Sam Altman, CEO of OpenAI

29:04It's like, yeah, you know, we were kids together. I'll like cancel my day and I'll like help you out of this mess. Even if people haven't talked to each other in like 10 years. I think that's like, it's such a deep and special part of YC. Anyway, on the Rippling point, I don't remember exactly when YC really got involved, but it was in a bad place by the time we did uh like i remember that meeting with mark and it was like i mean crazy threats were flying it was even even on the there was a thing that pg or jessica used to say about yc i forget who it was but it was like you know startups have all sorts of horrible problems you can bring us your problems it won't scare us we're like emergency room doctors if you come come in with like an arm cut off.

29:56We're like, well, do you have the arm? Yes. Yeah. Yeah. And so we were used to like a lot of really bad situations and I felt very unshaken by them at this point. But I remember when Parker was explaining his situation, it's one of the few times in all of I see where I sat back in my chair and I was like, whoa, this is a tough one. Yeah. Was it the worst you had seen as the operating doctor? I think so. I can't, at least I can't come up with a worse one right now. I would agree. I don't remember anything quite as dramatic and awful as that. So you said YC's got to step in and try to do something.

30:38Well, I think we had like either a regular partner meeting or an emergency one about it. And we were like, this is, you know, what are we going to do here first? And then we kind of like collectively decided that I would first talk to Mark and other people are going to do some other things too. and what kind of threats were going was it between you and mark there were threats no i remember that mark really wanted to solve it um okay but it was like i mean it was like a bunch of messy legal action by this point right yeah for sure so it was bad yeah so those were a lot of the threats but my memory of mark is that he was very good and wanted to solve it and said like okay you know this has like gotten to a bad place Let's see what we can do.

31:21But it stopped after that conversation, Parker said. I thought it went on a little bit. Maybe it did. That was like a pivotal moment, basically, I think. I remember the conversation. I think it did change the trajectory. I'm sure it didn't resolve overnight. But yeah, I think that maybe Mark getting involved and you getting involved helped a lot. Carolyn, do we have any other questions before we move on to OpenAI? Well, really quick, I just want to know what are some of your fondest memories of your time as president at YC? What stands out? We just talked about the bad. What are some of the good?

32:01I mean, it was mostly great. These are like things that came together in like a quick montage in no order. I think it was the first demo day after I took over, the first alumni demo day. either at the beginning or the end of it, I said something about PG and like how thankful I was and how much he had done for all of us and, you know, like grateful to be doing this. And like the whole room stood up and gave this like big standing innovation that went on for a while. And then after that night, I was at a dinner with a bunch of the founders and everybody went around and were just like, you know, it was cool that you did that.

32:42Here's my PG story and i was like very struck in that moment about how like how much like we had been affected by all of the partners but pg in particular had done like gone so like out of his way to do things for each of us that were kind of like great advice or fighting with an investor whatever and i sort of said right then like okay someday when i'm retiring from yc i hope that like i hope that there are people who like go around and like feel like I did something for them because this was like the level of gratitude towards what YC had done for all of us was this like special special thing so I was like you know a lot of things that can do at YC but the one thing I really want to do is like be there for founders like this and then you know like I still hear super frequently from founders that did YC while I was doing it about like hey you probably won't remember me you won't remember this thing you did but it like changed my company my career whatever and uh that's for me that is the most fulfilling part of being a yc partner yeah yeah the second best one for me is watching founders develop during the course of the three months of yc um it's like watching a child grow up in terms of how fast the development goes you can like watch someone that comes in and they're so unimpressive at the beginning of yc and so impressive by the time it's done.

34:09And it's like, how can that happen in three months? And can we do that for like a lot more people in the world? Yeah. Yeah. Can you call to mind any specific founders who you remember changed a lot in those three months? Tim Ellis just came to mind, Relativity Space. They 3D print rockets, but he was like in my group. He came in like, so kind of hapless. I was just such a force of nature by the end of YC. But again, it happens pretty often. That's a good example. Under the instruction of Sam Altman, he emerges. I don't think it's – actually, I don't think it's the partners that do it. I think it's the peers.

