OpenAI cofounder Greg Brockman on the scaling hypothesis and refactoring as a killer AI use case

Greg Brockman—OpenAI cofounder and Stripe's first engineer—joins John Collison to talk about research-driven product development, an early moment he thought OpenAI was doomed, S curves in AI advancement, and energy bottlenecks. Full episode transcript: https://cheekypint.transistor.fm/1/transcript Timestamps (00:00) Intro (02:51) Was OpenAI the first company to take the scaling hypothesis seriously? (04:53) Lessons from Dota about deep learning (08:08) What is a good new Turing test? (08:57) Personalization in AI (09:57) Research-driven product development (10:26) An early moment OpenAI felt doomed (15:01) OS limits on AI product development (17:59) When will AI make novel advancements in math or science? (20:03) Energy bottlenecks (22:30) S curves in AI advancement (24:00) AI coding (26:25) Refactoring as a killer AI use case (27:26) How OpenAI decides what products to built (28:53) Growing up in North Dakota (30:17) How far away is AGI?

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Speaker 00

  This is totally backwards from how you're supposed to do a startup, right? You're supposed to have a problem and we had no idea what the problem was. Is there a world where the AI becomes the manager and that gives you ideas and gives you some tasks to do? This was probably the hardest project that I've ever done because it felt totally doomed, right? It's like, I know like every instinct, every builder instinct of mine.

• 0:17John Collison?

  Did it actually feel doomed?

• 0:18Speaker 00

  Oh, it felt totally doomed.

• 0:20John Collison?

  Greg in 2010 dropped out of MIT to become our first engineer and went on to become Stripe's CTO. In 2015, after he left, he co-founded OpenAI. OpenAI is really cooking at the moment and Greg is one of the most productive people I know.

• 0:33Speaker 00

  Cheers.

• 0:37Speaker 00

  Wow, you really have some old school photos.

• 0:40John Collison?

  Some deep cuts. GDB trivia, so yeah, we figured we gotta put people at ease, you know, make people feel at home.

• 0:46Speaker 00

  Very nice, thank you.

• 0:47John Collison?

  Okay, well, I just dive straight into all the questions I have, which is the last.

• 0:51Speaker 00

  All right.

• 0:51John Collison?

  If you were not working in AI, how could one have known that something was about to start working? People were telling you that AI was the future in the 1970s, in the 1980s, in the 1990s. And then very quickly in the late 2010s, everything started happening.

• 1:04Speaker 00

  Well, you know, I was someone who was not in the field and so I remember very much what it was like. You know, 2013, 2014, it felt like every day on Hacker News, there'd be a new deep learning for X article. And I remember being like, what is deep learning? And, you know, I knew like one person in the field and I asked them to introduce me to more people in the field. And I just kept getting introduced to a bunch of my smartest friends from college. Now, if you actually look at the work that was being done, in 2012, basically, you know, image recognition for the first time, you could solve with a neural net much better than anything else. And it just blew all these traditional computer vision approaches out of the water. It's like this like learned system that is able to outperform 40 years worth of let's like write down all the rules and try to like, you know, sort of handcraft the algorithm for the task.

• 1:49Speaker 00

  And It's very easy to then be like, okay, well, this approach sure works for computer vision, but it's never going to work for machine translation. And you know, 2014, suddenly you're getting great results in machine translation. And I think that this pattern was applied in subfield after subfield.

• 2:05John Collison?

  One thing I've been wondering about is so many different things are finally working at the same time. And so we have L Ls, which are obviously amazing. Then we also separately have image models really working. And we also have text to speech and speech to text working way better than they were before. And so what's the common factor behind everything starting to work at the same time?

• 2:26Speaker 00

  Well, it's deep learning, right? I think deep learning is the core.

• 2:29John Collison?

  We had deep learning for a long time. Like, you know, why didn't deep learning work in the 1980s?

• 2:32Speaker 00

  Well, so, okay, so you look at the number of orders of magnitude of compute that we've, we've gone through from 1940 to today. I mean, it's, it's just astounding.

• 2:40John Collison?

  You think they're all explained by compute scale ups applied to the right algorithms?

• 2:44Speaker 00

  Of course, the type of algorithm changes and some of those results aren't even deep learning. But I think that fundamentally it is about compute and you need an algorithm that is scalable to actually absorb that compute.

• 2:52John Collison?

  Was OpenAI the first company to take the scaling hypothesis really seriously?

