Investigating Altman With Ronan Farrow and Andrew Marantz

0:00This is an iHeart Podcast. Guaranteed human. Premier Protein. It's for getting after life, not just fitness. With 30 grams of protein, 160 calories, and no sugar added. Helping people fuel their joyful lives. With Premier Protein, you can say yes to more. Whether it's crushing a big presentation at work, building an epic fort with the kids, or hitting the hiking trail with friends. Premier Protein offers delicious flavors like cafe latte, chocolate, caramel, vanilla, strawberry, and cake batter to name a few. Find your favorite flavor at premierprotein.com. Here's another mouth-watering recipe idea from Marikon, the world's finest rice vinegar.

0:39Try a Marikon hot chicken sandwich with pickled cucumbers made with chilies, garlic, and the vibrant, zesty flavor of Marikon Genuine Brewed Rice Vinegar. Or go sweet and savory with Marikon Seasoned Gourmet. A flavor bomb for veggies and grilled proteins. Get the green label for genuine brewed or the orange label for seasoned gourmet. Then check out ricevinegar.com for more delicious recipes. Because the food you love is better with Marikon. There's a fire inside you you can't ignore. Stand still, not a chance. You're a lifelong learner who's come this far. Now we're here to help you keep going further.

1:17Capella University, what can't you do? Visit capella.edu to learn more. Ever wonder how to make hosting look effortless? Here's a secret. When prepping for cooking and baking, get ahead of the mess with new Reynolds Kitchen's countertop prep paper. Just lightly wet the counter so the paper grips. Lay it down and drips and spills stay on the paper, not on your counter. Cleanup is as simple as lifting it away to reveal clean counters. Effortless? It is. Thanks to Reynolds Kitchen's countertop prep paper. Wet it, set it, prep it, done. Available in the Reynolds Wrap Aisle at Walmart. CallZone Media Hello and welcome to Better Offline.

2:06I'm, of course, your host, Ed Zitron.

2:20No monologue this week. Quit complaining because I've brought you all something far better. I am joined by Ronan Farrow and Andrew Marantz of The New Yorker, who just put out a massive piece about Sam Altman, the dampest man in Silicon Valley history, which is my comment, not theirs. And I think I want to thank you for joining me, both of you. First of all, I want to ask a really just blunt question, Ronan, considering your history of the people you've reported on why sam altman what drew you to him well this is fundamentally different than stories i've done that started with you know a single clear-cut lead of about criminality this was a situation where i was coming off of reporting you may be familiar with about elon musk and his whole empire and i had a lot of sources in the world of ai including sam altman who was on the record in that piece.

3:14And, you know, this was no novel, a genius projection on my part, obviously, it was at a moment when AI was becoming the kind of fulcrum of Silicon Valley. And it became apparent to me that there were underexamined questions about the trajectory of the technology and the integrity of the stewards of it. We really started looking at that with an open mind. You know, Andrew is on this beat of big tech all the time and is also well-sourced up in this world. And we both, I think, sat down and were having our conversations about what the most salient points of contention were in this world. And we're equally open to criticism about all of the labs, all of the figures.

4:07This question of why Sam Altman was fired what the specific alleged proof points were, and whether those questions were something we should care about. Were they right to care then? And should we still care now? Are those problems ongoing? Those really emerged as just a lingering, unanswered set of questions that hadn't been, you know, fully proved out and backed up in this kind of a forensic way. right so why was he fired like what did because the piece kind of got into it but it's all over like the answers are all over the place from everyone and each person seems to have a different interpretation andrew you want to start yeah i i think one of the abiding um frustrations and curiosities from this is the whole world saw sam altman get fired and then the whole world watched as he returned and that was never really adequately explained in the public narrative um and so we really go into with sort of more documentation, I think, than ever before, exactly what was going on.

5:15And as Ronan says, you know, it was not this sort of one smoking gun thing. Like it wasn't, you know, oh, you know, Sam Altman was strangling baby puppies in the office and nobody knew it until now, right? This is a much subtler, slower accumulation of what his critics say are manipulations, telling two different stories to people. The idea of telling different people what they want to hear comes up a lot. And so this is the kind of thing where you kind of need a New Yorker story, I say sort of self-interestedly. But I do think that you kind of need to put all the facts in one place for them to make sense.