34:52Yeah. Yeah. I think the magic of YC – a big piece of the magic of YC is the peer group is so effective. That's why it's so great that we're back in person because I think you just – you can't get that on Zoom. I am I thought it was totally crazy when all of the companies during the pandemic all the tech companies were racing to be like we're going to be remote we're going to be even more remote we're remote forever I was like this is I am confident this is going to get undone this stuff does not work over. Right. Yeah. Okay, we got to shift gears and move over to OpenAI because it's such a big thing now.

35:31OpenAI started as a nonprofit and you started it as a side project to explore AI. But a lot has changed since 2016. um so i had been an ai like a computer nerd in particularly ai nerd my whole life um i you know the summer before doing yc i worked in the ai lab at stanford and i thought nothing was working but i still thought it was the coolest thing ever and i kind of like wanted to be a ai researcher but nothing was working um and i went down a different path for a long time startups investing, whatever. And then in 2012, there was this real breakthrough called AlexNet that my co-founder Ilya helped to lead where we saw for the first time a neural network win in a competition.

36:25This happened to be image recognition. And then I think we all should have taken much more notice right then. It took a while for the world to really wake up to it. but we had this technology called deep learning that seemed to work and more importantly than that it seemed to get better with scale you could throw more compute at it and it just got better and this felt really crazy um there were other threads going on and i kind of had like gotten pretty deep in the whole like agis could be super dangerous if it goes badly kind of east bay whatever that thing was and and i felt like it was somehow worth exploring more uh but it wasn't clear what to do and also it wasn't clear at all that it would ever be a company like that there could be a company like opening i started as a research lab um we thought we were just gonna like write papers or actually we weren't even really sure we're gonna try to do research and And papers were what people knew as kind of the object to produce.

37:28But we imagined that we would just be like, you know, writing papers and maybe putting some open source technology into the world. And a nonprofit seemed like the right structure for that, especially given that we had these safety concerns about how the technology could evolve. We got to work and we were a sleepy research lab for a while. And we did a bunch of things that were sort of like impressive, but they didn't kind of like contribute to sort of this artifact that we hope to build someday. We made something that played video games really well or one video game really well. Can I stop you there?

38:08Do you remember – I don't know. Do you remember when you told us about the air hockey game? Is this what you're talking about? Maybe it wasn't air hockey. You told us about the arm, the arm that could play games. Oh, yeah. It wasn't air hockey. What was it?

38:28Something that was much slower. I don't remember, but it was not air hockey. Okay. Well, so you were just chatting with me about it, and you said something that still creeps me out. You're like, oh, yeah, last night the machine, the arm learned that if it's not winning, it can just pick up the table and tilt it a little bit. Do you remember telling me that? it's still creepy to this day that it learned how to do that should i not be creeped out by that because i am um well i i think no i think there is something something's coming full circle there so we had this like little simulator and uh we would try to have it do all these things and it would definitely occasionally learn that it could just exploit some bug like flipping over at the table or whatever.

39:11And that was, you know, that kind of learning, reinforcement learning, where you have a goal and it's just sort of randomly trying some stuff and then doing more of what works. That was a scenario in which you could see a lot of things going wrong. And at the time, this is before we discovered language models at the time, the kind of consensus view of how we were going to get to AI was we were going to build this reinforcement learning, AGI, was we were going to build this reinforcement learning agent and then put it in more and more complex environments until eventually it was like living on a simulated island trying to survive and reproduce and kind of rerun evolution or something like that yeah and the safety issues there are very easy to imagine like it would be much worse than flipping over a table thankfully we discovered a much better approach which became the gpt models and chatbots and things like that where we just try to like learn the themes of human output on the internet.

