1 00:00:02,720 --> 00:00:15,840 Speaker 1: Bloomberg Audio Studios, Podcasts, Radio News. 2 00:00:18,480 --> 00:00:21,320 Speaker 2: Hello and welcome to another episode of the Ad Thoughts Podcast. 3 00:00:21,440 --> 00:00:22,720 Speaker 2: I'm Tracy Alloway. 4 00:00:22,600 --> 00:00:23,800 Speaker 3: And I'm Joe Wisenthal. 5 00:00:24,079 --> 00:00:26,680 Speaker 2: So, Joe, we continue to publish some of the conversations 6 00:00:26,680 --> 00:00:29,720 Speaker 2: from our recent live show in New York, and I 7 00:00:29,720 --> 00:00:31,920 Speaker 2: hope people didn't mind. They didn't seem to, they seem 8 00:00:31,960 --> 00:00:33,760 Speaker 2: to enjoy it. But we did do a little bit 9 00:00:33,760 --> 00:00:36,280 Speaker 2: of media industry naval gazing. 10 00:00:36,080 --> 00:00:37,960 Speaker 3: Totally, so as we've been talking about it. If you 11 00:00:38,040 --> 00:00:40,720 Speaker 3: listened to past episodes from the show, you know, the 12 00:00:40,720 --> 00:00:43,760 Speaker 3: theme obviously had this big like sort of future of markets, 13 00:00:43,880 --> 00:00:47,640 Speaker 3: future of trading theme. But you know, I would say, 14 00:00:47,840 --> 00:00:50,879 Speaker 3: and maybe this is like I'm very biased here, but 15 00:00:51,080 --> 00:00:56,160 Speaker 3: I would say that information dissemination collection And I guess 16 00:00:56,200 --> 00:00:59,000 Speaker 3: you say, journalism is an important part of markets, right, 17 00:00:59,000 --> 00:00:59,920 Speaker 3: how people get informed? 18 00:01:00,120 --> 00:01:02,040 Speaker 2: Say journalism? Like that journalism? 19 00:01:02,200 --> 00:01:04,520 Speaker 3: No, But you know, like the thing like in the 20 00:01:04,520 --> 00:01:06,800 Speaker 3: way if you think about like the floor and the 21 00:01:06,880 --> 00:01:10,040 Speaker 3: faun of the market ecology, then the people who like 22 00:01:10,280 --> 00:01:13,919 Speaker 3: report the news, digest the news, explain the news, highlight 23 00:01:14,040 --> 00:01:17,080 Speaker 3: what news is relevant, what is not relevant are important 24 00:01:17,120 --> 00:01:19,760 Speaker 3: actors in that system and in a period of so 25 00:01:19,880 --> 00:01:23,280 Speaker 3: much change and uncertainty. The question of like A, how 26 00:01:23,319 --> 00:01:26,800 Speaker 3: you decide what's important and b how you convey that 27 00:01:26,880 --> 00:01:29,440 Speaker 3: to an audience. These are really tough questions. 28 00:01:29,560 --> 00:01:32,280 Speaker 2: Yeah, so, I think there's two major challenges here. One 29 00:01:32,280 --> 00:01:34,360 Speaker 2: of them is what you just said, which is like, 30 00:01:34,480 --> 00:01:38,160 Speaker 2: how do you explain these huge technological shifts to people 31 00:01:38,240 --> 00:01:40,840 Speaker 2: who may not you know, they might be outside of 32 00:01:40,880 --> 00:01:43,480 Speaker 2: the tech industry. How do you explain like an incremental 33 00:01:43,560 --> 00:01:47,200 Speaker 2: improvement in a model to the average lay person. And 34 00:01:47,280 --> 00:01:51,200 Speaker 2: then secondly the other big challenges. The media itself is 35 00:01:51,240 --> 00:01:54,240 Speaker 2: caught in the crosshairs of the debate over whether AI 36 00:01:54,360 --> 00:01:55,840 Speaker 2: is going to take all the jobs. 37 00:01:55,920 --> 00:01:56,120 Speaker 4: Right. 38 00:01:56,200 --> 00:01:58,440 Speaker 3: That's it, right, So any one of us who's in 39 00:01:58,520 --> 00:02:02,120 Speaker 3: the business of trying to consume, figure out what's important 40 00:02:02,160 --> 00:02:05,120 Speaker 3: and relay it is also thinking about well AI do 41 00:02:05,160 --> 00:02:08,120 Speaker 3: a better job than we do, and precisely that included 42 00:02:08,160 --> 00:02:10,519 Speaker 3: writing newsletters, are producing podcasts. 43 00:02:10,560 --> 00:02:11,000 Speaker 5: That's right. 44 00:02:11,040 --> 00:02:13,760 Speaker 2: So we had a trio of perfect guests for this 45 00:02:13,840 --> 00:02:17,240 Speaker 2: particular discussion. All of them have been on the show before. 46 00:02:17,720 --> 00:02:19,600 Speaker 2: We had James van Gelan. He is, of course the 47 00:02:19,680 --> 00:02:22,880 Speaker 2: founder of Satrini Research and the author of the viral 48 00:02:23,040 --> 00:02:27,560 Speaker 2: AI Jobs Doom scenario as well as Jasmine's son. She 49 00:02:27,680 --> 00:02:31,399 Speaker 2: writes a substack that's all about AI and the culture 50 00:02:31,440 --> 00:02:34,840 Speaker 2: of Silicon Valley and tech as well as samro He 51 00:02:34,960 --> 00:02:37,959 Speaker 2: is the author of one of my favorite newsletters with 52 00:02:38,080 --> 00:02:41,360 Speaker 2: the best name, the t kre So take a listen. 53 00:02:42,200 --> 00:02:43,000 Speaker 3: Where to even begin? 54 00:02:43,280 --> 00:02:43,840 Speaker 5: I'll start with. 55 00:02:43,840 --> 00:02:46,040 Speaker 2: James because he's next to me and I just want 56 00:02:46,040 --> 00:02:47,800 Speaker 2: to know what his life is like. Right now, do 57 00:02:47,880 --> 00:02:50,320 Speaker 2: people like recognize you on the street and scream at 58 00:02:50,320 --> 00:02:51,680 Speaker 2: you about AI job losses? 59 00:02:52,560 --> 00:02:56,639 Speaker 6: I've done two credible death threats, but not on the street. Okay, 60 00:02:57,000 --> 00:03:00,239 Speaker 6: I guess yeah, now yeah, I mean you know better 61 00:03:00,240 --> 00:03:04,359 Speaker 6: when it's no person right, But now I've I luckily, 62 00:03:04,440 --> 00:03:07,760 Speaker 6: I have been very much an internet personality and not someone 63 00:03:07,800 --> 00:03:10,480 Speaker 6: that does a lot of taped interviews. 64 00:03:09,880 --> 00:03:10,280 Speaker 7: Like this. 65 00:03:14,639 --> 00:03:17,880 Speaker 3: Percent of you. No, But in case if you randomly don't, 66 00:03:17,919 --> 00:03:20,440 Speaker 3: I assume you do. When was it that February or 67 00:03:20,480 --> 00:03:24,520 Speaker 3: something you wrote a post about a theoretical possibility of 68 00:03:24,639 --> 00:03:27,040 Speaker 3: you know, we're all going to lose our jobs but anyway, 69 00:03:27,680 --> 00:03:31,160 Speaker 3: so I could see why people were upset about that. Jasmine, 70 00:03:31,200 --> 00:03:33,880 Speaker 3: you write about AI and you sort of like try 71 00:03:33,880 --> 00:03:36,640 Speaker 3: to when I read you're writing it's like bridging a 72 00:03:36,720 --> 00:03:39,760 Speaker 3: gap right to some extent, because there are people, most 73 00:03:39,760 --> 00:03:42,120 Speaker 3: of them in San Francisco, often on a lot of 74 00:03:42,160 --> 00:03:45,440 Speaker 3: drugs or maybe they're on an occult or something like that, 75 00:03:45,680 --> 00:03:47,720 Speaker 3: and there's like these are like the people who are 76 00:03:47,760 --> 00:03:50,080 Speaker 3: like going to influence the rest of our lives. How 77 00:03:50,080 --> 00:03:52,560 Speaker 3: do you think about the question of, like what is 78 00:03:52,640 --> 00:03:56,000 Speaker 3: actually important to communicate to a broader public. 