1 2 3 1 2 3. Does that sound right? We're going to do all that personal drama in 90 minutes.

Oh, yeah. Oh, don't worry. Talk about building super intelligence. Is this part all going to be in the podcast?

No. Well, maybe not. Bear with me through my slightly silly intro, but uh, I think you guys are going to be okay. Welcome to the fortress of finance, the capital of capital. This is not that podcast. This is Core Memory. I am Ashley Vance, and I'm Kylie Robinson, and I think we have a hell of an episode for you.

Normally, we do an introduction of people at this part, but that's perhaps unnecessary. Today we have Sam Alman and Greg Brockman, the co-founders of OpenAI. You may have heard of it. Thank you guys for being here.

Thank you for having us. Thank you so much.

I think this is the first time you guys have ever done a podcast together.

That is amazing. But I think that's true. Certainly in a long time.

In a long time.

Maybe the first one.

And I will not fight you for doing it on our show, but you did buy a podcast. Is there? We just locked out.

You locked out.

Yeah. I'll take it. I'll take it.

I was curious why you guys bought a podcast. It's not really We don't have to go deep on it, but did you have quick thoughts on that?

I think the the people who who do TBPN are incredible. I think they're just very creative thinkers. And I think that in this world that we're moving to of building these AI systems that are so useful for people and helping people understand why why that's valuable for them in their personal lives and work lives, like these are the kinds of people that I think could help tell that message.

I see you on it. Have you been on TBNN?

I have.

Yeah. Okay. Okay. Um, I don't watch all the episodes, but

it's a fun pod.

Um, well, I thought, you know, since this is you guys, we haven't done this together at least in a while, we would I was going to go down nostalgia lane just for a little bit at the beginning. Um, Kylie and I have gotten to know both of you over the years. Uh, you know, I was just I we were reflecting as we were preparing for this. We're a little bit past the 10-year anniversary. You guys are two of the remaining co-founders. I think Voytech is the third. So you're, you know, you're this constant line that's been running through the company. You started as an underdog. You ended up as the the top dog. It's all through a lot of drama, undulations. Um, I was I mean I was I was genuinely curious how your relationship through all this has changed and how you guys have played off each other and if it's morphed over time.

It is extremely nice in look, we all wish there were less drama and we wish we just got to focus on the tech. But in a world of so much chaos and drama and you know tension and fighting and power struggles, it has been unbelievably nice to have a relationship with someone that's got the full context. We have all this history and to really depend on each other in amazing times, very tough times. Uh, it has been one of the nicest things about all of open air. You know, in many ways, the very first moment of OpenAI was right after this dinner that we did in July of 2015, and Sam and I were driving back to the city together and we looked at each other and we were like, we have to do this, right? There've been all this conversation of is it too late to start a lab that could go after Agi and have have a positive impact?

That seems so ridiculous now that we were so worried about that, but it was too late.

Yeah.

You missed it. You missed it. I remember feeling like that when you guys started though. I was like, "No, going to run away with this, you know."

Yeah. Yeah.

And, you know, the conclusion for dinner was it wasn't obviously impossible. And I think both of us just felt like, "Okay, this is just so important. We just have to do it."

Yeah.

And I think that spirit continues. And a lot of how we operated in the early days. Like I remember I was unemployed at the time, so I was full-time on it the next day. Sam actually had a day job. Um, but we were constantly on the phone like probably like five times a day.

Yeah. Exactly. Were you were you guys already were you like close friends already at that point or or not?

We had known each other for a super long time or it felt like a super I don't actually.

Yeah. 2010 2011.

Whenever you started at Stripe.

That's right. Yes. So we met through the Collison.

Yeah.

And so we'd been kind of casual social friends.

I guess it wasn't as long as I thought. Yeah. Maybe it was 2010 and this was now 2015. So five years.

Time compresses.

Yeah. And and and obvious I mean the being in the pressure cooker of all this doing this work I mean I would imagine it only you guys have only got closer over time.

Yeah. You you know people use the word trauma bonding. I hate that. I've I've liked I like the other things about like the people you're in the foxholes the foxhole with. the the one of the nice things about hard work no matter what, but certainly hard work in um stressful times is you really like forge these relationships that I at least have not seen get formed any other way.

Yeah. And and I do I do think the way that Sam and I work and relate is maybe different from what you'd expect from a typical co-founder relationship. Like I think that we are just in constant contact. That five calls a day, two minutes, five minutes each, that kind of spirit remains. Like I think we're just in constant sync and we don't always agree on everything, right? It's not like we come at the world from exactly the same point of view. But that's why we're so strong together, right? Is that I think we have very complimentary approaches that Sam will say here's an idea. I'll think about well maybe we could do it this other way or what about if we approached it with this angle or how does this relate to this other thing that we're thinking about. And one thing I deeply appreciate about Sam is that I think he always sees these connections between different ideas or just like keeps focused on here's the big picture that we need to get to and then together we figure out well how do we actually do it and I think connecting the grand ambition with the execution like that is what has always distinguished open AI.

Yeah. What are some of the points in these 10 years where you felt like it was really important that you guys diverge? Do you remember any key moments for you guys?

I I think one of the things that Greg has done the best, which is not my instinct, is uh really just push to focus on the most important thing in his own work and also in what the company's going to do. Um, so there have been times where I have wanted to do more things uh and Greg has just said, you know, is this the most important thing? Let's really just do this. Let's get the company focused. uh and we've diverged on that and that's been like a very helpful spirit of Griggs throughout the company.

Yeah. Yeah. And I would also add I think even for example thinking about compute and just constantly raising the ambition and sometimes I feel like okay I kind of logically know that yes like we're moving to this compute powered economy and yes that demand is always going to outstrip supply but like we've got all this like hard work to do and we already have all these big computers and we're operationalizing them and you know you still have all this like just physical infrastructure to build and you feel already swarmed in it and swamped in it and Sam is like no we even more. And I think that that actually has been a very important thing to really not like sometimes it's easy to lose sight of the high order bit of just the fact of this is going to be so important for not just the next 6 months but this is what's important for the next 2 years the next 5 years 10 years and I think that the keep like you need this balance of sometimes swimming in the details but you can't be swamped in the details and I think that that balance is something that I think again is what really has contributed to what OpenAI is and where we are going to go.

Is there there must be one product or strategy you guys have what's the one that you've disagreed about the most vehemently?

I was just thinking when Greg was talking about this this is not a a product but it was the thing that came to mind is that I was going to say even before you asked that uh we used to talk a lot about how to talk about safety. We never disagreed on the extreme importance of safety and what it will mean to get this right or get this wrong. But the the field has had a strange relationship with how we've talked about safety, how we've used safety, how much that becomes about power versus actually keeping things safe. And I I think earlier in our history, I got swept up more in the we got to really talk about this in a particular frame. And Greg was very disciplined about we're not going to fall into the traditional frame. We can't talk about that way. Now even then I think we have because this is so important we I would ever have fallen into the trap of still talking more in the wrong frame than we should. But I think one of OpenAI's greatest contributions to date has been finding a different way to talk about safety. Not just in how we build the products and how we talk about society needs to do but like what how we deploy them. The idea whole idea of iterative deployment a getting to maybe not the actually getting to a world where we're figuring out how to deploy products that get increasingly safe as the stakes go up. Greg really held a line there that I think has been quite important to the company against extreme pressure not to do that and I think that's been like quite quite important to our whole strategy not just how we talk about things but how we ship and build products.

Yeah. And if you look at for example the OpenAI Foundation which is the nonprofit that governs open AI and has a very large chunk of of equity. One of its pillars is AI resilience. And what that really means is thinking about how do we make AI be something positive for the world. And the answer is not any one intervention, right? It's not you have chain of thought monitoring and now you've achieved the mission. It's really a whole deep sequence of different ways that society should orient around this technology. And I think that this perspective of you're not going to solve AGI going well for the world in a paper, it has to be a worldwide effort from contributions from society, from many different people, from many different ways of really understanding what this technology is, how it will affect people, how it will affect the world. And this this is something that was not I think at all appreciated or understood when we were starting out 10 years ago because it's very easy just to fix it on you know we're technologists we're you know building technology that's the only problem we need to solve and I'm not saying anyone explicitly said it that way but I think sometimes you can fall into a mental trap of thinking about it that way and so a lot of what I think we have spent time on is being first principles thinkers and really thinking about how do you operationally deliver transformative technology to the world in a way that is going to actually help people in their daily lives. And one thing you realize is, well, if you have a very powerful piece of technology that will change things, probably it's going to go better if you've had a less powerful piece of technology, you've already helped change things in a positive way. And so, if you just think about it that way, you start to really be pulled down this road of thinking about resilience, thinking about iterative deployment. And again, I think this is a lot of the dynamic within OpenAI with between the two of us that we're always thinking about these questions of how do we actually achieve this mission and make it go better.