40:14Now, the reason I said it's full circle is we are, again, trying to take those language models we've learned from all of human history and do reinforcement learning on top of them. And this is where I think we have to be very careful about potential safety issues. You should be a little freaked out, but we got very smart people. Well, I want to say that was like 2016-ish. Is that right? Seems right. Yeah. Yeah. And I was like, ugh. Can I ask a real dumb, dumb question? Sorry, Sam. I'm just going to apologize in advance. But I have some questions that you probably don't get asked because they're so idiotic.

40:48But the LLMs grew out of this like transformer paper from Google, right? And OpenAI figured out how to make it work on a large scale. Why were you guys at OpenAI able to get more out of it than Google did? I think a couple of things. One, I think Google didn't realize the significance of what they had with the transformer. Other people, now, it would have been discovered at some point by somebody else. But a lot of people were publishing a lot of architectural ideas. And this was one really great one. But at the time, it felt like one out of, you know, many. And we had been looking at a previous generation architecture for language models for a while.

41:37So Google published this thing. We had kind of been like waiting for a much more efficient architecture and maybe had been thinking about related ideas for longer. But a really important thing is it was not at all obvious at the time, even though in retrospect, it looks like it should have been, that you were supposed to put all of your effort into one giant scale up. you go to a place like Google and there's a lot of smart researchers and they all want compute. And the easy thing to do is to say, we're going to give you each 1 % of our compute and you can run your experiments and find more breakthroughs.

42:20And it's really hard to say like, hey, actually, we see this interesting property of scaling. So most of the researchers are going to get very little compute. And we're going to pick one team to get 80 % of our company's computing power to do this one big training run. And I don't think Google could have gotten that done. Also, the leadership of DeepMind really just didn't believe at all in this approach. So they were like, no, this is not going to work. And they had a very different technical bet at the time. And in fact, not until the chat GPT moment happened many years later, and Google had their kind of famous code red, did they really get serious about language models.

42:58But they just, they made the wrong technical debt. So it's kind of a big company problem plus a decision maker problem. But they were also like, pretty delayed on launching things publicly, if I remember correctly. And I remember hearing from insiders like, oh, Google's made so many advancements in AI, but they just haven't launched it yet or whatever. Like, were they fearful? Do you think that some, they'd get some politically incorrect answer? Or do you think that AI was at odds with the advertising business, their cash cow? What do you think it was? I think it was those two things. So I think for sure, they were like a risk averse big company, where you had to go through like, this is what I'm terrified of happening to open AI someday.

43:47But you had to go through like 17 layers of management, any one of which could say no from a risk perspective you know it's there's going to be a bias issue there's going to be a copyright issue there's going to be a this there's going to be a that and i think all of those things happened and and also i believe although i have that that i've heard stories of this next one is only speculation i have no proof of it but if i were google's very senior team looking at this thing i would say like this breaks our business model and i would kind of like maybe even subconsciously hope it just wasn't like i would want to believe it wasn't going to work um i so so i i assume on speculation that there was something going on like that and then i i actually don't think they were making great technical progress.

44:44I think they had bet on this wrong direction. And now they do have very great technology. But I think people were saying it because they had all of these other research directions they thought were going to pan out. They just weren't that useful. Are there any queries that you use Google for at all anymore? I've stopped using Google. It's really weird. No, I have too, actually. I agree with that. See, Levy, do you only use chat GPT? Yeah, I do. Because it's, it's the lazy person. Yeah, chat GPT does all the work for you. Like Google gives you results you have to click on. I don't want to do that.

45:21I just want the answer. Right. You want the answer, not the web pages. Right, exactly. Yeah, you know, we're trying to roll out search to our free user tier pretty soon. Right now, it's only the really good search is only in the paid tier. But when we do, I would expect like a lot of people to start using it that way. I also think we're still just scratching the surface. Like, yes, I think ChatGPT is better than Google search for most stuff. Like not the like kind of effectively DNS queries where you're just trying to like get to facebook.com or whatever. Google's really fast at those. But for most other things, I think we're better.

