Bloomberg's Joe Weisenthal on AI monopoly risk, the open letter, and why Twitter drives the entire AI narrative

Speaker 2: spared executive amenity. Maybe they were gonna do a podcast in it. You never Anyway, we have Joe Weisenthal from Bloomberg, Odd Lots. You know him. You love him. He's back on the show. Welcome

Speaker 1: to this YouTube and Ultradumba. Joe. How are doing? Joe. It's been it's been too long. I miss you guys. Miss you too. You too. Okay. So I have a question for you. You've got you've had a podcast for a long time. Have you ever gotten, like, a negative review or a negative comment? Ever? No. Never. Never. No. Be honest. Be honest. Maybe you've got one. Have you ever thought about sending them a box of live cockroaches, maybe vandalizing There are actively people mad at Joe right now on Twitter about his playlist. I don't know. Hopefully, she's not watching right now, but my cohost Tracy would

Speaker 7: if if one of us would do it first of all, she is the one who reads the comments. I don't think she would. But I think if, like, you're to float with this idea, like, she might at least, like, spare a few mental I

Speaker 1: about doing that. I ran it down, but I I know the whole the the backstory with eBay. I try I try to be the voice of the team. It's a cautionary tale. You guys don't wanna have to, you know, down the line pay a $5,060,000,000 dollar settlement. Yeah. Definitely not. No. Sending a horse head to a former listener for leaving a bad review. We have that is not on the road map. No. Not at all. I'm trying to think if I can I don't I don't know if you caught the story right before this, but, basically, eBay had to pay a couple $56,000,000 because, like, a bunch of executives were were harassing this couple for just like criticizing the company in 2019? No is even really thinking about eBay in 2019. It's not like it's a big company. It's a company that everyone, lot of people have used and a lot of people love. But no one's, like, really it's not really that top of mind. No. And seven executives are, like, becoming criminals for a turf. I will die for this company.

Speaker 7: You know, in 2018, they might've said, oh, that would never be me. Maybe who among us, if you're late at night, you start going down a path, you have to exhibit really good cognitive hygiene because who among us knows the path to psychosis online? Many such cases, and maybe we all need a little bit of humility about what the set of conditions that could put us in that path. Yeah. Sometimes one prompt

Speaker 2: is all it

Speaker 7: Sometimes one prompt. I took a week of vacation last week, and I was like, oh, I'm not gonna post anything for a week. And I did. I actually dude. But I felt it all, like, bubbling out, like, Friday evening when I sort of mentally it's like, oh, vacation's over. It's the weekend now. And I was like, it's just like all was like, uncontrollably. So it's like, we all have to be mindful and self aware about what we're capable of because some maybe you don't really know, like, when it comes to Internet vendettas,

Speaker 2: what what could happen. Do you think social media is addictive? Because I can't it's so hard to tell when your job is social media. Like, I am the screen time. I have three hours of livestream time every day. Yeah. But it's like how I buy food. So it's like, am I am I addicted to my job? Like, I don't know.

Speaker 7: Well, I thought about the funny you mentioned food because this is how I justify it. Right? I've probably said it before. This is how I justify always being a lie. It's like, oh, it's very important for me to know what people are talking about. This is how I hone in my skill. But it is, like, you know, it's funny. I was thinking, like, food specifically because with other addictions, you just go cold turkey. Right? You're just like, I'm not you know? And with food, you can't, so you have to, like it's really hard to, like so even with me in social media, it's, like, it's hard for me like, cold turkey is not that hard. Mhmm. I'll just take a week off. Posting all the time is not that hard. The middle ground is when it gets really hard because there's just so many things to respond to, so many things waiting for my take, so many questions that only I need to So

Speaker 1: There's there's takes that will never exist. Right? Yeah. The takes that have been lost to history when you were on vacation. Yeah. You will never get those takes. That's right.

