Transcription
So, it looks like there's a book coming out that details the drama behind the firing of Sam Altman, all the way back in 2023. And it kind of shines a light onto what happened behind the scenes. I feel like there are a few points here that we haven't heard before.
The story interestingly begins with Peter Thiel. I actually was not aware that he was involved in any way, but now that I think about it, he's always somehow involved in some way, it seems like. Right now, a lot of people look up to Peter Thiel, especially in the tech industry. They follow his advice; he is a mentor to a lot of people. He was part of the PayPal Mafia, along with Elon Musk. So, if you're not familiar, Peter Thiel is kind of the dawn of the PayPal Mafia, right? Sort of like the head boss, the Godfather, if you will. I will not do a Godfather impersonation; let's just continue.
We also have Max Levchin, the CTO at PayPal; Elon Musk, of course; David Sacks, so he's on the All-In podcast and now the President's Council of Advisers on Science and Technology; we have Reid Hoffman of LinkedIn, and many, many others.
Now, interestingly, Peter Thiel runs Thiel Fellowship, where he gives some startup funds to people that have certain ideas. His goal is to find kind of the best and brightest, and if they wanted to skip college and instead go and build their idea, he kind of helps them do that. And his track record for finding talent in the tech space is kind of astounding. So, it's aimed at students aged 22 or younger, gives them a total of $100,000 over two years, as well as guidance and other resources. There are only 20 to 25 fellows selected annually.
So, some of the power player alumni of the Thiel Fellowship are, for example, Vitalik Buterin, founder of Ethereum and also a self-made billionaire. See if you can spot a pattern in all of these names. We have Dylan Field, the wealthiest person to go through Thiel's Fellowship, so he co-founded Figma, a challenge to Adobe's Photoshop. Looks like Figma got acquired for $20 billion, making Field a billionaire. We have Austin Russell, self-made billionaire at age 25. We have Joshua Browder, so he apparently started DoNotPay. Lucy Guo, so looks like she's worth $440 million. This is Lura Deming, so she was in the first sort of Thiel Fellowship in 2011. So, she started working at a lab at age 2, was enrolled at MIT by 14, and decided to drop out at 16 to pursue Thiel's Fellowship at 23. Her Longevity Fund raised $22 million.
Thiel apparently also mentored, or used to mentor, Sam Altman. In fact, in November 2023, right the same year, the same month that Sam Altman is fired from OpenAI, Peter and Sam are sitting and enjoying a fine dining experience in LA's Arts District.
So, by the way, so this book that kind of originates a lot of this, it hasn't been published yet. It sounds like it talked to a number of people that were kind of insiders that went through this, so we don't yet know all the sources quite yet. I believe the book launches in a few weeks. Meanwhile, there are multiple news outlets reporting on this, so I'm kind of just piecing everything together. So, kind of keep this in mind: I'm not saying any of this is true. I am saying that this has been published by various outlets; we're kind of just beginning to put all the pieces together.
So, back to our conversation at the Arts District. So, Peter Thiel is talking to Sam Altman, and Peter is worried about another AI-obsessed individual that he took under his wing, that he's helping to fund his institute that is largely concerned with AI alignment, AI safety. Now, this is about a year after the ChatGPT moment, the release of the ChatGPT app, so people have had time to sort of at least become familiar with it. So, this is where talks of AI safety and stuff like that, they're already taking place; it's in the news. And guess who that AI-obsessed prodigy with the AI safety institute is? Who that person is? It's none other than, yes, it's Eliezer Yudkowsky.
Here is Mr. Eliezer Yudkowsky. Uh, he decided to GIF himself as well, which, uh, my hats off to him. The Ghibli Studio style of things have been going kind of viral recently, as I'm sure you know, driven by the new ChatGPT, the GPT-4 model being able to produce native images. But Eliezer is the original AI alignment person. He is sort of, when people talk about AI doomers, he's kind of seen as the main person in that discussion. He's sort of like the poster boy, I guess, if you will, for that movement. If you haven't heard of his interviews, there are two great interviews that I've listened to; I can definitely recommend. One is the Lex Fridman podcast, number 368. The other one is Dwesh Patel. So, this is, if you've ever seen kind of like this thing where he's like pointing at him going, "Miss a line, miss a line," that's where that's from.