45:56But I think we still haven't done the magic. We still don't have something that can like go off and do like the kind of web search that Carolyn would delegate to somebody on the legal team at YC and expect to take like six hours of work of searching the web and reading stuff and putting it together. And, you know, like when we can do a search query like that, that's a whole new thing that is, I think, going to be really great. Yeah. Yeah. I can't wait. Sama, because we're the social radars, we've got to ask you, will you talk about November 2023? that crazy week when you yeah sure brockman got ousted by the board of directors you tell us that because i remember that fucking week it was so crazy you sent the nicest note or tweet right after in the middle of like the craziness i i think it was the 17th that it started and it was resolved by like five days later something um yeah i don't know what to say about it other than it was like totally crazy and surreal and felt like a dream.

47:01I was caught completely off guard. And I really did in that first moment, thinking I was having some crazy vivid dream that I was about to wake up from. This is when the board said, hey, Sam, do you have a sec? I mean, how did they even tell you? Because first of all, let's just set the stage here. This was the board that was part of the nonprofit organization, correct? Yeah. That is and was and is our only board. We only have a nonprofit board. And did they just one day say, Sam, we need to talk to you and you're out of here? What happened? Yeah. I got a text message from Elio the night before saying, can you talk tomorrow?

47:45Said, sure. A few minutes before he said, here's a Google Meet link. I was like, okay, let me go find a place to sit because I was like in a hotel and I was going to be a call. And then this happened. It was like very strange. I couldn't get any questions answered. And then they just put out like a blog announcement and like turned off my email and stuff. And so I'm out of town. My computer just got deactivated. I have a phone.

48:14And I kind of didn't think about what was going to happen next. I don't know if they did either but it went crazy like it was like national news for like days it was all anyone was talking about even over in England geez I mean the nerd Super Bowl is what I heard people call it and that still makes me chuckle a little bit um and I and then the feeling of that afternoon was just like this kind of fog of war like I I had no idea what was happening The team didn't know what was happening. Like they were trying to get answers. I was trying to get answers.

48:58And it was like nonstop calls. It started as like Sam must have done something horrible. Like this happens if you kill someone. In fact, I remember. So we were leaving like I was just on like call after call. Microsoft's trying to figure out what's happening. Like I'm like talking to a couple of board members, talking to the whole executive team. We're like trying to stabilize things. It's like feels very crazy. And so I'm on like these nonstop calls. Ollie is like packed up our hotel room. The first time I had a minute where I was not on a phone call is as we were like getting on the plane. And it was this very dramatic thing.

49:38It was like in Las Vegas. It was super windy. We were like standing right outside the plane. It was like he had been overhearing some of the calls. And I had like written him to us, but it was the first time we like really had a minute to talk. And he like, as we're about to like walk up the steps, he like puts his hands on my shoulders. And he's like, Sam, I'm with you 100 % no matter what, but what the fuck did you do? Wait, who is this that you're talking about? Who said this? This is my husband, Ollie. Oh, Ollie. He's like, what the fuck did you do? And I was like, I don't know. And he's like, okay, I'm good.

50:11And we like get on the plane. and then yeah it was great because the board of directors never clearly told you what did they say to you they said you're out of here because we don't trust you yeah it was like there were a lot of like personal power issues there were some legitimate ai safety issues that i i really disagree with um i'm very proud of our safety track record but i think other people have like a different opinion on what we've done well it's clearly a very polarizing issue i mean it's understandable that there's like this issue that comes up for you guys. I will say if you believe that AGI safety is an existential issue and that the company slash me was doing it wrong or otherwise unfit, then I admire the courage of convictions to say like, this is the logical conclusion rather than I think what most people do, which is say like, oh, AGI safety is this existential issue and it's coming really soon.