Speaker 2: Do feel like that you think of. Yeah. I I do feel like one benefit that you have and I have felt is when you have multiple, maybe dopamine reward paths, but, like, outputs, it is nice to be able to retreat because you can do a podcast. You can also write an essay. You can also rip off a post. And so if you only have one of those, then you can sort of get lost in the sauce of, oh, it's been rough since I had a great guest or it's been rough since I had a great viral banger post. If you can sort of move around and be like, okay. I'm gonna take a break from the from the the the knife fight at the dive bar and put my thoughts together into an essay,

Speaker 7: that could be quite This is very important to have multiple outlets for man's desire to to post, to put things out into the world. Yes. Yeah. Deepest desire, it's like in the hierarchy, is like to be someone we are listened to. Right? You could see it in all of the the billionaires who like cannot resist the impulse to get online and start posting. It's clear that money, yes, it's all great. Power is great. But what they really crave Yeah. Is to just be heard by someone out in the world, and so they start posting. The the the death of Twitter acts

Speaker 2: fully over now given that Oh, yeah. Jensen Huang joined, Mark Zuckerberg had to join. Like, it pull truly pulled everyone over Yeah. Slowly kicking and screaming because I imagine Mark Zuckerberg would prefer to promote his Wall Street Journal op eds just on his own platforms, but that's not where that community is. You gotta go.

Speaker 7: No. It's true, like, with AI especially specifically, which is that it's x. Right? Like, if every time someone leaves one of the big companies Yep. They have to post a big thread Yep. Or, you know, a really long tweet or whatever, and it's only x. I mean, they could post it on Instagram. They could post it on LinkedIn. They could post it they don't, though. Yeah. Mean, they might. I don't know. But they posted that is where the news happens. And you could feel like, I think, you know, the thing that was really striking last week with the the discourse as one says about the open letter, etcetera, it germinated on Twitter specifically. Why did the White House why did this why was this suddenly in that week a thing that the White House had some way needed to talk about Chinese open models? It was because of how much it was bubbling on social media, on Twitter specifically. I actually think the entire thing more or less originated from one tweet that said that Kimmy three was really good at front end design work. Yeah. That was that was it. It was that one post. Because if you look around, like, that actually that crystallized it in everyone's mind that it's like, here it is. It may be as good as the most advanced frontier models. No. There was one Yep. Category Yep. Of one overall thing, and that was the thing everyone locked into. And from there and I believe that that was the moment straight through to, okay, now the White House and all these companies need to make comments about open models can be traced to that specific

Speaker 2: all online, all on Twitter from that. Ameri front end built this country. If we don't have front end, we don't have any. This country needs front end development specifically. No. No. I mean, I I I agree with you. When I when I was looking at the benchmarks, I was like, the one that I'm actually interested in is the cybersecurity benchmarks. Because that's where we could actually have this meaningful discussion of, okay, you just actually democratized hacking into things. This is gonna get kinda crazy, and you're gonna have, like, script kitties out there that are, just hacking into the random flower shop on the corner and just making it a nightmare for them.

Speaker 1: And we couldn't really get there. I don't know. It hasn't been fully I got you posted this, but someone was sharing this morning. They're trying to figure out how much you need to spend to have your own Yeah. Setup at home to be able to Oh, bring Kimmy efficiently. It was like a $100,000. Yeah. Right?

Speaker 2: Just just to get to the point where you'd be able to Yeah. And it's a lot of energy. Sovereign way. You know, a sovereign hacker would need to have a 100 gram, which hopefully prevents, like, the script kiddies. It might be the new Lambo, though. You know? You you got the Influencer, the LARPer. They got the NBL 72 in the corner. You're like, yeah. What's up? I got the I I got the sovereign AI going. I don't know. How else did you process the the open letter? Were you surprised by the order that companies signed? I Ben Thompson had this interesting take that it's two of the last companies to sign. I don't think Apple signed yet. You'd think that they benefit from open source since they haven't been playing that game. Amazon was also a little bit late. But Amazon and Apple both have physical products like warehouses, and they sell physical things. And so they're maybe a little bit less nervous about AI disruption because they have a physical aspect. I don't know what I I don't know what what you thought about the the the sequence of events.