• 2:57Speaker 00

  I think that claiming the first is always difficult, but I think that it is clear that we sort of succeeded much more wildly sooner than anyone else. And so I think that we have real conviction behind what we needed to do. You know, some people think that OpenAI set out to prove the scale hypothesis, whereas it was almost the other way around. The scale hypothesis is what we observed as the thing that was working for us. And, you know, really saw it for the first time, actually during our Dota 2 project. We started out with 16 cores to train a little agent on, on, on Yakov and Shimon, who are getting the project from an ML perspective on their, on their desktop. And then they scaled to 32 cores. And it felt like every week I'd come back to the office and they'd scale it by another 2X and we had 2X performance.

• 3:41Speaker 00

  And just so clear, you just need to keep going. Like, where does this thing peter out? And it just never did.

• 3:45John Collison?

  Founders get too much credit because you have an initial product that's like a pretty reasonable idea. And then you listen to the customers and you follow what's working. And you're saying that was kind of OpenAI with the scaling hypothesis, where you started trying to make Dota AIs work and you noticed that adding more compute worked really well. And you said, where else? We'll just throw more compute at it, yield benefits.

• 4:07Speaker 00

  I think, I think that's like to, to first order correct. And I think one thing that distinguishes OpenAI from, you know, sort of the typical startup is we did everything in reverse. Right, it's like you're supposed to have a problem to solve. No one cares about the technology.

• 4:22John Collison?

  Form DMC up front.

• 4:23Speaker 00

  Exactly, yes. And for us, we really chased the technology without any idea of how it would be applied.

• 4:29John Collison?

  Yes.

• 4:30Speaker 00

  And a lot of pursuing a technology really is you have to let reality hit you cold hard in the face. Like, there's just, there's just no other way to, to achieve results. Like, you can't sort of will it into existence. You can't like sort of convince people that this is the thing. It's like, you have to actually make the system work. We have to just sort of figure out what is the right frontier? What are the problems? What are the things that are on the edge of working? And to really double down on those.

• 4:53John Collison?

  What else do you take away from the Dota work? Like, how else did it? You could, you know, you could say you could have just started with LLMs and, you know, we could have skipped that period in the wilderness, but it sounds like it was somewhat formative for the OpenAI organization.

• 5:04Speaker 00

  Well, so I think Dota had many lessons. One of which actually was a management lesson for, for me. I remember when we started out the project, I tried to set a list of milestones, right? It's like, okay, this date we're going to beat this player, this date we're going to beat this player. And

• 5:18John Collison?

  That didn't work?

• 5:19Speaker 00

  It did not work at all. I remember our first milestone came and

• 5:24John Collison?

  Went.

• 5:24Speaker 00

  Exactly, and so you realize that you cannot control the outcome, right? You cannot set outcome-based milestones. What you can do is you can control the inputs of, we're going to try these experiments by this date. We're going to implement this feature by this date. And that is what actually worked. And I remember it was, it was one of those things that was just like a story that I could not have sort of, you know, written it any better if we'd intended to. Where we had a, we beat our, our in-house like best player. And then we were playing against a semi-pro and he was just trouncing us, trouncing us. And then suddenly we're starting to get pretty good. So we showed up at the international, this, this, this tournament blind. First day, we had three players that we played against. We went 3-0, 3-0, and then 2-1. We're like, oh no, we lost. What happened?

• 6:09Speaker 00

  And it turned out that this pro that we're playing against, that he had used an item we'd never trained against. And we were like, oh no, like, we're totally going to be hosed. So what do we do? Well, we just need to change the training. And so people stayed up all night to, to get this done. They added, added this, this extra item in there. You know, 4:00 AM, they finally get the job running. That Wednesday, we're supposed to play against the number two and the number one player in the world. And our semi-pro plays against it. And he's like, this bot is totally broken. And we're like, oh no, we clearly had a bug, something terrible has happened. And you're like, look, it's taking all this damage it doesn't need to, I'm going to go kill it. He goes in to kill it, he loses. He was like, that was weird. He'd realized that what had happened was it had learned a baiting strategy. And then we realized, well, we have a super bot, but it's so bad at the beginning because it's trying to do the baiting.

• 6:54Speaker 00

  So what if we just stitch the two bots we have together? And then that bot was just undefeatable. And we played against this number one player and won. And to me, this is like the story of how deep learning works, right? It's like you kind of can't control where you're going to go. You can control everything that goes in. You can put these metrics and these measurements and you can have sort of the evaluations and the being able to gauge where you're at is almost as important as being able to make the forward progress. But if you get all those elements right, then you can do true magic.