But I think to sum up some of his views in a nutshell, and I think he would agree with me, is that superintelligence just kills you. And I don't think that's hyperbole. I think he's literally said that to people that are saying, "Oh, imagine all the benefits of superintelligence." He might reply with something like, "Superintelligence, you know, basically just kills you." And I think he does a great job of explaining sort of all the risks, all the potential risks behind developing AI. He strongly believes that it will turn out bad for the lives of humans. And he also plays a big role in the EA, the effective altruist, sort of movement. And of course, you might have heard of the website LessWrong. So, that was created by him, and he's definitely seen as sort of like the leader in the AI safety space. And apparently a mentor of his, somebody that funded some of his research, was Peter Thiel.
And again, November 2023, Peter Thiel is sitting with Sam Altman and warning Sam Altman about Yudkowsky's ideas. He's saying, "You don't understand how Eliezer has programmed half the people in your company to believe in that stuff," that stuff being that, you know, AI is going to kill everyone on Earth. And he's telling Sam, "Need to take this more seriously." And so, Peter, it seems like he has warned Sam Altman many times before about his company being overrun by EAs, right? So, EA, effective altruism movement; EAs are people that are part of that sort of movement. Because EA recently pivoted from trying to end global poverty and other sort of causes like that and focused a lot of the efforts on trying to prevent sort of this rogue AI from killing everybody on the planet, from murdering humanity. And Thiel is predicting that instead, EA will sort of end up destroying OpenAI. And apparently they have a quote of Sam responding, saying, "Well, that was kind of true of Elon, but we got rid of Elon."
So, at that time, this was OpenAI's board, their board of directors: you had Ilya Sutskever, Helen Toner, Adam D'Angelo, Tasha McCauley, Greg Brockman, and Sam Altman. So, the board of directors controls the OpenAI, the nonprofit entity, which is kind of like the top-line entity under which everything else is formed, right? So, the holding company, everything else is underneath the nonprofit, right? So, the employees and other investors own sort of portions of the holding company. Microsoft owns portions of this capped for-profit company. It's all very complicated, but the point is, at the top of the whole thing sits the board of directors. Now, there used to be more of them; some left due to various conflicts and for whatever reason they didn't introduce new ones. Now, some rumors are saying it's in part because Sam and Greg were kind of dragging their feet in terms of getting new people introduced, but whatever the case is, so Helen Toner and Tasha McCauley, they are sort of part of that EA movement, the EA community.
So, kind of during this time where Peter Thiel is warning Sam Altman about the potential EA infiltration into OpenAI, for members of the board, these people up at the top, so everybody except Sam and Greg, they're holding secret meetings to try to decide whether or not to fire Sam Altman. Some of the reasons why they wanted to get rid of Sam Altman was there were some safety breaches. So, for example, there were supposed to be an oversight board that would approve any releases, such as, you know, ChatGPT's new functionality that was going to be released to the people. According to some of the insiders, they're saying that Sam Altman claimed that all of them were approved by the joint safety board between them and Microsoft. It sounds like Toner found proof that only one has actually been approved. Seems like GPT-4 was launched earlier in India than the rest of the world for certain testing, and that also was without the approval of the joint safety board. There were also problems with the OpenAI startup fund. It was news to the board also that Sam Altman owned the fund personally. Now, supposedly this was just a temporary arrangement due to some tax structure, and Sam Altman earned no fees or profits from the fund. It was an unusual arrangement. And of course, one of the big things were the release of ChatGPT, which was supposed to be a research preview but became the fastest-growing app of all time, which, as later, you know, when when we saw interviews with Sam with Greg, they were saying that nobody could have predicted that. Certainly, I don't know if anybody saw it coming, right? That ChatGPT moment was rather sudden; it didn't seem like a product launch. It did seem like something they expected a number of people to try, but not to get, you know, global interest. And it seems like Ilya Sutskever was potentially the person that kind of pushed the first domino. So, he called Helen Toner and didn't quite tell her what was happening, but said, "You know, you should talk to Mira a little bit more," kind of giving her a hint as to where to go next. Mira Murati was the chief technical officer and was basically running the day-to-day operations of OpenAI.