Speaker 7: Yeah. I mean, I think there was, like, two things that really struck out to me, which is that the first wave of signatories were either companies that are behind the frontier or NVIDIA, which does not want to be left with only one or two clients. Like, this is the thing. So they're like, oh, like, this is let's start with NVIDIA. I think the nightmare scenario for NVIDIA or at least a nightmare scenario for NVIDIA is that it basically has it two one or two labs break so far from the pack that it's just, you know, almost monopsony

Speaker 1: conditions. Sure. And then those clients can either say and to me and and just to add to that. Yeah. Like, both the the the two front runners are already making their own chips. They're already working with multiple chip companies. Right? So they're already, like Totally. Actively trying to reduce

Speaker 6: their dependence on Yeah. Key supplier.

Speaker 7: So so first, they might have better pricing. They're like, no. We're just not gonna pay that margin that you expect, or they're already both diversifying, etcetera. So that is that is the first thing. And then the other thing that seems very obvious to me is that the all the first round of signatories was objectively falling behind the frontier. And so you're when you're in software categories, you have, you know, either companies that are trying to build a model, and they're behind. So Microsoft, at least on just we're talking LLM's clearly behind. Meta, clearly behind. Right? Mhmm. And then you get the and then you have the company, I don't know, the ServiceNows of the world who look like they could be in a position where they might have to pay a tax to the most advanced models. Yeah. And so, like and then you get delayed signatories, like, OpenNII or an Alphabet, etcetera, that are closed, but probably not at Fable five, like Max, like, look at the whole thing. And then they're the last ones to join. And so, like, I guess I had I was like, look. I love all that stuff about I love freedom. Like, I love freedom and I love tinkering. I love tinkering, and I wanted to, like, do all that stuff, etcetera. But when I think, like, what really, like, motivates these companies is, you know, I don't put a ton of stock on, oh, it's really important to diffuse AI into our main street businesses and hospitals and hackers. Sure. Great. I love that stuff. Lovely. But, like, you know, I think there is, you know, I think the thing that would really like, we're really not psychologically or sort of as an economy really equipped to think about what if one or two of these companies just runs away with it? What if Anthropic just runs away with it? Mhmm. What if, like like, we're we're sort of like, oh, yeah. Everyone's sort of in the game. And I think for most of us, I'm not a 100% sure of this because I'm not the most sophisticated, like, person technical person at all. But I have a feeling, like, most of us can't really get a good intuitive feel for, like, what is the difference between a GPT 5.6, a Fable five, or even the most advanced Gemini model or a Kimi or a DeepSeek four or whatever. Like Yeah. We're probably, like, not really able to personally grasp what the difference is at the extremes. But, like, what if we what if in a year from now, there's one company? What if or two that are just so, like, clearly, like, there is no longer any debate. These other ones just are not in the game. That is a very weird, I would say, very deeply uncomfortable situation we would find ourselves. I think none of us would like that. I think most of us feel comfortable in a world in which you have a handful of companies roughly jostling around the quote frontier. It's like, okay. This is, like, comfortable for us. But if we weren't in that situation, like, I think we would be rightfully, all of us would be quite anxious.

Speaker 2: So was Lena Khan right?

Speaker 7: Well, this so this is the thing. I think that, like and I think, like, the the Mark Zuckerberg op ed in The Wall Street Journal, I think, further hammers this point. And I actually think and there's a sense in which I find the Mark Zuckerberg op ed to be a little bit maybe more interesting than the open letter because the open letter is sort of this plausible deniability, where it's just we're advocating for a principle here. That principle is open. The Zuckerberg

Speaker 2: op ed And it's also like And it's and it's more their

Speaker 1: it's it's just, again, I think part of what you appreciated is, like, using the open letter format to just advocate for your own commercial interests. It's almost like stolen valor.

