1 00:00:02,400 --> 00:00:03,560 Speaker 1: As Media. 2 00:00:05,120 --> 00:00:07,680 Speaker 2: Hello, and welcome to Better Offline. I'm, of course your 3 00:00:07,720 --> 00:00:23,040 Speaker 2: host ed zitron no monologue this week. Quit complaining, because 4 00:00:23,040 --> 00:00:25,599 Speaker 2: I've brought you all something far better. I am joined 5 00:00:25,600 --> 00:00:27,960 Speaker 2: by Ronan Pharoh and Andrew Morantz of The New York 6 00:00:28,000 --> 00:00:30,400 Speaker 2: who just put a massive piece about Sam Altman, the 7 00:00:30,440 --> 00:00:33,160 Speaker 2: dampest man in Silicon Valley history, which is my comment, 8 00:00:33,240 --> 00:00:37,400 Speaker 2: not theirs, and I think I want to thank you 9 00:00:37,400 --> 00:00:40,080 Speaker 2: for joining me both of you. First of all, I 10 00:00:40,479 --> 00:00:43,159 Speaker 2: want to ask a really just blunk question, Ronan, considering 11 00:00:43,200 --> 00:00:46,479 Speaker 2: your history of the people you've reported on, why Sam Altman, 12 00:00:46,680 --> 00:00:48,560 Speaker 2: what drew you to him? 13 00:00:49,280 --> 00:00:53,360 Speaker 3: Well, this is fundamentally different than stories I've done that 14 00:00:53,600 --> 00:00:57,640 Speaker 3: started with, you know, a single clear cut lead of 15 00:00:57,680 --> 00:01:01,040 Speaker 3: about criminality. This was a situation where I was coming 16 00:01:01,080 --> 00:01:03,960 Speaker 3: off of reporting you may be familiar with about Elon 17 00:01:04,080 --> 00:01:07,800 Speaker 3: Musk and his whole empire, and I had a lot 18 00:01:07,880 --> 00:01:12,360 Speaker 3: of sources in the world of AI, including Sam Altman, 19 00:01:12,400 --> 00:01:15,280 Speaker 3: who was on the record in that piece. And you know, 20 00:01:15,680 --> 00:01:20,000 Speaker 3: this was no novel genius projection on my part. Obviously, 21 00:01:20,000 --> 00:01:22,959 Speaker 3: it was at a moment when AI was becoming the 22 00:01:23,040 --> 00:01:29,120 Speaker 3: kind of fulcrum of Silicon Valley and it became apparent 23 00:01:29,200 --> 00:01:34,560 Speaker 3: to me that there were under examined questions about the 24 00:01:34,600 --> 00:01:40,119 Speaker 3: trajectory of the technology and the integrity of the stewards 25 00:01:40,160 --> 00:01:43,960 Speaker 3: of it. We really started looking at that with an 26 00:01:43,959 --> 00:01:46,600 Speaker 3: open mind. You know, Andrew is on this beat of 27 00:01:46,640 --> 00:01:49,440 Speaker 3: big tech all the time and is also well sourced 28 00:01:49,520 --> 00:01:53,520 Speaker 3: up in this world. And we both, i think, sat 29 00:01:53,600 --> 00:01:57,680 Speaker 3: down and we're having our conversations about what the most 30 00:01:57,720 --> 00:02:02,680 Speaker 3: salient points of contention were in this world, and we're 31 00:02:02,760 --> 00:02:06,160 Speaker 3: equally open to criticism about all of the labs, all 32 00:02:06,200 --> 00:02:11,440 Speaker 3: of the figures. This question of why Sam Maltman was fired, 33 00:02:12,520 --> 00:02:16,920 Speaker 3: what the specific alleged proof points were, and whether those 34 00:02:17,000 --> 00:02:20,840 Speaker 3: questions were something we should care about. Were they right 35 00:02:20,880 --> 00:02:23,200 Speaker 3: to care then and should we still care now? Are 36 00:02:23,320 --> 00:02:29,079 Speaker 3: those problems ongoing? Those really emerged as just a lingering, 37 00:02:29,440 --> 00:02:33,760 Speaker 3: unanswered set of questions that hadn't been you know, fully 38 00:02:33,960 --> 00:02:36,680 Speaker 3: proved out and backed up in this kind of a 39 00:02:36,680 --> 00:02:37,440 Speaker 3: forensic way. 40 00:02:38,680 --> 00:02:40,560 Speaker 2: Right, So why was he fired? 41 00:02:40,560 --> 00:02:40,800 Speaker 4: Like what? 42 00:02:41,120 --> 00:02:43,560 Speaker 2: Because the piece kind of got into it. But it's 43 00:02:43,600 --> 00:02:45,640 Speaker 2: all over, Like the answers were all over the place 44 00:02:45,639 --> 00:02:48,000 Speaker 2: from everyone, and each person seems to have a different interpretation. 45 00:02:49,639 --> 00:02:51,799 Speaker 3: Andrew, you want to start yeah. 46 00:02:51,840 --> 00:02:56,720 Speaker 1: I think one of the abiding frustrations and curiosities from 47 00:02:56,720 --> 00:02:59,840 Speaker 1: this is the whole world saw Sam Altman get fired, 48 00:02:59,840 --> 00:03:02,560 Speaker 1: and then the whole world watched as he returned, and 49 00:03:02,600 --> 00:03:05,560 Speaker 1: that was never really adequately explained in the public narrative, 50 00:03:06,919 --> 00:03:10,760 Speaker 1: and so we really go into with sort of more documentation, 51 00:03:11,440 --> 00:03:15,000 Speaker 1: I think than ever before, exactly what was going on. 52 00:03:15,040 --> 00:03:17,320 Speaker 1: And as Ronan says, you know, it was not this 53 00:03:17,480 --> 00:03:20,200 Speaker 1: sort of one smoking gun thing like it wasn't you know, 54 00:03:20,639 --> 00:03:23,359 Speaker 1: Oh you know, Sam Altman was strangling baby puppies in 55 00:03:23,400 --> 00:03:25,480 Speaker 1: the office and nobody knew it until now, right. This 56 00:03:25,639 --> 00:03:31,960 Speaker 1: is a much subtler, slower accumulation of what his critics 57 00:03:32,000 --> 00:03:36,080 Speaker 1: say are manipulations telling two different stories to people. The 58 00:03:36,400 --> 00:03:38,600 Speaker 1: idea of telling different people what they want to hear 59 00:03:39,040 --> 00:03:43,200 Speaker 1: comes up a lot, and so this is the kind 60 00:03:43,200 --> 00:03:46,600 Speaker 1: of thing where you kind of need a New Yorker story, 61 00:03:47,040 --> 00:03:51,160 Speaker 1: I say sort of self interestedly, but like I do 62 00:03:51,200 --> 00:03:52,680 Speaker 1: think that you kind of need to put all the 63 00:03:52,720 --> 00:03:54,720 Speaker 1: facts in one place for them to make sense. And 64 00:03:54,760 --> 00:03:56,600 Speaker 1: there have been really good books about this. There have 65 00:03:56,600 --> 00:03:58,240 Speaker 1: been a lot out there. I mean, we're obviously not 66 00:03:58,280 --> 00:04:00,600 Speaker 1: first to this story, but I I just think it's 67 00:04:00,600 --> 00:04:04,040 Speaker 1: hard to understand what the allegations even are without lining 68 00:04:04,080 --> 00:04:05,560 Speaker 1: it all up in one place, and you kind of 69 00:04:05,560 --> 00:04:09,400 Speaker 1: have to centrally start from this premise of Sam Altman 70 00:04:09,480 --> 00:04:12,000 Speaker 1: did not go to everyone and say AI is going 71 00:04:12,080 --> 00:04:15,040 Speaker 1: to be a really big deal. It's going to be 72 00:04:15,080 --> 00:04:16,800 Speaker 1: a product we can try to sell. It's going to 73 00:04:16,839 --> 00:04:19,479 Speaker 1: be a good investment pitch for us. Let's all go 74 00:04:19,520 --> 00:04:20,559 Speaker 1: and make a bunch of money. 75 00:04:20,640 --> 00:04:20,800 Speaker 4: Right. 76 00:04:20,960 --> 00:04:25,320 Speaker 1: What he was saying was, you guys, you engineers, you 77 00:04:25,360 --> 00:04:27,600 Speaker 1: need to come along with me and build this technology 78 00:04:27,720 --> 00:04:31,240 Speaker 1: with me, specifically, because I will be circumspect, I will 79 00:04:31,240 --> 00:04:34,000 Speaker 1: be safe, I will argue for regulation, I will keep 80 00:04:34,040 --> 00:04:38,359 Speaker 1: it in this nonprofit AI safety research lab format. And 81 00:04:38,400 --> 00:04:41,119 Speaker 1: it was the betrayal of those promises that they felt 82 00:04:41,360 --> 00:04:44,599 Speaker 1: was the fireable offense more so than anyone smoking gun thing. 83 00:04:45,000 --> 00:04:48,080 Speaker 2: Except he was kind of brought back by capitalism. Like 84 00:04:48,120 --> 00:04:50,400 Speaker 2: it was very it was. I remember when it happened. 85 00:04:50,440 --> 00:04:53,520 Speaker 2: There was a journalist posting that they were crying with 86 00:04:53,560 --> 00:04:56,000 Speaker 2: happiness about it. It was genuine one of the darker 87 00:04:56,040 --> 00:04:58,640 Speaker 2: moments I can think in Silicon Valley history, because it's like, Wow, 88 00:04:59,120 --> 00:05:02,880 Speaker 2: Brian Chesky in Sacha Thedella got together and held hands, 89 00:05:03,320 --> 00:05:06,520 Speaker 2: got that Sam Wulman back up. It was just so bizarre. 90 00:05:08,320 --> 00:05:12,080 Speaker 1: It is weird, and it speaks to you know, this 91 00:05:12,160 --> 00:05:16,200 Speaker 1: thing again that people consistently allege in the documents we 92 00:05:16,240 --> 00:05:19,599 Speaker 1: saw and during our interviews of this sort of telling 93 00:05:19,640 --> 00:05:21,920 Speaker 1: different stories to different communities of people. I mean, one 94 00:05:21,920 --> 00:05:25,359 Speaker 1: of the most basic examples of this, right is turning 95 00:05:25,400 --> 00:05:28,080 Speaker 1: to the safety nonprofit people and saying this will always 96 00:05:28,120 --> 00:05:30,560 Speaker 1: be a safety nonprofit, and then turning to the VC 97 00:05:30,760 --> 00:05:33,039 Speaker 1: investment community and saying, let's go, guys, rock it to 98 00:05:33,040 --> 00:05:34,560 Speaker 1: the moon, let's go make a bunch of money, and 99 00:05:34,640 --> 00:05:37,800 Speaker 1: kind of simultaneously holding both of those communities seemingly in 100 00:05:37,839 --> 00:05:40,000 Speaker 1: the palm of his hand until he wasn't able to anymore. 