The Aboard Podcast

Paul Ford: Hi, I’m Paul Ford.

Rich Ziade: And I’m Rich Ziade.

Paul: And we’re seated closely today. You know what that means?

Rich: We’re getting along?

Paul: No, it doesn’t mean that. It means a guest on Riverside.

Rich: Oh!

Paul: Yeah. We get to look up at the screen and talk to a person. And that means we have to get, we get to snuggle. And that’s nice.

Rich: Nice.

Paul: It’s very nice. So this is The Aboard Podcast. It’s a podcast about how AI is changing the world of software, because it truly is. Rich and I have been in the software industry for at least 300 years.

Rich: Mmm hmm.

Paul: And lots of stuff is going down. And we’re going to talk about that today with someone who has really looked closely at one of the people who is absolutely at the top of the pyramid here, and figure out what the hell is happening.

Rich: They went in deep.

Paul: So let’s play the theme song and let’s tell people what we do, and then let’s have a real conversation.

Rich: Let’s do it.

[intro music]

Paul: Rich, what is Aboard?

Rich: Aboard is an AI transformation shop with incredible magical tools.

Paul: That sounds like nonsense to me.

Rich: We help you get stuff done faster, better with AI.

Paul: Let me give you an example. Recently, a big not-for-profit focused on children’s behavioral health came to us.

Rich: Unimpeachable.

Paul: And they said, “Could you help us with the dashboards? Just make it so we can do more with the dashboards.”

Rich: Yeah, that’s right.

Paul: And we made it—you can talk to the dashboards, you can ask questions of the data.

Rich: We bring the data in. We made it a lot simpler to understand where things can run better. Yeah, it’s great.

Paul: That org is called Make An Impact. There’s a nice case study on the website. But that kind of thing: If you have ugly, horrible data problems and you want to make them nice and fun and chatty, that’s what we’re for. That’s enough advertising. Let’s talk about Sam Altman. And we have, we’re very, very lucky and very, very grateful to have someone who is currently one of the Sam Altman experts on Earth. Andrew Marantz of The New Yorker. Hello, Andrew.

Andrew Marantz: Hey! Thanks for having me, guys.

Paul: It’s wonderful. In the April 13th edition of The New Yorker, The Future Issue, where you are a staff writer, correct?

Andrew: Yeah.

Paul: You wrote a quite lengthy, quite detailed sort of summary of Sam Altman’s career with your co-writer, Ronan Farrow.

Andrew: Mmm hmm.

Paul: You found him to be unimpeachable, an excellent human being, and a person of incredible trustworthiness.

Andrew: That’s right. Well, just like I heard you say in the intro, you know, a nonprofit comes to you, they must be unimpeachable. And OpenAI was started, famously, as a nonprofit. So there you go. That wraps up the story.

Paul: I mean, just two angels, Sam Altman and Elon Musk, came together [laughter] and said, we need to do something for humanity. And today we live in a bountiful world of joy and peace.

Rich: I do, I want you to get into this kind of slow burn of a story arc, because there’s no bombshell, really.

Paul: No.

Rich: It’s a bunch of little explosions, I would say, if we’re gonna run with that?

Paul: Go all the way back out. When did you both start working on this story? Give us a sense of the genesis.

Andrew: It was a long time ago. It was more than a year ago. We had both come to it from different angles. Ronan had done some reporting on Musk and other figures in this world. I had done some reporting on AI safety, or AI doomer people, and sort of trying to get inside the headspace of this sort of fundamental, I think, still unresolved question of, why do so many of the people who are closest to building AI find it to be an existential risk to themselves and everyone they love? And if so, why do they keep building it? Right?

So I had been working on that question. Ronan had been working on these sort of power in Silicon Valley questions. And we kind of converged on this question of, okay, if there’s one company or one person at the center of this mystery of, like, if AI is the most dangerous thing ever invented, who are the people inventing it, pushing it to the frontier so fast, why are they doing it? Really, the place to converge on that question was around Sam Altman and OpenAI.

So that was, I think, about 18 months ago at this point, that we started looking into it. And it was, as you say, it wasn’t one smoking-gun thing. It wasn’t like everyone who’s ever met Sam Altman knows that he secretly clubs baby seals, and once that is revealed to the public, he will be unmasked. It was a much more subtle sort of string of allegations. So it took a long time.

Paul: When’s the first time you ever heard of him? I have a very specific memory of him.

Andrew: Hmm. I don’t know if it was the first time. So one of the things—you mentioned that this was for The Future Issue or the, I think that’s what it was called. We had an online, we, The New Yorker, had an online issue called The Interviews Issue. And it was happening in November or December 2022. So I had the very clever, unique idea, I will interview ChatGPT for this Interview Issue.

And so I did a thing where I talked to the chatbot. And because I, you know, don’t work at Google DeepMind or one of these companies, it was pretty new to me. Unlike people who work closely with this stuff and they’re like, this is just GPT3 or GPT 3.5 with some dumb little API on it, I was like, oh, wow, this is a really impressive thing.

And it was around that time that it occurred to me to wonder, okay, I’m going back through all these interviews that Sam Altman has done with Ezra Klein and with, I think Tyler Cowen at that point, and a bunch of places where he had spoken publicly about his dire fears about this technology on his own blog. And I just remember thinking, like, I think that we are just structurally forced with him and with people in this position to take their word for it. Like, there’s no real collateral here. There’s no, like, put your money where your mouth is. It’s sort of just like, do they mean it? Well, do they sound like they mean it?