Speaker 7: For sure. So the open letter, the Zuckerberg op ed in The Wall Street Journal, I thought actually took it a bit further. He's like, we cannot have superintelligence controlled by one company. So that was not merely saying, let's let the market decide, let's keep light touch regulation. He was saying, no, this would be existentially bad. I think most people would actually agree with, if one company were to be the monopoly provider of, quote, superintelligence, unquote. So then you think, okay Which is funny too because

Speaker 1: he's like It's very funny. Well, he he's probably top three largest customers of Anthropic. Oh, Right? Right. So so he's the one fuel you know, fueling all And of then and then Yeah. Right. Two months later, he's like, okay.

Speaker 7: Actually But he said that, like, to me that, like, if that is a very telling fact when you think about, like, who are the largest customers, the competitors. So it's like that marginal difference. Like I said, like, most of us probably can't really tell a big difference between, like, advanced models. But when you have, like, a really big, very high skilled, very, like, sophisticated model maker, and they're a huge customer of the a major competitor that's better than them, like, people are going to the frontier. And this is something that, like, surprised me. Like, I thought maybe, like, there would be a bit more of a market for, like, sub frontier intelligence or, like, quote, cheaper intelligence. But look how fast Anthropic turned out Opus five. Mhmm. And I, you know, I haven't really used it that much, but it could be like that in a market where there's if there is one winner, that winner could emerge all up and down the price curve until you have, like, your Gemini Flash level. Frontier. Yeah. Yeah. Mean, that's what that's what Sam was on Invest Like the Best yesterday, and he's talking about how

Speaker 1: you, as as a company that sells intelligence, you want to have the best model at every point of the cost curve. Like, it's actually gonna be important to hit all of those different moments. Yeah. And I I I

Speaker 7: seems like and look, if things change by the week, so I'll come back out on a week and I'll tell you say something else. But it seems like a while like I started, you know, like I was just sort of poking around. Was like, oh, Gemini Flash. This is really fun for, like, high batch, low intelligence things. Right? Like you could do 30,000 queries or in two hours, and it cost you $20 or so. That's amazing. It could be pretty good for life as a golden tree for a high batch, low intelligence.

Speaker 1: Yeah. That's it. Just doing a lot of stuff.

Speaker 7: But probably or it seems at least plausible that the company that is has the most advanced model will probably also win for the sort of, like, low or could at least be in a good position to also win the high batch, low intelligence

Speaker 1: Yeah. But I would say but I would say that there has been a lot of there has been a lot of pushback around the RSI narrative in the last twenty four, forty eight hours because Opus five is actually not getting great reviews from the community. Think people had initially they felt like people thought it was like a great in between between like a fable and a five six soul. And then now I've seen a lot more negativity Well towards it. To

Speaker 7: me, again, goes back to our original point of sort of, like, the incredible importance of Twitter in all of these discussions. Because if you think about, like, each time a model comes out, okay, there's the benchmarks and it's like they but then it's like, you'll just see, like, someone will be like, there was some tweet when GDLM 5.2 came out that it he's like some guy was like, oh, I've been using GDLM 5.2 for the last eight hours. I've never seen an open model like this. Got, like, 2,000 retweets. It's like, eight hours? What are you doing? You're like, what are we doing here? I'm like, so I was like, oh, like, I do I got a sort of weird vibe from Opus five or well, then all these things. It's like Yeah. I don't know. I actually think this is probably a good reason for all of us to maybe log off a little bit more and to actually just read through benchmark reports

Speaker 2: rather than, like, one tweet or, like I've never seen a on front end dev like this. On the flip on the flip side of, like, the benchmarks, like, it actually makes me more strong on Mark Zuckerberg's, like, ability to produce a thoughtful opinion because he has cons he spent a billion dollars on frontier intelligence in a month deployed across 10,000 people and can look at, like, the needle how the needle moved across different models and and actually Yeah. Look at the, like, the efficiency. Like, the ROI is, at the end of the day, what matters in the in a business context. Yes. The vibe check matters. Yes, the benchmarks matter. But at the end of the day, in terms of moving the needle from $10,000,000,000 to $100,000,000,000 in revenue to $1,000,000,000,000 in revenue, a lot of that's going to be driven by our companies actually like, getting the bill and being like, yep. I'm happy to pay because it it it was worth it?