51:09And what are we going to have for dinner anyway yeah by the time i had got back to sf that evening we understood a little bit more about it um not much but it was like there was this kind of dramatic thing where i had tried to get one of the other board members off in a way that i think i didn't handle well and there were concerns about you know safety and who was going to be in charge of stuff and whatever but more than that, I was just like, whatever. I love open AI. I wish them the best. What I want to do is work on AGI research. I can figure out some other way to do that. I can do it at Microsoft.

51:45I can start a new company, whatever. The executive team had been trying to talk to the board, didn't get any answers. And the people started quitting the company. And I'm just like, so a bunch of us met up that night and more of us met the next day. I had a call with two of the board members the next morning where they were like, you know, we'd like to talk about you coming back. And my, I'm not proud of this, but my first reaction, well, in my head, my first reaction was absolutely not. And then, because I was just like hurt and mad and whatever. Yeah, of course. You know, excited. That's totally human.

52:29And then my second thing was, and this is probably not proud of, I was like, I will come back if all of you resign right now. I don't blame you at all because it was they were seemed like it was a shit show. And it was I think that was not a constructive thing. I think had I handled that differently and said, like, we can negotiate some version of the board where I feel like I can have a chance to actually function. It would have been done that morning. But I responded like with. With emotion. Yeah, I did. Yeah. And then it kind of like kicked off this weekend of like crazy nonstop negotiation, planning, counter planning, whatever.

53:15I was mostly talking with Adam who I think acted in good faith throughout it. But he was only one vote of several on the board. Adam D 'Angelo. Adam D 'Angelo. But it seemed all weekend like I was going to come back and we were going to figure out a new board. And, you know, it seemed like things were on a good trajectory. And then Sunday night rolled around. And to my total shock, they're like, we're going to appoint Emmett. Emmett Shearer, also from the 2005 YC summer batch. And I was like, oh, that's like, that was the angriest I had been in the whole time. because I was like the level of deception here.

53:57You know, I was like trying to keep things together. I was trying to like tell the company, hey, this is like heading to a good place. You know, don't go. Because all of our competitors, like these vultures were circling, trying to recruit whatever, whatever. And, but Mira had like told the board that because there were like no good answers, she was gonna just resign and not do this. And so then they like, you know, were like, well, we can't have Mira do this. So we got to pick somebody else. Oh, because she was appointed interim CEO immediately. I remember that. Yeah. I mean, she was really, really great.

54:32Emmett, they put Emmett in there. And at this point, I'm just like, what the fuck? I'm ready to go to bed. Like enough is enough. So I like announced that I am going to join Microsoft and work on an AI research project there. I stay up for like a couple of hours talking to people and I go to bed. And I hadn't slept watched in the last few nights. Of course.

54:55and i lay in bed and i'm this like weird thing of like very tired and still just like angry and full of adrenaline i can't sleep so i take a sleeping pill um and i go to bed probably like three in the morning something like that and and then i'm like rustled awake and this time it really feels like a dream because I'm in like a sleeping pill haze. I'm like wrestled awake and Ollie's like, you need to like look on your phone. And Ilya had said, I regret my participation in this. You know, didn't want to harm OpenAI. He didn't say we should undo it, but something like that. And like 700, you know, like 95 % or something of the OpenAI employees had signed this letter while I was asleep saying like, you know, this is a horrible decision.

55:49I'm going to we're going to like resign. We're going to resign if Sam does not come back. Emmett had also told the board like as he had this was maybe later in the day on Monday. I don't remember. But he had also told the board like, I'm going to resign, too. So finally, it still took a long time after that. Like, I cannot believe how long the whole thing went. But finally, I think it was Tuesday night. Maybe the Wednesday night, whatever. One of those couple of nights, Tuesday or Wednesday, we finally finished everything and announced that, like, we had sort of high-level agreement and that I was going to come back and kind of the company got put back together.

56:26That part was all a crazy thing to live through. But honestly, you can kind of do a lot for four or five days on a kind of crazy adrenaline rush. The couple of months after were just miserable. I was just exhausted. yeah um i was like in this like stupor i felt like i would like walk down the hallways and people would just like avert their eyes to me i felt like you know a patient with like some really bad diagnosis because they didn't know what to say they just felt bad yeah and it was like weird it was just so weird there was this like air of like something must have gone horribly wrong and like until the investigation was done everybody It was just like, ugh.