Yeah, I it feels like just yesterday I saw you at South by 2022 2023 and it was a completely different discussion about AI than what we hear today. And it feels like that's something that's changed in 10 years if like how you discuss safety and alignment. And I'm wondering how you reflect on that now. like what would you have changed in those panels and those news hits and like what have you learned about talking about safety

even before we get to safety uh I think we have fallen in the frame as tech nerds of talking about we're going to build super intelligence and dot dot dot it's going to be great for you and we've not filled in enough of the dot dot dot like we talk about we're building this amazing technology it's going to do all these wonderful things And there is like a sense in the world now of okay looks like you were right you are going to build this thing. Um, why like why do we want that? What's that going to do for us? And the thing that a lot of the field has said of oh it's going to you know cure cancer and you'll be so happy or like that's clearly not quite resonating. A lot of people are like sure cure cancer that'd be wonderful. Um, I think what people really want is prosperity, agency, that they're going to continue to have meaningful work to do. Um, there I saw an incredible post the other day that really stuck with me, which was like a a right to adversity. People actually, you want some challenges in life. You don't want every day to be perfect and everything done for you. And there there's a fear with AI that like let's say you're right. Let's say you build it. Let's say it like makes all this money and does all the work and whatever. Like what do I do? What's my kid going to do? what's life going to be like? Where's the growth going to come from? What's the what are people going to strive for? Um, and I think we've talked as a field and as OpenAI and as Greg and I, we've talked a lot about the amazing technology and what it can do and what it like the technological marvel and we have not connected the dots enough on here is what the future's going to be like. here is, you know, when when when I talk to parents with kind of school age kids, the most common question is like, "What what should my kids study? What's the future going to be like? What what will still have economic value?" And I I and I realize that that that's not quite the question they're they're really trying to ask. It's like, "How is my kid going to have a fulfilling life in this new world?" You know, you can answer it economically if you want, but that doesn't actually settle people. And I think it's because there's there's a deeper thing here. I was just going to say and I think I think we actually have a lot of perspective on this and I think that for example within ChachiBT we have so many people who say that my life or the life of a love loved one was saved through information I got through Chachi BT right whether it's there's someone who had a kid who had blinding headaches they were denied an MRI they used Chachi BT to try to research the symptoms and used to argue to be able to get insurance for the MRI. Turned out it was a brain tumor. They were able to intervene and save his life and just navigating that experience that that family says we have no idea how we would do this without chat GBT. And that's just one story. There are so many of those. I think people really understanding that this technology can help not just abstractly society but can help them, right? It can help them make money. Right now we're starting to see this wave of entrepreneurship. I think that's going to be a huge theme throughout this year. And I think we have a lot of perspective on well, we're going to build these AIs that are able to make rather than you contort yourself to the computer, right? You think about the way we do work. It's not natural, right? It's not it's not kind of what we were designed to do. Um, and instead the computer is going to do work for you.

And the question of well, what is the work? What is good? Like what is the thing that you actually want to be done? That's going to be something very deeply human, right? This agency, this empowerment. And so I think there is this very optimistic and not blindly optimistic, but I think this very positive change that this technology is bringing right now, but it's so much easier to just notice what's going to go away. Like what is the thing that you thought was solid and stable that's going to change, but it's much harder to see, well, what's coming? Like what's the new thing that you get? What is the the benefit that that comes as a part of this transformation? I think that's something that we're increasingly realizing that we also have to really spell out in addition to spelling out the other sides of it.

All right. What do we do at Core Memory? We cover innovative, fast-moving, forward-thinking companies, which is why Core Memory is sponsored by Brex because B is the intelligent finance platform for many of these companies. 30,000 companies from startups to the world's largest corporations rely on Brexit technology for their finances. They've got smart corporate cards, high yield business banking and expense automation tools that are fantastic. I hate doing my expenses. And Brex's AIS software run right through those expenses, figure out where we're spending money and take care of so much stuff for you so you don't have to waste your time on it yourself. Go to brex.com/corememory to learn more and just, you know, get with the program. Let's get going. Let's get out of this archaic finance software and move toward the future. core memory and BS. I have a couple questions based off what both of you were saying. Um, so I just went to Oberlin College with my son and we were in this room um with I don't know 600 700 people and it was fascinating to I mean, it's in Ohio and people had come from all over the country and the president was up there giving a talk and then opened it up to Q&A and so many of the questions were about AI, but they were um it reinforced for me that how much of a bubble we live in here because they were what I would consider um I mean it's not trying to like put anyone down. They they were pretty basic question. I just my my overwhelming sense of the room was like, "Wow, you guys kind of like don't know what's unfolding and about to to come across all of you." And I don't know, it wasn't I don't know it was like alarming. It was just it was eye opening for me, you know, and and they were asking um they were obviously asking stuff about like, well, how are you going to use AI in class? How are you going to stop kids from just relying on it? But then it was like, you know, they kind of advanced a little bit beyond there. that I felt like this room of pretty smart people were not not really in touch with what's going on. So I mean if this is a problem that people don't know these examples of of what could be done and we don't even know what shape this could take exactly that you just I don't see how we solve this and sort of um kind of prepare people in some way.

I do have one sort of positive take on this which is that I think that if you just read about AI it gives you one impression right it's again it's like very it just feels like you're trying to wrap your mind around this new technology what is it but when you use it it's so intuitive right and that is the purpose of AI in many ways it's like you think about back to how we've designed computers for the past you know 70 years it's really about the machine doesn't quite get you right it's like you have these goals in your head you have to break it down into the machine's language whether that's writing assembly code or okay fine now we get higher level languages now you can kind of talk to your computer but even if you think about how chat GBT works you still have to really understand concepts of language models right the fact like why do you have to create new conversations right new new new new tabs like why can't it just be a thing you talk to that remembers everything right why there's there's these limitations that are techn technological limitations but we're improving them and so I think that what we're building is the most intuitive technology We are building something that can torch the machine to you. And when people use it, they've realized, wow, this is what I can do with it. Like some of the stories that I love hearing are someone in the Midwest who, you know, one one of my friends is who his sister was telling him about this app that she wanted to see someone she wish someone had created. She described in a detailed way. While he was listening, he was like, "Uh-huh. Uh-huh." Typing into Codeex exactly what she was saying. Pushed enter. Few hours later, he shows her the result. It's app that's exactly what she described. And she's like, what the hell is this? Who built this? Like, this is exactly what I wanted. And he said, you built it. Right. And that is that kind of aha moment that I think that everyone is going to go through. And once you're once you experience what AI can do for you and the fact that you are now empowered like you have a image in your head you want to see in the world. If you look at some of the upcoming uh some of the the new image models we have absolutely incredible like you can create in a way that was never possible before and the fact that if you're like like another story that I think about so my grandmother had dementia Alzheimer's and that it was extremely tough for everyone. But one thing that really stuck out to me was she was actually able to use her Alexa to play music and she could still remember the lyrics and sing along to songs. And it was just a way that she was connected to who she was through technology. And that really stood out to me. It's like if you can build interfaces that are intuitive that anyone can connect to them in ways that right now it feels like we we have pieces of technology that are targeted to specific individuals or that you have to build all this skill that's not really related to deeply what you want but is just somehow specific to the tech itself.

