22 Ed Zitron? Okay. I mean, like, here's the thing. The automobile is not a comparison because we knew we had to go forward side to side and everything like that. Like, we had an actual use case for that. With generative AI, the way that these conversations happen, and large language models can be conversing with documents. It's one of the only use cases that is actually remotely useful because you can actually verify based on the parts of the document. Having a conversation with one of these, fine. It's a thing. I'm not amazed by it because there have been chatbots doing this with KMS since like, what, 2013, 2014?
17:52 Speaker 05 And the Eliza effect goes back to the 60s.
17:53 Ed Zitron? Yeah, and Eliza, even then, there's a Karen How's Empire of AI is a great thing about Eliza. The creator was just like, why is everyone so fucking impressed by it? Like, they're just like, really, it's like, what the fuck is this? I think what it is, is I am just not that impressed by it based on the larger discussion. Everyone acts like this is the fucking future, and it just feels like a growth of the past. What ChatGPT has become, for better or for worse, is what Google could have potentially been. It's insane that attention-
18:22 Speaker 04 I agree with you, but like, okay, I got a tick bite.
18:24 Ed Zitron? Okay.
18:24 Speaker 04 And you can literally, the second you get a tick bite, everybody says antibiotics, you gotta go. If you can't find the thing, I got a picture, I put it on there. The AI was immediately able to, and yes, I could verify it afterwards, because I called, the AI was able to say, this is the kind of tick it is, this is the area you're in, this kind of tick, you don't need antibiotics, you're not going to get Lyme disease. Here's why, sent it to my doctor, who, that, it's feedback, and they agreed. That'll be harder to do on Google.
18:52 Ed Zitron? Why didn't you send it to your doctor before you asked him?
18:55 Speaker 04 Well, he doesn't, his job isn't to like, identify what tick it is.
18:58 Ed Zitron? Oh, I, oh, sorry, I misunderstood what you said.
19:00 Speaker 06 Yeah, I want more clarity on the doctor thing. I raised my hand as you were talking because I'm like, did you trust the ChatGPT answer and leave it at that?
19:08 Speaker 04 No, then I, then I took that, when it said that, then I searched for what it said and compared, and it was right.
19:14 Speaker 06 Okay, so you-
19:15 Speaker 04 That it was a, it was a dog tick, not a deer tick.
19:18 Speaker 05 Out of curiosity, and Ed, you might be able to answer this question, is that-
19:22 Ed Zitron? And I don't actually know the answer. Is that iterative AI or is that generative AI? So we're using ChatGPT.
19:28 Speaker 06 It's a mix.
19:29 Ed Zitron? Either that or Claude, I don't know which one. So that would be generative AI. Okay. And so that would still be, that is, and by the way, sounds like a use case. It is the growth of, so what were you saying?
19:39 Speaker 06 No, I was gonna, I think, I think a version of Google could have done that for you in the past too, before ChatGPT. What ChatGPT, the generative side of it is providing is the LLM, the natural language interface, about pulling a lot of different sources of data and putting that together for you. That is ChatGPT. You could have done that with Google. Or I think there were apps, right, that you could do a photo sort of recognition of that.
19:59 Ed Zitron? For a while on Google.
19:59 Speaker 06 Yeah, and to my knowledge, the parts of Google and Google Health Researchers with their AI divisions were working on apps that could identify different things, like skin things. So like, is that a bug bite or is that eczema kind of thing through the Pixel phones? I don't think they've ever broadly released it, but they were experimenting way before ChatGPT was even a thing. I would argue that I think the LLM portion of this is more about how it converses with you and how it like understands what you're actually concerned about. So if you didn't give it the exact words of, look up this thing and should I see a doctor? Even if you didn't input the see a doctor request, it might, you know, divine that that's what you really want.
20:33 Ed Zitron? And this is the thing, like ChatGPT, I don't think would have been a big deal if Google had actually innovated in search at all. If they have.
20:40 Speaker 06 Instead of something adds into it?
20:42 Ed Zitron? Google search has been the same for like 15, 20 years, except worse. What you were describing there, valid, it's inference, basically. It infers the understanding from the image and all this and then spits out an answer. It's a use case. I think that in the way you used it, that was responsible. The problem is at scale, that is not going to work out so great because. Oh, I believe you know way more about this than I. No, no, no, no, no. This is, I'm actually really, no, I'm genuinely glad you brought this up because it's like, but what about that? Is extrapolating out to the greater AI replacing jobs thing? Because that's where, where my principal problem is. People are taking what is what Google should have been in 2017 and turned it into this is going to replace half of workers.
21:25 Ed Zitron? A quote from Dario Amade that was fucking made up, which he said off the top of his head. It's not, but sorry, just getting. Wario himself.
21:34 Ed Zitron? How are you, Amade?
21:35 Speaker 06 I think what you're also bringing up, Brian, is that there is this marketing, I guess, problem with AI, which is that AI has existed for a very long time. And the current version that everyone's really obsessed with is Gen AI. And generative AI is all about what it can generate for you. Using large language models, using art, like creating art, creating videos. All of that stuff is this current iteration of AI we're all talking about. But the previous stuff has existed before. And this is all just more of the same. I think that's why so many of us in the industry are so frustrated with it. Because there's a misconception. There's also this idea that the jobs that it's trying to replace are in that field of coming up with art, music and words that are not jobs that A, pay well to begin with, but B, are the parts of our life that we want to replace, right? We want it repetitive.
22:19 Speaker 04 When I talk about Yeddowsky, the reason I do is that he was somebody who, he's right on your side in a way. I mean, his thing always was, this is not going to be a net good for society.
22:30 Ed Zitron? That's not even where my argument is.
22:32 Speaker 04 And I'm sure you remember when Astro Teller wrote Exegesis. I remember reading that book like the day it came out. And it really did freak me out about what was possible. Now, we're not even there yet, right? It even isn't where that version of AI was.
22:47 Ed Zitron? But Yeddowsky's.
22:48 Speaker 04 Yeddowsky, yeah.
22:49 Ed Zitron? Yeddowsky, sorry. I find him quite distasteful. But on top of that, he's a fucking AGI doomer. And he's saying, he's just, yeah, sure, if a frog had wings it could fly. It's like, yeah, if the computer wakes up and does this, this could be scary. He's one of the few people that seems to actually think about what AGI could do. But also, we're so far more off from it that I can't see him as anything but a grifter. Because all he is doing is grift.
23:13 Speaker 04 And he's been on it long before there was grift. And then he was on it as a nonprofit. No, he.
23:18 Ed Zitron? Nonprofit grift exists.
23:19 Speaker 04 No, I mean, you're talking.
23:21 Ed Zitron? Look at OpenAI.
23:21 Speaker 04 Think about what I've written. You think I don't know nonprofit?
23:25 Speaker 04 What's the matter with you?
23:27 Speaker 06 You were looking at your solemnness.
23:28 Speaker 04 That's not in my domain, Exegesis. Brian won. Yeah. But he's, I have a different view. Now, I'm not saying all his opinions are correct. What I'm saying is that. That's somebody who flagged a bunch of these potential issues a long time ago. And of course, it shouldn't... It's horrible that these people in charge are so willing to slough off human beings. So excited. The moment they think there's this much of an edge. But also because of what I study in life, I can't be so... We've all of us who write about this stuff in fiction, right? You do it because, okay, you just have to report on something. You have to... I can take sort of like my partner and I can take what's in there and try to dramatize what might happen.
24:11 Speaker 04 And always would try to point out exactly that these motherfuckers will do all this shit. It's not surprising to me. It's horrific, but not surprising.
24:28 Speaker 00 Empower is all about helping you invest well so you can go out and live a little. And their Free Empower Personal Dashboard shows you your full financial life in one place. You'll be able to see your net worth, monitor budgets, analyze your portfolio, set retirement goals, and more. Whether you're saving for something big, want to treat yourself, friends, or family, or meet long-term goals, Empower helps you see where you stand. You've worked hard for your money, and Empower can help you make the most of it. Download the Empower Personal Dashboard or visit empower.com. Not an Empower client paid or sponsored.
24:59 Speaker 07 OnDeck is built to back small businesses like yours. Whether you're buying equipment, expanding your team, or bridging cash flow gaps, OnDeck's loans up to $400,000 help make it happen fast. Rated A+ by the Better Business Bureau and earning thousands of five-star TrustPilot reviews, OnDeck delivers funding you can count on. Apply in minutes at ondeck.com. Depending on certain loan attributes, your business loan may be issued by OnDeck or Celtic Bank. OnDeck does not lend in North Dakota. All loans and amounts subject to lender approval.
25:32 Ed Zitron? What bothers me as well with it, and I think you're completely right, and also like, I'm not completely saying everything you're saying is wrong. Like, you actually have well read on this, which is nice, because a lot of people who mention the AGI stuff don't read anything. Um, but I think the thing is as well is, so much about this AI AGI thing is not about what it can actually do. Yeah. The best fiction about AGI, my girlfriend Sarah showed me the Killswitch X-Files episode. Fantastic AGI episode. It's scary because it's not about the people making it, it's about the computer itself and the intentions spilling into it. A lot of what the AGI discussion now is, "And it will automate jobs," and that's the last time I'm gonna think about it before I say, "Here's Anthropic."