57:13Yes, because the overarching backdrop is what the fuck did Sam do? Totally. But it was kind of like the winter in SF is cold and dark and a little depressing. And that period, I remember it being very cloudy and rainy. And it was just a tough, really tough time to like, it's like, all right, I feel like these people did this crazy thing, really screwed me. And now I have to pick up all the pieces. The justice of it felt really bad. Like as if this job isn't hard enough, right? Because it's like a big company and it's a controversial, you know, thing you're doing. Like in everyone's crosshairs, like everybody either like wants us to go away or wants to be us or what.

57:54Like it's just like, you know, we're the, you know. Relentless. It's relentless. So that was really like it took a while to get out of the stupor there. You actually, if I recall correctly, didn't you and Ali actually have a ceremony in January? We did it in December, but then Ollie posted photos of it in January. Oh, that's right. Because I thought we were in Hawaii and I thought you guys were there too. Because I saw the photos on Instagram. I'm like, hey, we're here together. That was like the only nice moment of normalness in that. Yeah, I was going to say there must have been some bright spots.

58:26It was only our family. So it's just like really like quiet, intimate thing. Very cool. Do you think that that whole thing would have happened if Reid Hoffman had stayed being on the board? No. And not stepped off. You do not think so. In fact, had any of this. So there were three people that departed in really pretty quick succession. Reid Hoffman, Siobhan Zillis, and Will Hurd. Okay. Reid had a conflict. Bill went to run for president. And Siobhan had this issue where she had kids with Elon and Elon was starting a competitor. So three people had to leave the board in pretty quick succession.

59:01And had they, even one of them remained. Wouldn't have happened. Gosh. If there isn't already like a Harvard Business School analysis.

59:15I mean, I hope people really learn an important lesson about boards from this and that it doesn't happen again. Actually, that and what happened in March of 2023. I'm just thinking in terms of like what will we study later in decades, like the SVB crash and the OpenAI board issue. The SVB thing was totally insane. I remember that weekend had like a lot of flashbacks to me of the November weekend. Yeah.

59:46But I remember talking to some people in the government and I was like, you don't get there. Like, oh, it's like a bunch of startup. They was like, you don't you don't get this. This is like that felt close to a real disaster. How involved were you behind the scenes, Sam? Not that much. Like Ron Conway was a real. Ron was. Oh, gosh. And he asked me to do a few things. So I, of course, did whatever he asked. But that was really him. That was crazy. All right. We are coming up on time here. But I just want to make sure that I have asked. Oh, Elon. He just came up a second ago. I have to quickly tell you, I have a photo that I took of your calendar from April 1st, 2015.

1:00:31and your meetings for that day, it said Elon, Tim Cook, Larry Page, John Doerr. Do you remember this day? Quite a day. Oh, I think that's when we were trying to like get OpenAI set up. I bet it was. Yeah. Yeah. But I remember I was like, Sam, I have to take a photo of this because this is insane like in one day. But anyway. That's just a normal day for Sam. Normal day for Sam Holland. It was the original OpenAI fundraising was this like crazy period because everyone was kind of interested. So we could meet anybody or talk to anybody.

1:01:11But it was shockingly hard to actually raise money. Oh, yeah. This is a lesson about a nonprofit structure, too. Yeah. But at the time, you had no idea where AI would lead you was to all of this huge expensive compute time that you needed, right? We had no idea. Well, that's not fair to say. We knew it was going to be expensive. We were probably off by like 100x from how expensive it was going to turn out to be. And you can't get that from donations. Come on. Yeah. So again, had we known, we wouldn't have done. Like, had we known we were going to make a product and actually be able to be a company?