Speaker 7: This is something I noticed a while ago, which is that you read through, like, a lot of these, like, popular benchmarks that come out. Mhmm. And the thing that has surprised me is that up until recently, most of them actually don't talk about the dollar cost of completing a task. So it'll be like, can the model solve these 100 problems in engineering or whatever? It's like, oh, this one got 85, and this got 75, But we're really going to need is to see, what was the Okay, many dollars do I have to spend to get 90% of these questions solved? And we're just starting to see that. I have a theory, but I don't really know if it's right, that probably a lot of these benchmark organizations start off from a safetyist mindset. They're worried about, oh, like, bio bioengineering risk or how Can you do it? Binary. And can you do it at any cost? Right? Because you don't really care about your token cost if you're, like, trying to exert a lot of damage. You just do it. So I think that for, like, the bigger questions, like, hopefully, we'll see more, like, what was the dollar cost Yeah. Of achieving x at this model? And then maybe we'll, like, really those are the for I think for our conversations, those would be a lot more interesting than some of the sort of legacy benchmarks.

Speaker 2: How much do you think the open source it feels like we're in this moment of, like, this catch up. There was a the last six months pre GLM five point two. It felt like open source was sort of falling behind as the Frontier Labs released more models and more crazy demos, and there were more stories about mythos and solving math problems and all these crazy conjectures and stuff. Like, the the labs were sort of trading back and forth with, like, capability, new capability, new benchmark, crazy thing. And open source is kinda quiet. And then we got GLM 5.2, Kimi k three. Yeah. And all of a sudden, it was like, oh, okay. Like like, the the the the open source market is catching up for sure. Still on the same trajectory, maybe still behind, but, like, you know, like like, on track. How much is that moving the market, do you think? How much are are we in a replay of the deep seek moment, the Jevan's Paradox debate? Because it feels like there's this sell off, but I don't know. Like, it it feels weird to me that people would be processing it the exact same and making the same mistake that they made. So is this time different, or how how are people talking about it in New York? Like

Speaker 7: Yeah. Us people who just care about the lines going up. It is I I think, like no. For real, though. I mean, I do think, like, it's plausible that this on some level, this is just another, quote, deep seek moment in the sense, like, you get a little bit freaked out. I agree with you. Like, it's weird to think that, like, you would do that same playbook twice. I think, like, another factor is just generally setting aside, like, maybe the Kimi and GLM stuff as a little bit maybe it's a little overrated in terms of a story. I think people are looking at a situation in which they just do not foresee the end of these free cash flow losses. Right? That there is just like we're we're nowhere close. It's like, okay. When can you when is the off ramp for this spending? Yeah. That story isn't totally satisfying because if that were if it were just about, oh, they're spending a gazillion dollars on capital expenditures, you would expect the Korean chip stocks to do well. I mean, the Korean chip stocks have been absolutely getting crushed over the last three weeks. Yeah. So, like, there's something more to the story here. Like, I don't like, it could just be that a lot of manic it could be two separate things. It could be there were a bunch of manic traders in Korea who were buying these double and triple x leveraged ETFs that just trade SK Hynix. Yeah. And then that had to eventually flush out because there's a you know, there's so much volatility baked into that that you get a little selling and it ripples. And then in The US, it could be that investors are looking at like, I do not have any confidence that I could name the date in which the alphabets of this world will return to positive free cash flow. There is no reason to think that that off ramp is coming when they could say, like, oh, we're good. We have enough compute. So maybe there's a couple of things going on that I think are perhaps a little bit separate from the pure,