26:10 Speaker 06 I, I-
26:10 Ed Zitron? Kevin Rose. Uh, and it's, and I think what it is, is there are discussions to be had around what this stuff could do. If AI could do the things that they're saying it could do, if it could replace jobs, it would be doing it, but also, it would actually require a change in society. It's almost as if they're like, they want all of the profits from AI, all of the exposure and the ability to lay people off without actually doing anything to earn it. And it comes back to what you said about- Yeah. The people running this shit aren't even trying to build AI. They're not, Mark Zuckerberg's building a Manhattan-sized data center to build superintelligence.
26:44 Speaker 04 I wonder though, when you say, and you guys all speak to this, I wonder when you, when, um, when you talk about that Google could have done this, or should have done this, and then, it's a great point that you made about, um, that it is the way that it communicates, because what I wonder is, you're all experts. You're all native, early, um, adopters, native to, not only the internet, but native to tech.
27:09 Ed Zitron? Mm-hmm.
27:10 Speaker 04 And so, sure, you could, you have used Alte Vista to do this stuff.
27:15 Ed Zitron? Oh, yeah.
27:16 Speaker 04 Jeeves, for fuck's sake, right?
27:17 Ed Zitron? Yeah, it's a baby.
27:18 Speaker 04 But I'm saying, you could have used all that, but for generations of people who aren't literate, and millions, billions of people-
27:24 Speaker 06 Sure, sure. Exactly.
27:25 Speaker 04 That, even if that was possible through Google Image Search, but you ever try Google Image, watch people try to do a Google Image Search?
27:33 Ed Zitron? No, I, I agree.
27:34 Speaker 04 I'm thinking, this is all making it for people. Friendly to them, as you said. Yes. Easy for them. And I wonder if that's the.
27:43 Ed Zitron? That's exactly what I'm saying. That's why people like ChatGPT. It's not because it's this amazing product. It's because it does what people go to Google.
27:50 Speaker 04 That's hard, but communicating well. Mm hmm. In an inviting, seductive way. Mm hmm. Is very challenging.
27:58 Ed Zitron? Yes.
27:58 Speaker 04 Yeah. So for tech people to do that to regular people is the amazing thing.
28:03 Ed Zitron? What's insane is Google invented this tech. The attention, the 2017 attention is all you need paper was eight Google scientists. I just looked this up. I thought it was several. But it's like Noam Chizzi, they ended up paying like $2 billion for a character A just to bring him back. Google had this technology and had this movement just been, we're going to make inference of meaning better. Exactly what you're talking about. It would not be this big story. The thing is, they need it to be what you are describing are use cases, good, bad, whatever. They are things that people use it for. The actual things they're describing on I as a PhD level intelligence. It's going to solve physics. It's like, no, it's not. But if they say, what if Google search worked? They can't make a trillion dollar. They can't be like, this is going to be everything we need.
28:48 Ed Zitron? We're going to build data centers. It's just because what you're describing, Brian, is not the most evil thing in the world.
28:53 Speaker 04 Go back to Mike Drucker. Yeah. You really should be looking at it. Well, because I think what.
29:00 Speaker 04 I think in all these things, what I've learned, just being curious.
29:03 Ed Zitron? Yeah.
29:04 Speaker 04 Is to look at the incentives.
29:06 Ed Zitron? Yes, that's what I've been saying.
29:07 Speaker 04 And Google was disincentivized to disincentivized to make search work.
29:11 Ed Zitron? Uh huh.
29:12 Speaker 04 For the user.
29:13 Ed Zitron? Yeah.
29:14 Speaker 04 Because of the profit agenda, right? You said, oh, should these things should make a fucking profit? But here's the thing, in a weird way, them not making a profit is better for the users.
29:22 Ed Zitron? Yes.
29:23 Speaker 04 And Google wasn't, but I'm sure it's all in the Google paper that nobody's fucking reading.
29:29 Ed Zitron? I'm talking about the 1994 one.
29:31 Speaker 04 But I'm talking about for the user. I'm trying to, you're raising these amazing questions. But if I try to think about how to answer them. It's that these companies are like in the early days of social media and all this stuff in the beginning, they try to super serve the user to get you addicted to it so that then they can, but the incentive structure is what you gotta look at. And obviously, Microsoft's incentive structure has been locked in for a very long time. And that is for the people who have the most stock in Microsoft to make the most money. That's the incentive structure.
30:00 Speaker 06 Google too, to an extent. And that's why they had that stuff for that long, but they never, because they never found a way to integrate it into all the different parts of its businesses that matter, the ads part, the search parts, the like.
30:08 Speaker 04 It's also their monopoly. They just.
30:09 Speaker 06 And they have a monopoly. They have no reason to be, but I don't know if you remember this Google I O twenty seventeen, twenty eighteen, even twenty nineteen, when they first showed Kitchen Sync, which was their version of Chat G P T, but it wasn't as conversational. It was, you can ask this app to come up with ways to learn about a new hobby or to plan a thing for you. Before Chat G P T even was known widely to everyone, I think it was, OpenAI wasn't really even a thing. And it just wasn't a chat interface. And I think to Brian's point, the incentive that Chat G P T was, has brought to people and to my parents, who by the way, discovered Chat G P T last year, very annoying to me, but it, it's like, it's so much easier. It brings them into technology in a way that technology used to be kind of looking down on people for not knowing things. And, and you deal,
30:52 Speaker 06 you do away with that with the chat bot. My parents not only like feel so hip now, which, sorry, mom and dad, you're not, um, but, but also like, there's people who seek comfort in the companionship brought on by AI chat bots like Chat G P T.
31:05 Speaker 04 They're just differently hip.
31:06 Speaker 06 They are amazing.
31:07 Speaker 04 Yeah, they're real.
31:08 Speaker 06 They dance at their age. I love that.
31:09 Speaker 04 That's actually lovely.
31:11 Speaker 06 I, but the thing is, if you look at the use of Gen AI from the last, and I think I talked about this on that Kevin Roos episode I was actually here on, but, um, the use of Chat AI services has changed, right? It used to be like very interest based and very search based, but it's now companionship based as the top few uses. And that's why people are drawn to it. And I think that, I mean, one last point that came up really when I was like listening to you talk at is that like, the incentive for them to push towards like, yes, let's go towards A G I. It's not just like laying off people. It's also who can get there first. It's the race of like.
31:45 Speaker 04 It's not, it's.
31:46 Speaker 06 The tech AI ego thing, the tech AI, the tech CEO ego thing. And then from the ego standpoint, then they push down to profit.
31:52 Speaker 04 But so many businesses are like, but so many businesses are already. Sort of the industries are already acting in bad faith. Oh, yes. Yeah. Um, like, OK, look at the here's here's you want to I'll give you not a you like a real industry that I think will be transformed by it. And I think that's not a bad thing is like money management. Money management, which is a billions.
32:13 Ed Zitron? With generative AI?
32:15 Speaker 04 AI already can replace. AI works well with very set parameters, you're right.
32:20 Ed Zitron? Between AI, though.
32:22 Speaker 04 But, well, you can you can do that.
32:23 Ed Zitron? No, no, I'm saying.
32:24 Speaker 04 Look at that whole business. That whole business is about a front. Money management, but I'm not talking about the high net worth. I'm saying in general, people who use for their retirement accounts, money managers.
32:35 Ed Zitron? Wealth front future advisor.
32:36 Speaker 04 All that stuff. But within those companies, they're they're they're front facing language of humans who are just trying to keep you invested.
32:44 Ed Zitron? Oh.
32:45 Speaker 04 They don't know that stuff. They are not offering value really, and they're taking this big percentage from regular people trying to save for their retirement. And they're bleeding off my in the end, they're going. I was listening to like.
32:56 Ed Zitron? No, I'm smiling because you were right already. This was the 2015 through 2020 wealth.
33:01 Speaker 04 No, I know that all that stuff, but I'm talking about even now, like big banks, like the big banks. I'm not talking about their their AI front. I'm saying that I was listening to a talk given by Josh Brown is. And one of his partners, I think.
33:16 Ed Zitron? Okay.
33:16 Speaker 04 And he was talking about how essentially all of the.
33:24 Speaker 04 Back end of all that stuff, meaning you might still have a person talking to the user to the but everything else is going to be done by.
33:31 Ed Zitron? I just.
33:31 Speaker 04 Yeah. In the businesses. I think it was Michael Batnick from his company was talking about that.
33:36 Ed Zitron? Because here's the thing with that is from my knowledge of basically financial regulation, they don't want an LLM touching much of this. They there's a lot of stuff within financial research happening now with generative AI. There's a ton of companies doing like insanely high compute burn to do these massive. Kind of like evaluations of stuff. I don't know if anyone wants to touch the money with LLMs and they've actually been quite resistant to it, partly because they don't know how they work. Like, they truly don't know.
34:03 Speaker 04 Oh, I know the black box thing.
34:04 Ed Zitron? Yeah, yeah, yeah, that that and so it's like. A lot of these things also, when are they gonna happen though? Because they've been saying this for two years.
34:13 Speaker 04 But do you think there's a scenario, do you think there's a scenario where this, this goes, this goes away? I don't think it goes.