5:54And there have been really good books about this. There have been a lot out there. I mean, we're obviously not first to this story. But I just think it's hard to understand what the allegations even are without lining it all up in one place. And you kind of have to centrally start from this premise of Sam Altman did not go to everyone and say, AI is going to be a really big deal. It's going to be a product we can try to sell. It's going to be a good investment pitch for us. Let's all go and make a bunch of money. What he was saying was, you guys, you engineers, you need to come along with me and build this technology with me specifically, because I will be circumspect, I will be safe, I will argue for regulation, I will keep it in this non-profit AI safety research lab format.

6:38And it was the betrayal of those promises that they felt was the fireable offense, more so than any one smoking gun thing. Except he was kind of brought back by capitalism. I remember when it happened, there was a journalist posting that they were crying with happiness about it. It was genuine. One of the darker moments I can think in Silicon Valley history, because it's like, wow, Brian Chesky and Satya Nadella got together and held hands and got that Sam Altman back up. It was just so bizarre. It is weird. And it speaks to, you know, this thing, again, that people consistently allege in the documents we saw and during our interviews of this sort of telling different stories to different communities of people.

7:21I mean, one of the most basic examples of this, right, is turning to the safety nonprofit people and saying this will always be a safety nonprofit and then turning to the VC investment community and saying, let's go, guys, rocket to the moon. Let's go make a bunch of money and kind of simultaneously holding both of those communities seemingly in the palm of his hand until he wasn't able to anymore. You're exactly right to frame it as capitalism winning. And I think we were fascinated in this case of what's called the blip, you know, this brief firing and then Aldman undoing the firing because it does feel like it is about much more than Sam Aldman.

8:00The thing that OpenAI originally was when it was founded as a nonprofit and in the way Andrew described was pitched on the basis of these fears from the people developing the technology where the founders of this company were saying, listen, this is so dangerous that we're comparing it to nuclear weapons. and we need to create a noble-minded research lab without a profit motive. They recruited on that basis. People took pay cuts to go there on that basis. And this moment when the board that was empowered to remove a CEO if they felt that CEO could not be trusted with that mission, when they tried to do that and basically investors who really didn't know what the hell was going on at the time, That's one of the things we document.

8:51The board really fumbled the ball. We can talk about how. But the long and short of it is that those investors rallied around Sam Altman, who really stood to make them a lot of money. And they were worried that their money would go away if the company fell apart. And so it is a part of a bigger story about the way in which Silicon Valley builds companies based on hype and at least not solely real value and where the initial promises very often don't pan out. And what winds up governing is, I think it's valid when people in this piece have the criticism of, you know, we're now seeing a race to the bottom on the very safety issues that OpenAI was supposed to be founded around.

9:43And the blip was a turning point that encapsulated that. yeah with that in mind do you buy that sam walton gives a shit about safety do you actually think that the safety mission matters to him at all you've spoken to him a few times i think when he talks about still caring about and prioritizing safety look maybe i i am just um kind of too generous in my assumptions about people i assume that he believes the story that he is telling himself, which is very often has the same contour, namely what he cares about is winning. That's the word he uses. And he has set up the rationales in his mind and in the way he conveys it to others so that him winning and consolidating as much power as possible is the same thing as humanity winning.

10:37So for him, when he talks about these things, there is no daylight between them. You know, he says in the piece that his definition of winning is a crazy up level for everyone. You know, we're all going to be curing diseases and forming blockbuster startups. And I'm really, you know, paraphrasing, but barely exaggerating. He talks about it in these terms. And I think he really does think that him controlling the technology is the most direct path to that. He often says in conversations that accusations are the truest form of confessions or variations of this thought, you know, that when people launch criticism about someone else, they're very often projecting something about themselves.

11:24And it's interesting because in the course of my Elon Musk reporting, he talks about Musk in this very specific way where he says on the record, you know, I believe Elon is interested in saving humanity, but only if he's the one who can do it. And I thought about that quote about Elon a lot as I was in my conversations with Sam. I mean, the other thing I would say on this safety question, right, is a lot of these terms, safety, AGI itself, superintelligence, intelligence itself, these all are vague enough terms and they're poorly defined enough that there can always be this kind of slippage and equivocation.