Paul: Mmm hmm. I have a very specific memory because I’m a big nerd. I’m a big Alan Kay fan.

Andrew: Mmm hmm.

Paul: Alan Kay from Xerox PARC, for people who don’t know. If you’ve ever used a windowing interface, he’s partially to blame.

Andrew: Mmm hmm.

Paul: Like, just real deep.

Rich: Or thank.

Paul: Yeah, or thank, right? No, no, no. Just a brilliant guy. And so he was giving a talk and just out of nowhere he’s like, man, that’s—and I haven’t been able to find it, but I just remember it really specifically. He’s, like, that Sam Altman man, that’s just a guy right there. Like, he was like, literally those words, like a granddad who was like, there he is, the future of the Valley. And this would be like three years, it was like right when he was taking over Y Combinator.

Andrew: Mmm hmm.

Rich: Mmm hmm.

Paul: He’d been like first class. I knew Aaron Swartz a little bit. Like, I watched them all kind of come up. And so I’d never paid much attention to him. And then suddenly to just have like one of the great sort of figures in the Valley be like, there he is, that’s the future. And I’ve been watching him on and off ever since, and I still can’t fully figure it out. Like, he’s a very slippery human being that way. But whatever he has, boy does it charm the hell out of venture capital and investors, like, they just—and like, the true Valley just looks at him and they see themselves reflected. And I think it’s so, I think there’s a little bit of that. Like, he just, he represents what people want the Valley to be. Or he did at one point. And he kind of came out of that.

Rich: Yeah. And for those that haven’t read the article, Andrew, I mean, first off, go read the article. I ripped through it. It’s very… It’s strangely entertaining. It’s, like, the intellectual gossip—

Paul: That’s the job. The job—

Rich: You guys did a great job with making me feel smart and a little dirty at the same time.

Paul: That’s The New Yorker. You take really difficult shit and you add zazzle.

Rich: Yeah.

Paul: Just a little bit of zazzle.

Rich: For those that haven’t read it yet, go read it. But if you haven’t read it yet, give us a couple of, like, little vignettes. You guys diligently start to draw out a pattern. There was clearly, after a while, I was like, oh, here we go again. Give us a couple of little stories that sort of map out, where like, huh, there seems to be a pattern here.

Andrew: Well, you kind of. Paul, you just kind of touched on it, right? Where you have the investor VC types who are very charmed by Altman’s rhetoric, but then you also have the Alan Kay types, who are really, in many ways, kind of diametrically opposed to that camp, who are sort of more pure innovation for its own sake types. And then you also have this newer camp because of, it’s specific to AI, which is really, like, people who are terrified to summon this thing into existence.

And what emerges through this reporting, and I think through the piece, is that Altman does seem to have this way of talking to each of these camps of people in a way that kind of convinces each one that he is kind of, at bottom, really one of them or really sympathetic to their concerns.

And so, yeah, you do kind of get this accumulation of details. I mean, we start the piece with the place that everyone starts, which is, why was he fired in late 2023? And I think what emerged for us, which was kind of a hunch I had going into it, and then when we got all these documents, it was confirmed, is, as I said, if it was one simple thing, we would know it by now.

So what you get is, right, so we reviewed all these documents that Ilya Sutskever compiled with help from other people on the board that led to Altman’s being fired. But we also reviewed pages and pages of documents, of emails that he was sending to people, that Altman was sending to people before OpenAI was founded in 2015. We went back to his Y Combinator days and kind of investigated this old question that has swirled around Silicon Valley of was he pushed out of the top job at Y Combinator? And it seems from our reporting that he was, even though he continues to deny it.

But also just these little things, like you say, there is one, because OpenAI was founded as a nonprofit and then went through this very byzantine series of steps over a decade to gradually convert itself to a for-profit, which really only concluded last year, at one of those steps, there was a question of how the nonprofit should be compensated. And there were wildly different ways to create that valuation. This forms some of what will be argued about during this trial in Oakland, this Musk vs. Altman trial.

But, like, just this is one example of a kind of nugget of information that doesn’t mean that much on its own, but I think slots into a pattern. There was one board member who said, I don’t think the nonprofit is being compensated fairly. I know that the board is going to approve this package, but I’m going to vote no. And we reviewed documents of this board member saying, I’m going to vote no. And then we also saw documents where it seems that that vote was recorded in the official company minutes as an abstention rather than a no.

Now, it almost seems inconsequential, like the board package is going to pass either way. But there does seem to be this kind of almost meticulous drive to kind of tidy up narrative loose ends, which I think ironically ends up creating a lot of narrative problems for the company.

Paul: A flip side to this, which is wild, which really struck me, and it just struck me what you said, too, which is everybody keeps a diary like a 19th-century Russian princess. [laughter] Like, I mean, it’s just, it’s so shocking how much documentation is flowing your way as a journalist—

Rich: I think people know shit’s gonna might hit the fan.

Paul: Was that surprising to you when they’re just like, “Oh, I have this 140-page PDF of every—” Like, how did that play out?

Andrew: Yeah, I did find that surprising, honestly, because you do sometimes think, as a reporter, you know, we’re in this age of disappearing Signal messages and like, nothing will be preserved for prior generations. And actually, this is sort of a tangent but we had these for the 100th anniversary of the magazine, we had, a lot of writers sort of picked a favorite old piece from the archives and went back and wrote a sort of second take on it. And I, because I’m a weirdo and a masochist, I picked The New Yorker profile of Hitler. And I went back and read not only the piece, but all the correspondence around the piece. And I was just struck by how much of that is preserved. Right? You have intra-office memos, inter-office memos, cables back and forth from Paris and Berlin and it’s all just sitting there.