Three three things came up for me during what Greg was saying um which I I agree with all of one before we launched we used to try to talk to people and say this AI thing is coming you got to pay attention it's going to change everything it's really it's really important and it got kind of no attention we wrote these beautiful blog posts we sort of did all these incredibly impressive feats we won you know video game competitions we had a robot that could do a Rubik's cube with a hand we like did all this amazing stuff and we kind of felt pretty cool and we kind of said, "Oh, you know, the New York Times wrote about this. We must be doing great." But really truly no one cared. It got no actual impact. And then we launched Chad GBT, which was by far not the most impressive technological thing we had done by far. By far. And as soon as people could feel it, they're like, "Okay, I understand it." I think that was the moment probably most that the world has said, "Maybe this AI thing is real." like collectively all at once because people can use it and they can get value out of it. They can develop their own sense. It's very different than hearing about it. That's happened again as Greg said with coding models. Um, but those have really been the two kind of like large significant moments where I think the world has updated and said okay there's this thing happening. There will be more in the future, but so far the fact that you can ask a computer anything and get an answer or have a computer do anything anything with code for you. Um, th those have been really powerful and I I think that's how the world's going to update. Us saying super intelligence is coming is going to change everything. You know, maybe the people that listen to this podcast will sort of say, "All right, sounds sounds reasonable. Probably should pay attention to that." But it won't it won't have that big impact on the world. Uh so I think the most important thing we can do to help that audience that people think not just about like what does this mean for people not cheating in class but like what is this going to mean for the world is to ship great delightful products that create a lot of value and are easy for people to use and we will continue to do that. The the second thing is we have seen again and again in our history that when we put something out that's pretty good people say I can't really imagine what happens if this gets better. it can't get much better. This image model that Greg was mentioning, which we'll launch soon, was a real example of that for me. I kind of roughly thought image generation was solved. I was like, it's really good. Like, I don't need it to get any better. There's And then this new thing has been a real reminder of like, wow, this can go so much further and there's so much more I can do.

What does it do?

It makes ridiculously great images. Uh,

has all text? Because I try to make core memory merch mockups.

Try again very soon. Um, great job. The team really did a great job with this one. But even with chat GPT, like when we put GPT4 in ChatGpt, I remember a lot of people like sophisticated friends of mine saying, "This is it. This is AGI. Like if the model got smarter, I don't care. Just make it cheaper. Like I can't. This is amazing." Um, and if you go back and use that whatever it was, uh, I guess like March of 2023 version of GPT4, you'll be like, "This was terrible." But at the time, people were like, "It's solved. It's solved." I mean it's beat the touring test. It's done. It can't get any you know and and then it gets better and better and it like you can just keep raising the expectations and possibility of what you can do to say nothing of then you get into reasoning models and do code. I'm just talking about how much better chatbt got in that time period. So I think we see this again and again too where the world is like okay you made this amazing thing that's as good as it's going to get. That's as good as I need. And then month by month or at least quarter by quarter expectations and ability ratchet up in a huge way. And then the third thing that Greg was saying there is these models are still quite dumb relative to what they will be but more than that they have quite limited awareness of your life. You are still having to like massage them and cajul them and try to get the thing that you want. Uh, we are not no longer that far away from a model that just knows all of your context. It knows about you. It knows about your life. It knows what you're doing. It knows what you care about. It knows about the people in your life. It has access to your computer and your browser and if you want, of course, in the ways you want. It has access maybe increasingly over time to what's happening in in the real world around you. That is going to be a complete change to what it feels like to use a computer and what it feels like to use AI. And I am tremendously excited about that. But I don't think even we have a good intuition yet for what that's really going to feel like.

To that point, yeah, you think about how much time you spend right now just explaining to chat or whatever tool you're using what's going on. And think of how frustrating that is. Like if it's like you had a co-orker, you're constantly trying to explain to them, no, this is kind of what I want. this is what's going on. There's context I could package like that.

Exactly. Right. Like it's it's just not it's not really how you want to how you want these systems to behave. And and and just to say one more thing on on what Sam was saying is that I think that one of the biggest technical challenges we have at OpenAI is too much opportunity. Like AI is this like field of boundless opportunity. No matter what dimension you expand on, there's going to be something new and unprecedented and amazing. And so the important thing is having a vision for where can we really focus where we're going to get the most returns and the most benefit from this, you know, having multiple different efforts that all add up to something. And I think that when it comes to we have these coding systems now, but I think it's going to clearly expand to just all computer work. And it's funny by the way that you know you think about just the degree to which you have all of this like the work that you do it's like very seamlessly interlin between you're having inerson conversation you're typing this on your computer and trying to really figure out how to get the context in there it's going to be such an important problem all these things matter but then also in your personal life you really want this and we're starting to call where we want to go the personal AGI right this AI that really knows you has the context that you you can trust it right that you ask it or you ask it questions related to finances or health and it can give you trustworthy information. All these things are important and it needs that context as well. And so you start to really see that there's this blurry line between an AI that's used for this deep computer work and an AI that's used in your personal life at a technological level. Even if you want different systems because you want one that's really just knows your work context, one that knows your personal context. And so you can pursue those in parallel and build on the same technological foundations. And the thing that I find most amazing about the technology we're building on is it's all at the core one neural net, right? It's still deep learning. You're scaling it up. You're building one system and you're applying it to all these different amazing applications.

I can I ask a I want to ask a question sort of along those lines. I mean, I promised myself I wouldn't do this to you. Um, but I mean I was kind of like I wasn't like a non-believer, but I like I could see AI doing incredible things, but I was I was kind of like skeptical until relatively recently on

what what updated you?

Well, a couple things. I mean, one was I started playing with agents a lot more and really shaping them to do what I wanted and kind of feeling what both of you were talking about. It's like, oh, this is actually saving me a lot of time. It's doing my bidding. It's doing it quite well. And the other is, you know, I cover biotech so much and um just seeing some of the results that are coming out is is it strikes me that like coding and biotech seem like the most obvious places right now where it's really going. But then, you know, I was reading I was talking to you the other day at dinner and reading some of the things that you've written about, you know, I see this clear path um based off LLM to going to what you describe. I still have this part of me though. I think it's because I deal in language so much where the writing is still um not great where I this is nothing other than like anecdotal and intuitive is just it strikes me that the feeling I get back still is like no this is not super intelligence

we we are not there yet on personality

but like but like does LLM get you there but LM's will get us there'll get you there I would just think of it as it's very jagged Right. And I mean it's okay. For example, just in the past couple days, RAI solved this long-standing problem. Uh, so this mathematical mystery that's been, you know, of great interest for for a long time. A mathematician who spent a lot of time working on this many years thinking about this problem posted about his his views of of what was what what the contribution was. Terrence Tao also is saying that this looks like there's like maybe a connection between different fields of mathematics that this AI has discovered and we're starting to see real beauty come out of these machines. Now that's a particular domain, right? Mathematics is very different from creative writing and I think that the fact of these AIs being able to have interesting insights and being able to help like this it you know there's so much that these mathematicians are now saying well think about what more we can do. And so I I think that we have a jagged frontier in terms of what these AIs are good at. We know how to keep pushing that frontier back. But I think that the the thing that we're really looking for if you think about Alph Go move 37, right? That was something that not just it was this deep insight, but it really changed people's understanding of the game of Go and more people play Go now, right? So that it really increased the meaning of what people were doing. And so I think that what we're going to do is we're going to make something that is going to you'll have fewer complaints about it. But I think you're also going to be just like way more able to do the thing that you want in writing than you can imagine today.

I'm sorry to harp on this, but I was told GPD5 was going to be really good at writing. And my like update when I became less cynical was reasoning models. And when Claude delivered like really good research and then you guys had a research feature that was for me like oh my god this is amazing cuz it saved me a lot of time but I was promised before that I would have good writing by now.

The writing feels there's no there's no soul you know there there's something missing.

Well I I'll tell you so from a technical perspective the technology we have is you can train a model in this unsupervised way right. So it really we look at all the publicly available data and it learns to predict what comes next and it's really about putting in a new situation trying to figure out what's a reasonable thing to do. Then we do a reinforcement learning step where it actually tries out ideas and it gets rewards and punishments based on how well it did. You know just positive signal, negative signal, not not

you're not beating it.