12:06And sometimes that's unavoidable, but sometimes people can use it to their rhetorical advantage, right? So on questions of safety or alignment, what you would often find is that the Sam Altman of 2015, or even frankly, 2023, would be talking about the alignment problem as a literally existential problem. If we don't solve this, a rogue AI could kill literally every person on earth. And then you go back to him two years later, and he's saying, the alignment problem is this slightly annoying thing that kind of tempts you to, you know, spend more time on ChatGPT the way that algorithm, you know, Instagram's algorithm is out of alignment with the way you want to spend your time.

12:42That is absolutely not what he meant by it two years prior, but the word is the same. So he can kind of skate by with that. I mean, this is in the nicest way. Why do you believe him two years ago, but not, but like, what do you think he's changed? Do you think he just is talking differently? Like that's, I'm not accusing you of it. I'm just like Sam Altman's whole vibe is to me personally, just saying quite deceptive. So I'm just, I'm just wondering what you've seen. I'm not saying that I believed him. I'm saying that his employees believed him. Congress, to some extent, believed him. The public.

13:13I think actually that's a very important point, Ed. I think your analysis is correct. And actually, the perspective from which we write the piece is these have always been shifting assurances. And actually, part of what we emphasize and document is that even at the time of those older assurances where Sam Altman was the biggest doomer in the industry, he was saying and doing conflicting things at the same time in many cases. We document through a lot of internal communications a period of time in which many researchers were joining the company on the promise that this would forever be a noble nonprofit.

13:55But simultaneously, Sam Altman and Greg Brockman, his co-founder, were having very concerted conversations about how quickly and effectively they could get out of the nonprofit structure, essentially. We talk about cases where he's saying on the Hill, we need regulation. We need to start even a new government agency to regulate AI. And then simultaneously, he's working against regulations. Well, just to bolster the point you're making, what really does seem clear is that whether they should have believed it in retrospect, a lot of these early employees, early recruits, early co-founders really do seem to believe it.

14:31I mean, unless there was this kind of, you know, long game where they were like, we're going to write these emails to each other in 2015. And then 10 years later, they'll come out and it'll all look like this was, I mean, I guess that's possible, but from all appearances, it really seems like at least a lot of the key people really did believe this pitch at the time.

14:57Premier Protein. It's for getting after life, not just fitness. With 30 grams of protein, 160 calories, and no sugar added. Helping people fuel their joyful lives. With Premier Protein, you can say yes to more. Whether it's crushing a big presentation at work, building an epic fort with the kids, or hitting the hiking trail with friends. Premier Protein offers delicious flavors like cafe latte, chocolate, caramel, vanilla, strawberry, and cake batter, to name a few. Find your favorite flavor at premierprotein.com. Here's another mouthwatering recipe idea from Marikon, the world's finest rice vinegar.

15:32Try a Marikon hot chicken sandwich with pickled cucumbers made with chilies, garlic, and the vibrant, zesty flavor of Marikon Genuine Brewed Rice Vinegar. Or go sweet and savory with Marikon Seasoned Gourmet, a flavor bomb for veggies and grilled proteins. Get the green label for Genuine Brewed or the orange label for Seasoned Gourmet. Then check out ricevinegar.com for more delicious recipes. Because the food you love is better with Marikon. This is Martha Stewart from the Martha Stewart Podcast. Ever wonder how to make hosting look effortless? Here's a secret. Getting ahead of the mess with new Reynolds Kitchen's countertop prep paper.

16:12Just lightly wet the counter beforehand so the paper grips and stays in place. Then lay down the Reynolds Kitchen's countertop prep paper so drips and spills stay on the paper, not all over your kitchen counter. You can roll out dough, prep a party spread, or cook alongside family. When you're done, cleanup is as simple as lifting the paper and revealing that clean counter underneath. Effortless. You can use it for cooking and baking prep, and even crafting, especially when you need extra working space. because when the mess is already handled, you can focus on what matters, the food, the people, and the moment.