101 00:05:40,800 --> 00:05:45,000 Speaker 3: You're exactly right to frame it as capitalism winning, and 102 00:05:45,520 --> 00:05:49,919 Speaker 3: I think we were fascinated in this case of what's 103 00:05:50,040 --> 00:05:52,880 Speaker 3: called the blip, you know, this brief firing and then 104 00:05:52,920 --> 00:05:58,479 Speaker 3: all undoing the firing, because it does feel like it 105 00:05:58,560 --> 00:06:02,240 Speaker 3: is about much more than Sam. The thing that OpenAI 106 00:06:02,480 --> 00:06:05,880 Speaker 3: originally was when it was founded as a nonprofit and 107 00:06:05,960 --> 00:06:08,960 Speaker 3: in the way Andrew described, was pitched on the basis 108 00:06:09,040 --> 00:06:14,200 Speaker 3: of these fears from the people developing the technology where 109 00:06:14,200 --> 00:06:17,080 Speaker 3: the founders of this company were saying, listen, this is 110 00:06:17,200 --> 00:06:21,080 Speaker 3: so dangerous that we're comparing it to nuclear weapons, and 111 00:06:21,120 --> 00:06:27,080 Speaker 3: we need to create a noble minded research lab without 112 00:06:27,080 --> 00:06:30,080 Speaker 3: a profit motive. They recruited on that basis, people tod 113 00:06:30,120 --> 00:06:32,080 Speaker 3: pay cuts to go there on that basis. And this 114 00:06:32,360 --> 00:06:36,919 Speaker 3: moment when the board that was empowered to remove a 115 00:06:36,960 --> 00:06:40,120 Speaker 3: CEO if they felt that CEO could not be trusted 116 00:06:40,120 --> 00:06:44,400 Speaker 3: with that mission, when they tried to do that, and 117 00:06:44,960 --> 00:06:49,000 Speaker 3: basically investors who really didn't know what the hell was 118 00:06:49,040 --> 00:06:50,680 Speaker 3: going on at the time, that's one of the things 119 00:06:50,680 --> 00:06:52,920 Speaker 3: we document. You know, the board really fumbled the ball. 120 00:06:53,440 --> 00:06:57,760 Speaker 3: We can talk about how, but the long and short 121 00:06:57,800 --> 00:07:02,680 Speaker 3: of it is that those investors rallied around Sam Altman, 122 00:07:03,400 --> 00:07:05,760 Speaker 3: who really stood to make them a lot of money, 123 00:07:05,880 --> 00:07:09,160 Speaker 3: and they were worried that their money would go away 124 00:07:09,480 --> 00:07:13,480 Speaker 3: if the company fell apart. And so it is a 125 00:07:13,480 --> 00:07:16,800 Speaker 3: part of a bigger story about the way in which 126 00:07:16,920 --> 00:07:22,080 Speaker 3: Silicon Valley builds companies based on hype and at least 127 00:07:22,160 --> 00:07:27,720 Speaker 3: not solely real value, and where the initial promises very 128 00:07:27,760 --> 00:07:31,680 Speaker 3: often don't pan out, and what winds up governing is 129 00:07:32,440 --> 00:07:34,960 Speaker 3: I think it's valid when people in this piece have 130 00:07:35,040 --> 00:07:37,040 Speaker 3: the criticism of, you know, we're now seeing a race 131 00:07:37,080 --> 00:07:39,760 Speaker 3: to the bottom on the very safety issues that open 132 00:07:39,800 --> 00:07:43,920 Speaker 3: AI was supposed to be founded around, and the blip 133 00:07:43,960 --> 00:07:45,640 Speaker 3: was a turning point that encapsulated that. 134 00:07:46,720 --> 00:07:49,200 Speaker 2: Yeah, with that in mind, do you buy that Sam 135 00:07:49,160 --> 00:07:51,640 Speaker 2: Wilton gives a shit about safety? Do you actually think 136 00:07:51,680 --> 00:07:54,280 Speaker 2: that the safety mission Matt stem Atol, You've spoken to 137 00:07:54,360 --> 00:07:55,360 Speaker 2: him a few times. 138 00:07:56,680 --> 00:08:01,080 Speaker 3: I think when he talks about still caring about and 139 00:08:01,160 --> 00:08:07,600 Speaker 3: prioritizing safety, Look, maybe I am just kind of too 140 00:08:07,720 --> 00:08:11,880 Speaker 3: generous in my assumptions about people. I assume that he 141 00:08:11,960 --> 00:08:15,480 Speaker 3: believes the story that he is telling himself, which is 142 00:08:16,200 --> 00:08:19,120 Speaker 3: very often has the same contour. Namely, what he cares 143 00:08:19,120 --> 00:08:23,160 Speaker 3: about is winning. That's the word he uses, and he 144 00:08:23,240 --> 00:08:26,480 Speaker 3: has set up the rationales in his mind and in 145 00:08:26,480 --> 00:08:30,120 Speaker 3: the way he conveys it to others so that him 146 00:08:30,160 --> 00:08:34,880 Speaker 3: winning and consolidating as much power as possible is the 147 00:08:34,920 --> 00:08:38,520 Speaker 3: same thing as humanity winning. So for him, when he 148 00:08:38,559 --> 00:08:41,240 Speaker 3: talks about these things, there is no daylight between them. 149 00:08:41,360 --> 00:08:44,640 Speaker 3: You know. He says in the piece that his definition 150 00:08:44,679 --> 00:08:47,240 Speaker 3: of winning is a crazy up level for everyone. You know, 151 00:08:47,280 --> 00:08:52,240 Speaker 3: we're all going to be hearing diseases and forming blockbuster startups, 152 00:08:52,240 --> 00:08:56,520 Speaker 3: and he really, I'm really, you know, paraphrasing but barely exaggerating. 153 00:08:56,840 --> 00:08:59,679 Speaker 3: He talks about it in these terms, and I think 154 00:08:59,800 --> 00:09:04,760 Speaker 3: he really does think that him controlling the technology is 155 00:09:04,960 --> 00:09:09,560 Speaker 3: the most direct path to that. He often says in 156 00:09:09,640 --> 00:09:15,760 Speaker 3: conversations that accusations are the truest form of confessions or 157 00:09:15,840 --> 00:09:20,200 Speaker 3: variations of this thought. You know that when people launch 158 00:09:20,200 --> 00:09:23,960 Speaker 3: a criticism about someone else, they're very often projecting something 159 00:09:24,000 --> 00:09:26,200 Speaker 3: about themselves. And it's interesting because in the course of 160 00:09:26,200 --> 00:09:29,760 Speaker 3: my Elon Musk reporting, he talks about Musk in this 161 00:09:29,880 --> 00:09:33,840 Speaker 3: very specific way where he says, on the record, you know, 162 00:09:34,240 --> 00:09:37,040 Speaker 3: I believe Elon is interested in saving humanity, but only 163 00:09:37,040 --> 00:09:40,880 Speaker 3: if he's the one who can do it. And I 164 00:09:41,000 --> 00:09:43,559 Speaker 3: thought about that quote about Elon a lot as I 165 00:09:43,600 --> 00:09:45,319 Speaker 3: was in my conversations with Sam. 166 00:09:45,760 --> 00:09:47,360 Speaker 1: I mean, the the other thing I would say on 167 00:09:47,400 --> 00:09:51,000 Speaker 1: this safety question, right, is a lot of these terms safety, 168 00:09:51,720 --> 00:09:58,200 Speaker 1: agi itself, super intelligence, intelligence itself. These all are are 169 00:09:58,320 --> 00:10:03,920 Speaker 1: vague enough terms, and they're they're poorly defined enough that 170 00:10:03,920 --> 00:10:06,359 Speaker 1: there can always be this kind of slippage and equivocation. 171 00:10:06,600 --> 00:10:10,360 Speaker 1: And sometimes that's unavoidable, but sometimes people can use it 172 00:10:10,400 --> 00:10:13,640 Speaker 1: to their rhetorical advantage, right, So on questions of safety 173 00:10:13,720 --> 00:10:16,840 Speaker 1: or alignment, what you would often find is that the 174 00:10:16,880 --> 00:10:20,079 Speaker 1: Sam Altman of twenty fifteen, or even frankly twenty twenty 175 00:10:20,080 --> 00:10:22,560 Speaker 1: three would be talking about the alignment problem as a 176 00:10:22,600 --> 00:10:25,640 Speaker 1: literally existential problem. If we don't solve this, a rogue 177 00:10:25,640 --> 00:10:28,200 Speaker 1: AI could kill literally every person on Earth. And then 178 00:10:28,240 --> 00:10:30,240 Speaker 1: you go back to him two years later and he's saying, 179 00:10:30,440 --> 00:10:33,199 Speaker 1: the alignment problem is this slightly annoying thing that kind 180 00:10:33,240 --> 00:10:35,679 Speaker 1: of tempts you to, you know, spend more time on 181 00:10:35,840 --> 00:10:39,720 Speaker 1: chat GPT, the way that algorithm, you know, Instagram's algorithm 182 00:10:39,760 --> 00:10:41,199 Speaker 1: is out of alignment with the way you want to 183 00:10:41,200 --> 00:10:43,560 Speaker 1: spend your time. That is absolutely not what he meant 184 00:10:43,559 --> 00:10:46,120 Speaker 1: by it two years prior, but the word is the same, 185 00:10:46,200 --> 00:10:48,040 Speaker 1: so he can kind of skate by with that. 186 00:10:48,679 --> 00:10:50,400 Speaker 2: I mean, this is in the nicest way. Why do 187 00:10:50,440 --> 00:10:53,440 Speaker 2: you believe him two years ago? But not? But like, 188 00:10:53,480 --> 00:10:55,400 Speaker 2: what do you think he's changed? Do you think he 189 00:10:55,520 --> 00:10:58,199 Speaker 2: just well, he's talking differently, Like that's I'm not accusing 190 00:10:58,200 --> 00:11:01,800 Speaker 2: you of it. I'm just like Sam will Land's whole 191 00:11:01,880 --> 00:11:04,760 Speaker 2: vibe is to me personally, just saying quite deceptive. 