34:19 Speaker 06 I would love it too, but I think we're past that point.
34:21 Speaker 04 Is that what you think?
34:21 Speaker 06 No, no, no, no, no.
34:22 Speaker 04 You think it's gonna, you think it's gonna reveal itself as, as being fraudulent for what it
34:26 Ed Zitron? I'll explain.
34:27 Speaker 04 reports to be?
34:27 Ed Zitron? No, no, I think you're going to see, what, call him a shot. So I believe that OpenAI will, as an ongoing concern, eventually go into nothingness. Matt Hughes, my editor, believes they'll become a patent troll. I actually think it's an amazing thing. I think that what we experience of large language models will vastly pull back. I think there will be rate limits.
34:47 Speaker 04 What do you mean pull back?
34:48 Ed Zitron? As in, there'll be rate limits on ChatGPT. People are gonna be horrifyingly sad because those companions.
34:53 Speaker 04 Their companions are gonna go away.
34:54 Ed Zitron? No, they're not gonna go away. They're just gonna be much, much, much more limited. And I think that everything we see today, the kind of, and you look in any of the reds behind any of the serious, like, GPTs, they're all kind of saying like, yeah, we know that the abundance, the free ride is over. So, no, it's not going away, but you're not gonna hear about it constantly. And everything you use today is going to be so severely rate limited or dead. Those companies that are charging for generative AI things, beneath the surface, all of the API rates behind these companies. So the things that you plug in to run the models are vastly subsidized by big tech and by the companies.
35:29 Speaker 04 But so what's, so Aldebaran goes away and, but Google takes over?
35:32 Ed Zitron? Uh, well, Google Gemini exists, but the Gemini requests perhaps don't hit the LLM as much. They have.
35:37 Speaker 04 I'm using it as a parallel.
35:38 Speaker 06 Yeah, yeah, yeah.
35:39 Speaker 04 I'm saying the parallel, right?
35:40 Speaker 06 You're kinda saying there's consolidation that's going to happen.
35:42 Ed Zitron? But it won't be this big thing where you hear about it all the time. I think you were going to.
35:46 Speaker 04 It will just be the backend of lots of stuff?
35:48 Ed Zitron? It will be, oh no, it will just be something that sits there.
35:50 Speaker 06 There's a smaller in scale, right?
35:51 Ed Zitron? It's also not good as a backend. Large language models are not good backend.
35:55 Speaker 04 I understand, yeah, they're good at talking to you.
35:56 Ed Zitron? They're good at talking. And the one thing where. They can, they can divine stuff as well.
36:00 Speaker 05 One thing we're, we're kind of circling on the consumer level that we're not talking about is it's still a novelty. Now, I think it's gonna be, continue to be used. I do think that people are gonna continue to make, uh, lists and scheduling and.
36:11 Speaker 06 Summarize my.
36:11 Speaker 05 Summarize my thing or like, you know, what's a good vacation?
36:14 Ed Zitron? And there are on-device models that are able to do that.
36:16 Speaker 05 Right. I just think that right now it's such a big deal that everyone's using it because it's cool, because they, you know, your parents hear about it, because you're told you should use it for work.
36:24 Ed Zitron? Yep.
36:25 Speaker 05 And I do think that it'll stick around. I do think there'll be a contraction in the sense that, you know, it'll be a cool thing, but I think in 15 years, it's no longer to be as funny to produce like a picture of a pig with like Mickey Mouse's head and three boobs. And I feel like now that's a big, big novelty.
36:40 Ed Zitron? Some comedy is timeless.
36:41 Speaker 05 But like, but I just don't imagine that in like 15 years, having the same novelty as it does now.
36:46 Ed Zitron? 15 years?
36:48 Speaker 05 Or five years, 10 years.
36:49 Ed Zitron? If you take away all the headlines, if you take away all of the money and you actually look at what's there, it is everything you've, what you're describing is probably the most useful thing. It's like, do I know this? Uh, okay, this seems plausible. I'm going to double check it. It's kind of like what in Carter could have been. I don't even mean that sarcastically.
37:05 Speaker 05 No, that's a good point.
37:06 Ed Zitron? I, as a very cool child, I was very cool, would sit on in Carter for hours reading stuff because it was kind of good. Wow, you have access to everything. And I think human beings are curious. So of course they're gonna talk to it if it talks back. Yeah. I just think right now there is no business model. That's the biggest one. The biggest one is there really is no business model. Ads do not work. Ads are not going, how do you put, you inject ads within an L and M. Look at what happened with Grok. Grok happened because they tried to make a, let's make it just, how do we crank up this racism dial? And like, how do we mess with this system prompt? But the thing is, those subtle changes for even advertising will be bad. Perplexity's been talked about doing ads for a fucking year, not heard much of that, Aravind. And it's like, on some level, regardless of how
37:49 Ed Zitron? useful it is, the economics do not make sense. They're nothing like an Uber. Uber was such, you, yes, you well know. Still don't make sense. Uber was a complete fucking, and it still barely makes sense. They're raising another 1.2 billion dollars, but you can at least tell someone exactly what Uber is and why you'd use it. And it's just kind of chugging along through necessity. I don't know how necessary chat GPT is. A large language model, a Google Gemini, and Google is claiming they're doing efficiency stuff. That could last. I think it will be heavily rate limited.
38:21 Speaker 05 I, I'm, I'm certain you are all right about the business models and the viability of this from a business standpoint. You know so much more than I do. I'm, I'm learning and it's fascinating. As a, a user of it who is not on the inside of the business, I
38:36 Speaker 04 My prediction is you're dead wrong.
38:38 Speaker 06 No, Brian.
38:39 Speaker 04 What will happen? Oh, go, go. It's gonna become the dominant thing in most of society in lots of ways.
38:44 Speaker 06 I think here's.
38:46 Speaker 04 People are gonna use it and be, it's gonna be part of them. Um, like William Gibson predicted a really long time ago. Like, I think that it's fine, that I think it's bad at creative things, the thing people think it's good at. I don't, it's bad at that. I don't think it can write a story. Yet, none of them can convincingly write a story like Drucker could. Like, there's lots of that stuff that's not there yet. I, I have no idea, again, you all know way better than I do the science behind it. But you were asking about working out before. You could, there are 95% of people who are trainers. Can't do as good a job of programming. And you could play different AIs against each other and are good at asking questions.
39:23 Speaker 06 That's also because 90% of them are influencers with no serious background.
39:26 Speaker 04 Yes, but I'm saying even if you talk to science, but, but, because you can show programming. It can track for, it can just, now, you may, you may, and I'm sure you're correct. That's not the AI doing it, it's other people could have done. But the way AI can program and interact with you. And allow you to catch up. If someone asked me, all day long, people are asking me, wait, wait, look, questions about programming. Well, I can't program for you. I don't know you well enough.
39:53 Speaker 06 Not yet, yeah.
39:53 Speaker 04 But I, if we talk generally about what your goals are, I could definitely talk to ChatGPT or Claude. And build something that you could then iterate. That's a search engine. We're describing the iterations of search, though.
40:07 Speaker 06 Yeah.
40:07 Speaker 04 But it's packaged in this way now. Go ahead.
40:09 Speaker 06 The thing that three of us are going to tell you are at different levels. I think Ed's coming to you at the business model, the very like micro level.
40:16 Speaker 04 I'm sure he's right, yeah.
40:16 Speaker 06 And sure, and, and, and, yeah, that's the way, kind of how it's got to play out financially. And Drucker's come in with the like medium level of the like use case and everything. And I'm gonna tell you that at the top level, I'll draw another parallel for you, which is two things come to mind. One is how. Bitcoin and crypto, very exciting. Everyone found a novelty factor and it ran for everyone. Wanted to make NFTs and crypto a thing. And then now it's kind of dialed back down to a less fever pitch and more of a regular body temperature pitch. The macro metaphorical level is what I'm gonna say. I think AI will dial back down to that Norman normal regulatory sort of body temperature. And then to draw another parallel Tinder was everyone was making their app the Tinder of this Tinder of real estate. And what you're describing is like an interface that works
41:00 Speaker 06 really well for something like a use case. The chat GPT model is an interface that works really well for question and answers, seeking help, assisting you with things. That might never go away. That might just get built into every app as access to LLMs becomes easier for developers. They'll build it into the Bank of America chatbot. They'll build it into everything. And so I think that's where it balances out eventually over time, maybe through rate limits. I don't know that that's going to be the way. I think some consolidation might happen.
41:26 Ed Zitron? Yeah, but rate limits on actually accessing it.
41:28 Speaker 06 The question back and forth for the people who are using chat GPT, right?
41:32 Ed Zitron? Yes, yes.
41:32 Speaker 06 Yeah, I think that will come too. But I think eventually we're talking like five years with Druckers. I think with the rate limits maybe in the short term, but I think even longer term than that, we're seeing
41:42 Speaker 06 that might eventually go away. Those apps may not even exist standalone.