79 00:03:56,200 --> 00:03:57,880 Speaker 7: Yeah, I mean it's really interesting because I think the 80 00:03:57,960 --> 00:04:01,240 Speaker 7: vast majority of aimedia coming out of Silicon Valley is 81 00:04:01,280 --> 00:04:03,760 Speaker 7: buy AI people for other AI people, and it serves 82 00:04:03,760 --> 00:04:06,000 Speaker 7: that purpose really well, right, Like you got really in 83 00:04:06,080 --> 00:04:08,200 Speaker 7: depth podcasts, you got the Door Hush podcast, you got 84 00:04:08,200 --> 00:04:10,880 Speaker 7: this whole ecosystem of sub sex. One thing that I 85 00:04:10,960 --> 00:04:13,160 Speaker 7: really noticed in that I think most people who listen 86 00:04:13,240 --> 00:04:15,760 Speaker 7: to the comms coming out of the industry leaders will 87 00:04:15,760 --> 00:04:18,560 Speaker 7: notice is they're saying all these crazy things without like 88 00:04:18,800 --> 00:04:21,640 Speaker 7: four second thinking that like any normal person might hear 89 00:04:21,680 --> 00:04:23,320 Speaker 7: them say it, and they're like, oh, man, why don't 90 00:04:23,320 --> 00:04:25,279 Speaker 7: these people like AI very much? Just like I don't know, 91 00:04:25,320 --> 00:04:27,240 Speaker 7: You've been telling them you're going to take all their 92 00:04:27,320 --> 00:04:29,919 Speaker 7: jobs and kill everybody for the last several years, and 93 00:04:30,040 --> 00:04:32,359 Speaker 7: like you're on the record saying this stuff, and so 94 00:04:32,880 --> 00:04:35,640 Speaker 7: I think it's really interesting where it's not like crypto 95 00:04:35,680 --> 00:04:38,039 Speaker 7: where only a small fraction of sort of the broad 96 00:04:38,080 --> 00:04:39,800 Speaker 7: public ever deeply engaged with it. 97 00:04:39,880 --> 00:04:42,200 Speaker 5: AI is something that, whether people want to or not, 98 00:04:42,320 --> 00:04:43,279 Speaker 5: is impacting their lives. 99 00:04:43,279 --> 00:04:45,599 Speaker 7: It's on their social feeds, their kids are using it, 100 00:04:45,600 --> 00:04:47,880 Speaker 7: it's in their workplace, and so people have all these 101 00:04:47,960 --> 00:04:52,080 Speaker 7: questions about the intersection of AI and politics, AI and affordability, 102 00:04:52,120 --> 00:04:55,560 Speaker 7: AI and parenting and education, and the majority of I 103 00:04:55,640 --> 00:04:58,200 Speaker 7: think AI media historically has not really focused on that. 104 00:04:58,240 --> 00:05:00,560 Speaker 7: It's been more of the business and technology community. And 105 00:05:00,600 --> 00:05:03,039 Speaker 7: so I'm sort of trying to say, like, okay, but 106 00:05:03,080 --> 00:05:03,640 Speaker 7: what about the. 107 00:05:03,560 --> 00:05:04,120 Speaker 1: Rest of us? 108 00:05:04,640 --> 00:05:07,440 Speaker 2: Sam, you write a newsletter and the tag line is basically, 109 00:05:07,640 --> 00:05:11,000 Speaker 2: you know, stock markets usually go up over time. You've 110 00:05:11,240 --> 00:05:13,719 Speaker 2: you've been very right for the past couple of years 111 00:05:13,760 --> 00:05:17,280 Speaker 2: because they've certainly gone up, But like, are things starting 112 00:05:17,279 --> 00:05:20,159 Speaker 2: to get uncomfortable for you when people are talking about 113 00:05:20,200 --> 00:05:21,880 Speaker 2: AI bubbles and overvaluations? 114 00:05:21,880 --> 00:05:22,000 Speaker 5: Oh? 115 00:05:22,040 --> 00:05:25,560 Speaker 4: Yeah, things are always uncomfortable for me. I think that's 116 00:05:25,560 --> 00:05:28,039 Speaker 4: sort of like one of the most important things about 117 00:05:28,120 --> 00:05:30,960 Speaker 4: understanding this idea of stocks UJ go up is the 118 00:05:31,040 --> 00:05:35,240 Speaker 4: key word there is usually right, Like every way you 119 00:05:35,320 --> 00:05:38,560 Speaker 4: cut the data historically, even good times and bad times 120 00:05:38,560 --> 00:05:40,919 Speaker 4: and bull markets and bear markets, you always have like 121 00:05:41,000 --> 00:05:44,440 Speaker 4: these periods of volatility where you know, stocks do go down, 122 00:05:44,839 --> 00:05:47,000 Speaker 4: like that's the that's the catch share with stocks uj 123 00:05:47,160 --> 00:05:50,440 Speaker 4: go up is that they go down a lot often. So, yeah, 124 00:05:50,480 --> 00:05:54,200 Speaker 4: of course I'm nervous, especially when you're at at peaks. 125 00:05:54,440 --> 00:05:56,200 Speaker 4: I mean, of course, the data will also tell you 126 00:05:56,279 --> 00:05:59,400 Speaker 4: that the you know, twelve month returns after all time 127 00:05:59,440 --> 00:06:02,080 Speaker 4: highsac tend to be higher than when you're at low's. 128 00:06:02,400 --> 00:06:04,560 Speaker 4: But yeah, of course I'm nervous. I mean, I think 129 00:06:04,560 --> 00:06:07,479 Speaker 4: that's like very healthy for anybody in the investor class. 130 00:06:08,120 --> 00:06:11,080 Speaker 4: Is even no matter how optimistic you might be over 131 00:06:11,120 --> 00:06:13,039 Speaker 4: the near term or long term, you have to be 132 00:06:13,080 --> 00:06:15,600 Speaker 4: prepared for those big drawdowns because they do happen, and 133 00:06:15,600 --> 00:06:16,200 Speaker 4: they happen. 134 00:06:16,000 --> 00:06:20,120 Speaker 3: Frequently, Jasmine. You know, so James wrote his piece about 135 00:06:20,240 --> 00:06:22,880 Speaker 3: you know, the potential at least and it wasn't a forecast, 136 00:06:22,880 --> 00:06:24,520 Speaker 3: it was a prediction, but a scenario. 137 00:06:24,600 --> 00:06:28,440 Speaker 6: We've also written others right, right, but no one. 138 00:06:28,400 --> 00:06:31,159 Speaker 3: Likes the idea of like mass job laws. But I've 139 00:06:31,160 --> 00:06:33,320 Speaker 3: been reading some other stuff and there's a lot of 140 00:06:33,360 --> 00:06:36,400 Speaker 3: people who don't talk about AI job laws. They talk 141 00:06:36,440 --> 00:06:40,520 Speaker 3: about AI will literally kill every single person in the 142 00:06:40,560 --> 00:06:44,120 Speaker 3: world if the labs don't do a good job of 143 00:06:44,160 --> 00:06:48,640 Speaker 3: aligning the models. How seriously do you think the public 144 00:06:48,680 --> 00:06:54,159 Speaker 3: should take that specific element that if the models are misdesigned, 145 00:06:54,560 --> 00:06:56,760 Speaker 3: they will destroy humanity as we know it. 146 00:06:57,680 --> 00:07:03,599 Speaker 7: Ah Man contents, Oh god, they call this a pe 147 00:07:03,760 --> 00:07:06,800 Speaker 7: doom out in San Francisco. I won't give you a probability, 148 00:07:06,839 --> 00:07:09,479 Speaker 7: I suppose, but without getting to the question of like, 149 00:07:09,480 --> 00:07:12,160 Speaker 7: will literally every single person die? I do think there 150 00:07:12,240 --> 00:07:15,880 Speaker 7: is a very good reason to be concerned about misalignment 151 00:07:15,960 --> 00:07:18,800 Speaker 7: and ROGUAEI, because when you think about it, there's sometimes 152 00:07:18,840 --> 00:07:21,480 Speaker 7: people say that safety and the safety of a model 153 00:07:21,600 --> 00:07:25,440 Speaker 7: is maybe intention with these goals like acceleration, and is 154 00:07:25,480 --> 00:07:26,520 Speaker 7: the model really useful? 155 00:07:26,760 --> 00:07:27,840 Speaker 5: And I think the thing that's. 156 00:07:27,720 --> 00:07:30,120 Speaker 7: Really interesting is in the history of AI research is 157 00:07:30,160 --> 00:07:32,840 Speaker 7: often the people who make the biggest technical advances and 158 00:07:32,880 --> 00:07:36,200 Speaker 7: these foundational models who end up the misconcerned about misalignment. 159 00:07:36,240 --> 00:07:40,400 Speaker 5: And that's because actually the product functionality. 160 00:07:39,800 --> 00:07:42,080 Speaker 7: Of the models, their utility is only as good as 161 00:07:42,160 --> 00:07:44,160 Speaker 7: how controllable they are and how much. 162 00:07:44,080 --> 00:07:45,200 Speaker 5: They do what you expect. 