Exactly. Not at all. Yeah. Just just you know this this signal. Um, and then the hard part is well how do you judge? How do you decide if something was Yep. thumbs up or thumbs down. And so in math and science, much easier than in some of these more open-ended fields, but we also have AIs that are getting much smarter and much more able to provide that kind of reward signal. And so I think that the part of the challenge has been how do you expand the ta the set of tasks that can be graded and that's been a lot of the focus. And uh there is actually very interesting like a lot of this stuff dates back to the very beginning of Open AI. We kind of had a picture this is how it would go. But I would just say I think we're getting there. I think we definitely have a lot more progress to make. Um, but yeah, hopefully you can keep giving us feedback and we'll we'll be able to improve it for you. One embedded challenge in that is that the the writing you want is very different than the writing that most people want. And right now we have to, you know, make a model that about a billion people use and all kind of kind of like. And what we'd like to do is to get the model so good at personalization that you think it's a great writer and some other person that has a very different kind of life and set of needs than you also thinks it's a great writer for what they want and those are very different things. I think that's what I'm kind of asking though, like for the personal AGI, you would want it to have that sparkle, that soul, that sort of that magic, right? And and I just

if it's not

and I saw, you know, obviously Yan Lun's been arguing against this for a long time. I think I caught this Demis thing very quick where he seemed to be kind of aligned with you guys, but you know that LLMs will get us there, but he seemed to be saying we still need a couple more ticks maybe of something special to to but you guys seem so confident

if the model can solve. We haven't tried to make it be, you know, a what Ashley Vance will think is a great writer. We have tried to make it, you know, solve open math problems that the smartest mathematicians in the world can't solve. And I don't want to say anything about your intelligence relative to these mathematicians, but I I think

writers these writers are smart.

But I was going to say like, you know, solving math is also hard. So if we can do that, I'm pretty confident that this approach can also learn what you think is great writing and figure out how to do it for you.

By by the way, we also have some new new models coming on on this dimension as well. So after this podcast comes out, I can't wait to judge it. Let us know. Let us know if looks better.

Writing. I think personality and I think we're always improving like all the capabilities just moving up the jagged frontier. And so I think in this field it's always really important to judge the not the current position but what has that slope been and also fit it to an exponential. And so if you think about how does the writing stay compared to a year ago sorry we disappointed with DPD5. Uh, don't worry we're we're we're very motivated to uh, you know, to to to really deliver. Um, but I think that I think we really have a line of sight for how to improve it for for every application that people want.

Since day one, the Core Memory podcast has been supported by the fine people at E1 Ventures. They are a young and ambitious VC firm in Silicon Valley investing in young and ambitious companies and people. Thank you so much to E1 Ventures for all your support in the world that you guys have outlined. um this technology okay you know say everything is working great and this you know this is uh it's a quite optimistic scenario um diseases are being cured um resources are becoming more abundant problems are being solved so so humanity as whole is being lifted up it's I still don't see I mean it seems like very very some of the world's smartest people are developing this technology it strikes me it will disproportionately benefit um in almost all scenarios even though like everyone might get lifted up a bit that um you know things really get more extreme because every time um what you guys are talking about about how you would manipulate these tools use these I I just feel like I feel like this things will go even more extreme and the h the joke of the permanent underclass and that sentiment of like seeing these really powerful tools and feeling like oh my gosh What what am I even here for then? And feeling very disenfranchised.

Yeah. Like you can have all this time to do fun creative things. I just feel like other people will be manipulating this in such extreme fantastic ways.

This this is at the core of the OpenAI mission. Like this is really why we started this place because you see this powerful technology coming. It's going to be the most important technology ever created. How do you ensure it benefits everyone? Like truly all of humanity. Like that is in our mission. Ensure AGI benefits all of humanity. And we really mean it. And if you look at our corporate structure, we've tried multiple different iterations at trying to in the structure codify some of our values of how do we ensure that this is something that really does benefit people. You see it in the product choices we make, right? We decided to launch chatbt because we really believe this is technology that we need to be able to put into people's hands. And that was very controversial by the way that there was a different school of thought saying that the way to do it is you have to build it in secret that you can't give people access that you need to do in this other way and I think that when we look at the how do we actually make society resilient how do we actually benefit people it all kind of points to the direction that we we have been going and there's a lot of nuance to it um but I think that the way that we're we're headed is very much that I think the floor for everyone is not going to go up just a bit like I think we are going to head to this world where I mean you think about Even having a doctor in your pocket that's better than the best medical team that anyone in the world could get today accessible to anyone who has a smartphone like that is coming and that'll be free right that is a wild wild fact that's not a small it's going to be a little bit better that is fundamentally raising the floor in this massive way now I think the question of also raising the ceiling I think that's fair too right I think that the question of exactly how the distribution goes and we think about this a lot you look at the openi foundation That's that has you know somewhere between you know 25% 30% somewhere around there of open eye equity right that's like more than 150 billion dollars if it is the case that open becomes very successful that all of that value locked up in the nonprofit like that is something that will be for really benefiting the world and we think that's

well we were at dinner the other day just real quick and I was listening to you talk about what you do with agents and I've known you I mean look you're [ __ ] smart as [ __ ] and I was just like Jesus man Greg with this machine. And I like even though I feel like I'm doing interesting things that help me out, when I heard you describing your life and what you were doing, I was just like, "Oh my god, man. I mean, there's just going to be super people running around." Um, I don't know. It was like kind of intimidating really. I mean, it was cool, but it was also Yeah. quite intimidating. I'm like, "I don't know how to use this like you do at all."

Can I Can I try a less sanitized version of

Yes, please. Um, I hope I

Don't get in trouble for this. Uh, I can see three futures of the world.

Um, I can see one where, as Greg was saying, the floor comes way up. You know, everybody gets subjectively like 10 times richer. Just the sort of materially abundant material abundance, the prosperity is crazy huge. People are like, "Man, relative to my life a decade ago, I'm doing great." But also in that world, these people who really learn how to use agents and get a lot of compute together and whatever, we have some trillionaires, you know, maybe 10 trillionaires, whatever. So like the floor comes way up, like dramatically up. But because this is a lever that people can really use, uh, the, the sort of like most capable, most ambitious, the people who already started rich and have access to a lot of compute, inequality gets worse. So that's, that's one world I can see.

Um, I can see another world where, through many different kinds of things that could happen, um, the floor doesn't come up as much. We don't generate as much total prosperity, but also there's like less inequality. So maybe people in a decade feel like twice as rich, but inequality has come down.

Um, and actually, I'll just like stop and talk to those two because I think this is like the real crux of the issue. Was the third much scarier?

>> No, but it, it's like a distraction from the point I want to make. I think a lot of people in the world, like we think it's obvious, Greg, and I think it's obvious that people should prefer the first. I assume you also agree with that. Um, but emotionally, it's not where a lot of people are.

>> Well, I was going to say, I, I don't, I don't want to claim what is obvious or not. I, but I think it's, I think for me personally, I just see so much potential in this technology, and I think it's just so important from a societal level, just thinking about even American competitiveness. Like, you look at robotics, like I don't think that we are ahead in robotics at all.

>> We are not.

>> But we are not. With the software side, we do have a, we do have this opportunity, and I think that is something all of these factors together in my mind are, are worth considering at once.

Yeah, I mean, clearly, I mean, actually people who disagree a lot of other things, I think do agree that America needs to be competitive on chips and robots and AI and everything else. But, but there's going to be this huge question about how we organize society and the economy. And do we push for maximum prosperity and accept the inequality that'll come with that? Or do we say, like, actually, we're going to constrain it because we're going to be more focused on the relative nature and the fear that, well, if we don't do something, yeah, maybe the world will get more prosperous, but the people who are really good at using this are going to have all this power. And, you know, emotionally, I think intellectually, it's kind of seems clear to me. Emotionally, I really get why it's not clear, and I really get the, the fear that if we let inequality run here, the compounding nature of this tool is something that we don't yet understand.

Now, the last thing I'll say on this is, kind of no matter what, I think everyone should want much more compute, much more infrastructure, uh, and the cheapest possible access to AI, because otherwise, I think you really exacerbate inequality if there's a limited amount of this and the price goes up because of supply and demand, and only the rich people have it.

Yeah, I want, I wanted to just build on that because I think that to some extent, and some of this also plays back to, to earlier, just how Sam and I work together, how we think, right? Because I think that laying out those two options, I guess actually a really good point where it's like, well, are those the only two options, right? Is there something that we're not seeing in the space? And I think that the last thing Sam said to me is an unlock in terms of thinking about, so AI is really opportunity. AI is opportunity for everyone if you have access, right? If you have compute. If you don't have compute, you can't, right? No matter how good you are with agents, if you don't have the compute to run them, you're not going to be able to do much. And so, I think a world where if everyone does have access to compute, and, you know, I think for my generation growing up, we were much better at using computers than our parents, right? We grew up with it. We were, you know, native. And I think that the generation growing up now are going to be as good as I am at using agents, I think they're going to be 10 times better. It's wild to watch that.