• 7:21John Collison?

  And you're also describing something that works really well for an organization where it was motivating to stay up all night. Like, you know, if the prize was impossibly far away, it wouldn't have been as motivating. But the fact that there was a near term reward function and you were able to show concrete progress.

• 7:35Speaker 00

  I think so. Yeah, it's like, I mean, I think some of my favorite engineering stories have the same character. I mean, I remember you and I staying up all night to get our ISO 8583 integration done.

• 7:44John Collison?

  Something about staying up all night for like critical projects that actually have important history and.

• 7:48Speaker 00

  Turns out.

• 7:49John Collison?

  Yeah, all startups. I'm glad to hear the tradition of the live while I'm while opening eye. So Dota's fallen, chess has fallen, Go has fallen. We've passed the Turing test, I think by anyone's measure with, you know, people comment on how there was little fanfare when we did, but we seem to have pretty clearly done so. What's a good new Turing test?

• 8:11Speaker 00

  Well, I'll tell you two things. One is that if you look at the strict version of the Turing test, I would actually claim we haven't done it yet. So no one's really gone that extra mile to say, can we actually have an AI that like is fully indistinguishable from a human? And it's not clear if it's even a good task, right? But I think that the right question to your point is like, well, what is the milestone that we should be chasing in terms of capability? Like, I remember talking to one of our board members in 2018 and he said, I, you know, look, I get that you were all excited about near term AGI, but it just doesn't feel like it's on track. And I asked, well, what do you mean? He said, in a world with near term AGI, you would expect massive economic value to be delivered by AI already. And where is it?

• 8:52Speaker 00

  And in 2018, I think that was a very fair criticism and clearly that's starting to change now.

• 8:57John Collison?

  It feels like one thing that's.

• 9:00John Collison?

  May really change the AI market is personalization. Up to quite recently, when you asked ChatGPT a question, it was like walking into a shop off the street. They've never met you before. They know nothing about you, whatever. That's obviously not ideal for this, you know, close part of your digital life. I'm curious how you're thinking about personalization from a product point of view. Because it feels to me like the most meaningful change since the chat interface. Two and a half years ago.

• 9:28Speaker 00

  I mean, I think it's absolutely critical. And I think it is very rightly considered to be kind of a next frontier. Like, I am someone who always, when I just Google something, I go into incognito mode because I don't even want my computer to remember that history. And I always used to go for temporary chats on ChatGPT. But now my usage is totally reversed. I want ChatGPT to know, to remember everything. I want it to remember all of my interactions because it's useful.

• 9:52John Collison?

  Okay, so you guys figured out from a product point of view, how to make the memory actually work better.

• 9:56Speaker 00

  And this is actually, so a product point of view, but also really research point of view.

• 9:59John Collison?

  And I presume there's a flip flop between the product and research, where when you find something that's useful from a product point of view, then the product people say, and I am just a product person. Like, you researchers go actually make this good. And then that kind of kicks off more research.

• 10:10Speaker 00

  It's like, well, we actually, so to some extent, that's a failure mode in our mind, right? That I think that we really don't want to have that kind of silo. And we really want to blur the lines and have people cross collaborate. And so it's very different mindsets from how you would traditionally build a product versus how you do research. And part of what had happened, actually, was that we had GPT-3. We knew we needed to build a product in order to be able to continue to raise funding. And we're like, well, what product do we build? And we wrote down a list of like a hundred different products, right? We could do a medical thing. And then you're like, okay, well, now we have to sell the hospitals. We're going to have to hire doctors. And you just realize you give up on the G in AGI, right? You're going to like go for a specific thing. And so.

• 10:50Speaker 00

  Someone had the idea of saying, well, why don't we just make a API and let people figure it out?

• 10:58John Collison?

  Yes.

• 10:58Speaker 00

  And again, this is totally backwards from how you're supposed to do a startup. Right? You're supposed to have a problem and we had no idea what the problem was.

• 11:04John Collison?

  Yeah, yeah, yeah.

• 11:05Speaker 00

  And so.

• 11:05John Collison?

  We're back into the problem.

• 11:06Speaker 00

  And so this actually felt like this was probably the hardest project I've ever done because it felt totally doomed. Right? It's like, I know like every instinct, every builder instinct of mine.