16:54It may look effortless, but now you know it's Reynolds Kitchen's countertop prep paper. Take a tip from me, wet it, set it, prep it, done. Make it easy. Make it with Reynolds Kitchen's countertop prep paper. Available now in the Reynolds Wrap Aisle in Walmart. There's no denying it. Moms make the world go round. Celebrate mom with a gift that really pops. A personalized Funko Pop, where you'll select hair, clothing, and skin tone options to match her go-to look. Don't forget to show off her special talents, hobbies, and interests with accessories to match. You can even add her favorite furry friend.

17:32Show mom the love this year. Start creating now at Funko.com. Be unique, not anyone else. Express it now. Just pop yourself. That's Funko.com.

17:47Although my way of reading it personally is also, it felt like a lot of rich guys kind of farting around. It was just the conversations they had back and forth, just the banality of it all. Just like, oh, what if this happened and this happened? I believe Ilya Sutsukeva, he seems to, however people may feel about him in general, he seems to at least buy this. The Altman Elon Musk emails were very much like uh two guys at an airport bar after a few waiting for their flight it was very very interesting watching watching this unfold over time because the modern sam altman is very much more like a tech ceo capital t capital c like he's very much a i need like maximizing profit and growth and the like you said the safety stuff kind of seems just like on the side now it's more important that we just get this thing public?

18:36Yeah. Well, and I mean, it's sort of like two guys at the airport bar, except that the hundreds of millions of dollars of investment they're talking about actually are real. I mean, you know, so it is, I think you're right. There's a kind of combination of things going on here. And I would be prepared to believe that in that particular case of Musk and Altman emailing each other, that there is some kind of gamesmanship or brinksmanship going on and that even what they say in private might not be fully what they mean. But when you get outside of that tight of a circle. When you talk about the early recruits to OpenAI, the later recruits, the people who took massive pay cuts to, you know, give up$6 million salaries at Google to come work for them.

19:14Clearly, this pitch of We're the Good Guys was doing some work for those people. And I also think that we should keep in mind how effective the public pitch was. I mean, coming on the heels of Mark Zuckerberg and, you know, all these other, you know, sort of social media tycoons going out to the public and saying, you should be so grateful for these, you know, convenience machines that we're building and you should, you know, give us all your data and you should feel bad anytime you want to regulate us. Again, Altman's pitch was totally the opposite. It was, we're going to be conscientious. This is so dangerous that you actually have to regulate us.

19:46And we are just begging for our hands to be tied. Again, whether in retrospect, that seems like a genuine pitch or not is sort of separate from the question of how effective it was. So let's talk finances. Within the piece, there wasn't a ton of stuff about money, But there was this quote from a senior executive at Microsoft who said, I think there's a small but real chance Altman is eventually remembered as a Bernie Madoff or Sam Bankman free level scammer. Did you what kind of financial stuff did you hear about OpenAI, if anything? I mean, this is all very far out there already. We have a board member in the piece talking about how the company is levered up in a scary way right now.

20:27I mean, this is a company with one of the fastest cash burn rates in the history of startups. The spend level is almost unimaginable in the risk towards AGI. And so it really does require, it's almost not an exaggeration to say, all the money in the world. And this governs a lot of the different activities that we write about in the piece, right? Sam Altman for years courting Middle Eastern money. Everyone in Silicon Valley is courting Middle Eastern money and sidelining concerns about working with autocrats to various extents. Sam Altman is doing it on a unique level, agreeing to and catalyzing the building of massive infrastructure, even including in the Middle East.

21:16And over the objections, we talk about concerns that short-circuited a security clearance vetting process because there were so many red flags in the national security establishment about these dealings with the Middle East. We talk about, you know, gifts and trips on yachts with UAE officials. So, you know, this is a drive for more and more and more. And we talk to, you know, startup experts and economists who say one of the consequences of that is a whole lot of circular deals. None of this is new either, but this is a situation where it is the general Silicon Valley archetype of building companies on hype and promises before actual value, inflating a massive balloon of valuation.

22:08And it's putting it on a far grander scale where there's a whole ecosystem of partners that are just borrowing and borrowing and borrowing from each other on the assurance that they're all going to buy each other's products. It's an I'll scratch your back, you scratch mine situation over and over again. In some cases, Altman is running into problems because there are conflicting or at least allegedly conflicting deals being announced on the same day. We talk about tensions in the Microsoft relationship, particularly around this recent Amazon deal. Yeah, that was that Frontiers? Yes, it's to do with the exclusivity having been reaffirmed for Microsoft in terms of the underlying stateless models.