1:01:46and had we known we were going to need this much money, we would have, of course, been a different structure. And would that have changed anything that you did with the board makeup? We just would have been a normal company in general, which would have been healthier in so many ways. But I think it's like often if you're doing something really new, you have to feel your way through it and you just don't. You kind of don't know where it's going to, to a shocking degree. Like, Jessica, I'm sure when you started YC, you couldn't really imagine that it was going to become what it became. No. We have like 65 legal entities.

1:02:22Like it's a huge, huge thing. And so I think that just happens. And there's kind of somewhat unavoidable. But in our case, I think we made particularly bad decisions if we knew everything then that we don't know. But Elon was involved early on. I just have a random question. I believe like his superpower is that he's a genius engineer. He is a genius engineer. He also is incredibly good at moving fast and motivating people and getting stuff done outside of just engineering. Okay. Because he has such a wide variety of companies. It's amazing that he'd have such good judgment in so many different directions.

1:03:05It's super impressive. Do you think he bought Twitter to end wokeness? I'm sure that was part of the motivation. I mean, I think he loved, even before he bought Twitter, I think he was like, my memory is he was just tweeting all the time. I think he like loved the product. Okay. I'm sure he wanted to end wokeness. You know, it's like he's the kind of person that I imagine would like to sort of control the media platform that matters or whatever. So I bet it was like a lot of things. Okay. Sorry, that was a random aside. But I'm now in the time of my random rapid fire questions that I have. Because we've been through a lot since 2005.

1:03:51And I feel like when I finally write the YC book, I'm going to have to come to you to remind me of the crazy stories. Like, who did we used to think was like the worst reporter ever? Oh, man. Wow. That's a whole life. You have to go back to the Valleywag day. Yeah, it's got to be still in the Valleywag, right? Yeah. They were just like really nasty to YC. They were nasty. Like just mean, mean spirited. And didn't care if they lied. I was reflecting the other day. There were all of these people who kind of like made it their whole career or their whole personal brand to hate on YC. Like there were these people on Twitter that I used to think about all of the time.

1:04:35They just like got in my head. Yes. They were like, they knew how to do it. And they would just like make up all this awful stuff or like really just like, you know, rev up over like some little thing. And it felt like it took up so much mental energy. And it felt like such a big deal. And they all kind of like faded away and were doing good things for the world. Do you guys remember Secret? Yes. Oh, my God. That was a moment. Who did Secret? Who did Secret? I don't think we ever found out. I think we did know who made it. Oh, it was like a it was like a startup. You go on and you anonymously like say things about everyone else in the startup ecosystem.

1:05:20It was wild. Did they say bad things about us? I've probably blocked it out. It was bad things about everyone in every direction. I think they said bad things about everybody. Yeah. Anyway, but my lesson from that period is like these people who like make their whole thing about like negativity and trying to like tear you down. It probably feels easy for them in the moment, but they say it away. Yeah, I agree. I've garbage collected all of my hatred for these people, which was quite intense back in the day. All right. Who was the craziest YC interviewee like applying to YC? Do you remember any crazy stories?

1:05:58Mostly I just remember the things where someone would come in with like an obviously bad idea. You would kind of early on, because you're always going to these interviews with hope and wanting it to work. You would try to suggest a way that it could be made better. Yeah. And then you would go down this like rabbit hole of they would just fight you for the remaining nine and a half minutes. You're an idiot. You don't get why this is such a brilliant idea. and you would say, well, but there's this flaw. And those are the ones that I would just like, I would like think about those at night. And I was like, what was the strategy there?

1:06:39I don't know. I think I've blocked out a lot of the really bad interviews and I remember the really great ones. Like, you know, interviews are like a slog. And then once every 10 times, you're like, this is so great. These people are so great. Who was your favorite one then? We'll end on a happy note. um oh that's a tough one or among your favorites one of the companies that i kind of hand recruited was ginkgo bioworks yeah oh my gosh yes and smart smart i remember the first meeting i don't remember if it was i think it was like a special one-off interview we set up with jason and reshma and i was just like this is so exciting they're awesome this is exactly the kind of company I want to do.