>> It is crazy. I mean, I wrote, this is like the first massive story on core memory was that dude who built the nuclear fuser with with Claude. I just remember it was the fuser didn't even sort of matter as much as like what I saw him doing with his computer. I'm like, you just use this thing different, you know, he's like 22 and he has eight AI applications up. And if you can get compute to that kid, if you can get compute to every kid,

>> Then it is really going to be the most extreme version of the American dream, right? Where it's really anyone who has the desire, the willpower to really play with this technology, to really lean in, to build with it, to get the most out of it. I think you're going to see this outperformance, and I think you're going to see people have all sorts of mobility. And so I think that to some extent, the right way of answering this question of option one versus option two is to say exactly something else. And I think it's possible.

>> I feel like we're doing some broad stuff and and we have a bunch of specific questions. I'm gonna, I'm gonna be bad and ask one more broad one. That's okay. Just based on what we were talking, I mean, you guys probably know, maybe you don't. You know, I just cover hardware so deeply. And, um, I think I'm pretty, I would venture to say I probably been to more factories and hardware startups in the US than just about anyone and keep an eye on China. I mean, the thing that always plays through my head is that yeah, we're ahead on software, we're ahead on AI. Software has been this the story of the US for the last 40, 50 years of our our strength. But, you know, when I think about how this technology manifests itself in the physical world, I do not see a future where the US is competitive in any way, shape, or form, not just in robotics, but, you know, just all the componentry that goes into these hardware systems. I go to El Segundo, I see 30 startups. It's great. There's like this flourishing of ideas and people trying things, but it's like such small potatoes compared to what I see in China. Um, it, it just strikes me that there obviously will be a physical manifestation of all this in the world, and the US just seems like extraordinarily disadvantaged to to win that, to me.

>> Well, I think of people who are working incredibly hard to try and change that. I'd say Sam is, like on robotics in, well, I know through through all the investments in different hardware companies or,

>> No, no, we're we're trying to figure out how robots are an obvious part of this, and we're trying to figure out how to be very successful at robotics. Uh, the, I think if you could pick one thing to make the US competitive at manufacturing and the world of atoms in general, you would say we need a lot of robots that can build a lot, lot more robots. But like, we can't even make an actuator, like, you know, like,

>> We're gonna figure that out. We, we will.

>> Okay. Like through your robotics.

>> We have to. We have to. Uh, the, but there's other parts of this that really matter. If you think about, like, the number of gigawatts of power that the world is going to need just to support AI workloads, the number of chips we're going to have to fabricate, the, the boring stuff about like how we're going to just get enough racks assembled and enough network cables strung and made in the first place. Uh, the US is extremely behind here. I think robotics will be the solution if you just look at how fast this has to happen and what the US is currently good at and how much infrastructure we have to build up and how long that would take to do by hand. Um, but you are right in the diagnosis and the criticality of it. And,

>> I think luckily we have like a new, uh, like a new piece on the chessboard through OpenAI or the United States, United States.

>> The United. Tell, I mean, what do you mean?

>> I, I have a different feeling. I, I feel like we're play acting at hardware and we're gonna get completely hosed.

>> Well, on the code trajectory, yes. No, we totally agree with that. Uh, the, if we can make true general purpose robots, if we can have a Codex equivalent power thing for robots that you can say like, go configure a factory this way and make me more of this kind of robot, or go figure out how to mine and refine this kind of thing.

>> Then the playing field changes.

>> Okay. But you said the US and not OpenAI specifically, but I feel like you're pointing at something that you know, though.

>> Well, I don't think the US has a credible plan other than this kind of AI plus robotics to catch up fast enough.

>> I didn't know what the chess piece was. Did something you saw something recently?

>> Oh, I, I meant general purpose robots. Chess piece. Yeah. I feel like there's there's a real chicken and egg that's existed in robotics where if you don't have the robotic hardware, it's kind of hard to develop the software, and if you don't have the great brain, it's kind of hard to be motivated to develop the hardware. And we really saw this. We had a robotics project back in 2018. Do you remember the robotic hand?

>> Yeah. Yeah.

>> Super cool. The thing about that hand, though, so we train it through reinforcement learning, actually the exact same algorithm we used to solve competitive video games, right? Wild. Single piece of technology doing both. The one difference is that this hand would run for 20 hours before it had these strings as its tendons, and they would snap, and then you'd have downtime. Mechanical engineer would come in and fix it, you know, maybe be wake them up, have them come into the office, and you realize like, you cannot do ML that way. You just cannot. So actually, we ended up canceling that project. That team went on to go work on what became GitHub Copilot, you know, to contribute to that effort. And so you just realize how much faster software world has been moving. But I think we're at a point now where we have these amazing general purpose algorithms that can be used in all sorts of different ways. And we see how if you start to be able to apply those to the physical world, then I think you are going to have a very different story for how the hardware development and how the hardware tuning to the code design is going to happen. So I think that there's like real potential for change, but we as a nation need to have the willpower to really do this. But I think we really agree with you that without something like that, the current trajectory looks terrible.

>> We're so [ __ ]

>> He's convinced us.

>> It's pretty bad.

>> Taking a huge step back, you know, robotics, models, these are things that you guys are interested in building and really care about. But you guys have recently, it's been reported that you consolidated and focused. And I'm really curious. I'm sure like the audience wants to know like what's on the table now? What got cut? What do you guys care about? And why did you make those cuts? Why was it important?

>> So, before I'll let Greg answer, before he does, Greg has taken over really figuring out what our cohesive product offering and the research support that was going to be. Um, and it's been amazingly, uh, joyful inside of the company. Like, it's going to take a little bit longer for all the stuff to ship. He's only been in the role for a few weeks, maybe something like that. But, uh, the, the energy and excitement and the sort of like enthusiasm about what Greg is doing here is unbelievable. So, you can say what it's going to be.

>> And like, can you expand on? So like a few weeks ago you came in and because some of these cuts were already taking place, right? I mean, so, but you've come in and like assessed everything.

>> Uh, so I've, I've always been very involved behind the scenes with, you know, many parts of OpenAI. And so I think that kind of taking a, a foreground, uh, role is relatively recent, in this particular area. Although, fun fact is, I actually built the very first version of the API. So like, I've been doing product since product existed at OpenAI and always deeply cared about it. And there's a bunch of things I could say there. Like, I think that how core it is to our mission, uh, was something we didn't appreciate before we started building products, and afterwards, you realize how important it is. So that's why I've always been so close to it. So the place that we're at now is that we are clearly at a moment of transition to agents. No question. Right? People in software engineering, you've been feeling this for, let's say, the past six months. And that, you know, over the course of '25, I think there was a transition from, yeah, it's like kind of autocomplete to, okay, yeah, you have a sidebar in your editor and that you'll start talking mostly over there, but you're still mostly doing the same kind of software development you were doing before. To now, it's like, actually, you want a tool like Codex that is really an agent management platform, and the agents are going to take care of all the details. The agents are going to do all the, the nitty-gritty work. And there's still probably 20% of how you fit things together and the, the way that you structure your code and exactly some of these these higher level things that you would normally put in an architecture document that the human still really cares about and wants to manage. But the details of exactly the code, like nope, that's what agents are for. And so the question that we had is, first of all, how do we really rise to this moment? Because it's not just software, right? We see line of sight for every single vertical, right? For law, for finance, some of the mechanical skills there of writing, you know, creating spreadsheets and presentations. How to make our models extremely good at those, right? It's like you work with domain experts, you produce evaluations, you produce training data, you have the AI actually take its great domain knowledge and applied in these verticals and gets experience and get those, you know, yep, you did a good job in order to figure out what, what good looks like. So, we have the mechanical, the, the vision of exactly how to do this, but we need to make sure that we're building the right product surface to unlock all of it. And one thing that we have found is that the models have shifted from being the product to being a part of the product, right? That we used to have these very thin layers of software on top of them, and you didn't have to like think that hard about how it was architected. But now it's a very fat layer, right? That you have things like skills, connectors, you have exactly how you hook up to compute, use, how you manage context and memory, and all these things. And so there's just this deep layer of software, which is almost, you could think of it as the AI. It's kind of like we have this brain in the form of the model, and now we're building the body. Both are hard. They have to be co-designed together. And so a lot of what we're focusing on is number one, getting together an amazing agent platform. Like that is the number one focus that we are delivering on. We have teams that are executing extremely well on this. I'm super excited about what we'll be releasing over upcoming weeks. Um, the second is, where do you actually want to apply these agents? And that our priority there is really towards computer work, right? So that I use that term very deliberately, by the way. People like to talk about knowledge work, but no one thinks of themselves as a knowledge worker, right? Like that's not a thing that people do. It's kind of a term that's like almost removed from what the actual thing is. But the thing I like about computer work is it's like, I don't really want to do computer work. That doesn't sound like the thing that I want. But you realize how much of your time you do spend doing it, like chain behind your desk, you know, typing away, hunching over your shoulders, getting your carpal tunnel, all of those things. And so we are focusing on that, bringing Codex that exists today, not just for software engineers, but really making Codex be for everyone. And that's something that's coming very quickly. We'll have some updates even coming today as of this podcast filming. And, uh, there's a lot of exciting things that are still in the pipeline for that direction. And then the third thing is really thinking about personal AGI, which is about, think about ChatGPT right now, used by a billion users, and every single person on the planet is going to want an AI that represents them, that has their context, that they have built trust with, that it's not just something that you build, or it's not just something that you talk to one-on-one, but it can be out there doing things for you. For example, maybe it knows that you like a specific musician, and that musician's in town, and it notices it proactively, and tickets have just become available. There are some great tickets that are very cheap that you could get for you. It just goes and buys them, right? And maybe it knows it's built trust with you, so it kind of knows, yeah, I'm allowed to to do this without getting approval. Or maybe it realizes, I should probably, I'm not quite sure, I should check. And so we're building that as well. And if you think about these things, they all are kind of expressions of something that fits together into a cohesive whole. Like, in the end, you fast forward to where we're going, you really just want an AGI, right? You don't want a language model. You don't want threads. You don't want any of these like details. You just want something that is helping you, that is operating on your behalf, that is able to help you solve problems, that knows what your goals are and achieve those in work context, personal context. And so this is what we are prioritizing and building.