Speaker 2: like, oh, we're freaked out about margin compression from Chinese open models. Yeah. It's very odd because the industry revenues are going up. Like, and and the and the revenues are very, very diffuse. Everything from a ad supported use to a $10 subscription to a $200 subscription to a hyperscaler spending a billion dollars in a month on tokens. Like, the the demand is there in that it is this economy. It's not this total circular. There's just one like, it's it's like NFTs or something like that. It's not some dislocation, but I agree with you that as you play out the CapEx requirements, we're we've now drawn down all the hyperscaler cash flow. We're issuing more debt and equity. And at a certain point, you're like, okay. If there's another 10 x increase, if there's another 100 x increase, like, there's sort of not enough money in the world for this. Like and and so No. I see. Almost need a pause on CapEx more than you need a pause on, like, the capability development for because I do think that if you just if you just spent the CapEx that you're spending now, paused, and then just did the diffusion work to get these models into all the different businesses and use cases, like, the economic upside and the ROI would be there. But Yeah. Not when you're gonna spend 10 times next on CapEx. Yeah. I think my

Speaker 1: framework is like you have the whole world betting on a meaningful capability like continued capability increases, which I think is like okay to lean on because Sure. We've just been seeing that quarter over quarter. It is it is real. Yeah. And and then you also have some like hoarding, which is like, not only are the capabilities gonna get much more much better, I'm also gonna basically figure out a model that's gonna be so good. It's gonna just, like, hoover up a bunch of, like, new demand and I'll be able to ramp revenue and justify all of this CapEx. And if I but if I can't do that, someone else will. I'll be able to I'll be able to offload it. Right? But the concerning thing right the concerning thing right now is, like, the the incremental big buyers of compute, you effectively have a two players there. And and then at the same time, have the whole industry being like, woah, like, praying for, like, an oligopoly as opposed to a duopoly or a monopoly. Right? Yeah. Like, very interesting scenario to have the biggest companies in the world so scared of a of a of a

Speaker 7: monopoly that they're effectively praying for an oligopoly is, like, is my my view on it. Yeah. It's really wild. And and, you know, I don't know if you guys saw today the recent blog post from Dwarkash. Of course. Dwarkash had a great day about, like, expectations that compute could get a lot more expensive Yeah. Which, you know, we've one of the sort of stylized facts of the last few years has, of course, been the extraordinary drop in token prices, right? Like if you go back to like GPT-three token prices, they're down like 99.9%. But like, you know, so we think, okay, there is some iron law that token prices will get cheaper over time, but maybe not. I mean, we thought there was an iron law that memory prices would drop over time, right? Or that any sort of hardware would drop over time. And what the story of the last two years is that turns out that is capable of reversing. So, you know, it's not impossible to imagine that compute prices I think it's not impossible to imagine that compute prices get more expensive and get significantly more expensive. Then, of course, he posited the possibility. Maybe there is no, almost no thing that we do economically currently with AI is is economical where compute prices go, which is hard to imagine. And then the thing that I take away from it is, but that doesn't mean you could stop spending on it because you have to worry about hardening yourself for cybersecurity. Because it's like, all right, you know what? There's like, this is not economical for me anymore. I want to tap out of the whole AI risk. Well, then your competitor develops a stronger model and then you're done. You're really done. So, like, this is the scary thing, which is and I think even just to the sort of, like, hyperscalers is, like, you what if, like, free the situation is you're bleeding free cash flow, but you also don't have the option of saying we want to opt out of the game. Yeah. Which I think is a source of anxiety for investors. It is it is

Speaker 2: like, it's so weird that we went back and forth on, like, the doom narrative and the and and no matter where you sat on that, it was either, like, perfect utopia, everything's amazing, or, like, terminators kill everyone. And in fact, like, we are very much in this messy middle of, like, a lot of annoying things. Like, the cybersecurity equilibrium is I mean, maybe there's, like, this doom snarl, but most most likely, it's just, costs go up for everyone loosely because you have to harden yourself more against AI. And it's like, yeah, it's great to pick bugs, but there's gonna be more bugs introduced. And it's just like it's just annoying. It's annoying. And there's a lot of there there there's a lot of scenarios where it's like, they're okay. There's some more cost on AI slop and then you have to detect the slop and then you're annoyed. And we we have friend on our team who always laments when, like, AI does some amazing, like, math thing. He's like, no one cares. Like, that this is why you need to return to, like, you know, the the curing cancer, the doing something useful,