192 00:11:05,000 --> 00:11:08,120 Speaker 1: So I'm just I'm just wondering what you've seen. I'm 193 00:11:08,120 --> 00:11:09,920 Speaker 1: not saying that I believed him. I'm saying that his 194 00:11:09,960 --> 00:11:13,160 Speaker 1: employees believed him Congress to some extent belief in the 195 00:11:13,160 --> 00:11:13,720 Speaker 1: public life. 196 00:11:14,080 --> 00:11:16,760 Speaker 3: I think that's a very important point ed. I think 197 00:11:16,800 --> 00:11:20,000 Speaker 3: your analysis is correct, and actually the perspective from which 198 00:11:20,000 --> 00:11:25,120 Speaker 3: we write the piece is these have always been shifting assurances, 199 00:11:26,400 --> 00:11:31,160 Speaker 3: And actually part of what we emphasize and document is 200 00:11:31,200 --> 00:11:34,320 Speaker 3: that even at the time of those older assurances, where 201 00:11:34,360 --> 00:11:38,760 Speaker 3: Sam Altman was the biggest doomer in the industry, he 202 00:11:38,960 --> 00:11:42,040 Speaker 3: was saying and doing conflicting things at the same time. 203 00:11:42,120 --> 00:11:45,439 Speaker 3: In many cases, you know, we document through a lot 204 00:11:45,480 --> 00:11:50,360 Speaker 3: of internal communications a period of time in which many 205 00:11:50,520 --> 00:11:53,200 Speaker 3: researchers were joining the company on the promise that this 206 00:11:53,200 --> 00:11:57,760 Speaker 3: would forever be a noble nonprofit. But simultaneously, Sam Altman 207 00:11:57,760 --> 00:12:01,079 Speaker 3: and Greg Brockman and his co founder, we're having very 208 00:12:01,080 --> 00:12:05,080 Speaker 3: concerted conversations about how quickly and effectively they could get 209 00:12:05,120 --> 00:12:08,680 Speaker 3: out of the nonprofit structure. Essentially, we talk about cases 210 00:12:08,679 --> 00:12:12,760 Speaker 3: where he's saying on the hill, we need regulation, you know, 211 00:12:12,840 --> 00:12:15,800 Speaker 3: we need to start even a new government agency to 212 00:12:15,840 --> 00:12:19,760 Speaker 3: regulate AI. And then simultaneously he's working against regulations. 213 00:12:19,840 --> 00:12:20,360 Speaker 2: Well, just to. 214 00:12:20,320 --> 00:12:22,840 Speaker 1: Bolster the point you're making, Like, what really does seem 215 00:12:22,880 --> 00:12:26,640 Speaker 1: clear is that whether they should have believed it in retrospect, 216 00:12:26,960 --> 00:12:30,280 Speaker 1: a lot of these early employees, early recruits, early co founders, 217 00:12:30,480 --> 00:12:32,400 Speaker 1: really do seem to believe it. I mean, unless there 218 00:12:32,559 --> 00:12:35,800 Speaker 1: was this kind of you know, long game where they 219 00:12:35,840 --> 00:12:37,439 Speaker 1: were like, we're going to write these emails to each 220 00:12:37,480 --> 00:12:39,600 Speaker 1: other in twenty fifteen, and then ten years later they'll 221 00:12:39,640 --> 00:12:41,679 Speaker 1: come out and it'll all look like this was all 222 00:12:41,679 --> 00:12:44,240 Speaker 1: I mean, I guess that's possible, But from all appearances, 223 00:12:44,240 --> 00:12:46,160 Speaker 1: it really seems like at least a lot of the 224 00:12:46,200 --> 00:12:48,880 Speaker 1: key people really did believe this pitch at the time. 225 00:13:00,080 --> 00:13:03,080 Speaker 2: Although my way of reading it personally is also it 226 00:13:03,160 --> 00:13:06,360 Speaker 2: felt like a lot of rich guys kind of farting around. 227 00:13:06,520 --> 00:13:09,160 Speaker 2: It was just the conversations they have back and forth, 228 00:13:09,800 --> 00:13:12,520 Speaker 2: just the banality of it all, just like what if 229 00:13:12,559 --> 00:13:14,720 Speaker 2: this happened, and this happened I believe really a suits caper. 230 00:13:14,720 --> 00:13:17,720 Speaker 2: He seems to, however people may feel about him in general, 231 00:13:17,760 --> 00:13:20,520 Speaker 2: he seems to at least buy this. The Altman Elon 232 00:13:20,600 --> 00:13:24,520 Speaker 2: Musk emails were very much like to two guys at 233 00:13:24,520 --> 00:13:26,760 Speaker 2: an airport bar after a few waiting for their flight. 234 00:13:27,320 --> 00:13:31,760 Speaker 2: It was very very interesting watching this unfold over time, 235 00:13:31,760 --> 00:13:33,960 Speaker 2: because the modern Sam Mormon is very much more like 236 00:13:34,040 --> 00:13:37,520 Speaker 2: a tech CEO Capital T, capital C. Like, he's very 237 00:13:37,600 --> 00:13:41,480 Speaker 2: much a iron it like maximizing profit and growth and 238 00:13:41,559 --> 00:13:44,400 Speaker 2: the like you said. The safety stuff kind of seems 239 00:13:44,440 --> 00:13:46,959 Speaker 2: just like on the side. Now, it's more important that 240 00:13:47,000 --> 00:13:48,280 Speaker 2: we just get this thing public. 241 00:13:49,559 --> 00:13:52,040 Speaker 1: Yeah. Well, and I mean it's sort of like two 242 00:13:52,040 --> 00:13:54,840 Speaker 1: guys at the airport bar, except that the hundreds of 243 00:13:54,840 --> 00:13:58,000 Speaker 1: millions of dollars of investment they're talking about actually are real. 244 00:13:58,120 --> 00:14:00,920 Speaker 1: I mean, yes, you know, so it is. I think 245 00:14:00,960 --> 00:14:02,920 Speaker 1: you're right. There's a kind of combination of things going 246 00:14:02,920 --> 00:14:04,880 Speaker 1: on here. And I would be prepared to believe that 247 00:14:05,880 --> 00:14:08,800 Speaker 1: in that particular case of Musk and Altman emailing each other, 248 00:14:09,000 --> 00:14:11,800 Speaker 1: that there is some kind of gamesmanship or brinksmanship going on, 249 00:14:11,920 --> 00:14:13,920 Speaker 1: and that even what they say in private might not 250 00:14:14,000 --> 00:14:16,440 Speaker 1: be fully what they mean. But when you get outside 251 00:14:16,480 --> 00:14:18,480 Speaker 1: of that type of a circle, when you talk about 252 00:14:18,600 --> 00:14:20,920 Speaker 1: the early recruits to open AI, the later recruits, the 253 00:14:20,920 --> 00:14:23,800 Speaker 1: people who took massive pay cuts to you know, give 254 00:14:23,880 --> 00:14:26,760 Speaker 1: up six million dollars salaries at Google to come work 255 00:14:26,800 --> 00:14:29,320 Speaker 1: for them, clearly this pitch of where the good guys 256 00:14:29,480 --> 00:14:31,520 Speaker 1: was doing some work for those people, And I also 257 00:14:31,560 --> 00:14:33,520 Speaker 1: think that we should we should keep in mind how 258 00:14:33,600 --> 00:14:36,040 Speaker 1: effective the public pitch was. I mean, coming on the 259 00:14:36,080 --> 00:14:40,280 Speaker 1: heels of Mark Zuckerberg and you know all these other 260 00:14:40,440 --> 00:14:43,120 Speaker 1: you know, sort of social media tycoons going out to 261 00:14:43,160 --> 00:14:45,440 Speaker 1: the public and saying you should be so grateful for these, 262 00:14:45,640 --> 00:14:47,880 Speaker 1: you know, convenience machines that we're building, and you should, 263 00:14:47,960 --> 00:14:49,720 Speaker 1: you know, give us all your data, and you should 264 00:14:50,280 --> 00:14:53,560 Speaker 1: feel bad anytime you want to regulate us again. Altman's 265 00:14:53,560 --> 00:14:55,440 Speaker 1: pitch was totally the opposite. It was, We're going to 266 00:14:55,480 --> 00:14:58,240 Speaker 1: be conscientious. This is so dangerous that you actually have 267 00:14:58,320 --> 00:15:00,440 Speaker 1: to regulate us, and we are just begging for our 268 00:15:00,480 --> 00:15:03,800 Speaker 1: hands to be tied again. Whether in retrospect that seems 269 00:15:03,840 --> 00:15:06,480 Speaker 1: like a genuine pitch or not is sort of separate 270 00:15:06,480 --> 00:15:08,000 Speaker 1: from the question of how effective it was. 271 00:15:08,760 --> 00:15:11,800 Speaker 2: So let's talk finances. Within the piece, there wasn't a 272 00:15:11,800 --> 00:15:14,600 Speaker 2: ton of stuff about money, but there was this quote 273 00:15:14,600 --> 00:15:17,320 Speaker 2: from a senior executive at Microsoft who said, I think 274 00:15:17,320 --> 00:15:20,000 Speaker 2: there's a small but real chance Altman is eventually remembered 275 00:15:20,000 --> 00:15:22,680 Speaker 2: as a body made off of Sam Bankman Freed level scammer. 276 00:15:23,280 --> 00:15:26,600 Speaker 2: Did you what kind of financial stuff did you hear 277 00:15:26,640 --> 00:15:27,400 Speaker 2: about OpenAI? 