>> Yeah.

>> And so, you know, to me, one of the important questions, you know, people are asking, how do we think about consumer? How do we think about enterprise? And the answer is, if you take the definitions of these words as they exist today, we care a lot about consumer. We care a lot about enterprise. But I think that the meaning of these words will change and blur, because what we're doing is we are going to, so we're going to unlock this wave of entrepreneurship again. I think we're seeing the leading edges of it. Small companies be able to get tons and tons of revenue that was not possible before. And that's been a trend for a while. It's just going to really accelerate. Is that enterprise? Is that consumer? It's kind of neither, right? And so I think we are really focused on solving goals across all contexts. And that is the lens that we, we look at things through. And so that has meant, that has has meant we need to deprioritize other things that are also amazing on their own. So,

>> Well, Sora is the most, Sora is the most obvious one.

>> And why?

>> Well, because it's a different branch of the tech tree, right? So if you look at the models that actually power Sora, that they're not unified with the core GPT series. And secondly, the use case is not quite unified either, right? Doesn't fall as far under this goal. Like there's creative expression, there's something very important there. I think it was like an incredible model, the team does incredible work, and I think that technology will live on for other applications. But what we were really focusing on was this, the product suite that we want to be delivering, what do we want to be doing over the next 3 months, 6 months, 12 months? And the thing I described, by the way, is just step one, because we also see line of sight to much more powerful models. Like, you look at what we're doing right now in mathematics, it's okay. It's actually kind of mind-blowing where this result that I just mentioned of solving the new Erdős problem that seems actually really significant. I was just someone using GPT 5.4 Pro like two years ago.

>> Yeah. We used to like train our model. We had a team of 20 people to try to train our models to go solve a computing Olympiad, and we got a bronze medal. A team of 20 people for like two weeks and lots of compute. And now it's just this model that we trained very casually. Someone is able to point it at problems and get this kind of result. What if you point that at drug discovery? What if you do take that team of 20 people and all that compute and you really try to push it for scientific discovery? And that's something no one is pricing in right now. So I think it's rising to the moment of agents, really trying to make sure that the product investments we're making are sort of well structured, that we think about how all these pieces fit together, that we have connectors that work really well, and each of these pieces can be composed. And really also build an ecosystem, because it's not just about what we built, right? That we want to build some example agents. You think of Codex almost as an example agent, but it should be that if you're a developer, if you're someone with a creative idea, you can build your own agent, right? That you can build it for your application, for your purpose. You care about the specific math problem, you should be able to apply the agent to that math problem. And so, we're, we're enabling all of that.

>> But you, I mean, okay, just bear with me for one second, just to go quick through a couple things around this. I mean, so I mean, part of this was like you had to probably cut Sora for compute to get compute, right? So that comput-

>> It was reported that it was taking up a lot of compute.

>> I mean, every everything in this field that is successful, it's going to take a lot of compute.

>> I mean, the two of you in particular, and I think Ilya and some others, you know, kind of famous for going all in on where you were going to direct the compute in the early days, and now you have to make these very difficult decisions. You have to serve all these customers that you have. You got to make some money because you're spending a lot of money. And, you know, so you two are both wired to take the very biggest bet possible, but you're constrained now by the business in to some degree. So, um, that just seems very difficult. I feel like it's not in your nature. You're looking skeptical. It's a strange phrasing because I don't feel constrained by the business. I feel enabled by the business because the business is what has really allowed us to say we can scale compute, right? I remember when we launched Hatchp, and I remember we were talking right afterwards, and we're trying to figure out how much compute to buy, and I was just like, we got to buy it all. We just got to, we got to do it because it's just so clear there's so much demand. And I think that that has been a huge unlock in our ability to get lots of compute for,

>> I think without this incredible revenue machine, we would not be able to convince anyone that we should get all this.

>> Get this. But then the impression obviously when you see the stories about Stargate and things like that is that you guys are somehow pulling back on infrastructure.

>> I don't know where that's coming from. Uh, like there, you know, there will be like a site here and there we say, okay, you know what, this particular site maybe it only has air cooling, and so this is not as valuable to us as this other site. And that, like, there's specifics. And people really want to write the story of like pulling back. But very soon, it will be again, like, OpenAI is so reckless, how can they be spending this crazy amount? So like the media will go, will flip out either way, just because you all need something to write about, I guess. But we will keep building out as much compute as we possibly can.

>> One, one thing I think is also worth thinking about is that compute for us is not a cost center, it's a profit center, right? When you deploy it in the product. And so in many ways, our business is extremely simple, right? We rent or buy compute, and then we resell it at a margin. And as long as we have some positive margin on it, then it's scalable, right? Because the demand is just unlimited.

>> So, okay. So, data center hardware still all systems go? Or?

>> You mean like our own chip?

>> Yeah. Own chip, networking.

>> Yeah. Very, very excited about our, our chip.

>> Have an incredible team. And Titan when?

>> Uh, no, we're not talking about timelines, of course. But, uh, I'll just say that like, I spend a lot of time with that team, and I think it's,

>> They do a great job.

>> So that's how system go. Robotics sounds like it's all systems.

>> It's going to be a while till we have something to where you're like, ah, this is the ChatGPT moment. But,

>> But you haven't on that program. Social robots, are robots not the current thing, but are going to be so clearly important in the future.

>> And social network?

>> Not doing that. Not many. And then clearly like the super app, the browser, all that stuff's still still going.

>> Yeah. And one, one thing to realize about super app, cuz I, I feel like it's one of these things where it's like a catchy word, but,

>> I thought you guys like came up with this.

>> I mean, but used as internal short. Exactly. This is the thing to realize is that like, sometimes we're communicating to to our team, and then of course it ends up being communications to the world, and these are not intended to be the same thing. Super app in many ways, it's an iceberg, right? It's like, yes, we're going to have an app, and you'll see updates to what the Codex app is today to make it so its Codex is for everyone. And I think that ultimately, like what that becomes, that we have, we have a lot of steps to get to where we want to be. But it's really about saying we're building this unified agentic infrastructure that I described, and that is, I think, going to be a huge unlock for every single thing that people want to do.

>> Okay. Um, you and I have talked a little bit about, you know, you guys have had lots of drama. It slowed you down at times. You still have drama, sadly. Um, there's a lawsuit coming. Uh,

>> That'll be fun.