Paul: Oh, there’s filing cabinets at the office.

Andrew: Yeah, yeah. Well, and also, and also at the New York Public Library. I mean, there are boxes and boxes of stuff. So you go to the New York Public Library and ask for, you know, Janet Flanner’s cables in 1938 and 1939. And they’re all just there, like the original copies. And so it’s incredible. I mean, I did not need to do all that research for an 800-word piece, but I did and it was, I was overjoyed to do it. [laughter]

Paul: I don’t know, man. It’s The New Yorker‘s profile of Hitler. Like, every guy over, like, 30 is gonna double down on that one. “Hey, I’ll be back in a couple of days! Time to do a little Hitler.”

Andrew: It is, it is a rich text, as they say. But that’s, my only point there is just to say, like, there was so much that survived in its original form. And so my, my sort of reporter brain was like bemoaning how little of that will survive the moment we’re in. And so, to your point, I was pretty surprised by how many people kept a digital diary and, because I think even if they knew that the shit was gonna hit the fan, we wouldn’t that kind of militate in the other direction? Right?

Paul: This is an organization, this is a culture that apparently loves putting secrets into Google Docs and then sharing them. [laughter]

Rich: Yeah.

Paul: Which is wild, right? Because you’re like, here’s this guy and he very clearly, he’s very… Everybody in that role is very image-conscious. And everybody in the Valley is very image-conscious. And you know, they all have low body fat and they’re doing good.

Andrew: Mmm hmm.

Paul: And so like here you have this, like, just huge narrative engine, billions of dollars, let’s get it under control. And it’s just wild to me that everybody else is, like, I’m just going to this into Google Docs and sort of leave it in anonymous mode, so anyone can share it as an animal. [laughter]

Rich: It’s a URL.

Paul: Yeah, yeah. And so there is a. That is an unexplored aspect.

Rich: Question: I’m going to ask you to take a leap here. I think there’s two ways, and you guys sort of allude to him summing up, because you confront him with all these contradictions and all this kind of chameleon-like behavior. And his response was like, yeah, I’m pretty flawed. And you sort of roll with it. And then at one point I said, wait a minute, this guy pulled one over on Andrew and Ronan, too! What a plot twist. Like, he’s so good at it. It’s sort of a soft landing by the very end. Do you buy it or do you not buy it? Let’s start there.

Paul: I don’t know. It didn’t land as that soft to me.

Andrew: Hmm.

Rich: I don’t, I think—well, all right, the piece speaks for itself because. Yeah, but, yeah, tell me. There was a babe-in-the-woods vibe to his explanation.

Paul: Did you get charmed?

Andrew: So, yeah, I don’t think that many people read this as a soft piece, but I will say that we do try very hard to be open and influenced by all the information that comes our way.

Rich: Yeah.

Andrew: And it would be journalistic malpractice and sort of epistemic malpractice to take all the many pages of documentation and interviews and stuff from people who have these deep critiques without talking to Altman and his defenders and sort of, you know, steelmanning their position, as we say. So I do think that’s an important thing to do in any piece.

So, for example, to make it less personal, because I, you know, I think I didn’t go into this with like, I really want to psychologize this one individual. Obviously, like, all people at this level of power deserve a high level of journalistic scrutiny. But more to the point, as we’ve been talking about somebody at this particular juncture of technological history, it’s not just picking someone with a lot of power and money. It’s somebody whose entire pitch for himself is, “I alone can be in charge of the Ring of Sauron because I have this kind of quality of being uncorruptible.” I mean, this is a pitch that he’s made several times.

Rich: Right.

Andrew: I wasn’t particularly like, I need to, you know, get to a personal landing, you know, a place where I feel that I’ve unlocked this person’s personality. Although I think you get a lot, you know, in the piece that goes to that. I think that there’s more this structural question of why are we even asking about the personality flaws of an individual with this much power? Right?

Obviously, I think to raise that question in a way is sort of self-answering that like it’s absurd, if there is anything like the amount of power that Sam Altman says there is or will soon be in this technology, as many of the people we spoke to for the piece suggested, it seems sort of nuts to vest anyone with that amount of power. That, and again, you see this rhetoric cropping up again and again in the decade leading from the nonprofit to the for profit, which is Altman saying himself, I should not be trusted with this much power, begging Congress to regulate him, all these things.

So to the extent that I think it matters structurally, a lot of it hinges on the question of, well, were all of those just sort of ploys at regulatory capture, or were all of those just sort of ploys at recruitment and retention and getting what he wanted, or were those sincere? And either way, I think there are sort of troubling questions or conclusions to draw from that.

Just to finally sort of tie the knot on answering your question more directly, you know, we had several editorial conversations about this question of charm and persuasiveness, because we have quotes in the piece and we have anecdotes in the piece of people saying, you know, he’s like a Jedi Mind Master. He’s just incredibly next-level persuasive. And, you know, there are people who have interacted with him or have seen, you know, tape of him or whatever who are like, that guy? That guy’s the Jedi Master Persuader? It doesn’t seem that way.

And I think a lot of it has to do with what is the persuasive game being played, right. I think it’s not this kind of thing where it’s like the most Obama-level, Dalai Lama-level charisma, like, can work any room. [laughter] It’s not that kind of—that’s not the prompt, right? I think the prompt is, can you sit in a room with this person and feel that your deepest concerns are being mirrored back in a kind of sympathetic way?