1:07:23They're so smart. I remember that that was one that just came to mind of immediate like I'm so enthusiastic. Sam, also exciting personal news. You and Ali are expecting a baby in 2025, right? We are. Are you excited? I have never been more excited for anything. I'm sure I'll get even more excited soon. But this is what I think about in the shower now. I started buying all the baby stuff and it's... Oh, I love it. You're going to buy too much baby stuff though. I'm just going to tell you like it is now. You won't use half of it. You know, I had like a bunch of friends tell me that and say, here's the things you actually need.

1:08:02Just get these. Just get onesies. But I had no ability to like resist another cute onesie. Well, onesies are fine. Just don't buy like outfits because the baby will only be in onesies. Well, we're so excited for you. thank you so much for coming on the show and talking about all this stuff don't you feel like a cane is going to come and pull him away because we've gone five minutes over and one of your PR people you said hard stop is going to give us we're going to go launch this new thing you actually have a hard stop because you have to go do something and this will come out so much later but it's like the 12 days of open AI well I'm sure it will be a huge success so Sam thank you so much for coming on It was so much fun.

1:08:45It's going to be a great episode. Thanks. Thank you, Sam. Great to see you. Bye. Bye. Carol. That was so much fun catching up with Sam. no, I'm really glad he had time to come on. It was great to talk to him about, I know he squeezed us in before the launch of the new product, which he's doing for the next 12 days. Oh, I mean, it's pretty cool that both of us have known him since 2005. I know. I mean, yeah. It's, God, 20 years? I know. But he was always remarkable, whether he agrees or not. You know, he has his own little definition of formidable, but he's always been remarkable. His definition of formidable is scary, which is probably what most people think.

1:09:33But I think the YC definition of formidable is someone who is not intimidated by anything and will run through walls to get what they need to get done. And so obviously, like, we think he's the YC definition of formidable. Always has been. I loved hearing the Steve Jobs story. I didn't remember that. The little glasses, you know, shining in the and five, six, seven, eight. That's what I was thinking. Like, it's like a chorus line. I thought the same thing of him being upstage interviewing for a part or something. Yeah, like, yeah. Five, six, seven, eight.

1:10:12what look we started out with the giggles and now we're ending with the giggles i know i know was it mean that i told him he needed a stylist for his zoom background no that's not mean at all no well so do i look at my back i mean like who doesn't there's very few of us who've managed to like style our backgrounds. But he's Sam Altman of OpenAI. He deserves more than bricks. He deserves more than bricks. It's like a comic, you know, 1980s standup comic. Yeah. Yeah. It did. Yeah. It looked like a stage because those curtains that didn't have a window, the curtains had a brick wall. That was, that was a little confusing.

1:10:56There was so much we could have covered. I mean, I feel like this was like a lightning round of an interview with him because there's so much to cover. I know we never even asked him about when he was at Reddit, you know, exactly. Yeah, there's that. But actually, I think what would be more similar to what we've done with people since you mentioned Reddit, like Steve Huffman and all the other founders we interview is actually just kind of going deep on their actual startup. And looped, you know, we did spend a little bit of time on looped and kind of what he had to do to make that company work.

1:11:28But going through the whole trajectory of Looped, there's probably tons of stuff in there just from that would be interesting to other founders, like how that company got run, what happened to it, all that stuff. So if we ever go on the road, let's get him back and we'll ask him about more questions about Looped and for funny stories from YC. Yeah. And then the whole Reddit story, too. Yeah. And the Reddit story. Yeah. Because that was another. It's like Sam always gets involved with these crazy stories. yeah that's crazy a lot of people have in this industry have crazy stories but he has probably the most but i loved that he uh shared them with us and i had so much fun walking down memory lane with him yeah i loved it yeah all right that's gonna be a great episode i agree i will see you next time carolyn next time bye

1:12:28Trevor Продухa, 2013