22:54And then OpenAI announcing, well, now at the same time, in the same day, well, we have this additional deal on top of that with Amazon for our enterprise products that allow the building of agents, which they're saying is going to happen in a stateful environment. And that's going to be because they're creating new technology that will deconflict this with the underlying stateless models from Microsoft. The long and the short of it is a bunch of people at Microsoft feel this is not technically possible. And you have to rely on the underlying stateless models to create agents that are stateful, essentially with memory.

23:27That's a that's a lot of technical jargon to get at the point, which is there's a slew of deals where critics say they're circular and someone's going to have to pay up. maybe a lot of people are going to have to pay up sam altman's one of the people who has said we're looking at a bubble and and that a lot of people are going to lose a lot of money which is it's funny though whenever he talks about it it's never him never like it's never like i'm gonna lose i will be but it kind of gets back to what you were saying it's like the only one who can usher in this ai future is sam altman the only person who will not get destroyed in the bubble is sam altman makes me wonder if he's like a fantasist or a narcissist or a sociopath.

24:07It's like, I think at least one of those words was in the piece. Those words are definitely in the piece. And not from us, to be clear. Yeah, yes, yes, yes. These are quotes from people. A lot of people who have worked closely with the man. But it's remarkable. You're rarely on a story, I don't know about you, Andrew, where so many people use the word sociopath unprompted by us. And yeah, and look, I mean, again, I think it's very, I think readers should come to this with a lot of skepticism about how much of this is kind of industry competition. And we tried to not, you know, let people launder talking points through our story, right?

24:42We're very, everyone in this world is kind of a ruthless competitor, right? So it wouldn't be a very high bar to say, you know, I talked to this guy's corporate competitor, and they called him a sociopath. Like, it's not that kind of criticism that we're talking about. We're talking about these very long standing kind of documented patterns that people can dispute, but it's not just kind of these off-the-cuff observations. I think to your point about the bubble stuff, like your audience will be very familiar with this because I know you cover this stuff a lot. But this question of is it a bubble or is it actually a very useful or even potentially economically transformative technology is not really as clean a binary as that because a lot of the biggest bubbles in market history have come from infrastructure projects that ultimately got used, right?

25:28I mean, this is a kind of basic economic history point, but I think it's worth reiterating. The railroads, the canals, the even fiber optic cable, all of this stuff eventually gets used. The question is, does the speed at which the investors are building it match the usefulness in real time? Yeah. And I mean, I would debate that point and have done so many times because GPUs depreciate in three to six years and there's no way that this gets used afterwards. But putting that all aside, it is still this weird thing of, and it's something in my own work I've noticed it's weird we're discussing this still like it's weird that we have no clarity because even within your piece we don't really have clarity about what is going on like that very the one thing that stood out to me like they're still trying to work out a business model that whole thing about pitting world powers against each other was truly insane like the Greg Brockman you'll have to run through it again but it was like it as I understood it it would be Greg Brockman suggesting that they make something powerful enough to sell to different nations and create a bidding war like Mission Impossible, except with a chat bar.

26:33Yeah, yeah. This is, again, you know, like you say, there's a lot of kind of guy at the bar speculation going on. And so this is in the early days when, because OpenAI back then was a nonprofit whose legally binding fiduciary duty was the safety of all of humanity, they hired someone to come in and say, okay, how can we deploy this, you know, potentially dangerous thing safely? And this person, the way it was told to us, came in with all kinds of, you know, 50-page, you know, detailed white papers about how we can model it on the Baruch plan and the way that, you know, the nuclear weaponry was first brought safely into the world.

27:09And the way that that was told to us again was that that plan then somehow morphed into, okay, instead of doing this, you know, kind of safe international game theory, non-zero cooperation thing, why don't we just, you know, start an auction between China, Russia, the US, maybe a few other world governments and see who either wants to be given the technology or who can be sold the technology. Now, obviously, this is disputed. And also, even more obviously, this is not something that happened. But the fact that that kind of idea was being batted around, and apparently was taken outside the walls of the organization and was pitched to potential investors, it really does speak to how volatile.