278 00:15:27,600 --> 00:15:32,600 Speaker 3: If anything, I mean, this is all very far out 279 00:15:32,600 --> 00:15:35,400 Speaker 3: there already. We have a board member in the piece 280 00:15:35,520 --> 00:15:39,040 Speaker 3: talking about how the company is levered up in a 281 00:15:39,080 --> 00:15:41,920 Speaker 3: scary way right now. I mean, this is a company 282 00:15:41,960 --> 00:15:45,240 Speaker 3: with one of the fastest cash burn rates in the 283 00:15:45,280 --> 00:15:52,360 Speaker 3: history of startups. The spend level is almost unimaginable in 284 00:15:52,400 --> 00:15:56,240 Speaker 3: the risk towards AGI, and so it really does require 285 00:15:57,800 --> 00:16:00,440 Speaker 3: it's almost not an exaggeration to say the money in 286 00:16:00,480 --> 00:16:03,320 Speaker 3: the world. And this governs a lot of the different 287 00:16:03,360 --> 00:16:06,520 Speaker 3: activities that we write about in the Peace Right. Sam 288 00:16:06,560 --> 00:16:11,840 Speaker 3: Altman for years courting Middle Eastern money. Everyone in Silicon 289 00:16:11,920 --> 00:16:15,560 Speaker 3: Valley is courting Middle Eastern money and sidelining concerns about 290 00:16:15,560 --> 00:16:18,400 Speaker 3: working with autocrats to various extents, Sam Altman is doing 291 00:16:18,480 --> 00:16:22,920 Speaker 3: it on a unique level, you know, agreeing to and 292 00:16:23,000 --> 00:16:27,760 Speaker 3: catalyzing the building of massive infrastructure, even including in the 293 00:16:27,760 --> 00:16:31,480 Speaker 3: Middle East. You know, and over the objections we talk about, 294 00:16:31,520 --> 00:16:36,440 Speaker 3: like concerns that short circuited a security clearance vetting process 295 00:16:36,440 --> 00:16:38,960 Speaker 3: because there were so many red flags in the national 296 00:16:38,960 --> 00:16:41,480 Speaker 3: security establishment about these dealings with the Middle East. We 297 00:16:41,480 --> 00:16:45,440 Speaker 3: talk about you know, gifts and trips on yachts with 298 00:16:45,640 --> 00:16:52,400 Speaker 3: UAE officials. So you know, this is a drive for 299 00:16:52,560 --> 00:16:56,960 Speaker 3: more and more and more. And we talked to you know, 300 00:16:57,080 --> 00:17:02,320 Speaker 3: startup experts and economists who say one of the consequences 301 00:17:02,360 --> 00:17:05,399 Speaker 3: of that is a whole lot of circular deals. You know, 302 00:17:05,440 --> 00:17:07,480 Speaker 3: none of this is new either, but this is a 303 00:17:07,520 --> 00:17:12,199 Speaker 3: situation where it is the general Silicon Valley archetype of 304 00:17:12,600 --> 00:17:18,080 Speaker 3: building companies on hype and promises before actual value, inflating 305 00:17:18,119 --> 00:17:22,280 Speaker 3: a massive balloon of valuation, and it's it's putting it 306 00:17:22,359 --> 00:17:26,120 Speaker 3: on a far grander scale where there's a whole ecosystem 307 00:17:26,119 --> 00:17:29,200 Speaker 3: of partners that are just borrowing and borrowing and borrowing 308 00:17:29,280 --> 00:17:31,520 Speaker 3: from each other on the assurance that they're all going 309 00:17:31,560 --> 00:17:34,440 Speaker 3: to buy each other's products. It's an I'll scratch your back, 310 00:17:34,480 --> 00:17:38,360 Speaker 3: you scratch mind situation over and over again. In some cases, 311 00:17:39,359 --> 00:17:42,560 Speaker 3: you know, Altmann is running into problems because there are 312 00:17:42,600 --> 00:17:46,399 Speaker 3: conflicting or at least allegedly conflicting deals being announced, Like 313 00:17:46,480 --> 00:17:49,359 Speaker 3: on the same day we talk about tensions in the 314 00:17:49,400 --> 00:17:58,280 Speaker 3: Microsoft relationship, particularly around this recent Amazon deal. There is yes, 315 00:17:58,400 --> 00:18:01,840 Speaker 3: it's it's to do with you know, the exclusivity having 316 00:18:01,880 --> 00:18:06,960 Speaker 3: been reaffirmed for Microsoft in terms of the underlying stateless models, 317 00:18:07,160 --> 00:18:09,960 Speaker 3: and then open Ai announcing well now at the same 318 00:18:10,040 --> 00:18:12,360 Speaker 3: time in the same day, well, we have this additional 319 00:18:12,359 --> 00:18:15,000 Speaker 3: deal on top of that with Amazon, you know, for 320 00:18:15,119 --> 00:18:18,720 Speaker 3: our enterprise products that allow the building of agents, which 321 00:18:18,760 --> 00:18:21,919 Speaker 3: they're saying is going to happen in a stateful environment, 322 00:18:22,200 --> 00:18:25,159 Speaker 3: and that's going to be because they're creating new technology 323 00:18:25,200 --> 00:18:29,320 Speaker 3: that will deconflict this with the underlying stateless models from Microsoft. 324 00:18:29,440 --> 00:18:30,840 Speaker 3: The long and the short of it is a bunch 325 00:18:30,880 --> 00:18:33,720 Speaker 3: of people at Microsoft field this is not technically possible, 326 00:18:34,000 --> 00:18:36,479 Speaker 3: and you have to rely on the underlying stateless models 327 00:18:36,520 --> 00:18:41,040 Speaker 3: to create agents that are stateful, essentially with memory. That's 328 00:18:41,080 --> 00:18:42,960 Speaker 3: a lot of technical jargon to get at the point, 329 00:18:43,000 --> 00:18:46,520 Speaker 3: which is there's a slew of deals where critics say 330 00:18:46,600 --> 00:18:50,080 Speaker 3: they're circular and someone's going to have to pay up. 331 00:18:50,200 --> 00:18:51,679 Speaker 3: Maybe a lot of people are going to have to 332 00:18:51,680 --> 00:18:54,240 Speaker 3: pay up. Sam Altman's one of the people who has 333 00:18:54,240 --> 00:18:58,000 Speaker 3: said we're looking at a bubble and that a lot 334 00:18:58,040 --> 00:18:59,760 Speaker 3: of people are going to lose a lot of money. 335 00:19:00,080 --> 00:19:02,520 Speaker 2: Which is it's funny though whenever he talks about it, 336 00:19:02,520 --> 00:19:05,720 Speaker 2: it's never him, never let it's never like I'm gonna lose. 337 00:19:06,040 --> 00:19:07,680 Speaker 2: I will be. But it kind of gets back to 338 00:19:07,720 --> 00:19:09,399 Speaker 2: what you were saying. It's like the only one who 339 00:19:09,440 --> 00:19:11,560 Speaker 2: can usher in this AI future is Sam Mortman. The 340 00:19:11,600 --> 00:19:13,560 Speaker 2: only person who will not get destroyed in the bubble 341 00:19:13,600 --> 00:19:16,399 Speaker 2: is Sam Mortman. Makes me wonder if he's like a 342 00:19:16,440 --> 00:19:20,280 Speaker 2: fantasist or an arcist or a sociopath. It's like, I 343 00:19:20,320 --> 00:19:22,199 Speaker 2: think at least one of those words was in the 344 00:19:22,480 --> 00:19:23,080 Speaker 2: in the piece. 345 00:19:24,359 --> 00:19:25,920 Speaker 1: Those words are definitely in the piece. 346 00:19:26,040 --> 00:19:30,560 Speaker 3: And yeah, not from us, to be clear, Yeah, yeah, yeah, yes, yes, 347 00:19:30,080 --> 00:19:34,000 Speaker 3: these closely with the man. But it's remarkable. You know, 348 00:19:34,119 --> 00:19:36,159 Speaker 3: you're rarely on a story I don't know about you, Andrew, 349 00:19:36,160 --> 00:19:41,240 Speaker 3: where so many people use the word sociopath unprompted by us. 350 00:19:41,720 --> 00:19:43,600 Speaker 1: And yeah, and look, I mean again, I think it's 351 00:19:43,760 --> 00:19:45,840 Speaker 1: very I think readers should come to this with a 352 00:19:45,840 --> 00:19:48,520 Speaker 1: lot of skepticism about how much of this is kind 353 00:19:48,520 --> 00:19:52,479 Speaker 1: of industry competition. And we we tried to not you know, 354 00:19:52,680 --> 00:19:55,440 Speaker 1: let people launder talking points through our story, right where 355 00:19:55,560 --> 00:19:58,760 Speaker 1: very everyone in this world is kind of a ruthless competitor, right, 356 00:19:58,840 --> 00:20:01,240 Speaker 1: so it wouldn't be of very high bar to say, 357 00:20:01,400 --> 00:20:03,520 Speaker 1: you know, I talked to this guy's corporate competitor, and 358 00:20:03,600 --> 00:20:06,120 Speaker 1: they called them a sociopath. Like, it's not that kind 359 00:20:06,160 --> 00:20:08,800 Speaker 1: of criticism that we're talking about. We're talking about these 360 00:20:08,880 --> 00:20:13,240 Speaker 1: very long standing, kind of documented patterns that people can dispute. 361 00:20:13,280 --> 00:20:15,760 Speaker 1: But it's not just kind of these off the cuff observations. 362 00:20:16,520 --> 00:20:19,040 Speaker 1: I think to your point about the bubble stuff, like, 363 00:20:19,200 --> 00:20:20,920 Speaker 1: your audience will be very familiar with this, because I 364 00:20:21,000 --> 00:20:23,880 Speaker 1: know you cover this stuff a lot, But this question 365 00:20:23,920 --> 00:20:25,879 Speaker 1: of is it a bubble or is it actually a 366 00:20:25,960 --> 00:20:30,840 Speaker 1: very useful or even potentially economically transformative technology, It's not 367 00:20:30,960 --> 00:20:34,240 Speaker 1: really as clean a buying area as that because a 368 00:20:34,240 --> 00:20:38,000 Speaker 1: lot of the biggest bubbles in market history have come from, 369 00:20:38,160 --> 00:20:41,040 Speaker 1: you know, infrastructure projects that ultimately got used, right, I 370 00:20:41,040 --> 00:20:43,440 Speaker 1: mean this is a kind of basic economic history point, 371 00:20:43,440 --> 00:20:48,000 Speaker 1: but I think it's worth reiterating. The railroads, the canals, 372 00:20:48,080 --> 00:20:50,800 Speaker 1: the you know, the even fiber optic cable. All of 373 00:20:50,840 --> 00:20:53,840 Speaker 1: this stuff eventually gets used. The question is does the 374 00:20:55,040 --> 00:20:58,240 Speaker 1: speed at which the investors are building it match the 375 00:20:58,320 --> 00:20:59,440 Speaker 1: usefulness in real time? 376 00:21:00,119 --> 00:21:00,280 Speaker 3: Yeah? 377 00:21:00,280 --> 00:21:02,199 Speaker 2: And I mean I would debate that point and have 378 00:21:02,280 --> 00:21:04,639 Speaker 2: done so many times, because GPUs depreciate in three to 379 00:21:04,680 --> 00:21:06,960 Speaker 2: six years and there's no way that this gets used afterwards. 