Now, at the same time, there may or may not have been some conflict between Ilya and another OpenAI researcher, Jakub Pachocki. Rumors were that Sam Altman promised both of them to kind of lead the research direction of OpenAI. Also, around this time, Helen Toner published a paper in which she criticized OpenAI, kind of saying that Anthropic was doing a much better job kind of over AI safety. Sam Altman was very upset over this, and allegedly he told Ilya Sutskever that Tasha McCauley, another board member, said that Helen Toner should be fired over this. And apparently they all realized that this was not the case, that this was a lie, because they were at that time holding their own kind of secret meetings. So, whatever the case, they decided to pull the plug. And when Sam Altman is away from San Francisco, from the Bay Area, they call him on a video call and tell him that his services are no longer needed. They fired him. They also removed Greg Brockman, likely because Mira Murati did not want to work underneath him. So, since they wanted her to just run the whole operation for the time being, they removed Greg as well.
Interestingly, this is kind of an interesting bit of information: after that whole thing happened, the board members told Ilya Sutskever that they were pretty sure, or worried, or they thought that maybe he was sent as a spy to test their loyalty, right? So, maybe he sort of, uh, went to them to try to stir some trouble, saying, "Oh, you should talk to Mira; there might be some information that you need to know," and kind of got the ball rolling. What if he was doing all that as sort of this double agent working for Sam Altman? Now, I wasn't there; I don't know what happened, but there was one prediction that I made back in November 2023, and I stand by to this day; I'm willing to bet money on it, and that is this: this story, this will be a movie, a movie or a dramatic reenactment or some sort of TV series. Whatever this boardroom drama, I mean, come on, double agents, secret cult-like organizations, every one of these people is like a character basically. Ilya Sutskever would get all the employees together and chant "AGI," I mean, just that scene alone would make a movie. Somebody has to make this.
Anyway, now back then, when all of this was happening towards the end of 2023, I asked all of you what your thoughts on Helen Toner were. She seemed like she was sort of like one of the main drivers behind the whole behind the firing of Sam Altman. We got almost 4,000 votes, and only 4% of the people said that she did the right things for the right reasons. 27% said that she had good intentions, like she was trying to do the right thing, but she executed poorly and failed. And most people said, you know, 39% here said that all bad: so, just bad ideas, bad ethics, bad execution, just a complete fail. So, while we might sort of disagree about her sort of intentions, most people seemingly kind of say that the handling of the situation was a bit inept.
And this was seen almost in a day before the firing. So, the Thursday, the day before, the four board members called Mira Murati and said that they'll be firing Sam Altman the next day and asked her to step in as the interim CEO. So, she agreed. But when asked why they were firing him, they wouldn't tell her. Murati was concerned if they talked to Satya Nadella, the CEO of Microsoft, about this, if they gave him some advanced warning. They had not. They decided that Murati would tell Nadella right before the news broke. And of course, the next day they decide to pull the trigger; the whole thing becomes a worldwide headline. Everybody wants to know what happened. So, at this point they still haven't told anyone anything. They didn't tell Sam Altman or Greg or Mira, and the only thing that they're saying is that Sam Altman wasn't consistently candid with the board.
So, after the firing, the employees, the executives, they're all meeting with OpenAI's board, right, the four remaining board members, and they're baffled about what's happening. No one knows anything, right? So, Satya Nadella, running Microsoft, reports are that he was just mad; he's losing his mind because he's never just seen this sort of behavior. He has no idea what happened, why it happened. They notified him after it happened. Sounds like before sort of the news broke, so he was basically finding out just slightly before the rest of the world did, on a Friday, not really having any time to respond. But most people are baffled; they say they are not aware of any specific episodes that might warrant such an outcome, right? So, they're saying, "If he's been consistently candid, like, is there examples of that? Like, what are we talking about here?"