>> Yes. Yes. Well, let's talk about that in a second. Um, like, which company do you think has actually executed better over the last two years? OpenAI or Anthropic?

>> Look, I think that this is a hard thing to say from just the current moment, right? Because I think that the view in my mind is, we got to step back and really think about what is it that we're all here to do, right? And I think that from the OpenAI perspective, the things we've been talking about, in some ways are about, it's very clear that yes, selling to enterprises, like figuring out how to really deliver these coding tools. And it's been really, it's been actually, I think competition can really help elevate your own thinking and realize that, hey, we need to focus on this. And one example in coding was that I think we got late to the game of not just building models that were good in the abstract at coding, like we always had the best numbers on programming competitions, but you also need to apply them to messy repos, real world data, those kinds of things. And that was something that I think we had appreciated later than Anthropic did. So I think that's, that's kudos to them, but also something that has helped elevate our own execution. And now, head-to-head, Codex versus Claude, I think that that, that we get very favorable results. Um, and I think we've done an incredible, that our teams have done an incredible job across the whole company to really build a product that's not just competitive, but actually ahead in many, many ways. Um, but I think that the, the core, it's never about the ups and downs of the news cycle. It's about progress towards AGI, towards it benefiting everyone. And that is something I think we've been extremely focused on. The team's executing extremely well. And, uh, so I just want to say that you can't always tell from the outside, but if you think about the things we talked about, that there's this focus, but on many time scales that all add up to where we need to go.

Yeah, there's a million questions I want to ask with the time we have left, but one of them, we, we touched on this, like making sure a bunch of models are in the hands of a bunch of people and they have that powerful technology, but we've reached a point where now some of these models are too powerful. We're being told, and they're they're gated for only certain companies. Uh, Claude and Mythos have really made a lot of headlines and I think a lot of, a lot more fear. And I'm wondering what you guys think of this new moment. Are we getting to a point where like, we need to start keeping these powerful models behind the scenes rather than in the hands of everybody?

>> There are people in the world who for a long time have wanted to keep AI in the hands of a smaller group of people. Um, you can justify that in a lot of different ways, and some of it's real, like there are going to be legitimate safety concerns. Um, but I expect, but if what you want is like, we need control of AI, just us, because we're the trustworthy people, I think the, the fear-based marketing is probably the most effective way to justify that. Um, that doesn't mean it's not legitimate in some cases. Uh, but it is, you know, clearly incredible marketing to say, "We have built a bomb. We're about to drop it on your head. We will sell you a bomb shelter for $100 million. You need it to like run across all your stuff, but only if we like pick you as a customer." It, the way that we view balancing these new capabilities that are going to come with, uh, still our, our belief that the world needs to get and use and understand and come up with new ideas for this technology is not always easy. We, we have had, um, cybersecurity in our preparedness framework for a long time, and we have been building mitigations to figure out how we release this, um, how we put these models in the hands of a trusted access program, how, how we then put more capable models in the hands of everybody. Um, but there will be a lot more rhetoric about models that are too dangerous to release. There will also be very dangerous models that will have to be released in different ways. But to the point Greg was making about, the goal here is to benefit everybody, and also to, I don't want to say market this in a way, but but like get the world to come along on this journey with us, where where it's like, we are going to give you more powerful technology. There's going to be responsibility that goes along with that. We are going to help set up the world for success as much as we can. Um, but we will try to avoid the fear-based marketing as much as we can.

>> So I'll just ask you directly. Do you think Mythos is just a lot of marketing?

>> I'm sure it's a great cybersecurity. Like, again, we've been talking about this for a long time. Like this has been in our, in our model. But, but there's a way of saying like, like our version of this is to say, these models are going to get much better at cyber. We have this preparedness framework category. Here is our plan for how we deploy these into the world. Here's how this trusted access program looks like. Here's the mitigations we put on the models. Um, I think Anthropic has not had it as one of the categories in their preparedness framework. Uh, so I'm sure Mythos is a great model, uh, for cybersecurity, but I think we have a plan we feel good about for how we put this kind of capability out into the world.

When the Anthropic stuff was going down with the Department of War, I mean, from my perspective, it looked like, um, you had sort of David Sachs and people, um, with long ties to Elon. I could see putting pressure to make some of these things happen and have the government focus in on that. Um, you guys kind of came in pretty quick after and and had an announcement of your own that was more favorable. Um, I mean, you know, Elon's aggressive against you guys as well. And some of this is funny to me that, I mean, there is a world where like you guys and Anthropic are actually on this, this other side, and it feels like Elon and even, um, Zuck to some degree is my understanding, kind of pressuring things, um, as a side. Yeah. I mean, did you, did you feel like Anthropic was treated fairly in that? Do you feel-

>> No. Um, I don't, well, I think there was like a lot of bad behavior to go around there, but I don't think Anthropic was treated well.

>> Can you explain more like what was standing out to you about what was particularly wrong?

>> Um, I don't, look, with the things that have happened since with, you know, these models reaching this cybersecurity threshold that is clearly in the national security interest. I, I think it all has a little bit of a different flavor to it. But, you know, like threats of using the DPA and actually using supply chain risk designation, like this is, this is not the relationship that I think our government and our AI efforts need to have. We really care about supporting the US government. I think that's going to become increasingly important as these models get more capable. And I certainly don't think it's like a good stance for, uh, labs to say, you know, we have this super weapon. By the way, we're not going to work with you to help help you defend the country. Um, but I also don't think it's good for the government to be like fighting this stuff out in the press and using the big hammer that the government has, that needs to be used very rarely. So, I, a thing about OpenAI is we generally try to be like moderate and centrist and reasonable, and that is what we've tried to do here with the US government. But I see no future, no good future where leading AI efforts don't assist the US government. If everything we say is right, if we believe the things we're saying about where the models are going to go, which I certainly do, then the government needs our help, and we are honored to get to provide it.

You know, when I, I'm old as a tech reporter, I mean, when I was first writing about tech, you know, the big war was, it was basically Microsoft versus everybody else. And it was, they were the big evil, um, proprietary software company. You had the,

>> Open source companies counter. You had Mark Andreessen with Netscape as, you know, and, and, you know, like things would play out in the press, and it would be a little vitriol here and there, but, you know, and then people had their philosophical and religious camps. Obviously, like this stuff where we are in AI land is insane. That so much of this seems tied to the personalities of you, Elon, um, Zuck, Dario, and Demis, and various animosities and worldviews. Like, you wrote about this Shakespearean drama of it. Like, how do we possibly get past this? I mean, you guys are all so, um, I feel, I, I don't see a path out. And, and obviously, like we mentioned, there's a lawsuit coming that will only,

>> Wouldn't hold hands on stage, which I'm so sorry to say.

>> Some, some of the people involved only trust themselves to get it right. And because they think the stakes are infinite, and because they don't, you know, for different reasons, uh, they don't think anybody else can do it right, or they don't want anybody else to do it. Um, that leads to, I think, some very toxic behavior. We can't control the behavior of other people. Um, but we will continue to advocate for doing this as a collective project that humanity has to get right together, and not something where it should be about one person or one ideology winning or losing, or, you know, having a soul victory or something like that.

I mean, we'd be remiss, and I know it's super sensitive, but go ahead. I mean, it, you know, if you look at what happened to you personally over the last week, there's tons of science fiction books that have been written about what happens when, um, the factions that really don't want to see technological progress happen get, um, you know, things get more extreme. Um, I don't know, like, there's an argument to be made that this, this switch is now flipping, um, along those lines.

>> With people that want to like stop.

>> Yeah. I mean, you know, I've, I've read Richard Clarke's book years ago where a Bill Joy like figure decides AI can't happen, and he's running around destroying data centers. You know, this stuff has been out there forever. It just feels like, God, I mean, I don't see how things could get like more heightened. And so, um, it doesn't seem like it's getting better.

>> I, I assume it will go up and down, but directionally more heightened.