163 00:07:45,240 --> 00:07:47,480 Speaker 7: So when you say, hey, like AI agent like go 164 00:07:47,560 --> 00:07:49,520 Speaker 7: off and like make me a million dollars or whatever, 165 00:07:49,880 --> 00:07:53,280 Speaker 7: like whether that model goes and does so in a 166 00:07:53,360 --> 00:07:56,760 Speaker 7: law abiding and safe and non violent way. The safety 167 00:07:56,760 --> 00:07:59,040 Speaker 7: of the model, the alignment of the model is very 168 00:07:59,080 --> 00:08:02,320 Speaker 7: correlated with whether it is economically useful as well. 169 00:08:02,480 --> 00:08:03,400 Speaker 5: So I think that it's. 170 00:08:03,240 --> 00:08:06,480 Speaker 7: Important that we not hold them in total tension. And 171 00:08:06,560 --> 00:08:08,480 Speaker 7: I think there are really good reasons to be concerned 172 00:08:08,480 --> 00:08:10,080 Speaker 7: with whether these agents are doing the things that we 173 00:08:10,120 --> 00:08:10,760 Speaker 7: expect them to do. 174 00:08:11,280 --> 00:08:13,520 Speaker 2: Just going back to AI job losses, and I guess 175 00:08:13,560 --> 00:08:16,440 Speaker 2: any of you can answer this question, and please like 176 00:08:16,480 --> 00:08:19,240 Speaker 2: feel free to chime in on every question that we ask. 177 00:08:19,360 --> 00:08:21,040 Speaker 2: But you know, you also hear a lot of the 178 00:08:21,040 --> 00:08:26,520 Speaker 2: big tech CEOs talk about being genuinely concerned about the 179 00:08:26,560 --> 00:08:30,760 Speaker 2: future of society once AI is developed and starts like 180 00:08:30,960 --> 00:08:35,959 Speaker 2: taking people's jobs. Some people interpret that as AI CEO 181 00:08:36,040 --> 00:08:38,800 Speaker 2: is basically talking their own book and being you know, 182 00:08:38,920 --> 00:08:42,840 Speaker 2: like hyping up the capabilities of AI. And then you 183 00:08:42,840 --> 00:08:45,760 Speaker 2: have some people who argue that like maybe they're genuinely concerned, 184 00:08:45,800 --> 00:08:49,520 Speaker 2: not least because they're going to get like molotov cocktails 185 00:08:49,559 --> 00:08:51,440 Speaker 2: thrown at their houses and things like that. 186 00:08:52,120 --> 00:08:53,679 Speaker 5: Where do you fall in that spectrum. 187 00:08:53,760 --> 00:08:55,840 Speaker 2: Is this like a real concern when you talk to 188 00:08:56,080 --> 00:08:57,600 Speaker 2: senior people in Silicon Valley? 189 00:08:57,720 --> 00:09:00,280 Speaker 6: I do think that most of the people that I've 190 00:09:00,280 --> 00:09:03,520 Speaker 6: spoken to who are worried about it are genuinely worried 191 00:09:03,520 --> 00:09:08,240 Speaker 6: about it. Something that I'd add to it is pretty 192 00:09:08,280 --> 00:09:11,160 Speaker 6: much unanimously throughout history, when we've had any sort of 193 00:09:11,160 --> 00:09:14,640 Speaker 6: technological leap forward, it's been a positive thing. And I 194 00:09:14,760 --> 00:09:19,160 Speaker 6: think that AI will mirror that. It's just the question is, 195 00:09:19,679 --> 00:09:22,840 Speaker 6: we've never had a technological kind of advancement this quickly. 196 00:09:23,400 --> 00:09:27,120 Speaker 6: We you know, we go from having the Industrial Revolution 197 00:09:27,200 --> 00:09:31,240 Speaker 6: and the mechanization of agriculture and ninety five percent of 198 00:09:31,280 --> 00:09:33,960 Speaker 6: people used to work in agriculture, and then now I 199 00:09:33,960 --> 00:09:36,440 Speaker 6: think it's five percent too. And I'm very happy to 200 00:09:36,480 --> 00:09:41,079 Speaker 6: work in an air conditioned office rather than in a field. 201 00:09:41,440 --> 00:09:44,760 Speaker 6: A little bit. Yeah, so maybe not. But the thing 202 00:09:44,880 --> 00:09:48,880 Speaker 6: is that will happen. I think believe very strongly it 203 00:09:48,920 --> 00:09:51,599 Speaker 6: will happen with AI. It's just what happens in the interim. 204 00:09:51,760 --> 00:09:54,400 Speaker 6: What happens if the models get so good and and 205 00:09:54,840 --> 00:09:57,800 Speaker 6: can and people kind of adopt them with the same 206 00:09:58,480 --> 00:10:02,240 Speaker 6: approach and use them correct over five years instead of 207 00:10:02,280 --> 00:10:04,520 Speaker 6: one hundred. So I think speaking to some of the 208 00:10:04,559 --> 00:10:07,600 Speaker 6: senior people, they would echo the fact that over the 209 00:10:07,600 --> 00:10:10,840 Speaker 6: next thirty years, AI is going to be an immense 210 00:10:10,880 --> 00:10:13,880 Speaker 6: force for good and going to make people's lives more comfortable. 211 00:10:14,200 --> 00:10:17,160 Speaker 6: They're just worried about the pace of the trend. 212 00:10:17,320 --> 00:10:21,240 Speaker 4: Yeah, I think the transition is like super important because 213 00:10:22,040 --> 00:10:25,080 Speaker 4: as much as everyone likes this idea, like even on 214 00:10:25,120 --> 00:10:28,600 Speaker 4: the corporate side or the shareholder side, like the promise 215 00:10:28,679 --> 00:10:32,840 Speaker 4: of all this productivity that's unlocked by AI, and you know, 216 00:10:32,880 --> 00:10:35,199 Speaker 4: maybe some people are thinking quietly in terms of like, oh, 217 00:10:35,240 --> 00:10:37,480 Speaker 4: this is going to replace all of our workforce. But 218 00:10:37,760 --> 00:10:40,960 Speaker 4: you know, the economy sort of stops working when no 219 00:10:41,000 --> 00:10:43,280 Speaker 4: one has jobs and no has income. Right, It's like, 220 00:10:43,320 --> 00:10:45,480 Speaker 4: it's great that you have this huge profit margin, but 221 00:10:45,559 --> 00:10:47,920 Speaker 4: you have no revenue because everyone's unemployed and they have 222 00:10:48,000 --> 00:10:50,600 Speaker 4: any money. And you know, those people aren't shopping, those 223 00:10:50,640 --> 00:10:53,120 Speaker 4: people aren't buying you know, washing machines and the washing 224 00:10:53,120 --> 00:10:56,320 Speaker 4: machine manufacturer kit, you know, by you know, mental parts. 225 00:10:56,320 --> 00:10:58,200 Speaker 4: And suddenly we're digging up less copper, and the whole 226 00:10:58,240 --> 00:11:01,320 Speaker 4: economy shrinks because no one has jobs. Then the whole 227 00:11:01,360 --> 00:11:05,080 Speaker 4: point of this exercise becomes sort of meaningless, So which 228 00:11:05,080 --> 00:11:08,080 Speaker 4: is why I think you're increasingly hearing from folks on 229 00:11:08,160 --> 00:11:10,800 Speaker 4: this side, you know, talking about things like you know, 230 00:11:11,200 --> 00:11:14,040 Speaker 4: basic income and guarantee jobs and all these things. So 231 00:11:14,040 --> 00:11:17,520 Speaker 4: it's I think that's where you know, the conversation sort 232 00:11:17,520 --> 00:11:20,600 Speaker 4: of goes, is, all right, let's make this assumption that 233 00:11:20,640 --> 00:11:23,319 Speaker 4: we have all this productivity because of job loss. Well, 234 00:11:23,880 --> 00:11:24,600 Speaker 4: that doesn't work. 235 00:11:24,720 --> 00:11:24,920 Speaker 5: You know. 236 00:11:24,960 --> 00:11:27,000 Speaker 4: We people need to have money to go out and 237 00:11:27,000 --> 00:11:30,320 Speaker 4: spend and buy these products that are being produced so efficiently. 238 00:11:30,800 --> 00:11:32,800 Speaker 7: And I mean it's worth noting that, like for example, 239 00:11:32,840 --> 00:11:35,760 Speaker 7: like people politicians will point to say, retraining is the default. 