380 00:21:06,960 --> 00:21:09,960 Speaker 2: But putting that all aside, it is still this weird 381 00:21:09,960 --> 00:21:12,359 Speaker 2: thing of it's something in my own work I've noticed 382 00:21:12,880 --> 00:21:15,680 Speaker 2: it's weird we're discussing this still, like it's weird that 383 00:21:15,720 --> 00:21:18,399 Speaker 2: we have no clarity, because even within your piece, we 384 00:21:18,440 --> 00:21:22,200 Speaker 2: don't really have clarity about what is going on like that, 385 00:21:22,400 --> 00:21:24,200 Speaker 2: very the one thing that stood out to me, like 386 00:21:24,240 --> 00:21:26,320 Speaker 2: they're still trying to work out a business model. That 387 00:21:26,400 --> 00:21:31,000 Speaker 2: whole thing about pitting world powers against each other was 388 00:21:31,040 --> 00:21:34,040 Speaker 2: truly insane, like that Greg Brockman, You'll have to run 389 00:21:34,080 --> 00:21:36,280 Speaker 2: through it again. But it was like, as I understood 390 00:21:36,280 --> 00:21:39,080 Speaker 2: it would be Greg Brockman suggesting that they make something 391 00:21:39,080 --> 00:21:41,639 Speaker 2: powerful enough to sell the different nations and create a 392 00:21:41,640 --> 00:21:45,200 Speaker 2: bidding war like mission impossible except with the chat bart. 393 00:21:45,480 --> 00:21:48,359 Speaker 1: Yeah, yeah, this is this is again you know, like 394 00:21:48,440 --> 00:21:50,240 Speaker 1: you say, there's a lot of kind of guy at 395 00:21:50,240 --> 00:21:52,359 Speaker 1: the bar speculation going on. And so this is in 396 00:21:52,400 --> 00:21:56,040 Speaker 1: the early days when because Opening Eye back then was 397 00:21:56,080 --> 00:22:00,320 Speaker 1: a nonprofit who's legally binding fiduciary duty was the safety 398 00:22:00,320 --> 00:22:02,760 Speaker 1: of all of humanity, they hired someone to come in 399 00:22:02,840 --> 00:22:06,040 Speaker 1: and say, Okay, how can we deploy this? You know, 400 00:22:06,119 --> 00:22:08,960 Speaker 1: potentially dangerous thing safely, and this person, the way it 401 00:22:09,000 --> 00:22:11,359 Speaker 1: was told to us came in with all kinds of 402 00:22:11,400 --> 00:22:14,560 Speaker 1: you know, fifty page you know, detailed white papers about 403 00:22:14,560 --> 00:22:16,439 Speaker 1: how we can model it on the Barok Plan and 404 00:22:16,480 --> 00:22:20,000 Speaker 1: the way that you know, the nuclear weaponry was first 405 00:22:20,040 --> 00:22:23,080 Speaker 1: brought safely into the world, and the way that that 406 00:22:23,359 --> 00:22:26,040 Speaker 1: was told to us again was that that plan then 407 00:22:26,080 --> 00:22:29,359 Speaker 1: somehow morphed into Okay, instead of doing this you know 408 00:22:29,520 --> 00:22:34,520 Speaker 1: kind of safe international game theory, non zero cooperation thing, 409 00:22:34,960 --> 00:22:38,480 Speaker 1: why don't we just you know, start an auction between China, Russia, 410 00:22:38,600 --> 00:22:41,120 Speaker 1: the US, maybe a few other world governments and see 411 00:22:41,119 --> 00:22:44,360 Speaker 1: who either wants to be given the technology or who 412 00:22:44,359 --> 00:22:47,800 Speaker 1: can be sold the technology. Now, obviously this is disputed 413 00:22:47,840 --> 00:22:50,800 Speaker 1: and also even more obviously this is not something that happened, 414 00:22:51,119 --> 00:22:53,399 Speaker 1: but the fact that that kind of idea was being 415 00:22:53,480 --> 00:22:56,720 Speaker 1: batted around and apparently was taken outside the walls of 416 00:22:56,760 --> 00:23:01,840 Speaker 1: the organization and was pitched to potential investors, it really 417 00:23:01,880 --> 00:23:04,680 Speaker 1: does speak to how welid all, Yeah, it's it's really 418 00:23:04,680 --> 00:23:07,360 Speaker 1: a crazy situation. And the way, you know, we got 419 00:23:07,400 --> 00:23:09,879 Speaker 1: people denying this, but they didn't say like this is 420 00:23:09,880 --> 00:23:12,280 Speaker 1: made up. They just remembered the details of it differently. 421 00:23:13,119 --> 00:23:18,240 Speaker 2: So you you spoke to Sam Ortman a few times, right, yep, 422 00:23:18,400 --> 00:23:23,120 Speaker 2: like more than to what seemed to actually get under 423 00:23:23,119 --> 00:23:25,600 Speaker 2: his skin because it's over the last I've watched a 424 00:23:25,600 --> 00:23:29,080 Speaker 2: lot of Samuel I've I've watched the Orkman channel for hours, 425 00:23:29,160 --> 00:23:31,840 Speaker 2: and he seems fairly calm most of the time, but 426 00:23:31,840 --> 00:23:35,359 Speaker 2: occasionally something will really irritate him, like very suddenly, like 427 00:23:35,400 --> 00:23:38,080 Speaker 2: Brad Gertz. Brad getson. I think it is the venture 428 00:23:38,080 --> 00:23:40,840 Speaker 2: capitalist who was talking about valuations and he said, well, 429 00:23:40,880 --> 00:23:43,000 Speaker 2: get someone to buy all your shares, or when he 430 00:23:43,040 --> 00:23:45,000 Speaker 2: was on the hard Fork podcast on one of the 431 00:23:45,000 --> 00:23:47,920 Speaker 2: Glazing sessions, he got very angry about The New York 432 00:23:47,960 --> 00:23:51,840 Speaker 2: Times suing him and suing Open Ai. Even what was 433 00:23:52,160 --> 00:23:54,320 Speaker 2: his mood like, like, what was what were the questions 434 00:23:54,359 --> 00:23:56,160 Speaker 2: he was up for? What were the things he got 435 00:23:56,160 --> 00:23:58,879 Speaker 2: a bit worrisome about? 436 00:23:59,359 --> 00:24:04,640 Speaker 3: Well? He The truth is, I think both of us 437 00:24:04,680 --> 00:24:09,200 Speaker 3: picked up on from both Sam and institutionally from Open 438 00:24:09,240 --> 00:24:15,479 Speaker 3: Ai some degree of apathy about the integrity and lying 439 00:24:15,520 --> 00:24:21,959 Speaker 3: allegations that got him fired. I think that there is 440 00:24:22,520 --> 00:24:25,879 Speaker 3: real sensitivity and concern about the reporting in the piece 441 00:24:26,840 --> 00:24:31,720 Speaker 3: around the law firm investigation that was used to kind 442 00:24:31,760 --> 00:24:35,959 Speaker 3: of validate his staying on. And my background is as 443 00:24:36,000 --> 00:24:40,520 Speaker 3: an attorney in part, and I went very deep on 444 00:24:41,320 --> 00:24:43,840 Speaker 3: talking to people at the law firm and Andrew and 445 00:24:43,880 --> 00:24:49,240 Speaker 3: I really looked at this question. And look, the defense 446 00:24:49,320 --> 00:24:52,760 Speaker 3: of that outside investigation from some of the lawyers involved 447 00:24:52,800 --> 00:24:56,800 Speaker 3: is on full display in the story. But I think 448 00:24:56,880 --> 00:25:01,720 Speaker 3: the widespread criticism of it is and has merit. Which 449 00:25:01,760 --> 00:25:05,119 Speaker 3: is we report for the first time that nothing was 450 00:25:05,160 --> 00:25:08,160 Speaker 3: put in writing except for the eight hundred word press 451 00:25:08,200 --> 00:25:10,879 Speaker 3: release that open AI put out, which acknowledged only a 452 00:25:10,920 --> 00:25:15,600 Speaker 3: breakdown in trust. It is true that private companies sometimes 453 00:25:15,880 --> 00:25:20,919 Speaker 3: keep outside investigations when they want the outcome to be 454 00:25:20,960 --> 00:25:24,760 Speaker 3: exonerative out of writing, you know, to limit it liability 455 00:25:24,880 --> 00:25:27,760 Speaker 3: and because of privilege issues. But when you look at 456 00:25:27,760 --> 00:25:31,560 Speaker 3: the annals of high profile scandals like this, where the 457 00:25:31,560 --> 00:25:36,040 Speaker 3: goal is to restore trust and confidence, I think there 458 00:25:36,119 --> 00:25:38,360 Speaker 3: is a much better case that this should have been 459 00:25:38,359 --> 00:25:42,600 Speaker 3: memorialized in a more intensive way, that it should have 460 00:25:42,600 --> 00:25:45,560 Speaker 3: been released at least within the company, if not publicly. 