Paul: It’s not even in a room. It’s can you sit around a fire pit and out in the Valley, and can you sort of talk like a human spreadsheet?

Rich: Yeah.

Paul: You want to know my worst nightmare?

Rich: I’d rather not hear it, but we’re on a podcast. [laughter] Go.

Paul: Knowing that The New Yorker is profiling me for 18 months. [laughter]

Rich: It’s not going to end well.

Paul: That’s just like, I’m just thinking, like, you wake up on a Wednesday, you’re like, oh, sunshine and birds are out. I wonder if that article’s gonna drop. No, I mean—

Andrew: And that’s why, I mean, just to that point, like, to the point about is it sort of falling for someone’s persuasive wiles or is it steelmanning their position? I guess the one thing I would say to that, and you just reminded me of it, is like, I didn’t want this to be an exercise, and I know Ronan didn’t either, of like, if you spend 18 months rifling through anybody’s past, you’ll come up with something that you can frame to look bad. I really wanted to filter, we both really wanted to filter, against just competitor mudslinging, against kind of ambient sort of suggestive bullshitty things. Like, we really wanted to bulletproof it so that it was the stuff that really transcended that.

And I think, you know, your audience will get this, like, there’s often this critique of OpenAI that, you know, they started out as open source and that was what the open was supposed to be, and then they weren’t even open source anymore. You’ll notice that knock on them doesn’t appear in the piece, just as an example. And I think that’s because there are perfectly good-faith reasons, you know, agree, disagree, but there are perfectly good-faith reasons and explicable reasons for why that happened. And so we wanted to focus on things where there’s a less tidy explanation.

Rich: I think what Andrew nailed was this thing is so vast and its impact is so immense that to put it at the feet of one human being—who I don’t view as sinister, frankly, I view as flawed like we’re all flawed, frankly—who isn’t, you know, doesn’t have a PhD in ethics and moral behavior. Like, that’s not his—he’s been in the Valley, he’s seen…

Andrew: Or computer science, but that’s a separate… [laughter]

Rich: Or computer science, which I actually want to ask you about.

Andrew: Yes.

Rich: I think this is, this is someone that wants, he grew up in startup culture, where you really have to wake up each day and figure out how you’re going to navigate the next thing. Right? And how you’re going to get to the next deal, right? And I think that is so hardwired in him. And I don’t mean this as an, like, as an advocate for Sam Altman. I don’t think he woke up in 2021 and said, “I’ve got a master plan. We’re going to, we’re going to get into everybody’s minds through—”

Paul: I think it’s the opposite.

Rich: “—dot org—”

Paul: No—

Rich: And then we’ll switch, do a switch. I don’t think there’s any of that.

Paul: Not with this stuff. All these guys are made of his master plans. Sorry, Andrew, I know you’re on this podcast, too, but hold on. [laughter]

Rich: They’re curious to hear it. No, I’d love to hear your read on this.

Paul: Here’s how I see this. They’re made of master plans. Everybody wants to reinvent society from technology, and that is the ethos, and that’s what the Valley loves. When blockchain stuff scaled up.

Rich: Yeah.

Paul: They were like, okay, now we’re going to have our own Wall Street. We’re going to be our own place.

Rich: We’re going to disintermediate banks, and we’re going to give you more power.

Paul: And they always want to start, like, they want to get a couple old oil tankers, strap them together, and build a, build a seasteading culture. Right? There’s always this attempt to, like, we’re going to—

Rich: The seed is always altruistic, right?

Paul: Yeah. But it’s like, we, society has to get away from regulation, so we can just do what we want. Mom, get out of my room.

Rich: Yeah.

Paul: And so what I think is this particular technology scaled so quickly, it was so vastly interesting, and so surprising, that it actually scaled past all of the Valley’s own limits, to the point that they can’t get enough compute.

Rich: Yeah.

Paul: They have to go to Saudi sovereign wealth in order to get enough GPUs to do their stuff.

Rich: Yeah.

Paul: We’ve never seen anything in our lives scale completely outside of the Valley. Even the internet really kind of stayed anchored for a long, long time.

Rich: Yeah.

Paul: This thing is global so fast. And so the system of, like, Y Combinator, Paul Graham, all that stuff really broke down around the edges. And this guy kept himself right in the middle of it, and he got caught because there’s, he’s not the president, he’s not elected. He went and genuflected to Trump. Like, it all starts to look real bad. The Valley looks bad.

Rich: Okay, but, fine.

Paul: Okay.

Rich: But he didn’t have to do that. Like, he didn’t have to assert that, like, well, I guess competition kicked in.

Paul: I think you got—

Rich: Time to be for-profit, that’s the game.

Paul: I’m gonna say this. I’m gonna shut up, because I’m not the guest. But, like, I think when you are, your entire life is embodied in coordinating humans and deal-making and you feel you gotta go to the White House.

Rich: That’s the next deal.

Paul: You’re just gonna do it. Like, I don’t think there’s any wiring that would—like, he’s fine. He’ll just go buy another car.

Andrew: Well, and, or, I mean, some of our reporting made it seem like some of his most expensive cars may have been gifted to him by Emirati shieks. So there’s that element. I mean, some of it—

Paul: Well, it’s like me buying you a soda.

Andrew: Totally, totally. Which, you know, it’s been a minute. You haven’t bought me a soda recently.

Paul: Come on, come on, you come on up to Union Square. I got you. We got Spindrift in the fridge. [laughter] We’re gonna have a party.