All right, so this is where it gets good. This is where it gets interesting. This is where there's some like cloak-and-dagger stuff. The board again, so these four people, they sort of expected Mira Murati to just handle stuff for them. They expected her to calm the employees and explain exactly what was happening and why it was happening and just kind of like continue business as usual as the board searched for a new CEO. At the time, I believe there were also some sort of conversations about potentially doing a fire sale, selling everything to Anthropic and Dario Amodei, because he was seen sort of more AI safety conscious person. But the point is, they put Mira Murati in charge; they said, "You handle sort of like, like the masses; you handle everybody, the employees, while we figure out what to do with the company, if we're going to put a new CEO in or just sell it off." And they expect Murati to just handle everything. Instead, Mira Murati basically leads a company-wide revolt against the board. There's a narrative that's being spread in the company that Ilya Sutskever was mad at Jakub Pachocki because he received the promotion. They were also saying that Helen Toner was mad because Sam Altman didn't like her research paper that she published, that he tried to push her off the board. And look, here's the thing: because they failed to provide a narrative or an explanation of what was happening, of course, one was going to get developed somehow, one way or another. I recall at the time sitting there like just not knowing why, what was happening, why these things happened, anything like that. It was very frustrating. And of course, a lot of people are interested in AI development; their concerned. So, having four people do this with no explanation, first of all, the entire world is watching; there's no explanation for why it's happening. Like, you could have predicted that this wasn't going to end well.
So, what happens by Monday morning is that all of them signed a letter threatening to quit if Sam Altman isn't reinstated. So, Mira, most of the other people at OpenAI, Satya Nadella, meanwhile, is saying that, "You guys are all welcome to join the Microsoft Corporation." So, most people at the company are signing that letter saying they're quitting, they're leaving unless Sam Altman is brought back. Interestingly, among all the signatures are one Mira Murati and two Ilya Sutskever. At this point, Sam Altman is brought back. Helen Toner and Tasha McCauley, I guess they're just kicked out. Adam D'Angelo, I don't know if he temporarily left or what happened, but he's still on the board. So, here's the OpenAI website; so Adam D'Angelo is still listed as one of the board members. So, that's quite a wild story, right?
So, Peter Thiel, that same month that happened, is warning Sam Altman about the EA people that are going to destroy OpenAI. Within weeks they lead a coup against Sam Altman; they fire Sam Altman. Potentially Ilya Sutskever, maybe he's the one that started this, but largely it sounds like driven by Helen Toner again, as far as we can tell. I have no idea exactly. Then, over the weekend, the whole thing turns around. Now, Mira Murati and Ilya Sutskever are signing the document saying, "Hey, we're going to quit unless we bring Sam Altman back." And then a short while later, the people that are associated with EA are gone. Now, of course, since then, Ilya left OpenAI and started his own company, SSI, Safe Super Intelligence. Mira Murati left and started her own company. So, kind of a fascinating turn of events. It sounds like the book is going to have even more interesting details, but I'm curious to know what you think about this whole thing. Was this a failed coup? It was just people that were, you know, part of the EA movement that were concerned with AI safety; they found things where Sam Altman wasn't 100% consistently candid, like they said, and they figured they needed to kind of exercise their authority, their role as the board, to try to get him out and bring somebody else in, and this was just the, they fumbled the execution of that? Or do you think that this might have been something more nefarious? They just wanted to bring in somebody that kind of saw everything the way that they did, maybe somebody that was also part of the EA, EA movement, somebody that also believed that there's a high-X risk from AI? Or do you think this was some sort of a double-cross where Peter Thiel is like, "Hey, here's what you need to do to get rid of these people, get them off the board," and maybe Ilya Sutskever, maybe Mira Murati did have some role to play in that? Let me know what your take on this is. I'm very curious if you made it this far. Thank you so much for watching. My name is Wes Rth, and I'll see you next time.