41:45 Ed Zitron? Yeah, and I think I actually don't know if I fully disagree with you about everything you're saying. It's just that the scale of what we're talking about might be different. Everyone having a large language model to access, just saying that this is the future we're talking about, doesn't really change that much. I don't like the economics effect. I know I'm gonna do the business bullshit thing I do, but it's
42:08 Ed Zitron? the economic effects are quite limited right now. Now, if you're saying everyone will have a large language model and it will be hamstrung in some way or what have you, fine, I can I can buy that for good or for bad. I could see that happening. I just don't think that it goes much further than what we see today. And I think what you're describing is what Google search should have become. And like, that was what, I remember when Bard came out, I wrote about this. I was kind of like, surely what chat GPT is, is what search was meant to be, right?
42:34 Speaker 04 As the Coen brothers said, you know, sure, if a frog, you know, had
42:40 Ed Zitron? Had wings.
42:41 Speaker 04 Had wings, it wouldn't bump its ass hopping. You know what I mean? And so.
42:46 Speaker 05 And I'll say that, you know, I think that even as it's completely absorbed into our culture, we're still going to be on the phone, listening to an AI list medical options and going, "Human, human, operator, human." Like, I don't think that's going away.
42:59 Ed Zitron? Walgreen 771.
43:00 Speaker 05 Right, yeah, like, I don't think, like, I do think that people are gonna be like, "Yeah, okay, I'll talk you through my problems, until I hit this part with my insurance, and I just want a human right, right, right now." And I don't think that's going away.
43:10 Ed Zitron? No, not at all. I, and weirdly, I actually think we're really more on the same page than they may have seen, because it's like, my whole thing is, what you're describing is a use case. I think there are real harms, but I think we kind of agree where the dangers would be. My thing is, is that people are extrapolating from that to this insane level. Like, this whole, they keep talking about agents everywhere, you've got Matthew McConaughey.
43:32 Speaker 06 But agents is kind of what I'm describing, which is like, every app, every service has its own chatbot, more or less. They're just using a different word for it.
43:39 Ed Zitron? They're just using a different, yeah, and, I mean, that's kind of what, like, Bank of America already has a chatbot, and it does, it does not work. It's the bot you use when you're trying to search for a transaction.
43:48 Speaker 04 What Mike's talking about, for sure, the human, of course, we all do that. I guess I think it'll fool us better.
43:55 Speaker 05 That's fair, that's fair.
43:56 Speaker 04 I don't think, right.
43:57 Speaker 06 Oh, it's already fooled a lot of people very well.
44:00 Speaker 05 It's fooled people into killing themselves.
44:02 Speaker 06 Oh, god.
44:02 Speaker 04 Well, that's, tell you, that old character AI, of course.
44:05 Ed Zitron? It's so cool that Google paid $2 billion for them.
44:07 Speaker 04 I'm not making an argument that it's, this is a beneficent force.
44:10 Speaker 05 No, no, no, no, no.
44:11 Ed Zitron? No, I wasn't accusing you of that.
44:12 Speaker 04 I mean, in any way, shape, or form, like, I, like I said, I've been reading about this for so long. But I do think that to just sort of, that's why I brought up the horse and buggy, because, no, the, the people who wanted the horses, right? They were right about a lot of stuff, about the harms it would do.
44:28 Speaker 06 Yeah, the pollution.
44:29 Speaker 04 The pollution, the noise, the way it would take us away from our communities. Like, they were right about so much. What they were wrong about was the inevitable march of the future and time.
44:40 Speaker 06 That's, I think, the same, before coming to this podcast, I was reading Mike your piece, and I was like, "Oh, yeah, are we, like, in the Industrial Revolution, forgetting about the Agricultural Revolution, forgetting all the revolutions that came before?"
44:50 Speaker 04 Because I've seen it all from when I was, I remember when... AOL showed up and copy serve and I remember I was invited on message board when I was 14. I'm 59, like this guy I knew had a message board in New York and had to dial up and I mean, so I've seen this, so I'm an early adapter of stuff, even though I'm an old dude, I'm, I'm, but not from a tech side, from a user side.
45:10 Ed Zitron? I was, I was on use that as an 11 year old.
45:13 Speaker 04 This is the single, as a, just as a user, meaning I don't know how it works or why.
45:17 Ed Zitron? Yeah.
45:17 Speaker 04 But I can explain to you why people are so fascinated and NFTs, I was on, you could find the old tweets going, calling people. Buy one, are you fucking mad?
45:26 Ed Zitron? Hell yeah.
45:28 Speaker 04 I don't know what kind of early doctor you are. People is like literally like why I started doing what I do, but like, no, of course I recognize that as a con from the moment one. But this doesn't be like a con to you? But even when you say the Bitcoin thing, it's at 120 today. Yeah, I know.
45:42 Ed Zitron? Okay.
45:42 Speaker 04 So Bitcoin won. There is actual value, yeah. And NFTs were always huckster devices to separate suckers from their money. And look, Theodore Veblen said, and David Mamet quotes it all the time, every profession is a conspiracy against the laity. Every profession fucks over the regular.
45:59 Ed Zitron? Yes, I think that's super fair.
46:01 Speaker 04 And there's no doubt about that.
46:02 Ed Zitron? It's the big short speech about the people who get harmed.
46:04 Speaker 04 Yeah, Mamet was ahead of that by 20 years, but you gotta, you gotta look at it. And not me, Mamet.
46:09 Ed Zitron? And then, and then me.
46:11 Speaker 04 But, um, but so, uh, you, you gotta just understand that it is very effective for people.
46:18 Ed Zitron? I'm just trying to bridge from what you're describing to the automobile, which changed everything. I don't think large language models, I think they're going to create harms. I think there are going to be things that change, but it's like, what happens now? Because what we have right now is basically what we've had for two years. If you want to email me about reasoning, please do, I will email as much as you want. But the thing is, you look at this and people are going, okay, and then this will happen. It's like. That thing that with an L L M that goes and does something really basic thing, an L L M that you tell to go and do a thing online, they are bad, bad at it. There was a Salesforce study that said it's like thirty five percent. They like it was like thirty something percent. They fail,
47:02 Ed Zitron? or that was the only ones they completed.
47:04 Speaker 04 I don't think I've never asked an A I to do a task for me. I'm saying I've never asked it to do it. Do one of those agent kind of functions. I wouldn't. I agree with you. I wouldn't. Well, I don't think it's there yet. I mean, for again, as a user, but research, I've had good research done.
47:17 Ed Zitron? Really?
47:18 Speaker 06 Well, I mean, the A I and Gen A I has done a lot of good stuff in the medical fields too, right? CRISPR and all of that stuff. There's been a lot of discoveries about what sort of mutations you find in certain types of cancer that like I don't think humans could have done
47:31 Speaker 04 If I want to study an industry to consider writing about it, I can ask, you have to ask good. I mean, it's like in anything else, right? I can ask really good questions of an A I and send it, you know, for the $200 a month model one where they'll do that research thing. And if I ask it to do research and then you can, it'll like, it will come back. And maybe you have to send it back three times, but the speed and accuracy, yeah, but you can very quickly
48:00 Speaker 06 It accelerates the stuff, right?
48:02 Speaker 04 A lot of it because you can, like the book list thing that I can't fathom being that irresponsible, like those idiots who in the paper. But you literally, all you have to say to them is when it presents any kind of list, all you say to it is go verify that list, please.
48:15 Ed Zitron? How do you know what right is?
48:17 Speaker 06 Or verify it myself. I mean, I would verify it myself, right? At some point.
48:20 Speaker 04 But you go verify the list. It immediately goes, you're right. I was hallucinating. These three books don't exist. That happens. Then you go do it one more time and make sure that these titles are available in these stores. Then they'll give you links and then you can go look at.
48:35 Speaker 06 So it's about the prompting that you're doing.
48:36 Speaker 05 And I think sometimes we're splitting here, though, is you're, both of you are extraordinarily intelligent people who have done a lot of research in your life. So you know how to do those things. You know how to be like, I need to follow this up. I need to search this, make sure. I do think that one of the problems, again, coming from like the low level consumer level, is it's often being marketed as an impartial refer- impartial refer-
48:56 Speaker 04 Oh, you're totally right, Mike. You're totally right. It's not that. You cannot. You'll be fucked so bad if you use it for that.
49:01 Ed Zitron? But most people don't interact with it like you do, is what I'm saying.
49:04 Speaker 04 Well, but, but they can.
49:06 Speaker 05 They can. And I agree with you. They can. I think it's almost, oh, but it's, it's-
49:11 Ed Zitron? They don't lead them to do that. It's not like they have things that guide them. I would-
49:14 Speaker 04 This is really interesting here, what you're saying about this. Yeah.
49:17 Speaker 05 Well, I mean, I think that, you know, when you see people, and we all saw the Grok Nazi thing, but if you see people on Twitter, they don't use, they, I would say fewer use it to do Nazi shit as much as they go, "Hey, Grok, is this true? Is this true? Is this true?" And depending on what finger is on which scale, the answer's different each time, but I think that there's a lot of people-
49:37 Speaker 04 So I understand why you're saying, I, I did, I haven't been on Twitter since October, whatever that date is. So I don't know, but I, I would never, of course, you can't just say, "Let's go to the AI," as though that's a final arbiter, and certainly not today, but it's been 2025.
49:55 Ed Zitron? As that.