• 11:14John Collison?

  Did it actually feel doomed?

• 11:15Speaker 00

  Oh, it felt totally doomed.

• 11:16John Collison?

  It wasn't just like open-ended or something?

• 11:19Speaker 00

  No, it felt doomed.

• 11:19John Collison?

  You were still doing it.

• 11:21Speaker 00

  I mean, it's like, at some point, if you have, there was no, there was definitely no other path. There was no other path. It was the only shot we had. I remember someone also saying like, I can't imagine anyone paying for samples from this model. And I was like, you might be right. And.

• 11:35John Collison?

  I'm still trying to imagine it myself.

• 11:37Speaker 00

  Yes. And it was just not clear, were we above threshold or below threshold? And, you know, we showed it to people and people were interested, but it's very different from people being like, I will build my company on top of this.

• 11:48John Collison?

  Yes.

• 11:48Speaker 00

  And.

• 11:49John Collison?

  And so what was the first use case to get any traction?

• 11:51Speaker 00

  So AI Dungeon and.

• 11:53John Collison?

  What was that again?

• 11:54Speaker 00

  There you go. So AI Dungeon was a text-based, I, I, text-based adventure game.

• 12:00John Collison?

  Oh, sure. Yeah, yeah, yeah. Okay, but that was like, that was real revenue or that was non-zero revenue?

• 12:05Speaker 00

  Yes. And in fact, I believe they were our first paying user and.

• 12:09John Collison?

  And that gets you confused where you're like, ah, clearly the future of OpenAI is gaming, you know?

• 12:13Speaker 00

  I know, right? Well, it's back to our roots.

• 12:14John Collison?

  Exactly.

• 12:15Speaker 00

  Right? And, you know, it's, it's interesting too, because yeah, we had dreamed of all of these applications and, you know, medicine and all these things. And, you know, you start with the gaming application, but we could see signs of life on so many other things. I think in many ways, GPT-3 was like the best, world's best demo machine, right? When we released the API, people were coming with all these cool things you could do, but making them reliable, reliable was so hard. And it really wasn't until the next generation of GPT-4, until we started to figure out how to do post-training well, that then you were actually able to build real businesses on top of these things.

• 12:46John Collison?

  Bill Gates was saying recently that GPT-4 was the best demo he'd ever seen since Xerox PARC. You know this quote?

• 12:52Speaker 00

  Yes, he said it to me the night that he saw it.

• 12:56John Collison?

  Yeah, so that's high praise. I want to touch the medicine thing because you've mentioned it. So like you said, you know, I think you, your family has kind of personal stories. You've talked about getting very valuable diagnostic help. We ourselves actually, it's much more minor in our family, but we managed to fix a cat thanks to, you know, debugging it with an LLM. And I think this is an interesting example, right? Because so many people that I know have had some kind of experience like this. And maybe it's because you actually don't get that much time from a doctor. Are there other examples like this medicine application where you're seeing a lot of success that many people

• 13:39John Collison?

  have similar stories, but we just hear less about?

• 13:42Speaker 00

  Yeah, I think it's a great question. And by the way, I think like medicine is an example of one where I, I kind of thought it was going to be one of the last domains that we successfully be able to add value in. But it turns out that the bar is so low that you just need to exceed WebMD. And so I think that we have seen sort of, you know, other areas that are like a real common theme. Like one that's very interesting right now is like the life coach, like life advice kind of application, where we just talk.

• 14:06John Collison?

  That's actually really taking off.

• 14:07Speaker 00

  Yeah, it really is. And so I think that there's, you know, things like education is another area that just like clearly like is really having an impact. And there are studies coming out now that actually show that people are able to learn better through the use of these tools.

• 14:22John Collison?

  Let's be expected, right? Like it is the Bloom to Sigma effect in a product.

• 14:25Speaker 00

  Yes, yes. And that, for example, is like why, you know, Saul Khan, that's why he started Khan Academy was to think about if you can give personalized tutoring to everyone.

• 14:33John Collison?

  Yes.

• 14:34Speaker 00

  And we showed him GPT-4. He's like, this is the thing.

• 14:35John Collison?

  Yes.

• 14:36Speaker 00

  Like, we need to become a GPT-4 app. And so I think that there are these really amazing applications that are affecting everyone's like daily lives. You know, obviously programming is another one. I think that people are seeing all across the board in professional contexts. We're, we're heading to a world where just like, you know, if you want to do productive work, you don't have access to computer, like you're going to be hampered. And so similarly, not having access to AI is heading in the same direction.