461 00:25:45,640 --> 00:25:47,640 Speaker 3: And and look, you don't have to turn to us. 462 00:25:47,840 --> 00:25:50,639 Speaker 3: There was a whole set of executives and stakeholders in 463 00:25:50,680 --> 00:25:54,120 Speaker 3: this company who were shocked that there was no actual 464 00:25:54,200 --> 00:25:58,680 Speaker 3: report released, so you know, there was a real effort 465 00:25:58,960 --> 00:26:02,040 Speaker 3: to obfuscate some of this stuff in the view of 466 00:26:02,119 --> 00:26:04,520 Speaker 3: many people around it, and we report on a number 467 00:26:04,560 --> 00:26:07,359 Speaker 3: of alleged irregularities there. That's a good example of a 468 00:26:07,440 --> 00:26:11,680 Speaker 3: nuts and bolts matter where there is concern, and honestly, 469 00:26:11,840 --> 00:26:16,040 Speaker 3: my legal analysis is there should be because if this 470 00:26:16,119 --> 00:26:20,320 Speaker 3: company goes through its IPO, there are actually legal standards 471 00:26:20,320 --> 00:26:23,920 Speaker 3: by which shareholders could demand more information about that. And 472 00:26:24,000 --> 00:26:26,320 Speaker 3: we talk to people around this who are saying, well, 473 00:26:26,359 --> 00:26:28,480 Speaker 3: there might need to be a new investigation at some point, 474 00:26:28,760 --> 00:26:32,439 Speaker 3: so there are some practicalities like that. Obviously, he was, 475 00:26:32,640 --> 00:26:37,199 Speaker 3: as you might imagine, also very concerned about all of 476 00:26:37,240 --> 00:26:42,000 Speaker 3: the elon rivalry stuff in the piece and all of 477 00:26:42,040 --> 00:26:44,800 Speaker 3: the scuttle butt that is so far out there in 478 00:26:44,880 --> 00:26:48,840 Speaker 3: Silicon Valley about his personal life, you know, which we 479 00:26:48,880 --> 00:26:54,560 Speaker 3: looked at very incisively and fairly. But I think your 480 00:26:54,600 --> 00:26:59,240 Speaker 3: point is well taken, which is the core allegations about 481 00:26:59,280 --> 00:27:04,520 Speaker 3: integrity and honesty don't seem to much get under his skin. 482 00:27:04,840 --> 00:27:06,920 Speaker 3: And when I asked him, even on a personal level, 483 00:27:07,520 --> 00:27:09,840 Speaker 3: from the perspective of someone where if this was set 484 00:27:09,840 --> 00:27:12,199 Speaker 3: about me so widely. You know that I had a 485 00:27:12,240 --> 00:27:14,919 Speaker 3: real problem with honesty and was, in the view of 486 00:27:14,960 --> 00:27:18,240 Speaker 3: many people around me, compulsively lying and telling people conflicting 487 00:27:18,240 --> 00:27:23,399 Speaker 3: things that would be devastating and would trigger a lot 488 00:27:23,440 --> 00:27:26,600 Speaker 3: of deep self reflection and work on myself. And so 489 00:27:26,680 --> 00:27:30,080 Speaker 3: even when I asked him within that personal framework, you know, listen, 490 00:27:30,080 --> 00:27:33,400 Speaker 3: are are you? Is this something that you have talked 491 00:27:33,400 --> 00:27:36,320 Speaker 3: about in therapy? How do you talk to yourself about 492 00:27:36,320 --> 00:27:38,080 Speaker 3: this trait? How do you carry the weight of that? 493 00:27:39,440 --> 00:27:42,080 Speaker 3: I don't know about your impressions you were in that interview, Andrew, 494 00:27:42,560 --> 00:27:45,720 Speaker 3: I did not sense a lot of like deep bracing 495 00:27:45,840 --> 00:27:49,960 Speaker 3: self reflection or self confrontation. There was kind of a 496 00:27:50,000 --> 00:27:53,240 Speaker 3: slightly anodyne discussion of like, yeah, I've tried therapy on occasion, 497 00:27:53,320 --> 00:27:57,320 Speaker 3: you know, I love breath work, and not a lot 498 00:27:57,359 --> 00:28:00,720 Speaker 3: of I mean, look, and maybe he is just choosing 499 00:28:00,760 --> 00:28:01,479 Speaker 3: not to reveal. 500 00:28:02,160 --> 00:28:20,879 Speaker 2: Yeah, but that's all very potentially true. Yeah, it's I 501 00:28:21,320 --> 00:28:23,359 Speaker 2: was literally talking to a friend of the show, Steve 502 00:28:23,359 --> 00:28:25,040 Speaker 2: Burke from Games Next Us about this the other day. 503 00:28:25,040 --> 00:28:27,439 Speaker 2: It's like if someone if everywhere on I don't know, 504 00:28:27,440 --> 00:28:31,080 Speaker 2: if you saw the Iran was threatening Stargate UA. So 505 00:28:31,200 --> 00:28:34,639 Speaker 2: the olken Aidata center in the Middle East. Everyone was 506 00:28:34,760 --> 00:28:37,440 Speaker 2: talking about it like yeah, fuck him up, destroy it, 507 00:28:37,480 --> 00:28:40,640 Speaker 2: blah blah. And it's like, now that you said that, 508 00:28:40,680 --> 00:28:43,040 Speaker 2: I'm wondering if he cares about that either, because if 509 00:28:43,080 --> 00:28:46,560 Speaker 2: I'd made a product and I went online and like, Okay, 510 00:28:46,600 --> 00:28:49,320 Speaker 2: a large contingent, very annoying people liked it, but tons 511 00:28:49,360 --> 00:28:51,880 Speaker 2: of people like disgusted about I would be a little 512 00:28:51,960 --> 00:28:56,240 Speaker 2: more concerned. And it sounds like he doesn't really care. 513 00:28:56,360 --> 00:28:59,080 Speaker 2: Like you brought up like the person who killed killed 514 00:28:59,120 --> 00:29:02,640 Speaker 2: themselves because the reason chat GPT is well, did he 515 00:29:02,720 --> 00:29:05,280 Speaker 2: have any reaction to that or the negativity against open 516 00:29:05,320 --> 00:29:07,680 Speaker 2: aye or was it the same kind of anodyne reaction. 517 00:29:09,400 --> 00:29:09,800 Speaker 3: Andrew. 518 00:29:10,320 --> 00:29:14,120 Speaker 1: I mean, we got statements to the effect of, you know, 519 00:29:14,360 --> 00:29:17,640 Speaker 1: this responsibility weighs very heavily, and you know, I have 520 00:29:17,680 --> 00:29:21,600 Speaker 1: no judgment about what's in his head or heart. I 521 00:29:22,120 --> 00:29:26,880 Speaker 1: think that one way that this might seem different from 522 00:29:27,000 --> 00:29:31,640 Speaker 1: like a sort of more standard micro human scale reaction 523 00:29:32,120 --> 00:29:34,960 Speaker 1: could be that a lot of these people, based on 524 00:29:35,000 --> 00:29:38,240 Speaker 1: their public statements alone, see themselves as kind of playing 525 00:29:38,280 --> 00:29:42,040 Speaker 1: in this big sort of enders game, kind of galactic 526 00:29:42,120 --> 00:29:46,680 Speaker 1: simulation scenario. Right, So I don't you know, this is 527 00:29:46,720 --> 00:29:48,520 Speaker 1: not something that Sam Altman said to us, and this 528 00:29:48,600 --> 00:29:51,000 Speaker 1: is not something that I'm imputing to him. But you know, 529 00:29:51,120 --> 00:29:54,480 Speaker 1: if you you know, follow the way that Musk and 530 00:29:54,560 --> 00:29:58,560 Speaker 1: Altman and a lot of these people think they are 531 00:29:58,920 --> 00:30:02,479 Speaker 1: avowedly playing this galactic scale. Now, there's a possibility that 532 00:30:02,480 --> 00:30:04,360 Speaker 1: that's just hype and that's just to get the next 533 00:30:04,360 --> 00:30:07,160 Speaker 1: investment round, and that's a totally plausible possibility. But to 534 00:30:07,200 --> 00:30:09,560 Speaker 1: the extent that that stuff seeps into their self conception 535 00:30:09,640 --> 00:30:11,880 Speaker 1: at all, there's a way of reading it that says, 536 00:30:12,280 --> 00:30:14,240 Speaker 1: you know, maybe some eggs are going to get broken 537 00:30:14,280 --> 00:30:15,800 Speaker 1: to make the big galactic omelet. 538 00:30:16,600 --> 00:30:20,200 Speaker 2: Yeah, it's is. I keep thinking about how many times 539 00:30:20,200 --> 00:30:22,360 Speaker 2: people call to a sociopath as well, because it's just 540 00:30:23,000 --> 00:30:25,360 Speaker 2: this profile is very interesting because a lot of it 541 00:30:25,520 --> 00:30:28,080 Speaker 2: was going through stuff that being reported, but like getting 542 00:30:28,080 --> 00:30:30,760 Speaker 2: it deeper and deeper and deeper. But in the end, 543 00:30:31,320 --> 00:30:34,160 Speaker 2: compared to like other monsters you have covered, Rona just 544 00:30:34,240 --> 00:30:38,840 Speaker 2: being blunt, Almond almost seems quite tepid, despite the scale 545 00:30:38,880 --> 00:30:42,480 Speaker 2: of all of this, despite how like how much is 546 00:30:42,520 --> 00:30:46,840 Speaker 2: theoretically or otherwise at stake, he just seems and all 547 00:30:46,880 --> 00:30:51,200 Speaker 2: of these people Dariya Ramaday included just seem like business guys. 548 00:30:51,640 --> 00:30:54,720 Speaker 2: They talk about the big philosophical things, but when you 549 00:30:54,760 --> 00:30:58,280 Speaker 2: get down to it, business guys, regular business guys. There's 550 00:30:58,720 --> 00:31:02,760 Speaker 2: like a distinct lack of emotion to it. But not 551 00:31:02,840 --> 00:31:06,120 Speaker 2: in not in an interesting not interesting is not the work, 552 00:31:06,120 --> 00:31:08,360 Speaker 2: but like colorful way, it's just very cold. 553 00:31:08,760 --> 00:31:11,840 Speaker 3: Yeah, I hear you. I think so. First of all, 554 00:31:11,960 --> 00:31:14,840 Speaker 3: I really stay away from in my own mind in 555 00:31:14,880 --> 00:31:17,680 Speaker 3: how I frame things these terms like you know, monster 556 00:31:18,160 --> 00:31:22,080 Speaker 3: or villainy. Even by the way, when reporting on you know, 557 00:31:22,400 --> 00:31:27,200 Speaker 3: someone who's like very clearly guilty of like serial violent crime, 558 00:31:28,480 --> 00:31:33,040 Speaker 3: I think my job, almost especially when the fact pattern 559 00:31:33,200 --> 00:31:37,040 Speaker 3: is really demming, is to kind of excavate what's sympathetic, 560 00:31:37,800 --> 00:31:42,640 Speaker 3: be strenuously, strenuously fair. This is more in line with 561 00:31:43,160 --> 00:31:45,680 Speaker 3: you know, reporting I did on Musk, reporting I've done 562 00:31:45,680 --> 00:31:49,080 Speaker 3: in other cases that's you know, more about national security 563 00:31:49,600 --> 00:31:56,560 Speaker 3: issues where the centerpiece is not some portrait of damning criminality. 