This must, like you wrote about. I mean, clearly this must be horrifying. Horri-

>> Yeah. Yeah. I mean, I don't think I have anything deep to say here. Uh, that was a crazy way to wake up. Um, that the first day I was sort of in this like kind of adrenaline shock about it and just trying to like figure out logistics. And then the day after I was just like, you know, there's going to be more stuff like this, and it's incredibly disheartening. I went through a real depressive cycle about it. Um, but it's very scary, and I, yeah, I don't think I have anything super deep to say. I think the doomism talk hasn't helped. Uh, I think the way certain other labs talk about us hasn't helped. I actually don't want to make like a side. I think the way Anthropic talks about OpenAI doesn't help. Um, and, you know, I, like, I hope that cooler times will prevail. One, one thing I do want to say, just to, you know, to the point of we can't control other people, we can control ourselves, like one thing I, I've been just really impressed by and astounded by with Sam is just how throughout this time, like that very day, he was there doing things that absolutely required him. Like, he just, like, keeps pushing on the mission. And I think that is something that I don't take for granted at all. Like, I just think that the degree of resilience that is represented by Sam is, uh, is extreme, dream, and I think very underappreciated.

>> The, I mean, I've seen the same thing. I don't, well, people be like, oh, you're taking it easy or something like that. I don't know. I, I am, um, it's hard to think of like another individual over the last three years who's been through more dramatic, uh, business and and personal good.

>> The narrative has been quite dramatic.

>> People won't be sympathetic. They're like, you're,

>> Won't be sympathetic, then that's fine. Um, I, I, I said just the night before that this thing happened at my house. Um, I had some people over for dinner, and we were talking, like, some work colleagues, and we were talking about the next phase, and they were like, ah, it's been like a, you know, brutal time for you in the press, but I, I guess it's like good in some sense. And I was like, h, you know, at least no one tried to kill me.

>> Um,

>> You said that the night before. Uh, and, and it, but it really did put into perspective the, like, you know, people can say all the mean things they want, and as long as they only say the mean things, it's like, it's really not that big of a deal. You keep going.

>> You always told me your dream is to like retire out in Napa. I mean, like, so like why not?

>> We got a lot of work left.

>> But like, somebody's going to get, it's clear somebody's going to get to AGI now. We got like five companies that could reasonably do this.

>> Which company would you most like to get there first? Is that, is that what it comes down to?

>> No, cuz look, I think that the point, because the missing thing is this point of how do we help people really understand what it is that this technology can do for them. And I think that's something that every company can contribute to, right? That I think is something we have a perspective on. We've talked a lot about in this podcast. And fundamentally, I think that that is something that is kind of the responsibility and on all the people who are trying to create this technology is also to show its benefits and why people should want it. Why should people be protecting and defending the ability to create it? Like, why does America need to have leadership here? Like, why is all this good for not just the country, but you personally, your future, the future of your kids? And that's something that we wake up thinking about, like, like, I would say every day. Like, I don't think it's an overstatement. Like, this is something we talk about constantly. We think about whether it's weird creative, you know, legal structures for our company, we've done that many times because we're trying to solve for this mission, for this thing that we think is so important. And that's something where if other people want to contribute to that, the merrier. Like, that's something we should all be doing. But it's something that drives us uniquely. We so believe that if we can deliver technology that enables prosperity for everybody, if we can give people more agency over the future, uh, that will, through some pits and starts, lead to a better world. I don't think all people working in the field believe that. But anyone who does, we are delighted to work with, and that is the mission that, you know, we want to move the world towards.

Going back,

>> I think, well, I think we have time for like a couple questions.

>> Okay. Do you have wish list questions?

>> I've got, can I'm going to answer real fast. Yeah. I actually want to end on the one that I'm about to ask. So, I'll pick this one.

>> Okay.

>> All right. We're ready. We're ready.

>> I mean, how, there is like a, how existential do you view this trial? I, I actually think it's a real opportunity for us to tell our story, because if you look at the course of OpenAI, we have really let the other sides when there's splits, tell the story.

We really have done our best not to sort of say, "Well, well, that's not really what happened." Like, let's talk about the truth. And that, in this case, we finally have no choice, right? Because we have to defend ourselves. We have to tell the truth. We have to tell what happened. And I'm extremely proud. Like, I've spent a lot of time looking back at the history, at a bunch of different messages. And of course, there's always things that you can like try to like cherry-pick, be like, "Aha, you said this thing."

"Your diaries are are famous."

I know, right? But the thing is that they're not. And first of all, incredibly personal documents. Extremely painful to have something so personal be sort of taken from you and then, you know, attempted to be weaponized. But those particular, you know, sentences are kind of the worst thing that the opposition could find. You're like, "Really?" And the thing is, they're all taken out of context, right? The question of, you know, "Hey, like, we're in the middle of this negotiation, right? We've all agreed the only path forward for OpenAI is a for-profit. Sam, Ilia, Greg, Elon, we all agreed on this. We all said this is what we got to do. This is literally the thing for the mission."

And now you're on this crazy negotiation, right? Elon's like, "Need majority equity. Need to be CEO. Need full control." And we got so close. It's like, "Okay, fine. Okay. We're not going to be equal partners. You need this this mass amount of equity. If that's what you say you need, like, we would like you to be involved. We can get there. Okay, fine. You know, Sam will be CEO. Elon will be CEO. He needs it so everyone knows he's in charge. Fine."

But absolute control. Absolute control over OpenAI. Even if you say, "Well, I'll dilute down. I'll give it up in the future." And you're like, "What is our mission? Like, do we really believe in our mission? Do we really care about this picture of we want this technology to benefit everyone? And there shouldn't be one person in charge of the whole future. Doesn't matter who that person is." Like, that was the breaking point. That was the thing that caused us to say no. And so we've never told that story for years, but now we will. And so I, I think that it is a real opportunity for people to understand what truly motivates us, what we truly stand for. I think it's insane that he's doing this, but I'm kind of my fear at this point is he decides to like drop the case right before the trial and we don't get to do all this. But I am happy to like explain all this to the world and have this chapter behind us.

"Okay. So my my question is rounding back to the beginning of this podcast, like I was talking about this narrative. I feel like I, as an AI reporter, grew up with hearing of, you know, we're all going to die and we're all going to lose our jobs. I feel like I heard that time and time again. And now portions of your personal notebook are public. Um, I'm curious again, how would you have talked about this differently with hindsight being 20/20?"

Well, I think that the way that we think about it today is the way that I wish we had talked about it in some sense, but I don't know we could have with the knowledge we had at the time. For example, even some of it is related to the technology itself. We had a picture in 2017, 2018, the way we would build AGI was through a competitive multi-agent simulation. Like imagine an island of like a thousand agents that all are in a battle to survive and replicate. And you could see that if you put a ton of compute into it, maybe it would build something very smart. But that smart thing would not be connected to human values at all. You'd have to have a separate step of figuring out, "How do you even talk to this thing, right?" Didn't grow up at all in the real world. Doesn't have any concept of language. Doesn't have any connection to our reality. It's just smart and powerful. That's a very scary system, right? You start from a place of of trying to think about how could you possibly align it.

But instead, we have this language model route, which is rooted in our values, which is rooted in understanding humans. And we have chain of thought that actually you can monitor, that actually if you approach things right in a technical sense, you have a path to make that be faithful. So it really represents what is truly motivating to AI. And you just realize we have a totally different technological path to get to the outcome we're talking about, and it's a much more optimistic one. And so I think that there was some technical learnings we had to have for what is this technology truly going to be? How will it be created? What are the ways in which you make it useful? And that is something I think we could not have appreciated, but we do today.

"Um, this has been a fascinating discussion. It's um, really kind of you guys to both show up and do this together. Um, we we have but a humble podcast. We're honored um for for you guys to spend this time. And I guess I mean the..."

"Yeah."

"Obviously we cover a lot of ground, but the headline is in relatively short order, new models..."

"Really good new models."

"Yeah."

"Um, well, thank you guys. Thank you guys so much. Thank you for having..."

"New Models useful for everyone."

"Thank you guys."

"Thank you."

"The Core Memory podcast is hosted by me, Ashley Vance, andor Kylie Robinson, or both of us together. It is produced by me and David Nicholson. Our theme song is by James Mercer and John Sortland. And the show is edited always by John Sortland. Thank you so much to Brex and Ewan Ventures for all your support. And thank you most of all to everybody for listening or watching. We love you. Please leave us a like, a review, a subscribe, all those tremendous things. Thank you and we'll see you again."

Get Our Mobile App

The OpenAI Founders On Their Plan To Battle Elon, Compute And Everything Else

Transcription