• 15:02John Collison?

  Okay. So speaking of not having access to AI. I will posit that these days it feels like AI product development is mostly OS limited. Is that how you feel? Like, are we stuck at the moment?

• 15:18Speaker 00

  I do feel a little, little of the stuckage, but it don't, not to worry, it is overcomeable. But yeah, I think it is true. Look, like, two years ago, we've released plugins in ChatGPT. Do you remember, do you remember those?

• 15:30John Collison?

  Yeah.

• 15:30Speaker 00

  Yeah, and that was like trying to make it so anyone could write apps that then ChatGPT could access. And the models were just not that good, right? Then we limited it to like three plugins at a time, you could have only so many functions and stuff. And it just wasn't that reliable. And now we're in a world where MCP basically, you know, is really taking off and is a way to hook up your AI to different tools and very much like kind of try and take that same type of idea and really make it work. Now, the world that we're in is very similar where there are certain interfaces we don't have, right? Being able to like access your phone and all those APIs. And there's a question of is the model above threshold to actually use them or not? And my observation has been that basically, I think that there is maybe a lag of like six months

• 16:15Speaker 00

  of different interfaces that are hard to access. But once we have a model that's good enough, we will find a way, people will find a way. And so I think that we're in a world where I have every expectation that we will get the future that has been promised, but it's just gonna take some work.

• 16:30John Collison?

  I feel like there are many moments where I'm using my phone and I want a single button where it just, it's like, you know, ChatGPT, what do you think of this? Like, I need your comment, I need your fact check, I need your explanation, something like that. And you take a screenshot and you're like, go into ChatGPT and you click upload photo. And, you know, it feels like very 1993 versus the button on my phone that just says, "Hey, ChatGPT, what do you think about this?" And obviously, you guys are not empowered to go build that. That's what I mean by it feels somehow like we're a little operating system.

• 17:00Speaker 00

  Well, I definitely get it, but I'll say, I think that there are two dimensions. And this is kind of how I've been thinking about things since we released the API back in 2020. There is capability and convenience. What you're referring to is the convenience, right? It's like pretty inconvenient to do the screenshot and paste it. But the thing is, if the capability is good enough, you are willing to accept any sort of inconvenience, right? It's like if this, you know, by taking a screenshot and showing it to ChatGPT, it could give you amazing insight. It could tell you like how to, you know, build Stripe in some way. And it takes you like a month to do it. You have to crawl to the top of the mountain. Like you'll do it, right? Like the convenience will not stop you. And so the point that I'm trying to make is that if the capability is high enough, people will start doing a specific.

• 17:45John Collison?

  They'll discover the use cases and the convenience will just catch up.

• 17:47Speaker 00

  And then in the convenience, there's so much pressure, right? There's pressure on the phone manufacturer, there's pressure on us, there's pressure on everyone in order to to bring down the.

• 17:54John Collison?

  So I just need to be patient and it'll be great in three years' time.

• 17:56Speaker 00

  Yes, and and really lean in, use the AI.

• 17:59John Collison?

  A criticism people like to levy of AI is, yeah, it's great and handy and all, but it hasn't come up with a single novel advance in mathematics.

• 18:10Speaker 00

  Have you?

• 18:10John Collison?

  Or science. Well, you could have if, you know, you'd become a mathematician. But, uh, you know, but humanity has, you know, for keeping the score, uh, the scoreboard. What do you make of that criticism?

• 18:21Speaker 00

  Uh, just wait.

• 18:23John Collison?

  Okay, so you think like, uh, take, uh, um, uh, you know, one of the millennium prizes, you think we plausibly will see that?

• 18:30Speaker 00

  I think for sure. I mean, there's no question.

• 18:31John Collison?

  Like two years, five years, 10 years?

• 18:33Speaker 00

  I think that is the question. It's just timing, right?

• 18:35John Collison?

  That is my question.

• 18:37Speaker 00

  I mean, I would, I would put it two to five years is the right, the right number. Um, and, uh, I think ultimately that this comes back to the question of benchmarks, right? Is that that actually being able to solve a millennium problem is pretty high bar.

• 18:49John Collison?

  Yeah.

• 18:50Speaker 00

  Pretty high bar. And once you can do that, there's so many other things that will definitely be possible.

• 18:56John Collison?

  Yeah.