49:56 Speaker 05 I think that's my problem, is it's being positioned that way. And I think you're absolutely right. I think you're absolutely right how useful it is, especially if you have the skill to use it. I think the problem is, it is being marketed as, "This is a catch-all solution. This is a panacea to your knowledge problems." And, you know what I mean?
50:11 Speaker 04 Of course, there's a dumbing down. Well, it is like believing Rockstar Games that, that they're gonna get fucking Grand Theft Auto out, right? I mean, it's no different. I mean, I literally would, look at the actuarial table. If you had to build one on my life expectancy verse when the next fucking iteration of GTA, I might lose.
50:30 Ed Zitron? But here's the thing, here's the thing though, with you describing the research. It isn't an invalid use case. What you're describing is how people use Google Search to do research. They pull up a bunch of stuff, they go through them, they look at it, and they go, "Is this right?"
50:42 Speaker 04 But that's a slog.
50:43 Ed Zitron? It is a slog. What you're describing is a slog.
50:45 Speaker 04 No, no, it is not. It's a pleasure. It's not, no, you gotta be honest about it. It is a total pleasure, not a slog.
50:51 Ed Zitron? I should be clear. I'm also, I have actually used these things. I've genuinely tried, because I'm, I love my dude, that's Magismos, and I've really, I've sat down and been like, "What am I missing?"
51:01 Speaker 04 You just became Paul McCartney.
51:02 Speaker 05 That was a true Paul McCartney moment.
51:06 Ed Zitron? He has not accepted my invite for the show, though. No, he had not invited me. My mom would be so happy. No, it's just-
51:16 Ed Zitron? I really feel like this keeps coming back to the you are using it in a totally fine way. I'm mad at the fact that everyone's like, and this is the only thing you'll need. You can fully trust this. This is the best thing. The information's the best. You're looking at it and going, this is a way of digging through information and passing stuff and having a conversation with the information.
51:31 Speaker 04 Lowest comment in the broad society.
51:33 Ed Zitron? Yes, the lowest.
51:34 Speaker 04 Lowest comment, but unfortunately, yes, you're right. Our educational system is really fucked up. Disadvantaged people have no chance. Mike went to a school that allowed people from disparate areas to get. We gotta, yeah, we gotta reform the education system in the country so that everyone has a fair shot. I agree.
51:49 Ed Zitron? But there's a very basic.
51:50 Speaker 04 So let's do it.
51:51 Ed Zitron? Yeah, but there's actually a very basic thing we don't have to.
51:53 Speaker 04 What's that?
51:53 Ed Zitron? There should be regulation that says that these things need big fucking disclaimers that say, hey, check everything. They won't do it because the incentives we discussed, but that would actually be, I think, or the stealing is also bad. I think the environmental damage is bad, but I think we agree on that.
52:08 Speaker 04 Yeah.
52:08 Ed Zitron? It's just.
52:09 Speaker 04 Safeguards are always great, but safeguards, but then you gotta figure out who decides what those safeguards are and who's. Like you talked to Bill Gurley, he'll talk about regulatory creep, right? So where do you wanna? I'm just saying, you smart guy.
52:19 Ed Zitron? No, no, no, I agree with you.
52:20 Speaker 04 He's a smart person and he's thoughtful about this stuff. And so how, who do you want? Do you want the, I mean, with this, you want the current government deciding?
52:31 Speaker 04 Guardrails, who's to do?
52:32 Ed Zitron? But here's a very basic guardrail. It's just a disclaimer that says everything with generative AIs, blah, blah, blah, that you cannot.
52:38 Speaker 06 Some of them have it now, but they're all in preview, right? And that's probably what they're couching against.
52:42 Speaker 04 They do say right on, um, chat GPT now, it says.
52:44 Speaker 06 GPT Gemini.
52:44 Ed Zitron? Double check every single time.
52:46 Speaker 04 Yes, it says it every single time.
52:47 Ed Zitron? Really?
52:48 Speaker 04 Every single time.
52:49 Speaker 06 And Apple Intelligence.
52:49 Speaker 04 I saw it today, actually, for sure.
52:51 Speaker 06 Yeah, because they, they have been criticized, but to Brian's larger point, there are guardrails put in place into a lot of these. I'm most familiar with the Gemini's and the Apple intelligences and the Amazon ones of the world. And they, their guardrails are around like, um, CSAM, right? Child sexual abuse material or like not presenting people's faces or like trying to avoid photorealism because then you get very deceptive very quickly. You don't see any of that in Grok, maybe it depends.
53:17 Ed Zitron? I just looked.
53:18 Speaker 04 I know, I mean, I.
53:18 Ed Zitron? I just used.
53:19 Speaker 06 Which one, GPT? I don't use.
53:20 Speaker 04 GPT does really, it does, but I'm not gonna turn on my phone.
53:23 Ed Zitron? No, but I'm like, I'm just like, here's the. Thing, if they're there sometimes and not others, that's also bad. Yeah. Like, it's like, it's in, it's because when you've got people who are killing themselves, people that are having, it was Miles Clay, I think it was, or Rolling Stone wrote this piece about people having psychotic reactions. Yeah, I agree, this, this administration have probably done well regulating this, but the answer being let's regulate nothing is terrifying.
53:46 Speaker 04 Well, it's really, but it is confusing because if you think about, let's say, Facebook.
53:51 Ed Zitron? Right.
53:52 Speaker 04 There's no doubt that Facebook was used in Myanmar to foment a genocide. People were warned inside, who knows where it got to, this is very well documented.
54:07 Ed Zitron? Yeah.
54:08 Speaker 04 How could one, after the fact, it's, okay, we, it's really hard to sort of figure out these use cases. And then should all social media have gone away? Like, some people think it should be all social media.
54:19 Ed Zitron? I feel better about that if Andrew Bosworth, the CTO, yeah, yeah, yeah, I mostly say it for the list.
54:26 Speaker 04 Yeah.
54:27 Ed Zitron? Because also the yellow.
54:28 Speaker 04 No, but if I know, I'm a Luddite.
54:29 Ed Zitron? Well, that's good.
54:30 Speaker 04 If I know, your listeners know.
54:32 Ed Zitron? He did an internal letter in 2016, 2017, where he said that all things were justified for growth, including a terror attack. And that's kind of how they approach everything.
54:40 Speaker 04 No, it was monstrous. Like, when I saw that Myanmar thing, it made me say, like, I should never use Facebook. I mean, I think that is as bad a thing as I've ever seen. I mean, literally.
54:50 Ed Zitron? But this is still an example, though. And Meta's LLM allows children, and Jeff, Jeff Horowitz reported this at the Journal, allowed children to, like, have sexual conversations with John Cena, very peculiar. Like, there was like.
55:05 Ed Zitron? Let's be super clear with not John Cena.
55:10 Speaker 04 You said with John Cena.
55:12 Ed Zitron? I just want to say it was not John Cena. It was just John Cena. I mean, I was in a jail for a hundred years.
55:19 Speaker 04 On the bear, I was in a scene with John, lovely guy, and he definitely wasn't doing that.
55:24 Ed Zitron? No, no, he seems like a guy. I'm saying it was the voice of John Cena you're allowed to have pedo conversations with. But again, it's this lack of restriction because no particular technology is evil at its core.
55:35 Speaker 04 That's fair.
55:46 Speaker 00 Empower is all about helping you invest well so you can go out and live a little. And their Free Empower Personal Dashboard shows you your full financial life in one place. You'll be able to see your net worth, monitor budgets, analyze your portfolio, set retirement goals, and more. Whether you're saving for something big, want to treat yourself, friends, or family, or meet long-term goals, Empower helps you see where you stand. You've worked hard for your money, and Empower can help you make the most of it. Download the Empower Personal Dashboard or visit empower.com. Not an Empower client paid or sponsored.
56:18 Speaker 07 OnDeck is built to back small businesses like yours. Whether you're buying equipment, expanding your team, or bridging cash flow gaps, OnDeck's loans up to $400,000 help make it happen fast. Rated A+ by the Better Business Bureau and earning thousands of 5-star TrustPilot reviews, OnDeck delivers funding you can count on. Apply in minutes at ondeck.com. Depending on certain loan attributes, your business loan may be issued by OnDeck or Celtic Bank. OnDeck does not lend in North Dakota. All loans and amounts subject to lender approval.
56:51 Speaker 06 I think what Brian was saying, and I don't know if I'm misinterpreting you, but like it's, there is some fatigue at seeing these things throughout, all the pharmaceutical warnings you get the end of every commercial, every warning you're gonna get from Gemini from now on, every Apple intelligence warning that's like, notification summaries can be wrong sometimes, like sometimes you just, I see them on my phone.
57:09 Ed Zitron? Where are you seeing the notification warning?
57:11 Speaker 06 I'll show you later.
57:11 Ed Zitron? No, no, no, I believe you. I'm like, why am I not seeing?
57:14 Speaker 04 What I'm saying is that, but why I brought up Myanmar is, no, we, as a society, unfortunately, we default to the convenient, fun, easy thing. Yeah, that's accurate. What we all should have done when Facebook, and I'm not even blaming, exactly, like you can blame Boz, I don't know enough to blame anybody there. We know that the technol, that Facebook itself was used by these generals in Myanmar. Right, yes. And nobody took pause, like, like, um... I think very few people left Facebook as a result of that. I I think so. I just don't know what the fix is for these problems because we gravitate like bees to honey.