Andrew: I don’t want to just do the, like, you guys do antithesis/thesis and I have to do synthesis, but I actually do think that the synthesis is correct here, which is yes, a lot of these guys, Altman definitely included, seem very hardwired to do the next deal. That, I think, is on display several times in the piece. You know, most notably the, you know, Pentagon blacklists Anthropic, Altman’s ready to swoop in and do the deal. Right?

Paul: That’s how you get the billions. You don’t get—you can make money, but you can’t do that unless you are that guy. That’s how you get the growth.

Andrew: And what I would also say though, to differentiate this, which has to do with the other point Paul was making about how fast this has scaled and how fast it sort of escaped, slipped the bonds of Silicon Valley. Like, we have a quote in the piece of someone saying, the amount of capital you need for this is just hard to comprehend. There’s basically only five or six entities in the world that can fund this kind of growth.

Rich: Yeah.

Andrew: It’s basically the U.S. government, the four or five biggest tech companies in the world, the Saudis, the Emiratis. That’s it.

Paul: Sovereign wealth.

Rich: That is a boring cocktail party. [laughter]

Paul: Unless you’re bombing something.

Andrew: Yeah, unless you’re bombing a data center.

Rich: I mean, he’s chasing the realization, how do I make this, how do I fully realize this?

Andrew: Yes. And I think all these guys are, I think one sort of wrinkle in the comparison to Bitcoin or seasteading or any of these other kind of world-dominating sort of plans is again, like, in a lot of those cases, think what you will about the master plan, but there is a kind of statement of it that is pretty straightforward. Right?

Paul: Mmm.

Andrew: We want to get outside of, you know, international laws, so we’re going to try seasteading. We want to get outside of regulation. Right? They’re saying what they want to do. We want to replace fiat currency. There is a fundamental contradiction at the core of this, which you kind of sound, like, naive if you keep harping on it, because, oh, that’s old news. That happened years ago. But the fundamental contradiction remains, which is they did not go to people and say, we want to build AGI and we’re going to take hundreds of billions of dollars in order to do it, and we’re going to scale up these models, and then we’re going to turn around and have one of the most powerful businesses in the world. They didn’t say that. They said, this is not a business. This is an AI safety research lab.

Rich: Right.

Andrew: We forget it because we think, oh, everyone says stuff to make the deal happen, but this is really a difference of kind, not a difference of degree, in terms of what the pitch was. The pitch was fundamentally actually at odds with what it ended up being.

Rich: Yeah. Which led to, obviously, the attempt to remove him. And this was, the one part of the piece that jumped out at me is once he saw that leverage had tipped back in his favor and he was going to be put back in place? Boy, that sort of sheepish guy who’s really kind of sweet, like, boy, that mask came off. And it was just the final scene of The Godfather. Right? He cleaned house in a pretty aggressive way. And that, to me, was the most revealing thing.

Paul: I mean, and I agree with you, but what choice does he have? He comes back in, he’s been ousted. It’s humiliating. It’s bad.

Rich: Yeah.

Paul: And what are you gonna do? What would you do?

Rich: Well, it wasn’t that. I’m not surprised that he did it.

Paul: Okay.

Rich: I think, I found myself trying to reconcile that episode from how he’d been portrayed until then, which is like, hey, I aim to please. That’s kind of my nature and sometimes gets me in trouble.

Paul: But it’s not just that. We have to protect everyone in the world from a difficult, complicated future where robots come alive.

Rich: Yeah.

Paul: Versus I need to go talk to my good friends the Al Waleeds about how to build a data center in Jordan. I think that is a hell of a jump.

Rich: It is a hell—

Andrew: I mean, they keep using—so people have alternatives.

Paul: I mixed up my Arabs, by the way, and I work with a lot of Lebanese. I just need to be out in front of that.

Rich: You’re forgiven.

Paul: Back to you, Andrew.

Rich: Go.

Andrew: Well, this was actually something that came up is, like, when we kept asking so many questions about the Middle East, we were sort of met with like, “Oh, do you have something against the Middle East?” [laughter] And I was like, “It’s really not that. It’s just asking questions about all the yachts you’ve been on and how you’re getting your money.”

Rich: That’s really funny.

Andrew: Look, this, again, it’s like, yes, people say things. People say, don’t be evil, or people say, we want to connect the world and all that stuff. And yeah, we can, as savvy consumers of information, we can kind of read between the lines. The fundamental disconnect of this story of like, please tie my hands was so, it wasn’t an offhand comment one time. It was something that Sam Altman said in every room he was in for eight years. Under oath—

Paul: He loves regulation, is what he says.

Andrew: Yes.

Paul: He just loves it. Can’t get enough, give me more regulation, until it’s actually regulation time, at which point they start to get a little, real squirrely.

Andrew: And make of this what you will, but it wasn’t just, you know, going before the Senate, you know, in 2023. The first email he sends that we have a record of to Elon Musk raising this idea in May of 2015, says, “We need a Manhattan Project for AI—” I’m quoting from memory, so I may be getting this wrong. “—that will aggressively comply with all regulation.”

Paul: Yep.

Andrew: And if you take the Manhattan Project part of that even remotely seriously—there are some people who don’t, and again, they think it’s just all hype and bullshit. But if you even take the possibility of that seriously, and then you imagine what it would be if Oppenheimer were like, “I’m running this secret lab in the desert. It’s very important for the following reasons. We need to beat the Nazis, et cetera, et cetera.” And then, you know, they try to kick him out of the lab, he comes back, and he goes into founder mode and goes, actually—

Rich: I’m going to sell them.