• 18:56Speaker 00

  And I think that we're starting to see the leading edges of this. And to me, you know, if we look at our definition of AGI, you know, we recently started talking about this framework of thinking about levels of AGI, you know, starting from chatbots to reasoners, to agents, to innovators, to organizations, five levels. And we're basically somewhere in level three right now.

• 19:16John Collison?

  Yeah.

• 19:16Speaker 00

  And level four, this innovators, like that's going to be different. You know, I recently posted some pictures of our visit to Abilene, Texas, where we're building these big data centers together with with our partner Oracle. And. Imagine taking that whole data center and just thinking hard about one problem. Right? Imagine just thinking about how to solve a millennium problem or how to cure a specific kind of cancer. Maybe it needs access to some apparatus. Maybe it needs access to robotic wet labs. You know, maybe it needs access to different tools in the world. But that level of computational power, coupled with the ability to experiment and learn from your ideas, like that is going to be something the world has never seen.

• 19:55John Collison?

  Okay, so yet again, we just, we haven't put a respectable amount of compute on these problems compared to what we will be doing.

• 20:02Speaker 00

  Yeah, we still have these tiny little computers.

• 20:04John Collison?

  So that actually gets to, in terms of these scaling laws, do they eventually run out because we just run out of compute? Or like, do we eventually get to the point where we're inventing new kinds of nuclear energy? And so that is what unlocks the next level? A lot of energy that comes online now is for data centers, which was not true when you guys started training GPT-2. And so isn't that just the upcoming bottleneck?

• 20:28Speaker 00

  Right. It's as it should be, right? It really should be that it's energy manufactured into intelligence and that that's your only bottleneck.

• 20:36John Collison?

  But I'm saying that'll be like quite a plateau compared to the exponential growth we've seen over the past few years.

• 20:42Speaker 00

  Well, I think.

• 20:44John Collison?

  Unless things really change in terms of permitting and plans for building and everything like that.

• 20:48Speaker 00

  Yes, this is, I think, the core, right? Is that if you look at every trend in this field, there's these exponentials, these S-curves that sum up to exponentials.

• 20:57John Collison?

  Sure, but these exponentials were mostly existing in like tech, you know, Silicon Valley space, where it's pretty easy to have exponential growth. It's pretty hard in permitting and real estate and damming rivers and building nuclear power plants. It's harder to have exponential growth.

• 21:12Speaker 00

  Well, let's see, let's see how fusion pans out.

• 21:15John Collison?

  Yeah, okay, but even fusion, you know, most industry observers would say is still kind of five years away. And so I've heard it's the next five years of power growth.

• 21:22Speaker 00

  So, look, I think, I think that it is very possible that we end up bottlenecked on, on energy. And that's actually one reason that we've been spending a lot of time really trying to advocate for the fact that we just need far more power.

• 21:32John Collison?

  Yeah.

• 21:33Speaker 00

  And I think that like my observation of the market is that ultimately the. Capitalist markets do provide. Yeah, I think there's this like absolute tsunami of demand that is coming our way. But I feel some confidence that, again, the like, when there's enough pressure, when there's enough clarity of this is the bottleneck, and it's not just for any company, right? It's really for national competitiveness. And you look at other countries that are just building huge amounts of power, far more than we are. I think that actually for America to remain competitive, there is just no choice but to build. We've got to forget power. We do.

• 22:07John Collison?

  Speaking of bottlenecks, everyone was talking about the data wall in 2023. I think this is an interesting thing where no one is talking about the data wall anymore. And yes, it doesn't feel like AI progress has slowed down. Is it just test time compute? Is it like people were wrong about the data wall and where it presented a bottleneck? Like, is there actually still a data wall, but it's two years away?

• 22:29Speaker 00

  It's basically all of these things, right? It's really, it truly is, right? It's like you keep changing the paradigm, right? That is, that is the real core of the Kurtzwall view of the world, right? Is that, fine, this one way of doing things taps out. And if you just look at that one way of doing things, you feel hopeless, right? You feel like this is it. But somehow you will find a new S curve. And I think that's what's happened. For example, synthetic data, for example, reinforcement learning, right? If you think about the RL paradigm, fundamentally, that's a data production mechanism, right? And just that the AI happens to be training on its own data. And then you learn on it very rapidly, and then you, you learn on that. And each of these has taken us much further. And so I think there are lots of algorithmic ideas, lots of techniques, lots of ways of even using the existing data better.