57:50 Speaker 06 But there there are also like tools that can be used as weapons and it depends on the perpetrator and the person using these, right? And it's an age old question. Again, coming back to the industrial and agricultural evolution. This can be just a tool for hacking a plant off of its husk, or it can be used as a murder weapon. Back to AI can be very informative, very helpful for people who need companionship. It can be used like people will send me scam texts all the time. The the technologies keeps keeping up with them and filtering them out, so they keep changing their spam. Bad actors are gonna bad act. That's just the way it's gonna be. What can we do? We can. I stopped using Facebook. I try to educate my parents every single time they use a GPT answer with me. I'm like, Mom and Dad, stop using that. But then they just keep using it because they're the sort of person that's susceptible to this.
58:36 Speaker 06 They will just use it because convenience and they don't want to do the actual work of maybe the Google search method, which you were saying at. Or they just want something that's easy and they don't care if they get it wrong, maybe.
58:47 Speaker 04 I guess I love the idea of you using your brains to figure out how to make these things safer and more useful as you agitate within the industry. I think trying to
58:58 Speaker 04 trying to find a way for them to disappear seems
59:01 Speaker 06 It's impossible.
59:01 Speaker 04 Impossible. And and again, like NFTs were obviously gonna disappear. But the underlying Bitcoin thing wasn't because there's too much. Well, the moment this election happened. The moment this election happened, Bitcoin was going to 200. Bitcoin was going because
59:17 Ed Zitron? And what sucks about that is we could have stopped crypto. I was writing about it at the time. I was and the only
59:22 Speaker 04 How are you gonna stop crypto?
59:23 Ed Zitron? Uh, by informing people about the inherently criminal aspect that underlies everything, the fact that tether is more than likely in the hands of multiple different group. I tried and I failed. And you know what? It sucks, but you try, you can put ideas out there, you can see if people pick them up. And I mean, that's kind of what you do there. Um, in the case of crypto, it's such a weird thing as well because it's there, but it isn't. It doesn't really do anything other than fund things or be funded. And it just kind of exists where it's just like they kind of, they don't even, I kind of respect the fact they don't even try and be like
59:55 Speaker 04 You're literally, you're talking the Charlie Munger and Warren Buffett's book. That's what they always, I know, but that's what they always said, um, because of that. That that's their whole Point that it, forget the blockchain, it doesn't really do, you know, it can't really do anything.
1:00:08 Ed Zitron? And the thing is, though, there was real money in that, which there isn't in generative AI. And I think that what's funny is this convenience thing you're talking about may actually be their downfall. Because those 500 million weekly active chat GPT users are costing them billions of dollars. It's probably all gonna.
1:00:23 Speaker 04 Fascinating. That is fascinating.
1:00:24 Ed Zitron? Yeah, that is interesting.
1:00:25 Speaker 04 They mean their popularity is gonna destroy those companies, not, not, not the underlying tech, but that the companies themselves, that's fascinating. I'm really, like I'm saying, you're teaching me something. I had no idea about that.
1:00:38 Ed Zitron? I'll explain it very simply, which is OpenAI last year spent $9 billion to lose $5 billion. Anthropic spent $5.3 billion. No, sorry, they, they spent like nine, seven, eight. They spent billions to lose $5.3 billion, of which a chunk of that was just given to Amazon for servers. It's very fucking weird. They lose money, their conversion rate on ChatGPT is awful. So 500 million, I think they have 15.5, 60 million paying subscribers. They don't publish monthly active users, because if they did, you could do the math and see it's trash. And on top of that, they just can't find a way to make money. They lose so much money. So what's more than likely is, yeah, these companies might die. LLMs will hang around because there are use cases.
1:01:23 Ed Zitron? And Google is like, Jeff Dean over at Google is one of the least evil tech people. There are actually people there who like the tech and give a shit about it. And there's more efficiency stuff coming out of Google's models. The thing is, what we see today, I do not believe is, I think the most annoying scenario is gonna be the longest life large language model with unlimited access is gonna be on fucking Meta. Because of their unrestricted tripe that's on every fucking app. But I think things like ChatGPT are just gonna be limited. You may still have people who use them in the way with the gym.
1:01:52 Speaker 04 But it's interesting what you're saying about, um, maybe the, maybe an answer, and you think about the industrial revolution is eventually. Is that people are gonna have to be trained on instead of firing nine thousand people, train nine thousand people.
1:02:06 Speaker 06 Become prompt engineers.
1:02:07 Speaker 04 And yeah, become prompt engineers and and the fact that they didn't, I mean, look, no one is less surprised than me by incentive structures, making these motherfuckers act like evil, um, completely non caring, you know, monsters. But if they can be convinced that there's a profit motive in training people.
1:02:27 Ed Zitron? There's no profits.
1:02:28 Speaker 04 No, if they can be convinced of it.
1:02:30 Ed Zitron? No, but I must be clear, though.
1:02:31 Speaker 04 But eventually, but in all these businesses, there was no profit until there was profit.
1:02:34 Ed Zitron? Nothing even.
1:02:35 Speaker 04 How long did it take Amazon to become profitable?
1:02:37 Ed Zitron? Amazon took about eleven years with AWS, but AWS was the concern that reduced costs for Amazon itself to run their.
1:02:42 Speaker 04 But in two thousand, a lot of people thought it was never gonna become profitable, right?
1:02:46 Ed Zitron? Yeah, but that's not the same logic.
1:02:48 Speaker 04 It is the wrong.
1:02:48 Ed Zitron? Yeah, but no, but the economics are completely. It's completely different economics.
1:02:51 Speaker 06 Can I ask though, Brian, how much would you pay to keep using chat G P T a month?
1:02:55 Speaker 04 I'm a I'm a bad, I I'm I'm older and I've done well and I've I've, do you know what I'm saying?
1:03:01 Speaker 06 Do you pay anything right now, I guess, is the question?
1:03:03 Speaker 04 I paid two hundred dollars a month.
1:03:04 Speaker 06 Two hundred a month, okay, so Google one's like two fifty a month, right?
1:03:07 Ed Zitron? Why do you pay two hundred?
1:03:08 Speaker 06 That's a lot of money.
1:03:09 Speaker 04 Because for the supercharged research.
1:03:11 Ed Zitron? You can get the research on the twenty bucks a month one, though.
1:03:13 Speaker 04 Not the same level of the research.
1:03:15 Speaker 05 That's an interesting thing to point out, too, is that you're getting better quality by paying for a higher.
1:03:19 Speaker 04 Yes, I'm aware of it, it's unfair.
1:03:21 Ed Zitron? No, but seriously, is there a quality difference?
1:03:22 Speaker 04 Yeah, if you look at it, we could, yeah, I mean.
1:03:24 Ed Zitron? Oh, my God.
1:03:25 Speaker 06 But Mike's saying this is the how's the profit?
1:03:27 Speaker 04 They say there's a quality, they would tell you there is an and in the amount of them, whatever the user, what is it, the rate?
1:03:31 Speaker 06 So they sell you shit products to make you buy the less shit product?
1:03:34 Speaker 04 Well, I don't know if they sell you shit product, but you're saying the use rate, I, someone told me, someone I trust a lot, a person who's tech savvy was like, this is when that happened, they used it for a while. And a couple of months ago, um, I was in a meeting with someone and they were like, if I was gonna tell you to spend money on anything, spend two hundred bucks a month.
1:03:52 Ed Zitron? You want to know something crazy, though?
1:03:53 Speaker 04 Yes.
1:03:53 Ed Zitron? They lose money on every 200 buck a month customer.
1:03:56 Speaker 04 I believe you because they're because I'm using because when they do that search, you're saying it's burning so much.
1:04:01 Ed Zitron? But that's the thing.
1:04:02 Speaker 06 Yeah.
1:04:02 Ed Zitron? How likely do you think that that will continue? Because the deep research stuff.
1:04:07 Speaker 06 Is it charging you more or are you gonna quit, I guess?
1:04:09 Speaker 04 Yes, well, of course, there's a number very soon. I'm already at the.
1:04:12 Speaker 06 Yeah, you're very high end of it. Yeah.
1:04:14 Speaker 04 Yeah, for sure. Double this. I would not. I would not pay double.
1:04:17 Speaker 06 Right.
1:04:17 Speaker 04 Because I guarantee I would not pay.
1:04:19 Ed Zitron? And I will be honest, and this is not even me being like a hater or anything. It may not be that cheap, uh, because right now, it's the big story in my newsletter, please.
1:04:27 Speaker 06 What's it? 500 a month?
1:04:29 Ed Zitron? For a 200 a month power user who's already lose using, pardon me, losing the money. They lose so much money on them that it's like 400, 500, a thousand dollars a month. Claude code right now, which is slightly different because the way they do context stuff. Nevertheless, on the 200 dollar a max user on Claude, they are, they could be losing. They, they had someone on, I saw on Twitter, they spent 10,000 dollars in compute on a 200 month subscription. These are the majority of power users. Power users go nuts, as you well know. This is why I'm so pushy on the economics, because what we are seeing today, it's like if every Uber weighed 40,000 pounds and the fuel was giraffe blood. It was just like this insane economic, and I sound like I'm kidding, but
1:05:13 Ed Zitron? the economics are completely bonkers. So as much use as you're getting today, I just don't know how practically they'll provide that. And they might be cheaper models, but the cheaper models not might not be able to do. I assume you use O3?