240 00:11:35,760 --> 00:11:38,720 Speaker 7: We have never had a successful large scale reskilling program 241 00:11:38,760 --> 00:11:41,360 Speaker 7: in the history of the United States. During say, you know, 242 00:11:41,400 --> 00:11:44,400 Speaker 7: the de industrialization. You lost all these factory jobs. It 243 00:11:44,440 --> 00:11:46,959 Speaker 7: wasn't that many jobs, and we created way more net jobs, 244 00:11:47,000 --> 00:11:49,160 Speaker 7: say in Silicon Valley, but the steel workers did not 245 00:11:49,240 --> 00:11:51,640 Speaker 7: learn to code, right, And so people are right to 246 00:11:51,640 --> 00:11:54,800 Speaker 7: be concerned, both morally and about the political backlash that 247 00:11:54,840 --> 00:12:15,040 Speaker 7: could occur even with relatively small numbers of net job loss. 248 00:12:15,080 --> 00:12:17,920 Speaker 3: Sam like, I love your newsletter, I read it and 249 00:12:18,120 --> 00:12:23,000 Speaker 3: paid subscribe. Where all that I think I could train 250 00:12:23,120 --> 00:12:26,080 Speaker 3: a model to write a newsletter that say stocks go up? Yeah, 251 00:12:26,120 --> 00:12:28,320 Speaker 3: I think I could, like I believe or not? 252 00:12:28,520 --> 00:12:33,120 Speaker 4: My uh my, my sister. My sister recently got into 253 00:12:33,120 --> 00:12:36,680 Speaker 4: claude coding and she did this project with herself and 254 00:12:36,679 --> 00:12:38,600 Speaker 4: and thought that, you know, I find this really interesting. 255 00:12:38,640 --> 00:12:41,640 Speaker 4: And basically she created this app that was like a 256 00:12:41,720 --> 00:12:46,199 Speaker 4: sam robot that was informed by you know, stuff that 257 00:12:46,240 --> 00:12:48,760 Speaker 4: I've published and like by by bilin all this free 258 00:12:48,760 --> 00:12:51,280 Speaker 4: stuff that's out there, and it's like, I was kind 259 00:12:51,280 --> 00:12:54,400 Speaker 4: of offended first of all, but yeah, it's a problem. 260 00:12:54,559 --> 00:12:57,480 Speaker 3: Do you worry about it? Is it absolute proprietor of 261 00:12:57,520 --> 00:12:59,960 Speaker 3: a yeah, the company, you know, it's one. 262 00:13:00,160 --> 00:13:02,520 Speaker 4: It's like there was this time where you could go 263 00:13:02,559 --> 00:13:06,240 Speaker 4: out and say, well, you know, something that you can't 264 00:13:06,440 --> 00:13:10,199 Speaker 4: really replace is you know, the individual's voice, the personality 265 00:13:10,280 --> 00:13:12,800 Speaker 4: or whatever. And it's like I, you know, I run 266 00:13:12,840 --> 00:13:14,439 Speaker 4: some of this stuff too, and run these queries and 267 00:13:14,480 --> 00:13:16,680 Speaker 4: say like, well, how would Samua write this? And it's 268 00:13:16,760 --> 00:13:20,600 Speaker 4: like it sounds like me? And sometimes it actually you know, 269 00:13:20,760 --> 00:13:23,200 Speaker 4: uses the language and comes up with the words faster 270 00:13:23,600 --> 00:13:26,280 Speaker 4: than how I would. So it's like, you know, what 271 00:13:26,360 --> 00:13:28,640 Speaker 4: I do has to be about something more than just 272 00:13:29,000 --> 00:13:33,200 Speaker 4: having the samual voice right, so to answer the question 273 00:13:33,320 --> 00:13:35,200 Speaker 4: like the problem and the challenge and I you know, 274 00:13:36,000 --> 00:13:38,240 Speaker 4: James and Jasmine and we talked about this a little 275 00:13:38,240 --> 00:13:40,720 Speaker 4: bit in the back is like, you know, where do 276 00:13:40,800 --> 00:13:43,000 Speaker 4: you add that value? And it's like it's it can't 277 00:13:43,000 --> 00:13:45,560 Speaker 4: be something that's like just sort of replicable or or 278 00:13:45,600 --> 00:13:49,200 Speaker 4: something that you know, someone who might be interested in 279 00:13:49,240 --> 00:13:51,920 Speaker 4: what you're reading or writing about can put into a 280 00:13:52,000 --> 00:13:55,640 Speaker 4: query and get back like, you know, the toughest part 281 00:13:55,800 --> 00:13:57,840 Speaker 4: about my job, and I think a lot of our 282 00:13:57,960 --> 00:14:01,480 Speaker 4: jobs is being able to come up with those ideas 283 00:14:01,600 --> 00:14:05,120 Speaker 4: and those angles that no one's asking for. Yeah, and 284 00:14:05,400 --> 00:14:07,560 Speaker 4: you know, again like I write about the stock market 285 00:14:07,640 --> 00:14:09,920 Speaker 4: usually going up, which, by the way, even before AI, 286 00:14:10,040 --> 00:14:11,960 Speaker 4: it's kind of a ridiculous thing to be writing about 287 00:14:12,160 --> 00:14:15,200 Speaker 4: because everyone who has money in their fore one ks 288 00:14:15,240 --> 00:14:17,640 Speaker 4: and iras, I've already been told this. They already know this. 289 00:14:18,000 --> 00:14:21,400 Speaker 4: So it's like why do I exist? And you know, 290 00:14:21,440 --> 00:14:24,000 Speaker 4: it's a relatively small market of people who are reading 291 00:14:24,000 --> 00:14:26,920 Speaker 4: what I'm writing about, But yeah, I guess there's something 292 00:14:26,960 --> 00:14:29,400 Speaker 4: about And not to toot my own hornets, this is 293 00:14:29,480 --> 00:14:32,440 Speaker 4: reader feedback that I'm getting, but I've been told that 294 00:14:32,560 --> 00:14:36,200 Speaker 4: there's something about whether it's the cadence or the timing 295 00:14:36,600 --> 00:14:38,760 Speaker 4: of things I do write about, or the way I 296 00:14:38,760 --> 00:14:41,800 Speaker 4: frame it that feels a little bit more human than 297 00:14:42,600 --> 00:14:45,160 Speaker 4: the research or the analysis that you know, the chief 298 00:14:45,160 --> 00:14:47,800 Speaker 4: investment strategists or whatever is sending out to their financial 299 00:14:47,800 --> 00:14:49,960 Speaker 4: advisor who's regurgitating it back to them. 300 00:14:50,320 --> 00:14:52,560 Speaker 2: James, one of the things you've been doing that I 301 00:14:52,600 --> 00:14:56,160 Speaker 2: don't think a bot can replicate just yet is sending 302 00:14:56,240 --> 00:14:59,080 Speaker 2: actual human beings or a human being to the strait 303 00:14:59,080 --> 00:15:03,080 Speaker 2: of hormone. Is that like the edge in media? Is 304 00:15:03,120 --> 00:15:06,640 Speaker 2: it like first person experiments and data gathering? 305 00:15:06,800 --> 00:15:09,200 Speaker 6: Yeah, I mean you just have to look at this 306 00:15:09,280 --> 00:15:12,080 Speaker 6: from I'm pretty confident that investment research won't exist in 307 00:15:12,120 --> 00:15:14,440 Speaker 6: the same way that it does now. The Internet kind 308 00:15:14,480 --> 00:15:18,160 Speaker 6: of made it democratized, so to speak. The distribution. It 309 00:15:18,240 --> 00:15:21,160 Speaker 6: was like the only the banks had distribution. Now we 310 00:15:21,240 --> 00:15:24,000 Speaker 6: have distribution and we can utilize that. But if we're 311 00:15:24,000 --> 00:15:26,160 Speaker 6: just utilizing it to kind of do the same thing 312 00:15:26,480 --> 00:15:29,680 Speaker 6: that the banks are doing, well, there is so much 313 00:15:29,680 --> 00:15:32,600 Speaker 6: investment research that these models have been trained on, and 314 00:15:32,720 --> 00:15:34,600 Speaker 6: as they get better, they will pretty much be able 315 00:15:34,640 --> 00:15:37,640 Speaker 6: to do the same thing. So it becomes very much 316 00:15:37,640 --> 00:15:40,800 Speaker 6: about doing something different and I think that there's like 317 00:15:40,840 --> 00:15:43,440 Speaker 6: two ways you can do that. One you can be 318 00:15:43,560 --> 00:15:47,480 Speaker 6: right all the time, so hopefully you know fingers crossed well, 319 00:15:47,760 --> 00:15:50,960 Speaker 6: you know, I think that that one aspect is like 320 00:15:50,960 --> 00:15:54,080 Speaker 6: like when I came on Ale's was the only media 321 00:15:54,120 --> 00:15:56,080 Speaker 6: that I did after that piece, which I was very 322 00:15:56,120 --> 00:15:58,640 Speaker 6: happy to thank you. I said, one of your questions 323 00:15:58,680 --> 00:16:01,240 Speaker 6: show was like, why do you think this so viral? 