• 23:11Speaker 00

  And so I think that fundamentally, the S curves continue. And if you zoom out, it all looks smooth and uninterrupted.

• 23:19John Collison?

  So it's kind of like chip miniaturization, where, you know, each generation, people are like, okay, well, that's the smallest you could possibly make a chip. I mean, that's it, we're done with miniaturization. And somehow we figure it away.

• 23:28Speaker 00

  Yes. And now, one difference with chips is I think at the end of the day, there is some conch limit, like.

• 23:33John Collison?

  Right, but we've never been that close to that.

• 23:34Speaker 00

  Right, right, yes, yes.

• 23:35John Collison?

  And where does AI coding go? And in particular.

• 23:43John Collison?

  Vibe coding is all the rage right now. It's kind of the term of 2025. It's sort of working. It's very impressive. No one is really fully letting AI software engineers run end to end in production. So I'm just curious, what are your one to two year predictions on what happens with AI coding?

• 24:02Speaker 00

  Well, my general observation is that once something kind of works in this field, the next gen is going to be great. And so I think that's where we are right now for AI coding. And so I think what we're going to see is AI is taking more and more of the drudgery, more of this like pain, more of the the kind of parts that are not very fun for humans. Now, one thing that's very interesting is that I think that so far, the vibe coding has actually taken a lot of code that is actually quite fun and left behind the review and the deployment and these things that are not fun at all. And so I'm hopeful that we're actually going to be able to make a lot of progress on these other areas as well. But fundamentally, we should really end up with a full AI coworker. And I think it really will be anything you want to create. You can be the manager, right? And you can really have this team of software engineering agents.

• 24:46Speaker 00

  Now, the thing that I think will be very interesting to see is is there a world where the AI becomes the manager and that, you know, gives you ideas and gives you some tasks to do. And that's something that, again, is just like totally backwards in terms of how we think about it. But are there are there ways in which you can actually, you know, have have outcomes for companies and actually have people whose jobs become much more meaningful because they have an AI who really deeply understands them in the same way that your, you know, AI doctor really deeply understands all of your all of your needs?

• 25:14John Collison?

  Well, and isn't part of the common thread that we're talking about here, often places where AI tools underperform, it's because they're trying to do something generally, you know, like a voice recognition is not that good because it's trying to recognize all voices as opposed to trying to recognize, you know, my voice in particular. And similarly with AI coding, they work well in places where you need no context at all. And, you know, we're single shotting an app based on, you know, publicly available libraries and in places where you have to understand a million line code base. You haven't fully figured out how to. Do a good job of that. Like, is that a fair parallel to draw between all these challenges?

• 25:52Speaker 00

  Well, I think there's two things in there. One is that I think this is already changing, right? So if you look at something like Codex, it's actually great at operating in a big codebase. Like, I ask it for where functionality is implemented, and it's better than I am at finding it. Right? Which is kind of a wild fact. It's like, super cool to see it, like, grapping around and like, you know, just going and exploring.

• 26:14Speaker 00

  And actually, this is one thing that we really shot for with Codex was to build a tool for software engineers who are not necessarily vibe coding, right? It's not about building a new app from scratch, which is a cool demo. But like, that's not actually how most software gets written. And actually, I think maybe the killer enterprise feature is refactors, right? It's like rewriting your KOBLO app or, you know, changing your-

• 26:33John Collison?

  When Facebook did hip hop, you know, to do static PHP.

• 26:37Speaker 00

  Exactly, right? If you think about it, like, the amount of, of like, deep sophisticated thought that is required to accomplish a refactor is actually not that high. There's a lot of mechanical work that's just the sheer volume of it that's hard. And like, that's an AI-shaped problem, for sure. So I think we're going to see a lot more productivity on all sorts of AI tasks as a result. Now, we are in a world where, you said a second thing, which is, you know, like, maybe you need to kind of narrow down more. Like, I think that the way these models work is you actually do want one model that knows more and more things, and you want it to have some personalization to you. But the fact that it, you know, they kind of have this one base model that kind of knows everything is actually a very useful starting point. So I do think that you're going to see a world where we'll have more and more capable base models

• 27:23Speaker 00

  and figuring out how do you really connect it to all of your organization's code and context and history.

• 27:28John Collison?

  How does OpenAI decide what products to do? I'm just curious how you think about when to develop specific products? Or when you think, oh, you can truly do that with ChatGPT, and that's good enough?