Andrew: I’m going to make some deals with the Japanese or whatever. [laughter] Like, people would have questions. Either you were not serious about your pitch then, or something is missing now. But, like, the contradiction is too big.

Paul: Well, we’re in this very funny zone where the ideologies that guided them, so much money showed up that we don’t hear about them anymore. We’re not hearing about effective altruism as much anymore, which was very much—

Rich: This is a question I have for Andrew is…

Andrew: Mmm hmmm.

Rich: And maybe this is, you probably observed it because it’s outside the purview of what you focused on for the piece. But it seemed to like just kind of coast—like, OpenAI gradually shifting from nonprofit to clearly a for-profit, kind of, the world kind of rolled with it.

Andrew: Mmm hmm.

Rich: And I’m wondering if—

Paul: It makes you a picture of a teddy bear astronaut. [laughter]

Rich: Which I—fine.

Paul: Yeah!

Rich: Fine. Do you get the sense—

Paul: You can’t do that at a not-for-profit. Have you ever tried to do that with a not-for-profit? [laughter]

Rich: I’ve never tried to do that.

Paul: No.

Rich: It’s, like, six weeks. Very true.

Paul: It’s, like, six weeks and twelve board meetings before you can see even one hair on the teddy bear. [laughter] Anyway.

Rich: Do you think the Valley, and maybe Altman specifically, sort of sensed it’s going here anyway?

Andrew: Yeah. I think that, look, we have people in the piece representing both of these views. The kind of, it was always a master plan, 3D-chess view. And the it was improvisatory checkers all along view. And I do think that it may be that nobody knows, including even Altman himself. Right? Because you can convince yourself of lots of things post-hoc and all that sort of stuff.

Rich: Yeah.

Andrew: I think to some extent, though, what’s incredible is how well it worked. Right? You don’t have—and who knows, maybe some of this will be unwound through this trial or maybe other things are afoot. But as of now, this sort of strategy to put yourself at the top of the hierarchy that nobody really had tried or thought of before.

There was one analogy—this ended up cut from the piece because we had so many AI analogies in the piece that we were sort of ditching them at the last minute for ballast. [laughter] But there was one that sort of gets at this that I kind of liked, which is, so one of the big watershed moments where the public understood that AI actually was heating up and it wasn’t in another sort of research winter was AlphaGo beating the world Go champion.

Paul: Mmm hmm.

Andrew: One of the ways that happened, this was broadcast live on Korean TV, and there was this move called Move 37, where AlphaGo makes this move that was so outside the realm of what any human would ever contemplate doing that all the commentators just thought it was a mistake. They’re like, okay, finally the model is glitching out. We finally hit the limits of what the model can do.

Rich: Yeah.

Andrew: And then only in retrospect, after the model beats the human, could they piece together how that was actually the winning move. It was so out there, it was so creative and aberrant that Lee Sedol couldn’t figure out what was going on until it was too late.

Paul: Mmm hmm.

Andrew: Now there are people, and I’m not going to say conclusively that we know for sure, but there are people who think that these eight years of Altman being the “please regulate me” guy and the “I’m so scared that this technology is going to kill everyone on Earth guy,” that that was his move 37. That it was at the particular juncture of a Congress that was tired of giving carte blanche to tech CEOs, a public that felt really burned by social media in particular, a recruitment pool of high-quality AI researchers who were intensely, disproportionately scared of the thing they were building. And then investors who, according to our reporting, Altman could still go to and say, like, “Trust me, we are going to do fine here.” Like to the point where there was a little purple box on the investment sheet that said, “We’re going to cap your investment at 100x because otherwise the rocket ship growth will be too intense.” The fact that all of those pitches were made simultaneously, even with their mutually contradictory elements, is something that I do think is sort of historically unique.

Paul: But to me there’s like a nerd Occam’s Razor that kicks in, which is this is, this is kind of typical nerd behavior. You believe in the deep—you’re the very spiritual person when it comes to technology and you really, you really feel that, you know, super-intelligence is right around the corner, but you also just really love deals. I think that there’s a maturity of strategy in what you just outlined that I actually don’t see anyone possessing. Because when we’re on the other side of it, the execution was pretty flawed, right?

Andrew: Mmm hmmm.

Paul: It just—this enormous thing showed up and they went, oh my God, there’s so much money to be made. And then my thesis is just like, okay, now Dario left and now I gotta compete with that. And it’s a little baby, but he’s pretty smart. I gotta keep my eye on it. And then Google, who came up with the cool technology that actually turned out to be the one that’s going to make us all the money, they’re going to come eat us at any moment. I got to stay here, aligned with Microsoft.

Rich: Here we are, right?

Paul: I got to stay aligned with Microsoft. But they’re kind of annoying and I just sort of like, I see it, I can see it bubbling out of a sort of generic Veep-style incompetence as much as I can see it coming out of a more, like, slightly conspiratorial, strategic, move 37 vibe.

Andrew: Totally.

Paul: I favor incompetence because it fits to my own view of the world and myself. But maybe they are that strategic.

Andrew: Well, and look, just, I think totally to echo what you’re saying, nobody is the villain of their own story. Nobody is twiddling their mustache and saying, “How can I do something bad?” And so I don’t think the conspiratorial—look, let’s put it this way. Even in the move 37 analogy, it’s being done by something definitionally without an organizing sentience. Right? So it’s not saying this is this, you know, evil supervillain thing to do. It’s just saying this is a move that people haven’t really tried before and it turns out to be a successful move. But that’s not at all…

Rich: There’s no end game.

Andrew: Yeah. It’s just. It’s not at all mutually exclusive with, I’m going to try to make deals and see what works.