Speaker 1: actually helping you with some problem that people do experience. But a lot of times, it's just these abstract Going back going back to the Zoc op ed, he has he's in the position where he doesn't have a he didn't do a bunch of podcasts about AI and AI doom in, like, 2019 That's when no one cared about AI. Whereas a lot of the other key players were doing podcasts and they were talking about the risks and all these things. So he can kind of come out and just say like, look, it's gonna be awesome. Just gotta make sure there's fair competition. The pushback there is that like the average He's not the best messenger for this because the average person believes that social media, while it is great in some ways, has damaged birth, you know, hurt birth rates, right? Like, damaged teen mental health, right? Just draining the attention, right? All these things. So, like, So he's maybe the best messenger that we got, but he's not the best. So

Speaker 7: the issue that I arise when I think about this is that, okay, you have two executives. One is like, it's all going to be okay, And one of them, it's we're gonna have doom. Right? So and then it's like, well, which one of them has been paying attention and taking seriously AI for the last ten years, and which one of them three years ago was talking about the metaverse? Well, it's like, I'm gonna trust the one who has been taking AI seriously for the last ten years, and then, unfortunately, that happens to be the doom guy, or that seems how it is. Yeah. That, unfortunately, that, like, if you just go by these heuristics we have of, like, well, who was like, there have been people who are, like, very right about AI who are very nervous. Yeah. And this is, like, what makes me I don't have any ability to adjudicate these questions. Like, I'm just, like, we're just reading stuff on the Internet. I'm, like, you know, I'm not so but, unfortunately, a lot of the people a lot of the doom people have had much better intuitions about just the broader trajectory of the technology

Speaker 1: than the people who are very optimistic, who three years ago or four years ago were talking about NFTs. Well, and there's some line, I think it's from Demis, where he's saying, like, he met with Zoc Yep. When they were selling DeepMind. It's a great Zoc is Zoc. Of Zoc, you know, viewed, like, the metaverse as, like, or VR as, like, equally interesting to AI, and that was part of what made him decide to not go to Meta. Yeah.

Speaker 2: Has Dennis never watched a movie in VR? It's electric. It's the best experience. No. I saw just that. Wanted to you you got it. It's amazing. Apple Vision Pro. It's it's fantastic. But the the there was a doom narrative around crypto, and there was a doom narrative around metaverse. The the crypto one was that every criminal is gonna be able to transact anonymously, there's and gonna be a huge increase in drug Well, doom scenario was actually that the dollar would just completely collapse. But that was something that people were cheering about. They were excited about that. Yeah. And then and then with the metaverse, it was like wire heading. So everyone's gonna be in the metaverse, and we're gonna lose any and all interaction in the real world. And, of course, Mark Zuckerberg never addressed that. He was never on podcast being like, yes. It is there is a risk that it goes into the matrix, and we're all in these virtual worlds and, like, I'm not gonna do that. Apple sort of, like, forced his hand with that with the crazy AR and, like, the screen on the outside. And Apple's messaging was like, we don't wanna be an isolating device. We wanna be a a mixed reality device. And and they were sort of being, like, the anti doom, but Zuck never really went full, like, doomer or even addressed it. And and and then, of course, we just, like, sort of got the good ending because it just, like, kinda flatlined and never really went anywhere. I don't know. I don't even know if that's the good It's, like, a middle ending. No. No. Totally. And it's interesting because when you read, like, that Zuckerberg