1:05:25 Speaker 04 Yeah, of course.
1:05:26 Ed Zitron? Yeah, so.
1:05:27 Speaker 04 Which takes a lot. I, yeah, and sometimes it still takes such a long time.
1:05:30 Ed Zitron? And that's the thing.
1:05:31 Speaker 04 To even have a conversation.
1:05:32 Ed Zitron? And that's the thing.
1:05:33 Speaker 04 An incredibly long.
1:05:33 Ed Zitron? How much of this is sustainable long term? And I know the business stuff is annoying because you still have your experience, which you like.
1:05:40 Speaker 04 No, it's not annoying. It's fascinating. I'm fascinated by the whole thing. And this is great.
1:05:44 Ed Zitron? It's just the long term here, and I'm talking long term as in 18 months, is I don't know if we will have deep research at any price point approaching the one we'll have in that period.
1:05:55 Speaker 04 Well, the deep research is the only reason anyone should pay the money.
1:05:57 Ed Zitron? And that's the thing, though. The only reason that people should pay is a thing that is not sustainable and has no. So when you talk about how we bring this towards like an AWS situation.
1:06:08 Speaker 04 Yeah.
1:06:08 Ed Zitron? AWS is like profitability was built on infrastructure expansion. It was because they were building the rails of cloud writ large. It was never in the billions and billions a year with no, there is no path to profitability here. Uh, there was one model that OpenAI said they were going to deliver to OpenAI, uh, Microsoft in 2023 called Arrakis. Uh, and and it, and they failed to do it. They have yet to discover a, or make, because it may not be mathematically possible, a really good large language model that can do the kind of reasoning like that, that would be reliable and have the web search tool.
1:06:42 Speaker 04 And I want to say you do have to, when you talk about prompts, sorry, I was thinking about this when I was learning about complexity theory. I read this book by a guy named Neil Thiess that I love. He's a professor at, at NYU and I'm not good at physics and I really want to understand quantum physics. And so I, I, I would ask questions, right? Do research on this and then find a way to be able to explain it. So go read these books, go find documentaries. And then I just quickly realized at some point that the, that the AI hadn't, it was clear to me it hadn't read something. Yeah.
1:07:14 Speaker 06 Cause it doesn't have access to.
1:07:15 Speaker 04 But I could figure it out. It hadn't. And I just said, like, did you really look at that video? Like it was video. I go, did you really look at that video? That speech? And then it immediately went, no, you got me there. I didn't look at the speech, but I'm going to go now, find it a different way. So you do, and you do have to be
1:07:29 Speaker 06 Vigilant.
1:07:30 Speaker 04 acculturated to having those kinds of.
1:07:33 Ed Zitron? Crazy.
1:07:33 Speaker 04 Uh, I like, you have to maybe have advanced degrees or have trained yourself. I, so I'm, I'm not, this is fascinating, right? Because I take for granted the steps I, I take for granted.
1:07:44 Speaker 06 Your learned behavior, right?
1:07:45 Speaker 04 To, um, and make my way through the world.
1:07:48 Speaker 06 Yes.
1:07:49 Speaker 04 And like the way that I would interrogate something like that to get to an answer that's useful.
1:07:56 Speaker 06 Yeah. As opposed to maybe my parents would be like, oh, okay. You watched that video? Oh, okay. Yeah. Okay. Right. Right. That's your answer. Right?
1:08:01 Speaker 04 Yeah. And they might get wrong information from it.
1:08:03 Speaker 06 Exactly.
1:08:04 Speaker 04 I can't argue with that. You're right about.
1:08:05 Ed Zitron? And I think that a lot of this, a grand theory of this comes down to. I think people have a lot of questions and not a lot of people to ask them to. Questions about their life, how they're feeling, then.
1:08:16 Speaker 06 That's the loneliness thing.
1:08:17 Ed Zitron? I think it starts with the medical side where it's. It's quite hard to ask a doctor a question in any country. It's hard to know whether they and also doctors regularly make you feel annoying. I'm not talking about my doctor, he loves me. Uh, but you can't go to a doctor regularly with little questions. They don't have the time because they must maximize profits or they are busy, one of the two.
1:08:37 Speaker 04 Yeah.
1:08:37 Ed Zitron? And ultimately, we are sitting there going, I've got this weird rash or like I might leg itched in this place in the same day, three days running. What could that mean? And you can't really Google that.
1:08:48 Speaker 06 It's just cancer, that's what WebMD tells you.
1:08:50 Ed Zitron? But you can't really Google that, but ChatGPT, whatever, can do an impressive impression of it. Or, in your case, Brian, and I think this is reasonable, lead you in a direction towards like a Mayo Clinic article about a particular thing. Something good raised to your doctor. That makes sense. People are lonely. People just have weird questions and I think that there is partially a bad side where it's like everyone wants everything immediately. We must have everything we want immediately. But also, people are curious and we are more disconnected as people. We are more decentralized as people. We don't meet people. We don't have, we are by at scale overworked, underpaid. So we don't have the time to be generous with our time.
1:09:29 Speaker 04 Yeah. I'll give you one other use case though, because and it goes to the, the question of training, training. But you could use this for, you said your parents dance or whatever. So I'll tell you that, um, if you are somebody who is all use like deadlifting or squatting with a barbell as an example. If I put a thirty second clip into ChatGPT and I say, please watch this. Tell me, is this form at risk of injury? How should I modify it? Is it good enough? What do you think about the load of this weight? The answer it will give is, and I, I have checked it against like the world, the answer it will give is outstanding. And that's, and that's how would you get that in another way? I don't think there is, and that's a very small example.
1:10:14 Speaker 04 Well, that's different than search. There's an expert, no, that's.
1:10:17 Ed Zitron? It's searching using a video.
1:10:18 Speaker 04 No, it's not.
1:10:19 Speaker 06 The matching is a bit less sophisticated in the regular Google search.
1:10:24 Ed Zitron? No, I'm saying that, no, that is.
1:10:26 Speaker 04 It's not merely matching it against a perceived perfect form. It's looking at your femur. Literally, it'll go with your femur size. This is the kind of squat that you could do if I low bar versus high bar. Here's why. Here's what this looks like. You're over mid.
1:10:38 Ed Zitron? I'm agreeing with you. I'm just saying that search as a term has grown to look at this image and compare it to these sources, which is theoretically what search.
1:10:46 Speaker 04 But then in the good language, right?
1:10:47 Ed Zitron? Yeah, yeah, yeah. Oh, no, no, it's the natural language.
1:10:49 Speaker 06 Simplifies the steps, yeah.
1:10:51 Speaker 04 How to fix it.
1:10:52 Speaker 05 But I do worry if you ask a bad quest, like if you not ask a bad question, phrase a question incorrectly. And I'm not saying you. If one phrases, let's say they ask a fitness question, but they phrase it a little bit weird and the answer they get is harmful, it almost, I'm worried about.
1:11:08 Speaker 04 Nightmare.
1:11:09 Speaker 05 Situations like that. And also it feels like we're removing an element of human responsibility or accountability where it's like, well, the machine answered weird rather than like a doctor answered weird.
1:11:20 Ed Zitron? And I think that lots of corporations love that idea.
1:11:22 Speaker 05 They love that because it's like, if I, if my AI accidentally denies your insurance, it was the AI, it wasn't us.
1:11:29 Ed Zitron? It's the algorithm. It's the same algorithmic pass on thing. But I think you're right in the, that is kind of cool. I also have used those three because I do pay for this. I'm not a baseless hater. I did ask, what's the distance between this bottom of this photo frame and the floor? And it spent 10 minutes to give me a completely insanely wrong answer. They're not good with numbers.
1:11:52 Speaker 04 Oh, fascinating. See, the other day in this example I just gave you, it said, I can't, it just said like, I can't see.
1:11:58 Ed Zitron? Which is good.
1:11:58 Speaker 04 That was great.
1:11:59 Ed Zitron? They should do that.
1:12:00 Speaker 06 What version are you using for this?
1:12:01 Ed Zitron? O3.
1:12:02 Speaker 06 Both of you using the same model?
1:12:03 Ed Zitron? I was using O3 on ChatGPT plus though. So now I, now my.
1:12:07 Speaker 06 Try the pro.
1:12:08 Ed Zitron? Now my, I don't know how I feel about giving Clammy Sammy 200 bucks.
1:12:11 Speaker 04 Just try it for a month and see what it does.
1:12:13 Ed Zitron? Yeah, I don't.
1:12:14 Speaker 05 You can write it off.
1:12:15 Ed Zitron? I don't want to write.
1:12:16 Speaker 04 I already gave 200 bucks to Wario Am a day. No, because I would love you to do that and then call me and say it's the same. No, no.
1:12:21 Speaker 06 Try the same result.
1:12:22 Speaker 04 No, I would love you to tell me. I'm now actually really curious. I would love you to say to me, dude, just spent 20 bucks, the same, that would be.