• 27:41Speaker 00

  Yeah, this is a really tough question, right? It's something we really struggle with. And I think that over time, you know, when we've, I guess, actually, when we first launched ChatGPT, we were left with this, well, we're an enterprise business and we're a consumer business. And that seems terrifying, terrifying as a startup. And- I remember talking to one of my board members who said, "It just feels like what you have is an unfocused strategy at first, right? Because you're just doing all these different things. But if you think about it, maybe an analogy is to a company like Disney, where you make one core asset, like Little Mermaid, right? And then you productize it in all these different ways, right? You think about Little Mermaid, the ride, the lunchbox, the T-shirt. And I think that we have some element of that, that we have the core model. And then we have a question of, well, what are the applications

• 28:26Speaker 00

  that this can add a lot of value to quickly, right? With like a small amount of additional work. And so I think the question of what areas to go into are, how far does it take us off the general path? The return on how important is this domain area, especially for achieving the bigger goal? How much synergy is there across it with respect to other things we work on? So coding is one where there's very clear synergy, very clear ROI, right? Because if we can speed ourselves up, that's something that accelerates everything.

• 28:55John Collison?

  How is being from North Dakota, how's it shaped you?

• 29:00Speaker 00

  Look, North Dakota was an amazing place to grow up.

• 29:02John Collison?

  Was it actually? Come on.

• 29:03Speaker 00

  It actually was.

• 29:04John Collison?

  I've been there.

• 29:06Speaker 00

  And it was great. Look, it was incredibly safe. Our doors didn't even have working locks. Like, it was that kind of place. I had a lot of freedom academically. So sixth grade, my dad taught me some algebra. Seventh grade was the first time they split you into advanced math. And so I was going to be taking pre-algebra. So my mom took me to go see the teacher. The teacher, and we asked, "Can you skip?" And the teacher looked at us very condescendingly and said, "Every parent believes that their child is special. I can guarantee your son will be plenty challenged in my class." And so after a month, me sitting in the back just playing games on my calculator, and she calling me randomly to try to trip me up. And I just look at the board and be like, "2X." She said, "Okay, fair enough. Your son is nothing to learn in this class."

• 29:50Speaker 00

  And so they moved me into eighth grade algebra. But then eighth grade rolled around, and I had no more math left in my middle school.

• 29:57John Collison?

  And so you went to college, right?

• 29:58Speaker 00

  Well, so I did in high school, in high school, start going to University of North Dakota and took a bunch of classes there. But also, I was connected to a lot of people who were the top math kids in the country through things like Math Camp and the math competitions.

• 30:11John Collison?

  So you're saying the social scene was not too distracting in North Dakota?

• 30:15Speaker 00

  Not too distracting, but it was, yeah, it was definitely fun.

• 30:18John Collison?

  Last question. Do you remember we were going to Camp Y C in 2017? And

• 30:27John Collison?

  I asked you how far AGI was away and you said two or three years. And

• 30:32Speaker 00

  Did I say that?

• 30:32John Collison?

  You did, yeah, yeah, yeah.

• 30:34Speaker 00

  I don't, I don't see the recording.

• 30:36John Collison?

  I'm just trying to think, like, was I right? Were you right? How should we grade that? Because we didn't get AGI, but we didn't not get AGI either. And so I'm curious if you have any reflections from your own AGI prediction journey.

• 30:53Speaker 00

  I think that we are, I will say, I think that AI is surprising. I think that that is like the single most consistent theme is that the thing we were picturing, we got something different, but we got something better, more magical, something that is more helpful. And so I'm actually quite happy with that. Now predicting where you go, it's again, it's really hard to manage the outputs here. One goal of OpenAI that we have successfully achieved is every year to have at least one result that just feels like a step function better than anything before.

• 31:24John Collison?

  You know, when you see us kind of, you just want to have one really awesome AI feeling thing each year.

• 31:29Speaker 00

  That kind of thing.

• 31:30John Collison?

  I like that.

• 31:30Speaker 00

  Yeah.

• 31:31John Collison?

  That's a good way to tie it back, which is you, you know, the way we grade that prediction is that you've stopped setting metrics based on outputs.

• 31:38Speaker 00

  Yeah, exactly, exactly.

• 31:39John Collison?

  Yeah.

• 31:40Speaker 00

  Yes, but it does feel we're getting really close to something pretty, pretty magical.

• 31:44John Collison?

  I agree. Thank you.

• 31:46Speaker 00

  Thank you.