Rich: I think you’re nailing it, Andrew, which is, I think, look, the truth is, the one thing we could say conclusively is out of all the people he, all the people he courted, the team, he would not have been able to assemble that stellar team if he had told them, “We’re going to make gobs of money. And there might be some consequences.” Like, there’s just no way. Right? You nicely draw out these personalities that were deeply concerned and felt a moral obligation to not just do whatever. And—

Paul: I mean, in this way, he, I don’t buy his—he wants to be the next Steve Jobs. I don’t fully buy that, but that was Jobs getting, like, the Macintosh team together. We’re going to, we’re going to put a pirate flag up in this one building, and we’re going to change the world by taking everything we can from Xerox.

Rich: Look, man, master plan I don’t know about. What I do know is he goes into a meeting with an outcome in mind, and if that outcome has to drive a particular narrative, he’s coming in with that narrative.

Paul: I think that’s real. Right? There’s an extremely high—

Rich: And I think it gets compounded.

Paul: There’s a very high likelihood of that outcome, if he goes in.

Rich: In the meeting, he knows that. And so I think—

Paul: There’s also, there’s the motive force of someone having $20 billion. It tends to line people up.

Rich: He tends to kind of keep that momentum going.

Andrew: There is also, I mean, look, I think you guys know this as people who are in tech but are not, you know, just maximizing your own sort of personal wealth at all moments. There are really people in this world who are constantly getting gobs of money thrown at them and turning it down. And so we have records in this piece of, you know, just, Ilya Sutskever is one example of, you know, Google wants to keep him. They give him a $6 million counteroffer, and he says, no, I’m going to go work at the pirate ship good guy shop, which is OpenAI, and make a fraction of that. And look, it’s not at all—

Paul: That’s a good point, though, because when. When it all starts to melt, he’s like, well, what have I done?

Andrew: Right. And one of the things, you know, you were asking before about were there, you know, claims of Altman’s that, you know, we were taken in by or found persuasive or whatever. I think it’s totally plausible, you know, Altman told us and has told other people he’s not primarily motivated by money. I think that’s totally plausible. There’s no way for me to know. We don’t come down on it one way or another. But I think it’s important to remember that there are people in the world who are not primarily motivated by money. Now, they might, you know, be motivated by power or something else that we might find to be suspicious. But I do think that there are clear indications, proof, I would say, of people in this story who are not maximizing their own sort of personal optimization function at every moment.

Rich: Yeah.

Andrew: The question is like, what are they optimizing for?

Rich: I have a fun closing question.

Paul: All right.

Rich: Is Musk the good guy in this?

Paul: No, don’t do it.

Andrew: [laughing] I think a very clear, so we’re all going to be watching the trial, or I am going to be watching the trial for the next few weeks.

Rich: Yeah.

Andrew: There’s a lot that has come out in discovery. A lot of it is in the piece, obviously not all of it, that does make it look like Musk has an argument when he says that he was kind of hoodwinked out of his money. Now, how much money and how willing he was to be hoodwinked and how much his concerns were more sincere than other people’s concerns, these are all questions now for a jury to decide. And I… But one thing I will say is, you know, if, I think it would be a deep mistake, and I hope readers are not making this mistake, if they walk away from this piece thinking, you know, I don’t know about this Sam Altman guy being AGI dictator. I really think AGI dictator should go to Elon instead or to Dario instead [laughter] or, you know, I really think the AGI dictator should be Demis.

Rich: Yeah.

Andrew: Like, the whole, again, this whole structural thrust of the piece is the fact that we’re throwing around concepts like “AGI dictator” with even a 5% shred of possibility that that has meaning attached to it in a real sense is insane.

Paul: I will say Amodei is starting to, like, they’re putting some money into actually lobbying for regulation against OpenAI. So, like, I will, he’s put at least a little bit of juice into…

Rich: Yeah,

Paul: But I mean, you’re right—

Rich: Andrew’s point is no one person should hold those keys.

Paul: No, my God. This is banana cakes.

Rich: Yeah.

Paul: I mean, look, I have a certain amount of empathy reading the story. It’s not that I’m like—I actually don’t particularly ever want to interact with that guy. Like, Valley people make me very nervous. And he’s probably a really good persuader. So God help me if you, and I’m a gullible human. So God help me if he ever wanted anything I have. [laughter] But if you were suddenly sitting astride something and you look at it and Bill Gates says, “That’s the most impressive thing I’ve ever seen.” And you have been riding this for your whole career, and you are a very good boy, and you see a trillion dollars in the road in front of you? It’s very, just genuinely confusing. Like, even someone who’s good at deals, even good at structure, even a president would have trouble dealing with the level of attention and focus and drama.

Rich: Yeah.

Paul: And there’s no infrastructure, because they’re not government. They can’t regulate. They barely, really, they kind of live in very much a bubble.

Rich: Yeah.

Paul: And I think that you just end up in this situation where they don’t, no one has the skills for dealing with the change that they are bringing about.

Rich: Yeah.

Paul: And it keeps blowing up in everybody’s face.

Rich: It’s a pretty common technology runaway train pattern. Right? It’s happened a few times.

Paul: But never at this scale.

Rich: Never.

Andrew: Not at this scale, not at this speed, and it does make you think that some of that rhetoric from 2019, 2020, Sam Altman saying, “Someone really ought to stop me from having this amount of unchecked power,” seems like there’s something to it.

Rich: Yeah, yeah.