1:12:28 Speaker 06 Then you could save your 180 a month.
1:12:30 Speaker 04 Yeah, that'd be great news, please.
1:12:31 Speaker 06 I am, what Mike was bringing up, I thought was going to be similar to the point I had been mulling over when you were talking about the training stuff, which is that we already deal with due to like, you know, capitalism, lower barriers to entry, an influx of individuals who may not actually be fully equipped for the jobs they purport to do. So whether it be journalists like myself, or like, again, fitness influencers or trainers that say they have whatever types of physical health degrees that are just the result of a 10-hour course online, that sort of thing.
1:12:59 Speaker 04 Yeah.
1:12:59 Speaker 06 We're already dealing with the like, quality dilution of like, information coming from sources like that. To throw AI into the mix feels like making it even worse, like harder than ever to tell what the truth is. And I don't know about you all, but I find myself gaslighting myself all the time now, regardless of like, my own life, whether it's the truth in the world, whether I'm being too sympathetic to multiple different perspectives. I don't know what the truth is anymore. I can't tell you what the cold, hard, scientific truth of anything ever is. And that's where it's led me.
1:13:29 Speaker 04 You also have to be willing, I agree with, that's a brilliant point. I think one of the things I would say to people, if someone asked me, what, how do you, how should you communicate?
1:13:35 Speaker 06 Mm-hmm.
1:13:36 Speaker 04 I, I would say, and it's really painful for people, like, because I've seen them talk online about what they love about, um, conversing with AI. You gotta say, click every toggle that says, be mean, tell me the truth, don't tell me I'm smart. Yeah. Like, you gotta instruct it to really be withholding in that way, if you want to actually engage, so that, so that you're right, you're not getting gaslit. Because yes, I agree that, and this is dangerous, the default setting is to gaslight you. You gotta actually go, I don't need to hear that. Glaze you, I think the young people.
1:14:10 Speaker 06 Glaze you.
1:14:10 Speaker 04 Sure. Wait, can we talk about one topic before we're done? And, okay. You didn't hear the producer just laugh out loud. Just one, can we talk about just one, I think they used to say glaze, but can we just talk about one, um, positive, purely positive tech thing that happened the last week? Okay. July 4th. Okay. Who, who's on TikTok and knows all about the anti-pasto party in Texas? The biggest story on TikTok? Please.
1:14:32 Speaker 06 Was it the one that one person went to? Yeah, okay.
1:14:35 Speaker 04 It's the greatest thing. Please, run this through this. Okay, this, they're in Texas? These people have a July 4th party. There's a woman, she's just moved there recently. Her kid becomes friends with someone else's kid and this woman, Sarah is the parent of one boy, says, "Come with me over to these people's party." Woman A makes the apotheosis of all antipasto salad, the greatest salad you've ever seen in your life, goes to their house.
1:15:03 Ed Zitron? Yeah.
1:15:04 Speaker 04 And these people are like, "Who's this stranger in our house?" And they kick her out, even though she brought this incredible. And she goes home, gets on TikTok and she's crying. And she's like, "I brought this salad and they kicked me out of their house." And the entire internet defended her.
1:15:21 Ed Zitron? Yeah.
1:15:22 Speaker 04 And loves her so much. And it's like, it's an incredible story. Everybody I know is like, everybody of all ages, like nieces and nephews of mine, and then older, people older than me are all sending, and it's an amazing thing. It's a huge community has rallied to hate these fucking people and to love her and her homemade mozzarella and homegrown tomatoes that she brought to their house.
1:15:46 Ed Zitron? That's like scaling local news.
1:15:48 Speaker 06 I saw something else, but yeah. No, I saw something else altogether.
1:15:51 Speaker 04 It's really worth finding.
1:15:52 Speaker 06 But like Reddit does things like that too.
1:15:53 Ed Zitron? And that's the thing, Reddit, like this is, I liked ending this on a positive note. That's lovely. Reddit does, you can say something.
1:15:59 Speaker 06 Well, yeah, no, community and social forums like that, that's what the internet is great for. And not a lot of AI is present in a lot of that.
1:16:07 Ed Zitron? And it's almost Reddit, especially right now, has got good because it isn't there. The CEO keeps thinking of shoving it places, but even on the better offline Reddit, we've got nearly 9,000 of you now. Hey, guys. But it's great because I, one of my biggest stories I wrote recently was on Cursor and them falling apart. And it was because someone on the forum was just like looking through their stuff like, "Ah, you", and everyone had this full conversation about it. You've got these people out there in this morass of fake stuff or generated stuff or SEO stuff. You've got genuine people. There is still a joy to all of this crap. I love Blue Sky as well, but Reddit has really just, I'm shocked.
1:16:45 Speaker 04 I spend a lot of time on Reddit. I read it.
1:16:47 Ed Zitron? I'm shocked by how much I love it.
1:16:48 Speaker 04 There are groups I won't ever go to on.
1:16:49 Ed Zitron? Oh, sure. Yeah, of course.
1:16:50 Speaker 06 Pick your own.
1:16:51 Speaker 04 I mean, I know, but I'm saying, you know, like, but, but there where people have a hobby or a thing like whether it's music, you know, you're like, I Take stuff. Yeah, it's great. And you must love the like, the Claude Reddit, because they hate every day. Someone's like, why does Claude suck today? You must love that.
1:17:09 Speaker 06 I love r slash Google, yes.
1:17:11 Ed Zitron? I, I like r slash SAS because it's all just people like running SAS. No, it's all people's software. No, no, no, you're thinking of a different SAS one. I'm talking S A A S.
1:17:21 Speaker 06 Oh, software as a service.
1:17:22 Ed Zitron? Yeah, I'm a loser. So, um, no, you watch people being like, my app has been up for six months and it's made seven dollars. Should I continue? A bunch of people are like, yes.
1:17:30 Speaker 06 And those subscribe.
1:17:31 Ed Zitron? No, I, I kind of love them.
1:17:33 Speaker 04 Yeah.
1:17:33 Ed Zitron? That you've got these niches, but I hate to say it. I do need to call this episode.
1:17:37 Speaker 06 No.
1:17:38 Ed Zitron? Brian, where can people find you?
1:17:40 Speaker 04 Oh, uh, Instagram.
1:17:42 Ed Zitron? Yes, we'll have a link to there as well.
1:17:44 Speaker 04 Yeah.
1:17:44 Speaker 05 Uh, you can find me on Instagram at Mike Drucker is dead and on Blue Sky at Mike Drucker. And, uh, buy Good Game No Rematch. It is a book that's available digital, hardcover or audio with the audio read by myself.
1:17:57 Ed Zitron? Hell, yeah.
1:17:57 Speaker 06 Very nice.
1:17:57 Ed Zitron? Shirlen?
1:17:58 Speaker 06 Uh, I'm at engadget.com or on threads at Shirlen Instagram, C H E R L I N N S T A G E R E N.
1:18:04 Ed Zitron? Type in Google, the man who destroyed Google search. That's me. I'm Ed Zichon. Thank you so much for listening as ever. This is recorded in the wonderful New York studio in iHeart Radio. Daniel Goodman, of course, is our producer. Thank you so much, Daniel. Thank you all for listening.
1:18:25 Ed Zitron? Thank you for listening to Better Offline. The editor and composer of the Better Offline theme song is Matt Osowski. You can check out more of his music and audio projects at mattosowski.com, M A T T O S O W S K I dot com. You can email me at ez@betteroffline.com or visit betteroffline.com to find more podcast links and of course my newsletter. I also really recommend you go to chat dot where's your ed dot at to visit the Discord and go to r slash better offline to check out our Reddit. Thank you so much for listening.
1:18:58 Speaker 02 Better Offline is a production of Cool Zone Media. For more from Cool Zone Media, visit our website, coolzonemedia.com, or check us out on the iHeart Radio app, Apple Podcasts, or wherever you get your podcasts.
1:19:36 Speaker 00 Empower is all about helping you invest well, so you can go out and live a little. And their Free Empower Personal Dashboard shows you your full financial life in one place. You'll be able to see your net worth, monitor budgets, analyze your portfolio, set retirement goals, and more. Whether you're saving for something big, want to treat yourself, friends, or family, or meet long-term goals, Empower helps you see where you stand. You've worked hard for your money, and Empower can help you make the most of it. Download the Empower Personal Dashboard or visit empower.com, not an Empower client, paid, or sponsored.
1:20:07 Speaker 08 Cash flow crunch? OnDeck's small business line of credit gives your business immediate access to funds, up to $200,000, right when you need it. Cover seasonal dips, manage payroll, restock inventory, or tackle unexpected expenses without missing a beat. With flexible draws, transparent pricing, and control over repayment, get funded quickly and confidently. Apply today at ondeck.com. Funds could be available as soon as tomorrow. Depending on certain loan attributes, your business loan may be issued by OnDeck or Celtic Bank. OnDeck does not lend in North Dakota. All loans and amounts subject to lender approval.
1:20:37 Speaker 07 This is an iHeart Podcast. Guaranteed Human.
Previous Make Fun Of Them, Pt. 2 Next Monologue: The Updated Pale Horses of the AIpocalypse Page updated November 1, 2026 · transcript and summary are machine-generated