324 00:16:01,760 --> 00:16:04,720 Speaker 6: And I think it's like, this is not like tooting 325 00:16:04,720 --> 00:16:06,320 Speaker 6: my own horn. It's just sometimes you go on a 326 00:16:06,320 --> 00:16:08,560 Speaker 6: hot streak. And when we started that was very much 327 00:16:08,600 --> 00:16:11,280 Speaker 6: in like a year of just like one hundred percent 328 00:16:11,360 --> 00:16:13,480 Speaker 6: correct calls, and you know, obviously I've had a bunch 329 00:16:13,520 --> 00:16:15,480 Speaker 6: of wrong ones since then. But it was like a 330 00:16:15,520 --> 00:16:17,560 Speaker 6: bunch of people were reading something and they would read 331 00:16:17,600 --> 00:16:19,680 Speaker 6: it and they would say, well, if I hadn't read 332 00:16:19,720 --> 00:16:21,480 Speaker 6: this piece, I wouldn't have bought in video in twenty 333 00:16:21,480 --> 00:16:24,120 Speaker 6: twenty three, or wouldn't have bought Eskehinich something. So I 334 00:16:24,480 --> 00:16:27,960 Speaker 6: generated value from it. And then if you can add 335 00:16:27,960 --> 00:16:31,040 Speaker 6: into that something that the model can't do with a 336 00:16:31,120 --> 00:16:33,640 Speaker 6: query they can't be out in the real world, then 337 00:16:33,760 --> 00:16:35,840 Speaker 6: it's going on in the real world. If there's if 338 00:16:35,920 --> 00:16:37,680 Speaker 6: you know in five years, the robots come and then 339 00:16:37,720 --> 00:16:39,120 Speaker 6: the models can go out in the real world. We'll 340 00:16:39,120 --> 00:16:40,600 Speaker 6: have to find something else that they can't do. 341 00:16:40,840 --> 00:16:43,520 Speaker 3: Judgment, How do you think about this question of like 342 00:16:43,920 --> 00:16:47,120 Speaker 3: what the human writer can still bring to the table. 343 00:16:47,240 --> 00:16:49,280 Speaker 3: The one time you were on Autolaus before was nothing 344 00:16:49,280 --> 00:16:52,880 Speaker 3: about you were talking about Chinese peptize. 345 00:16:53,280 --> 00:16:56,400 Speaker 2: But you did go to a peptid rate now, yeah. 346 00:16:56,360 --> 00:16:58,320 Speaker 3: So like that is sort of your equivalent of sending 347 00:16:58,360 --> 00:16:59,920 Speaker 3: an analyst to the Strait of Homers. 348 00:17:00,800 --> 00:17:02,760 Speaker 7: We were talking about this, I would be the San 349 00:17:02,800 --> 00:17:06,120 Speaker 7: Francisco spy that Jay sets up. I guess I'm I've 350 00:17:06,160 --> 00:17:08,840 Speaker 7: doxed myself already, but I can get away. 351 00:17:09,000 --> 00:17:10,600 Speaker 3: But is that like, do you feel like that's a 352 00:17:10,640 --> 00:17:14,040 Speaker 3: big like yes, margat that like being in it in 353 00:17:14,160 --> 00:17:16,399 Speaker 3: some way physically in it is that it is like 354 00:17:16,520 --> 00:17:19,439 Speaker 3: what you can do at least currently that the AI can. 355 00:17:19,640 --> 00:17:22,560 Speaker 7: Basically, I mean, I'm very bullish on secrets and I'm 356 00:17:22,640 --> 00:17:24,560 Speaker 7: very bullish on gossip, right and and. 357 00:17:25,920 --> 00:17:28,000 Speaker 5: I mean that's kind of what journalism is in the end. 358 00:17:28,320 --> 00:17:30,400 Speaker 7: It's you have what is not in the trading data, 359 00:17:30,480 --> 00:17:32,280 Speaker 7: Like data is like one of the most valuable things. 360 00:17:32,280 --> 00:17:35,360 Speaker 7: It's like chips, data algorithms, right, and the data's out 361 00:17:35,359 --> 00:17:37,520 Speaker 7: there in the world, and they've scraped the entire Internet. 362 00:17:37,560 --> 00:17:39,200 Speaker 5: They've torn the covers off all these books. 363 00:17:39,359 --> 00:17:41,600 Speaker 7: Their markre is like paying hundreds of dollars for these 364 00:17:41,640 --> 00:17:43,760 Speaker 7: experts to like feed all their knowledge into the models. 365 00:17:43,880 --> 00:17:46,520 Speaker 5: But the thing is, the world's constantly changing. It's very dynamic. 366 00:17:46,680 --> 00:17:49,560 Speaker 7: They're parts of the world, parties u straits that are 367 00:17:49,640 --> 00:17:51,840 Speaker 7: unexplored by humans. And if you go and you are 368 00:17:51,840 --> 00:17:53,720 Speaker 7: the first person to see that, or you're the first 369 00:17:53,720 --> 00:17:55,720 Speaker 7: person to be there at a critical moment in time 370 00:17:55,920 --> 00:17:58,440 Speaker 7: as news is happening, as news is breaking, that kind 371 00:17:58,440 --> 00:18:01,720 Speaker 7: of actual reporting is like more valuable than ever. One 372 00:18:01,760 --> 00:18:03,919 Speaker 7: way that I think about reporting is that's you have 373 00:18:04,040 --> 00:18:06,480 Speaker 7: this private knowledge that's maybe known in whisper networks, in 374 00:18:06,520 --> 00:18:09,119 Speaker 7: the you know, in some party scene or whatever, or 375 00:18:09,160 --> 00:18:11,760 Speaker 7: you have tacit knowledge that nobody's really written down yet. 376 00:18:11,840 --> 00:18:14,280 Speaker 7: And as a reporter, my job is to take the taseit, 377 00:18:14,400 --> 00:18:16,280 Speaker 7: or to take the private and to turn it into 378 00:18:16,280 --> 00:18:19,520 Speaker 7: public knowledge. Like that particular task I think is extremely 379 00:18:19,600 --> 00:18:20,040 Speaker 7: robust in. 380 00:18:20,040 --> 00:18:20,640 Speaker 6: The era of AI. 381 00:18:20,960 --> 00:18:23,720 Speaker 2: Okay, so speaking of secrets and gossip and going out 382 00:18:23,760 --> 00:18:26,359 Speaker 2: into the real world, you recently came back from a 383 00:18:26,400 --> 00:18:29,000 Speaker 2: trip to China, right where you were sort of comparing 384 00:18:29,119 --> 00:18:32,520 Speaker 2: I guess how China is thinking and undertaking AI versus 385 00:18:32,560 --> 00:18:34,400 Speaker 2: the US. What's your big takeaway? 386 00:18:34,480 --> 00:18:37,000 Speaker 7: Oh yeah, I mean China is one of those places 387 00:18:37,080 --> 00:18:40,159 Speaker 7: where it's hard to get a sense of without going 388 00:18:40,200 --> 00:18:42,120 Speaker 7: in person, because it's so restructed. 389 00:18:42,160 --> 00:18:44,840 Speaker 5: What's on the public Internet, and we have these off internets. 390 00:18:45,680 --> 00:18:47,639 Speaker 7: One interesting thing is the way that I think about 391 00:18:48,200 --> 00:18:50,800 Speaker 7: USAI research is it sort of had three eras of 392 00:18:50,840 --> 00:18:53,520 Speaker 7: American AI. You had the academic era, maybe you'd call 393 00:18:53,560 --> 00:18:55,879 Speaker 7: the boom kicking off with image net or alphag. You 394 00:18:55,960 --> 00:18:58,560 Speaker 7: had the commercial era, let's call it kicked off with CHATGBT, 395 00:18:59,000 --> 00:19:02,040 Speaker 7: you know, the maybe geopolitical era that's sort of Pentagon 396 00:19:02,040 --> 00:19:05,120 Speaker 7: in Mythos drama. And China is weirdly still in the 397 00:19:05,160 --> 00:19:08,720 Speaker 7: academic era. More so, it's very collaborative, very pro open source. 