Speaker 7: op ed in the Wall Street Journal, it reads like you know? So you mentioned, like, everyone is super cynical about social media. Right? You mentioned the birth rates. Everyone's got so, like, so it just feels so, like, kinda gross about the whole thing now. But, you know, in, like, 2010, 2011, people were, like, really excited. It's like, we're gonna connect. It's gonna be great for democracy. Citizens are going to, like, be able to, like, collaborate and stuff like that. And so when I read the Zuck Up Ed, I was like, this is a kind of optimism that I haven't read from anyone in over ten years about where digital technology is going. Unfortunately, people like you know, very few people are as optimistic about social media as they were in, say, the beginning of the twenty tens. But it it did hearken back to that this is gonna be very democratizing, etcetera, and I hope it is. Well, I

Speaker 1: expect, like, personally, social media, I do think has made my life my life better. Like, net like, if you net it out, it's like I you a know, I work Me too. I work in media, but even if I didn't, you know, pre being in in this kind of business, met a lot of people, I learned a bunch of stuff that I otherwise wouldn't, I think it's been a had a net positive impact on my life. I think that AI, hopefully, one of the good scenarios is that it is the same way, right, where like net net makes everyone's lives a bit better even though it has these things that are

Speaker 7: Yeah. Well, I think are we quite should bad, right? I think we should be open minded to that possibility. I mean, you know, look. We get things everyone's dooming so hard these days about it, and you just go down the list. You know, I can we could easily rattle off 10 very three or four big things right now. But, you know, usually the we're we're wrong. We're pretty bad about predicting the future. And so, like, we should be open minded, I think, towards, like, you know, right tail scenarios where things go well because I don't think, like Yeah. I put a lot of stock or I don't think there's any good reason to put a lot of stock on what the Twitter conventional wisdom is Yeah. Or the editorial board of any publication, what they're all anxious about right now. Like, you know, like, I don't know what that great outcome looks like per se. You know, I do look. We could get great technological breakthroughs. I'll say one thing, which is that, like, in the circles of people I talk to, everyone is, like, reading more philosophy these days probably because of AI. It's pretty good. Like, I'm pretty happy. Like, that's a pretty cool thing that, like, people are sort of, like, a reignition of a certain curiosity about things that I think was sort of dormant for a while. So, like, that seems like a pretty great music time. For the first time in years. Making music even make a lot easier at the beginning. Like, we should pay attention to, like, the I actually do think we should pay attention to the good things that are you know, because it's so easy to get caught caught focused on a negative narrative. In the meanwhile, like, various, like, positive things are happening. I, you know, changed a flight recently, and it was much easier because I used Chad GPT to, like, walk me through the process of, like, turning Let's go. I'm nervous. Of, you know, to to a a round trip flight to a three leg flight, which I was very anxious about. It's easy. So, like, we should pay attention Yeah. I think, to things in our, like, day to day life that we can say, you know what? AI is sort of making this better even if, like, it doesn't feel like in the whole even if we're, like, anxious about where it's all going.

Speaker 2: Last question. You have a podcast. We have a podcast. That's right. There's a lot of new podcasts. Would you support an open letter that we're writing? We would like you to sign in to ban new podcasts. Would require a of people talk ban, but it would there would be a multiyear

Speaker 1: sort of approval process, bunch of safety checks. Yeah.

Speaker 2: So, like, a new government agency to issue RSS feeds. Yeah. I this is really important. It is. I'm I'm so glad you brought this up. Because someone Such an important topic. Someone could just launch an RSS feed that, you know, ends the entire world. Who knows? We don't know it. Can't happen. Or or removes viewers and listeners from mod lots and TVP, and we cannot let that happen. I that would that's ridiculous. Are you saying that I would do this for my own financial gain? This is crazy. You're saying I think believe you would even go there. That's wild. I would never think that way. I'm purely thinking altruistically. Okay, Joe? Let's leave it. America needs to win. On podcasts. We do. Yeah. This is very important. We actually are objectively winning. We are winning on just talked about. We have been winning on entertainment and have no No. There's no competitors yet. Yes. Well, thank you so much for signing the letter. I love it. I love it. We'll circulate. Great to see you, Joe. Have a great rest of your week. We'll talk to you soon. Take care. Bye. Bye. Let me tell you about