Paul: I think that was a fun thought experiment. But then, after you take that sovereign wealth, you gotta show something on the balance sheet. And what are you gonna do?

Andrew: Yeah.

Paul: What are you going to do?

Rich: Get me those returns.

Paul: Yeah, exactly.

Rich: Give me your own personal sentiment on AI and what the—you went into the belly of the beast here. Like, these are people that are really shaping something profoundly important in our future.

Paul: And your job is to put words into sequence, making sure they’re backed by facts in a way that is exciting and interesting to an audience.

Rich: How are you… Well, your sentiment.

Andrew: Well, I mean, so to the latter point, I try and I think succeed at not doing the thing that a lot of these tech people often accuse us journalists of doing, which is to just be motivated by pure pique and outrage that they’re coming for our jobs. Like, I don’t feel, I’ve sort of searched my soul on this because I used to get this critique when I would write critically about social media as well, you know, “Oh, you’re just salty because we’re, you know, disrupting your dinosaur industry.” And I was like, I think you are doing that and I think that’s bad. But I think I believe those things independent of whether I think you’re coming for my specific job.

Paul: Oh, I like when they just say clickbait over and over while hitting the Reddit post button. It’s just great. Yeah, okay.

Andrew: Exactly. So I don’t think that I’m sort of just lashing out at my sort of tech overlords because I’m worried that they will replace me. I do think that… Look, a lot of times with some of, especially the more far out, kind of sci-fi, Skynet-y kind of arguments, I would go in kind of, I think, tacitly assuming, like, okay, there will be some obvious seam I can pull at here logically that will make this whole thing fall apart. And if I just sort of learn enough of the technical jargon, I’ll be able to see where the logical flaw is.

And a lot of times I was like, you know, I don’t, you know, I’m not going to end up with a kind of (P)doom of 99.9% and I’m not going to, you know, sort of cash in my—I don’t even know what one does at that point. But I was, I didn’t end up there. But I think I ended up thinking, okay, what would it mean to go from a null set of, like, just not even knowing what the gravest arguments are to knowing what they are and not being able to offhand refute them. And that does feel different.

And that goes for the literally existential things. That also goes for what if big chunks of our transportation infrastructure and energy infrastructure are given over to generative AI bots that we don’t fully understand? That goes for all kinds of things like human enfeeblement and AI slop everywhere. Those things do worry me, in a way that I don’t have an easy solution for. I don’t have, like, if you could give me a magic wand and make me a 60-vote, filibuster proof Senate majority, I would know how to solve. I don’t know that it’s even within the realm of government to solve at this point.

But I do take seriously the fact that this stuff is moving fast enough that even its creators don’t really know what it’s capable of. And that that’s not something that we can just sort of wish away by saying, oh, those sci-fi nerds, they’re just trying to get our attention. Like, I think they have our attention for a reason.

Rich: Yeah.

Paul: You ever seen these guys get interviewed by Congress? Congress, it’s amazing, because it will be like, Congress, they’ll be like, [kindly grandpa voice] “Will you come help us regulate it?” [laughter] Like, they’re just… “You’re such a good boy.”

Rich: It’s a rough scene.

Paul: Oh, it’s terrible.

Rich: It’s a rough scene.

Paul: Let me leave us with this, because I was reading this piece and as I got to the end, I was like, this is good. It helps me understand something. And then I had this thought: 40 years to go, at least. He’s not going anywhere. He’s got billions of dollars. This person is going to be part of our lives for the rest of our lives. And I just want you to react to that. You ready?

Andrew: Well, I mean, there are lots of people working there who don’t think we have 40 years left. So that’s encouraging in a way. [laughter]

Rich: Ah, lord.

Paul: All right, damn it. That was a good setup then. Good for me, because that was beautiful.

Rich: You guys collude on that one?

Paul: No, that was really, you’re absolutely right. Many people do not, that is probably—and if anyone from that world is listening when they heard me say 40 years, they’re like, “We have like six months, man.” [laughter] What are you, what are you even talking about? Andrew Marantz. This has been—

Rich: Great discussion.

Paul: Edifying. The piece is great. People should go read it. And that’s really it. Like, it’s good to sit and read. Maybe even on paper, you could print it out and have an experience where nothing, the words can’t talk back to you. I think that’s good.

Andrew: Mmm. Totally. We even have a service where we print it out for you and ship it to wherever you are. So…

Paul: Oh, my God.

Rich: Stapled?

Paul: No, it’s—I do it. I do it, and it’s, I read every single word assiduously, and they never pile up.

Andrew: Totally. That’s the thing, well, what I will say is, do not worry about the pile. You don’t worry about the pile of Netflix you haven’t watched. You don’t worry about the pile of…

Rich: I like that.

Andrew: Just don’t—

Rich: No pressure.

Andrew: I absolve you of the pressure. It’s all good.

Paul: I needed that. I really did. [laughter] Because, you know, it’s right there, and then you’re just sort of like, I got it. Well, boy, that’s, that’s a lot about Iran.

Rich: I tried The Economist for a year. [laughing]

Paul: No, no, that’s, The New Yorker is The Economist for people who actually like to read. [laughter] You’re not… The Economist is punishing. It’s like something your father-in-law gives to you.

Rich: Andrew, this is great. Thanks so much.

Andrew: Thank you, guys. I’m coming in person next time for a Spindrift.

Rich: Amazing.

Paul: Oh, you can have two.

Rich: Welcome anytime.

Andrew: Awesome. Thank you, guys.

Rich: Take care.

Paul: All right.

Andrew: Cheers.

Paul: Thank you.

[outro music]