398 00:19:09,200 --> 00:19:12,760 Speaker 7: The labs and the researchers themselves seem unconcerned pretty much 399 00:19:12,760 --> 00:19:13,320 Speaker 7: with the sort of. 400 00:19:13,240 --> 00:19:16,840 Speaker 5: Big philosophical and geopolitical questions, and that's there's sort of. 401 00:19:16,800 --> 00:19:19,960 Speaker 7: A division of labor where because the party and the 402 00:19:20,000 --> 00:19:23,440 Speaker 7: government is so active in shaping exactly what AI's role 403 00:19:23,440 --> 00:19:26,399 Speaker 7: in society ought to be. Then the companies themselves have 404 00:19:26,480 --> 00:19:29,320 Speaker 7: sort of abdicated that responsibility, and they're also much more 405 00:19:29,359 --> 00:19:32,440 Speaker 7: focused on collaborating because they'd see collaboration as the only 406 00:19:32,480 --> 00:19:34,399 Speaker 7: way to sort of maybe have a chance against the 407 00:19:34,520 --> 00:19:35,680 Speaker 7: US front here, just. 408 00:19:35,640 --> 00:19:37,720 Speaker 3: To follow up on this a little bit, because so 409 00:19:37,800 --> 00:19:41,959 Speaker 3: much of the American AI conversation is suffused with both 410 00:19:42,400 --> 00:19:45,720 Speaker 3: the job loss question and then the even more sinister 411 00:19:45,920 --> 00:19:48,879 Speaker 3: question of like will the models turn against us? Is 412 00:19:48,920 --> 00:19:51,239 Speaker 3: that to mention there as well, that's sort of the 413 00:19:51,320 --> 00:19:54,320 Speaker 3: safety alignment part is that a big part of the 414 00:19:54,440 --> 00:19:55,800 Speaker 3: Chinese AI culture. 415 00:19:55,600 --> 00:19:57,520 Speaker 7: Not a lot, And I mean we ask researchers that 416 00:19:57,560 --> 00:19:59,840 Speaker 7: a lot of the top Chinese labs how much they 417 00:20:00,000 --> 00:20:01,760 Speaker 7: thought about safety, and it was clear that it was 418 00:20:01,840 --> 00:20:04,679 Speaker 7: less of a propriety. One is that they're compute constrained, right, Like, 419 00:20:04,760 --> 00:20:07,919 Speaker 7: so one open AYE researcher is allocated more compute than 420 00:20:07,960 --> 00:20:10,320 Speaker 7: like an entire Chinese lab, oftentimes. 421 00:20:09,840 --> 00:20:11,120 Speaker 5: Because of the chip controls. 422 00:20:11,480 --> 00:20:14,320 Speaker 7: And so when you're that constrained, you're probably not going 423 00:20:14,400 --> 00:20:17,840 Speaker 7: to devote as many as much compute to safety and alignment. 424 00:20:17,880 --> 00:20:18,640 Speaker 5: So that's one thing. 425 00:20:19,000 --> 00:20:21,360 Speaker 7: The other thing I noticed is just that China has 426 00:20:21,400 --> 00:20:23,600 Speaker 7: a little bit of this more I almost like I 427 00:20:23,640 --> 00:20:26,679 Speaker 7: call it like it's not techno optimism, which sometimes people confuse. 428 00:20:26,720 --> 00:20:30,240 Speaker 7: It's more like technodeterminism or this pragmatic approach, which is 429 00:20:30,280 --> 00:20:34,600 Speaker 7: that technology has always progressed, automation like mechanization, like the 430 00:20:34,720 --> 00:20:37,040 Speaker 7: nutural revolution, like technology has always progressed. 431 00:20:37,040 --> 00:20:39,639 Speaker 5: It's overall made life better. There's no way to stop it. 432 00:20:39,680 --> 00:20:41,879 Speaker 7: There's not really a culture of protests and resistance in 433 00:20:41,960 --> 00:20:44,479 Speaker 7: China because of the political environment, and so as an 434 00:20:44,480 --> 00:20:47,320 Speaker 7: individual or as a company, there's no point in being like, oh, 435 00:20:47,320 --> 00:20:49,160 Speaker 7: we don't want this, We're going to regulate it, We're going. 436 00:20:49,160 --> 00:20:50,000 Speaker 5: To refuse it. 437 00:20:50,040 --> 00:20:52,280 Speaker 7: Like it's like you adopt or else you are going 438 00:20:52,320 --> 00:20:53,879 Speaker 7: to fall off, You're going to get left behind. 439 00:20:53,920 --> 00:20:57,240 Speaker 5: And so every individual is much more concerned with how can. 440 00:20:57,160 --> 00:20:59,840 Speaker 7: I adopt like open claw or whatever it is, as 441 00:21:00,080 --> 00:21:02,600 Speaker 7: as I can so that I get this job, because 442 00:21:02,640 --> 00:21:04,520 Speaker 7: if I don't do it, there's a million people behind 443 00:21:04,560 --> 00:21:05,840 Speaker 7: me who are going to take it instead. 444 00:21:06,359 --> 00:21:10,159 Speaker 2: James, I know you're approaching the US versus China AI question. 445 00:21:10,640 --> 00:21:13,200 Speaker 2: I guess both from an investment perspective and also from 446 00:21:13,280 --> 00:21:18,520 Speaker 2: a social economic perspective. Per your doom scenario, not Jesus 447 00:21:19,040 --> 00:21:22,119 Speaker 2: doom scenario, but like, where do you stand on this 448 00:21:22,160 --> 00:21:23,080 Speaker 2: particular question. 449 00:21:23,520 --> 00:21:26,240 Speaker 6: I think that it's very important that we continue to 450 00:21:26,280 --> 00:21:29,480 Speaker 6: sell chips to China. I think that if you were 451 00:21:29,520 --> 00:21:33,080 Speaker 6: to cut China completely off from chips or crack down 452 00:21:33,119 --> 00:21:36,480 Speaker 6: on some of the smuggling, it would It's there's like 453 00:21:36,560 --> 00:21:38,919 Speaker 6: a sun Zoo thing about you, like you build your 454 00:21:39,040 --> 00:21:41,560 Speaker 6: enemy at Golden Bridge upon which to retreat. I think 455 00:21:41,600 --> 00:21:45,639 Speaker 6: it's very similar here where we don't really want to 456 00:21:45,840 --> 00:21:50,400 Speaker 6: encourage even more capacity for China to take the reins 457 00:21:50,520 --> 00:21:53,960 Speaker 6: in Ai. And if we can control certain aspects of 458 00:21:54,000 --> 00:21:57,560 Speaker 6: the infrastructure stack, then the US absolutely should. I think 459 00:21:57,640 --> 00:21:59,480 Speaker 6: that the next thing that we'll see, and this is 460 00:21:59,480 --> 00:22:02,000 Speaker 6: like a really I think, kind of spicy take if 461 00:22:02,040 --> 00:22:04,440 Speaker 6: you were to look at the price action of the 462 00:22:04,760 --> 00:22:09,400 Speaker 6: memory complex, I think that we will see within the 463 00:22:09,400 --> 00:22:13,000 Speaker 6: next like Chinese deep seek moment will not be about 464 00:22:13,040 --> 00:22:17,280 Speaker 6: a model, It will be about hardware. That's so, is it? 465 00:22:17,359 --> 00:22:17,439 Speaker 5: Oh? 466 00:22:17,520 --> 00:22:17,680 Speaker 1: Really? 