564 00:31:57,800 --> 00:32:01,680 Speaker 3: The analysis of Altman's elege edged you know, pathological or 565 00:32:01,720 --> 00:32:09,080 Speaker 3: compulsive lying is very subtle. It is undertaken with if anything, 566 00:32:09,120 --> 00:32:11,960 Speaker 3: I think maximum generosity to him. You know, in the 567 00:32:12,000 --> 00:32:14,040 Speaker 3: course of any piece like this, you have a lot 568 00:32:14,080 --> 00:32:16,760 Speaker 3: of conversations with the subject of the reporting about what's 569 00:32:16,800 --> 00:32:19,840 Speaker 3: in and what's out, and I really erred on the 570 00:32:19,920 --> 00:32:24,840 Speaker 3: side in a lot of cases of generosity towards Sam. 571 00:32:25,080 --> 00:32:28,920 Speaker 3: I feel for anyone who's under this kind of a microscope. 572 00:32:28,960 --> 00:32:30,880 Speaker 3: I think it's correct, you know, if you have this 573 00:32:31,040 --> 00:32:34,400 Speaker 3: much power to be held accountable, but I want to 574 00:32:34,440 --> 00:32:37,800 Speaker 3: do it in a way that is not unkind, and 575 00:32:37,840 --> 00:32:40,640 Speaker 3: that is tough but really fair. So I've been struck, 576 00:32:40,720 --> 00:32:42,920 Speaker 3: actually by the range of reactions to the piece. I 577 00:32:42,920 --> 00:32:45,360 Speaker 3: think Andrew and I did our job well in the 578 00:32:45,440 --> 00:32:48,520 Speaker 3: sense that I see people who are like, this is, 579 00:32:49,320 --> 00:32:52,160 Speaker 3: you know, the most horrifying thing ever, and this is 580 00:32:52,200 --> 00:32:55,720 Speaker 3: a completely untrustworthy person who needs to be kept away 581 00:32:55,720 --> 00:32:58,600 Speaker 3: from any position of authority. And I also see people 582 00:32:58,760 --> 00:33:01,320 Speaker 3: who are like, you know what, this feels actually like 583 00:33:01,320 --> 00:33:04,960 Speaker 3: a sympathetic portrait. My mom was calling me last night, 584 00:33:04,960 --> 00:33:06,920 Speaker 3: I having read the piece, saying, you know, I just see 585 00:33:06,960 --> 00:33:09,360 Speaker 3: interviews with him, and I sense this kind of charm 586 00:33:09,400 --> 00:33:12,080 Speaker 3: and vulnerability, and you know, I read the piece and 587 00:33:12,600 --> 00:33:15,200 Speaker 3: he's a complicated guy. But I think you did a 588 00:33:15,240 --> 00:33:18,480 Speaker 3: really good job actually also making him sympathetic. So I 589 00:33:18,520 --> 00:33:21,360 Speaker 3: think the fact that the fact pattern can accommodate this 590 00:33:21,560 --> 00:33:26,320 Speaker 3: range of perceptions is telling. And my hope is that 591 00:33:26,360 --> 00:33:29,400 Speaker 3: the power of what we've done is precisely in that 592 00:33:29,520 --> 00:33:33,840 Speaker 3: kind of forensic nuance. The facts are very unimpeachable. I 593 00:33:33,880 --> 00:33:36,959 Speaker 3: think it is very difficult to argue with the idea 594 00:33:37,120 --> 00:33:41,520 Speaker 3: that there is a serial honesty problem here and a 595 00:33:41,560 --> 00:33:44,000 Speaker 3: preponderance of people around him who consider it to be 596 00:33:44,040 --> 00:33:46,000 Speaker 3: a problem. And to your point of like, is this 597 00:33:46,040 --> 00:33:50,120 Speaker 3: all milk toast? Doesn't matter? I think what it comes 598 00:33:50,120 --> 00:33:52,680 Speaker 3: down to is, yes, he is a businessman, and this 599 00:33:52,760 --> 00:33:54,840 Speaker 3: is a world of businessmen that we are looking at. 600 00:33:56,160 --> 00:34:03,719 Speaker 3: Sam Altman's degree of dissembling appears to be so noteworthy 601 00:34:03,760 --> 00:34:07,480 Speaker 3: that it transcends and breaks through and becomes a constant topic. 602 00:34:07,520 --> 00:34:08,600 Speaker 2: What do you mean by dissembling? 603 00:34:08,719 --> 00:34:13,400 Speaker 3: Sorry, the allegation that Sam Altman says conflicting things to 604 00:34:13,600 --> 00:34:18,879 Speaker 3: different people, and at times says things that are simply untrue. 605 00:34:19,160 --> 00:34:23,880 Speaker 3: It is so pervasive that it breaks through and becomes 606 00:34:23,920 --> 00:34:27,000 Speaker 3: a topic of conversation around him across his career, over 607 00:34:27,040 --> 00:34:29,720 Speaker 3: and over again. Is one thing we document, even against 608 00:34:29,760 --> 00:34:32,560 Speaker 3: the backdrop of all of the Silicon Valley businessmen, all 609 00:34:32,600 --> 00:34:37,200 Speaker 3: of the expectation which is, you know, to a dismaying extent, 610 00:34:37,520 --> 00:34:42,520 Speaker 3: increasingly just what people assume is normal, that there will 611 00:34:42,560 --> 00:34:45,640 Speaker 3: be some degree of pitchmen, you know, hyping based on 612 00:34:45,680 --> 00:34:49,120 Speaker 3: things that aren't totally grounded in fact. Even with all 613 00:34:49,239 --> 00:34:51,120 Speaker 3: of that, even with the kind of race to the 614 00:34:51,120 --> 00:34:53,760 Speaker 3: bottom in terms of truth and trust in Silicon Valley 615 00:34:53,800 --> 00:34:58,960 Speaker 3: and America right now, particularly in American business, Sam Altman 616 00:34:59,080 --> 00:35:03,480 Speaker 3: is noteworthy, and that to me made him actually an 617 00:35:03,520 --> 00:35:08,040 Speaker 3: extraordinarily interesting and challenging subject to write about, because while 618 00:35:08,080 --> 00:35:11,880 Speaker 3: it does exist in subtlety, the accumulation of facts that 619 00:35:11,960 --> 00:35:17,560 Speaker 3: got him fired is striking even above and beyond that 620 00:35:17,600 --> 00:35:19,160 Speaker 3: backdrop and that set of norms. 621 00:35:20,400 --> 00:35:23,040 Speaker 2: So, as we wrap up, you spoke with Dario ama 622 00:35:23,080 --> 00:35:25,200 Speaker 2: Die of Anthropic a few times, right, or at least 623 00:35:25,400 --> 00:35:28,719 Speaker 2: it wasn't wasn't clear how many conversations happened. How does 624 00:35:28,760 --> 00:35:32,600 Speaker 2: he compare to Sam personality wise and as far as 625 00:35:32,680 --> 00:35:33,680 Speaker 2: believing it all goes? 626 00:35:34,800 --> 00:35:37,200 Speaker 1: I mean, I think personality wise they're quite different, and 627 00:35:37,239 --> 00:35:39,640 Speaker 1: we've seen them sort of clash, you know, the sort 628 00:35:39,680 --> 00:35:42,640 Speaker 1: of like memes of them being unwilling to even you know, 629 00:35:42,840 --> 00:35:44,680 Speaker 1: touch hands and all of that stuff. I think that's 630 00:35:44,719 --> 00:35:50,000 Speaker 1: pretty real and you know, there are many different styles 631 00:35:50,040 --> 00:35:52,480 Speaker 1: of this. I think, you know, Sam and Elon have 632 00:35:52,760 --> 00:35:56,560 Speaker 1: obviously very different public personas Ilia Setskiver has a very 633 00:35:56,560 --> 00:36:00,840 Speaker 1: different public presentation. The Dario thing, I think is complicated, 634 00:36:00,920 --> 00:36:05,359 Speaker 1: right because Anthropic was spun off from open AI. They were, 635 00:36:05,440 --> 00:36:08,600 Speaker 1: you know, the Open AI safety people who left to 636 00:36:08,719 --> 00:36:11,879 Speaker 1: form their own company, and now there's this danger of 637 00:36:11,920 --> 00:36:15,040 Speaker 1: them perpetuating this race to the bottom with you know, 638 00:36:15,120 --> 00:36:18,000 Speaker 1: getting the same you know, Middle East autocracy money that 639 00:36:18,040 --> 00:36:21,560 Speaker 1: they critiqued other companies for doing of. You know, there 640 00:36:21,640 --> 00:36:23,840 Speaker 1: was this whole dust up around the Pentagon stuff, but 641 00:36:24,280 --> 00:36:26,959 Speaker 1: that only was able to happen because Anthropic was doing 642 00:36:27,000 --> 00:36:29,480 Speaker 1: so much classified work for the Pentagon in the first place. Right, 643 00:36:29,560 --> 00:36:33,680 Speaker 1: So it's not that there are these very clean distinctions, 644 00:36:34,120 --> 00:36:37,440 Speaker 1: at least to an outside the industry kind of perspective. 645 00:36:38,840 --> 00:36:41,239 Speaker 1: And look, I mean, I think your point about like 646 00:36:42,120 --> 00:36:43,920 Speaker 1: how much of this is just par for the course 647 00:36:43,960 --> 00:36:47,000 Speaker 1: business stuff is a really important one. And also your 648 00:36:47,280 --> 00:36:49,360 Speaker 1: your question is the right one, like why should anyone 649 00:36:49,480 --> 00:36:52,280 Speaker 1: have invested so much in this rhetoric in the first place? 