467 00:22:18,920 --> 00:22:24,919 Speaker 6: I thought Wellex that I think you know, everyone's kind 468 00:22:24,960 --> 00:22:28,480 Speaker 6: of piling into uh, the bottleneck trade so to speak, 469 00:22:28,560 --> 00:22:30,880 Speaker 6: and and uh, I think there's kind of a new 470 00:22:31,000 --> 00:22:34,920 Speaker 6: vintage of investors that are doing this bottleneck investing without 471 00:22:35,040 --> 00:22:39,240 Speaker 6: the awareness that bottlenecks are made to be widened. And 472 00:22:39,320 --> 00:22:41,680 Speaker 6: I think that what's going to happen if we continue 473 00:22:41,680 --> 00:22:45,679 Speaker 6: to see this kind of meteoric rise in de ram prices, uh, 474 00:22:45,920 --> 00:22:49,080 Speaker 6: is that just like every other technological bottleneck, there will 475 00:22:49,080 --> 00:22:51,200 Speaker 6: be a bunch of nerds in their basement that are 476 00:22:51,400 --> 00:22:54,640 Speaker 6: very very incentivized to fix this. And there's there are 477 00:22:54,680 --> 00:22:56,399 Speaker 6: many ways that you can do that. One is, you know, 478 00:22:56,760 --> 00:23:00,399 Speaker 6: you match flash bandwidth d RAM. It's one hundred times cheaper, 479 00:23:00,600 --> 00:23:04,159 Speaker 6: like like it's so And I think that with China 480 00:23:04,960 --> 00:23:09,360 Speaker 6: CXMT is there like a Micron skha, you know, And 481 00:23:10,320 --> 00:23:13,159 Speaker 6: with the IPO happening this year, there will be a 482 00:23:13,200 --> 00:23:14,199 Speaker 6: lot of capital that will I. 483 00:23:14,400 --> 00:23:16,879 Speaker 3: Wondering about this because it's like the models are so 484 00:23:17,240 --> 00:23:21,640 Speaker 3: like why even store any photos or images or whatever 485 00:23:22,119 --> 00:23:25,160 Speaker 3: in a big bank of memory data? The models could 486 00:23:25,200 --> 00:23:28,120 Speaker 3: just recreate it on command. And so I've been wondering 487 00:23:28,240 --> 00:23:30,479 Speaker 3: if like in the end, like storage is not going 488 00:23:30,520 --> 00:23:32,720 Speaker 3: to be that big of a deal because the model 489 00:23:32,880 --> 00:23:36,119 Speaker 3: just like, yeah, let's see that photo of like, you know, 490 00:23:36,240 --> 00:23:38,280 Speaker 3: me and my son's birthday, and the model will just 491 00:23:38,359 --> 00:23:40,960 Speaker 3: create it and why did I need to save it 492 00:23:41,000 --> 00:23:41,680 Speaker 3: to my iPhone? 493 00:23:41,680 --> 00:23:45,960 Speaker 6: I think that the reason why it's it's interesting because 494 00:23:45,960 --> 00:23:48,520 Speaker 6: if you look at like memories like a key component 495 00:23:48,560 --> 00:23:51,280 Speaker 6: of the GPU. Rightah, so but then why didn't you 496 00:23:51,320 --> 00:23:53,639 Speaker 6: know in video was rallying for a year and a 497 00:23:53,680 --> 00:23:57,000 Speaker 6: half before memory started rallying, right, So why did that happen? Well, 498 00:23:57,160 --> 00:23:59,520 Speaker 6: it's because of the event of agentic AI and having 499 00:23:59,560 --> 00:24:00,639 Speaker 6: to remember Sam. 500 00:24:00,680 --> 00:24:03,880 Speaker 3: I have a question for you, Like, for the theme 501 00:24:03,920 --> 00:24:07,640 Speaker 3: of your substack, generally speaking, stocks go up? Is the 502 00:24:07,720 --> 00:24:11,560 Speaker 3: sub theme? Just ignore the everyone else on the stage 503 00:24:11,840 --> 00:24:14,640 Speaker 3: because you're just gonna get distracted and you're gonna get 504 00:24:14,680 --> 00:24:18,760 Speaker 3: freaked out and whatever. And then allimately ignore all the 505 00:24:18,800 --> 00:24:22,640 Speaker 3: odd lacked episodes, ignore all the news, ignore the doomers, 506 00:24:22,640 --> 00:24:24,800 Speaker 3: because in the end, the only thing that can happen 507 00:24:24,800 --> 00:24:27,359 Speaker 3: if you paint to the news is you do something stupid. 508 00:24:27,400 --> 00:24:28,400 Speaker 3: Then you missed the long run. 509 00:24:28,520 --> 00:24:31,160 Speaker 4: No, I'm actually the complete opposite of that, okay. And 510 00:24:31,240 --> 00:24:33,520 Speaker 4: you know, like I said before, like you know, I'm 511 00:24:33,560 --> 00:24:36,119 Speaker 4: always worried, okay, And this is like start of the 512 00:24:36,119 --> 00:24:38,520 Speaker 4: message I try to, you know, communicate out to the world. 513 00:24:38,760 --> 00:24:40,560 Speaker 4: Like you know, I do a lot of things and 514 00:24:40,600 --> 00:24:43,560 Speaker 4: communicate a lot of things that are kind of counterintuitive, Like, 515 00:24:43,680 --> 00:24:45,640 Speaker 4: you know, I do check my prow in K plan 516 00:24:45,760 --> 00:24:49,840 Speaker 4: every single day, you know, I do. You know, Instead, 517 00:24:49,920 --> 00:24:52,560 Speaker 4: you know a lot of advisors and professionals and stuff 518 00:24:52,600 --> 00:24:54,680 Speaker 4: will go on TV and say, you know, forget your 519 00:24:54,880 --> 00:24:59,159 Speaker 4: you know, for and K password or ignore politics or 520 00:24:59,240 --> 00:25:01,480 Speaker 4: ignore what's go on, and I run. You know, we 521 00:25:01,560 --> 00:25:03,320 Speaker 4: have a long history of getting past all this stuff. 522 00:25:03,359 --> 00:25:06,000 Speaker 4: I think that's all incredibly silly. I think you really 523 00:25:06,040 --> 00:25:08,280 Speaker 4: do have to think about how bad things are at 524 00:25:08,280 --> 00:25:10,399 Speaker 4: a given time. That way, you sort of build up 525 00:25:10,400 --> 00:25:13,080 Speaker 4: those memories when you know, ten years from now, when 526 00:25:13,119 --> 00:25:15,520 Speaker 4: there is another war, you do remember how bad things 527 00:25:15,520 --> 00:25:18,400 Speaker 4: were in the past and how you know, markets evolved 528 00:25:18,400 --> 00:25:20,920 Speaker 4: out of all that stuff. So yeah, like I think 529 00:25:20,960 --> 00:25:24,399 Speaker 4: it's you know, to ignore things makes you sort of 530 00:25:24,440 --> 00:25:27,240 Speaker 4: more vulnerable to making mistakes. So it's like, yeah, be 531 00:25:27,400 --> 00:25:30,600 Speaker 4: really conscious of like where there are job losses, where 532 00:25:30,640 --> 00:25:33,600 Speaker 4: there will be industries that fall apart, because all the 533 00:25:33,680 --> 00:25:37,359 Speaker 4: lessons you learn today from all the bad things that happen, 534 00:25:37,760 --> 00:25:39,920 Speaker 4: and when you lose your job and when your neighbors 535 00:25:39,960 --> 00:25:41,840 Speaker 4: lose their job, and then you know, a couple of 536 00:25:41,880 --> 00:25:44,000 Speaker 4: years from now, you know, maybe the market's higher. You know, 537 00:25:44,080 --> 00:25:46,000 Speaker 4: ten years from that, it's going to happen all over again, 538 00:25:46,040 --> 00:25:49,040 Speaker 4: and you're more robust when that stuff happens. So yeah, 539 00:25:49,080 --> 00:25:51,640 Speaker 4: I think it's a complete mistake to ignore terrible things 540 00:25:51,680 --> 00:25:55,040 Speaker 4: going on as someone who's optimistic in the long run. 541 00:25:55,080 --> 00:25:55,240 Speaker 6: Good. 542 00:25:55,680 --> 00:25:57,960 Speaker 2: I have one last question for all three of you, 543 00:25:58,080 --> 00:25:59,919 Speaker 2: and it ties into something that Joe and I have 544 00:26:00,080 --> 00:26:02,400 Speaker 2: experienced on the podcast, which is a lot of our 545 00:26:02,440 --> 00:26:05,760 Speaker 2: episodes are starting to feel very surreal. You know, we're 546 00:26:05,800 --> 00:26:09,120 Speaker 2: talking about like space elevators and data centers in space, 547 00:26:09,240 --> 00:26:13,359 Speaker 2: and companies with like trillions of dollars of market capitalization 548 00:26:13,800 --> 00:26:17,480 Speaker 2: and lines that always seem to go up seemingly forever. 549 00:26:17,640 --> 00:26:20,240 Speaker 2: Everything feels very sci fi and surreal at the moment. 550 00:26:20,960 --> 00:26:25,479 Speaker 2: What's the most surreal thing or like data point that 551 00:26:25,520 --> 00:26:28,719 Speaker 2: you have seen or witnessed in person in recent months? 552 00:26:29,240 --> 00:26:31,520 Speaker 4: The first thing that comes to mind, like I said before, 553 00:26:32,080 --> 00:26:35,600 Speaker 4: like prompting, I think this is actually Gemini asked this 554 00:26:35,760 --> 00:26:38,280 Speaker 4: to like, you know, how would I write this? Oh yeah, 555 00:26:38,400 --> 00:26:41,480 Speaker 4: and came back and I remember thinking, gosh, not only 556 00:26:41,560 --> 00:26:44,119 Speaker 4: would I have sounded like that, but it's actually using 557 00:26:44,320 --> 00:26:46,360 Speaker 4: you know, slightly you know, words I would have run 558 00:26:46,359 --> 00:26:49,720 Speaker 4: through at the sarus and actively chosen to fold into 559 00:26:49,760 --> 00:26:52,359 Speaker 4: my writing. That was really, you know, kind of scary. 