650 00:36:52,400 --> 00:36:55,239 Speaker 1: And I think one way of approaching that is, like 651 00:36:56,200 --> 00:36:58,880 Speaker 1: to the people who are really close to this I 652 00:36:58,920 --> 00:37:01,919 Speaker 1: think this does go be on like narcissism of small 653 00:37:01,920 --> 00:37:05,440 Speaker 1: differences stuff and like who puts the nicer wrapper on 654 00:37:05,800 --> 00:37:09,839 Speaker 1: the you know, who does better branding? Like these people again, 655 00:37:09,880 --> 00:37:12,800 Speaker 1: in their private correspondence when they don't think anyone's looking, 656 00:37:13,280 --> 00:37:15,680 Speaker 1: they talk about this like it's Lord of the Rings literally, 657 00:37:15,719 --> 00:37:17,640 Speaker 1: like they talk about you know, and so again, you 658 00:37:17,680 --> 00:37:20,720 Speaker 1: can think that's all diluted, and many people do, And frankly, 659 00:37:20,719 --> 00:37:23,200 Speaker 1: a lot of people we spoke to said in retrospect, 660 00:37:23,239 --> 00:37:25,240 Speaker 1: I can't believe I was so naive that I bought 661 00:37:25,239 --> 00:37:28,720 Speaker 1: into all this stuff. But when you have convinced yourself 662 00:37:28,760 --> 00:37:30,680 Speaker 1: that that's what's going on, whether it's right or wrong, 663 00:37:31,280 --> 00:37:35,520 Speaker 1: the stakes feel extremely high. And if you believe the rhetoric, 664 00:37:35,680 --> 00:37:37,520 Speaker 1: you know, it's it's kind of you know, it's funny. 665 00:37:37,520 --> 00:37:39,480 Speaker 1: I was listening when you were talking about this, Like 666 00:37:39,840 --> 00:37:42,640 Speaker 1: you could say the same thing about someone who was 667 00:37:42,880 --> 00:37:46,399 Speaker 1: really taken in by a political movement and say, right, oh, 668 00:37:46,440 --> 00:37:48,319 Speaker 1: did you really believe Obama when he said he was 669 00:37:48,360 --> 00:37:50,000 Speaker 1: going to go after the banks, you know, on the 670 00:37:50,040 --> 00:37:52,239 Speaker 1: campaign trail? Or did you really believe Trump when he 671 00:37:52,280 --> 00:37:54,160 Speaker 1: said he was going to not start any new wars? 672 00:37:54,719 --> 00:37:55,000 Speaker 2: Yes? 673 00:37:55,200 --> 00:37:57,839 Speaker 1: People believe it, and the betrayal of that belief can 674 00:37:58,120 --> 00:38:01,520 Speaker 1: hit really hard. And so you know, yes, politicians lie, 675 00:38:01,560 --> 00:38:05,120 Speaker 1: and yes, you know, business pipe men tell stories to investors, 676 00:38:05,120 --> 00:38:07,200 Speaker 1: but like doesn't mean that people don't get taken in 677 00:38:07,239 --> 00:38:07,560 Speaker 1: by it. 678 00:38:08,280 --> 00:38:12,000 Speaker 2: So final question, actually in line with that, do you 679 00:38:12,080 --> 00:38:15,720 Speaker 2: think sam Oltman sees himself as a businessman or as 680 00:38:15,800 --> 00:38:17,920 Speaker 2: a as a statesman? 681 00:38:18,160 --> 00:38:23,600 Speaker 3: Almost that's a really interesting question. I think that you know, 682 00:38:23,880 --> 00:38:27,759 Speaker 3: he I honestly since in my conversations with him that 683 00:38:27,760 --> 00:38:34,560 Speaker 3: he sees himself as something quite singular, and the grandiosity 684 00:38:34,840 --> 00:38:38,080 Speaker 3: of his claims about his mission, you know, when he 685 00:38:38,160 --> 00:38:42,399 Speaker 3: was telling recruits that he believed that this technology could 686 00:38:42,440 --> 00:38:46,279 Speaker 3: eventually what was the exact quote, Andrew capture the light 687 00:38:46,400 --> 00:38:50,840 Speaker 3: cone of all the value in the universe. I think 688 00:38:51,360 --> 00:38:55,480 Speaker 3: he exhibits a trait that I have seen writing about 689 00:38:55,680 --> 00:39:01,360 Speaker 3: various Silicon Valley moguls of this era, which is they 690 00:39:01,520 --> 00:39:04,879 Speaker 3: do have a kind of Messianic quality to how they 691 00:39:04,960 --> 00:39:07,880 Speaker 3: see themselves, you know, not as businessmen, not as statements 692 00:39:07,920 --> 00:39:13,440 Speaker 3: but statement, but as everything to everyone, as super governmental. 693 00:39:13,960 --> 00:39:17,560 Speaker 3: I think Elon is a much more extreme example of that. Psychologically, 694 00:39:17,760 --> 00:39:21,279 Speaker 3: the kind of Messiah complex. There's much more mania there. 695 00:39:22,120 --> 00:39:25,960 Speaker 3: Sam is someone who you know, as evidenced by the 696 00:39:26,000 --> 00:39:29,720 Speaker 3: fact that he participated in this story so deeply, still 697 00:39:29,760 --> 00:39:33,040 Speaker 3: wants to kind of play by the rules and within 698 00:39:33,080 --> 00:39:37,480 Speaker 3: a system to some extent. But the reality around all 699 00:39:37,520 --> 00:39:43,120 Speaker 3: of these guys is, we have let Silicon Valley grab 700 00:39:43,400 --> 00:39:46,320 Speaker 3: all of the levers of power in the United States. 701 00:39:47,560 --> 00:39:51,839 Speaker 3: It is such a center of gravity economically, it is 702 00:39:52,719 --> 00:39:54,719 Speaker 3: I mean, one could say more or less the entire 703 00:39:54,719 --> 00:40:00,960 Speaker 3: economy at this point, and it is bankrolling a huge, 704 00:40:01,120 --> 00:40:05,920 Speaker 3: huge swath of politics now. So it is very hard, 705 00:40:05,960 --> 00:40:08,359 Speaker 3: I think for the people at the forefront of that 706 00:40:08,880 --> 00:40:13,680 Speaker 3: to consider themselves anything but messianic. And you add to 707 00:40:13,719 --> 00:40:17,160 Speaker 3: that the specific contour of AI in which these guys 708 00:40:17,200 --> 00:40:21,640 Speaker 3: really believe and not without reason, that they are birthing 709 00:40:22,120 --> 00:40:27,680 Speaker 3: the future of the earth, that this may supersede the 710 00:40:27,800 --> 00:40:33,120 Speaker 3: human species as the next dominant thing. I think that 711 00:40:33,200 --> 00:40:37,120 Speaker 3: really magnifies the messianic quality. Yeah, and so that all 712 00:40:37,120 --> 00:40:39,879 Speaker 3: goes to the responsibility question. Like yes, to Andrew's point. 713 00:40:39,920 --> 00:40:45,080 Speaker 3: You know, business people, they fib and they dissemble and 714 00:40:45,280 --> 00:40:47,680 Speaker 3: they are guilty of empty hype. And that is the 715 00:40:47,719 --> 00:40:53,520 Speaker 3: whole saga of Silicon Valley from the beginning the fact 716 00:40:53,560 --> 00:40:57,279 Speaker 3: that Sam Altman is not, in my view, you know, 717 00:40:57,440 --> 00:41:00,600 Speaker 3: a villain in any clear cut way, and the piece 718 00:41:00,640 --> 00:41:03,840 Speaker 3: goes to pains to be generous to him. To me 719 00:41:04,000 --> 00:41:06,600 Speaker 3: makes this a more interesting story because both things can 720 00:41:06,640 --> 00:41:09,680 Speaker 3: be true. You can have someone who is not a monster, 721 00:41:11,120 --> 00:41:13,799 Speaker 3: but who is still in a position of so much 722 00:41:13,920 --> 00:41:17,359 Speaker 3: power that if they have this particular trait of no 723 00:41:17,440 --> 00:41:21,879 Speaker 3: one can trust them, and what they say is going 724 00:41:21,920 --> 00:41:25,319 Speaker 3: to happen with the most dangerous technology on earth might 725 00:41:25,360 --> 00:41:29,400 Speaker 3: not be what happens. That is something that we should 726 00:41:29,520 --> 00:41:31,400 Speaker 3: all care about, and I think it just speaks to 727 00:41:31,440 --> 00:41:35,960 Speaker 3: this wider set of structural issues where this is the 728 00:41:36,000 --> 00:41:39,120 Speaker 3: most acute need we have as a species for proper 729 00:41:39,160 --> 00:41:42,200 Speaker 3: governance and regulation, and that is all falling away and 730 00:41:42,239 --> 00:41:45,439 Speaker 3: it's all in the hands of these private, unaccountable individuals. 731 00:41:46,280 --> 00:41:49,719 Speaker 2: Well, thank you so much Andrew Rohan for joining me this. 732 00:41:49,840 --> 00:41:52,839 Speaker 2: It's been an awesome conversation. The peace be linked. Yeah, 733 00:41:52,880 --> 00:41:53,719 Speaker 2: thank you for joining me. 734 00:41:53,880 --> 00:41:55,880 Speaker 1: Thank you so much, Ed, Thank you so much, Ed. 735 00:42:04,719 --> 00:42:06,360 Speaker 2: Thank you for listening to Better Offline. 736 00:42:06,520 --> 00:42:08,920 Speaker 4: The editor and composer of the Better Offline theme song 737 00:42:09,000 --> 00:42:11,680 Speaker 4: is Matasowski. You can check out more of his music 738 00:42:11,680 --> 00:42:15,319 Speaker 4: and audio projects at Matasowski dot com, M A T 739 00:42:15,320 --> 00:42:19,799 Speaker 4: T O S O W s Ki dot com. You 740 00:42:19,800 --> 00:42:22,320 Speaker 4: can email me at easy at better offline dot com 741 00:42:22,440 --> 00:42:24,760 Speaker 4: or visit better offline dot com to find more podcast 742 00:42:24,800 --> 00:42:28,160 Speaker 4: links and of course my newsletter. I also really recommend 743 00:42:28,160 --> 00:42:30,080 Speaker 4: you go to chat dot Where's Youreed dot at to 744 00:42:30,160 --> 00:42:32,520 Speaker 4: visit the discord, and go to our slash. 745 00:42:32,200 --> 00:42:35,359 Speaker 2: Better Offline to check out our reddit. Thank you so 746 00:42:35,480 --> 00:42:36,200 Speaker 2: much for listening. 747 00:42:37,040 --> 00:42:39,520 Speaker 1: Better Offline is a production of cool Zone Media. 748 00:42:39,680 --> 00:42:42,520 Speaker 2: For more from cool Zone Media, visit our website cool 749 00:42:42,560 --> 00:42:45,959 Speaker 2: Zonemedia dot com, or check us out on the iHeartRadio app, 750 00:42:46,000 --> 00:43:05,640 Speaker 2: Apple Podcasts, or wherever you get your podcasts.