560 00:26:52,560 --> 00:26:56,000 Speaker 4: It really made me think about separating out, like exactly 561 00:26:56,040 --> 00:27:00,399 Speaker 4: what is it that I'm offering to subscribers. Yeah, but 562 00:27:00,880 --> 00:27:04,520 Speaker 4: being able to sort of see increasingly see yourself in 563 00:27:04,560 --> 00:27:07,200 Speaker 4: these machines that are getting better at, you know, replicating 564 00:27:07,280 --> 00:27:08,080 Speaker 4: human behavior. 565 00:27:08,840 --> 00:27:13,960 Speaker 6: I think that probably for me, there's like a quiet 566 00:27:14,080 --> 00:27:17,760 Speaker 6: kind of robotics boom that's occurring outside of like you 567 00:27:17,920 --> 00:27:21,159 Speaker 6: like people are kind of waiting until the humanoid robot 568 00:27:21,200 --> 00:27:24,960 Speaker 6: comes and folds their laundry and everything. But the dream Yeah, 569 00:27:25,320 --> 00:27:28,760 Speaker 6: for me for sure. But the thing is like Amazon 570 00:27:28,880 --> 00:27:32,199 Speaker 6: this year will employ more twice as many robots as 571 00:27:32,240 --> 00:27:35,040 Speaker 6: they do humans. You're starting to see in a lot 572 00:27:35,040 --> 00:27:38,480 Speaker 6: of the earnings reports, like from a Panic for example, 573 00:27:39,200 --> 00:27:43,360 Speaker 6: that like factory automation is really experience a huge inflection 574 00:27:43,440 --> 00:27:46,560 Speaker 6: that's not commensurate with other segments of that are selling 575 00:27:46,560 --> 00:27:51,040 Speaker 6: into factories. And I think if you just imagine, like 576 00:27:51,760 --> 00:27:54,320 Speaker 6: what level is robotics going to be at when it's 577 00:27:54,359 --> 00:27:57,960 Speaker 6: fully like like automated the factory, automated, the warehouses, by 578 00:27:58,000 --> 00:28:00,680 Speaker 6: the time it actually comes into your home that's kind 579 00:28:00,680 --> 00:28:02,760 Speaker 6: of that's like, that's going to make things so much more. 580 00:28:03,520 --> 00:28:06,159 Speaker 2: Just to be clear, Amazon is not employing the robot. 581 00:28:06,040 --> 00:28:10,760 Speaker 6: Right, you just know that, like when this happens, you 582 00:28:10,800 --> 00:28:12,960 Speaker 6: will be on a podcast talking about robot rights. 583 00:28:13,040 --> 00:28:13,280 Speaker 4: Yeah. 584 00:28:13,960 --> 00:28:14,919 Speaker 6: I can already hear it. 585 00:28:15,280 --> 00:28:15,840 Speaker 5: I mean already. 586 00:28:16,080 --> 00:28:17,919 Speaker 7: It's perfect that James said that because I was going 587 00:28:17,960 --> 00:28:20,840 Speaker 7: to talk about my trip to the Unitary office in 588 00:28:20,960 --> 00:28:23,680 Speaker 7: Hungjo and being both face to face with humanoids, which 589 00:28:23,720 --> 00:28:27,680 Speaker 7: is an extremely surreal, Uncanny Valley experience. They are good dancers, 590 00:28:27,720 --> 00:28:30,320 Speaker 7: they are good fighters, They're incredibly mobile. 591 00:28:30,840 --> 00:28:33,000 Speaker 5: The quadrupeds are they look like bug dogs. 592 00:28:33,000 --> 00:28:35,080 Speaker 7: They're like climbing all over and those are actively being 593 00:28:35,119 --> 00:28:38,160 Speaker 7: deployed in factories for inspection and surveillance. We saw a 594 00:28:38,160 --> 00:28:41,120 Speaker 7: twenty four to seven pharmacy that was in operation where 595 00:28:41,160 --> 00:28:43,920 Speaker 7: a humanoid robot would take stuff off and drop it 596 00:28:43,920 --> 00:28:46,560 Speaker 7: in a box and delivery drivers, Chinese delivery drivers were 597 00:28:46,560 --> 00:28:47,240 Speaker 7: coming in and out. 598 00:28:47,440 --> 00:28:48,760 Speaker 5: It was fully in operation. 599 00:28:48,960 --> 00:28:51,240 Speaker 7: And so China's pushing as fast as they can on 600 00:28:51,280 --> 00:28:54,400 Speaker 7: the factory automation and the physical AI stuff. And when 601 00:28:54,440 --> 00:28:56,480 Speaker 7: you see it, you just look at it and you're like, 602 00:28:56,600 --> 00:28:58,760 Speaker 7: that's clearly where the feature is headed by. 603 00:28:58,640 --> 00:29:01,200 Speaker 3: The way I always do the test in a new model, 604 00:29:01,840 --> 00:29:03,880 Speaker 3: LLLLM comes out, I say, like, write ten tweets in 605 00:29:03,880 --> 00:29:06,360 Speaker 3: my voice and there's still I don't think I can do. 606 00:29:06,480 --> 00:29:08,600 Speaker 3: But there was one from Claude and it said the 607 00:29:08,680 --> 00:29:11,440 Speaker 3: tenure Yield doesn't care about your feelings and frankly, it 608 00:29:11,480 --> 00:29:13,560 Speaker 3: doesn't care about mine either, and I was like, you 609 00:29:13,600 --> 00:29:15,560 Speaker 3: know what, It's not really a thing I would say, 610 00:29:15,840 --> 00:29:17,800 Speaker 3: but that's kind of a good tweet anyway. Thank you 611 00:29:17,920 --> 00:29:20,840 Speaker 3: all so much, James, Jasmine and Sam for coming on 612 00:29:20,840 --> 00:29:22,440 Speaker 3: a lad flag. That's a lot of fun. 613 00:29:37,000 --> 00:29:39,680 Speaker 2: So that was our conversation with James van Gelan of 614 00:29:39,720 --> 00:29:44,440 Speaker 2: Satrini Research, Jasmine's son, the Substack author, and Samro, the 615 00:29:44,560 --> 00:29:47,960 Speaker 2: editor of the newsletter t K. I'm Tracy Alloway. You 616 00:29:47,960 --> 00:29:50,000 Speaker 2: can follow me at Tracy Alloway. 617 00:29:49,680 --> 00:29:52,760 Speaker 3: And I'm Joe Wasnental. You can follow me at the Stalwart. 618 00:29:52,960 --> 00:29:56,920 Speaker 3: Follow our guest James van Gielan he's at Satrini, Samro 619 00:29:57,160 --> 00:30:00,880 Speaker 3: at Sam Row and Jasmine's son at Jazz New Sun. 620 00:30:01,360 --> 00:30:04,560 Speaker 3: Follow our producers Carmen Rodriguez at Carmen armand dash Ol 621 00:30:04,560 --> 00:30:08,360 Speaker 3: Bennett at Dashbot, Keil Brooks at Kelbrooks and Kevin Lozano 622 00:30:08,480 --> 00:30:11,400 Speaker 3: at Kevin Lloyd Lozano and from our Odd Lots content. 623 00:30:11,440 --> 00:30:13,840 Speaker 3: Go to Bloomberg dot com slash odd Lots, we're the 624 00:30:13,920 --> 00:30:16,560 Speaker 3: daily newsletter and all of our episodes, and you can 625 00:30:16,600 --> 00:30:18,640 Speaker 3: chat about all of these topics twenty four to seven 626 00:30:18,720 --> 00:30:22,240 Speaker 3: in our discord discord dot gg slash od logs. 627 00:30:22,280 --> 00:30:24,160 Speaker 2: And if you enjoy odd Lots, if you like it 628 00:30:24,240 --> 00:30:27,720 Speaker 2: when we occasionally indulge ourselves in talking about the media, 629 00:30:27,760 --> 00:30:30,240 Speaker 2: then please leave us a positive review on your favorite 630 00:30:30,240 --> 00:30:33,560 Speaker 2: podcast platform. And remember, if you are a Bloomberg subscriber, 631 00:30:33,640 --> 00:30:36,560 Speaker 2: you can listen to all of our episodes absolutely add free. 632 00:30:36,680 --> 00:30:38,680 Speaker 2: All you need to do is find the Bloomberg channel 633 00:30:38,720 --> 00:30:42,400 Speaker 2: on Apple Podcasts and follow the instructions there. Thanks for listening. 634 00:31:00,320 --> 00:31:08,120 Speaker 6: In e