The Power of Quality Human Data with SurgeAI Founder and CEO Edwin Chen

In the generative AI revolution, quality data is a valuable commodity. But not all data is created equally. Sarah Guo and Elad Gil sit down with SurgeAI founder and CEO Edwin Chen to discuss the meaning and importance of quality human data. Edwin talks about why he bootstrapped Surge instead of raising venture funds, the importance of scalable oversight in producing quality data, and the work Surge is doing to standardize human evals. Plus, we get Edwin’s take on what Meta’s investment into Scale AI means for Surge, as well as whether or not he thinks an underdog can catch up with OpenAI, Anthropic, and other dominant industry players. Sign up for new podcasts every week. Email feedback to [email protected] Follow us on Twitter: @NoPriorsPod | @Saranormous | @EladGil | @echen | @HelloSurgeAI Chapters: 00:00 – Edwin Chen Introduction 00:41 – Overview of SurgeAI 02:28 – Why SurgeAI Bootstrapped Instead of Raising Funds 07:59 – Explaining SurgeAI’s Product 09:39 – Differentiating SurgeAI from Competitors 11:27 – Measuring the Quality of SurgeAI’s Output 12:25 – Role of Scalable Oversight at SurgeAI 14:02 – Challenges of Building Rich RL Environments 16:39 – Predicting Future Needs for Training AI Models 17:29 – Role of Humans in Data Generation 21:27 – Importance of Human Evaluation for Quality Data 22:51 – SurgeAI’s Work Toward Standardization of Human Evals 23:37 – What the Meta/ScaleAI Deal Means for SurgeAI 24:35 – Edwin’s Underdog Pick to Catch Up to Big AI Companies 24:50 – The Future Frontier Model Landscape 26:25 – Future Directions for SurgeAI 29:29 – What Does High Quality Data Mean? 32:26 – Conclusion

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

• 0:05Sarah Guo?

  Hi listeners, welcome back to No Priors. Today, Elad and I are here with Edwin Chen, the founder and CEO of Surge, the bootstrapped human data startup that surpassed a billion in revenue last year and serves top-tier clients like Google, OpenAI, and Anthropic. We talk about what high-quality human data means, the role of humans as models become superhuman, benchmark hacking, why he believes in a diversity of frontier models, the scale meta-not-M&A deal, and why there's no ceiling on environment quality for RL or the simulated worlds that labs want to train agents in. Edwin, thanks for joining us.

• 0:38Speaker 01

  Great, great. Seeing you guys today.

• 0:40Sarah Guo?

  Surge has been really under the radar until just about now. Can you give us a little bit of color on sort of the scale of the company and what the original founding thesis was?

• 0:51Speaker 01

  So we hit over a billion in written revenue last year. We are kind of like the biggest human data player in this space and we're about 100, a little over 100 people. And our original thesis was, we just really believed in the power of human data to advance AI. And we just had this really big focus from the start of making sure that we had the highest quality of data possible.

• 1:14Speaker 02

  Can you give people context for how long you've been around, how you got going, et cetera? I think, again, you all have accomplished an enormous amount in a short period of time. And I think, you know, you've been very quiet about some of the things you've been doing, so it'd be great to just get a little bit of history and, you know, when you started, how you got started, and how long you've been around.

• 1:30Speaker 01

  Ah, yeah, so we've been around for five years. I think we just hit our five-year anniversary. So we started in 2020. So before that, so I can give some of the context. So before that, I used to work at Google, Facebook, and Twitter. And one of the, like, basically the reason we started Surge was I just used to work on ML at a bunch of these big companies. And just the problem I kept running into over and over again was that it really was impossible getting the data that we needed to train our models. So it was just this big blocker that we faced over and over again. And there was just like so much more that we wanted to do. Like, even just the basic things that we wanted to do, we struggled so hard to get the data. It was really just the big blocker. But then simultaneously, there were all these more futuristic things that we wanted to build. Like, if we thought of the next generation AI systems, if we could barely get the data that we needed at the time to solve,

• 2:15Speaker 01

  like, just building a simple set, no, it's a classifier. If we could barely do that, then like, how would we ever advance, advance beyond that? So that really was the biggest problem. I can go into more of that, but that was what we faced.

• 2:27Speaker 02

  And you guys are also known for having bootstrapped the company versus raising a lot of external venture money or things like that. Do you want to talk about that choice in terms of going profitable early and then scaling off of that?

• 2:37Speaker 01

  In terms of why we didn't raise, so I think a big part of it was obviously that we didn't need the money. I think we were very, very lucky to be profitable from, from the start. So we didn't need the money. It always felt weird to give up control. And like, one of the things I've always hated about Silicon Valley is that you see so many people raising for the sake of raising. Like, I think one of the things that I often see is that a lot of founders that I know, they don't, they don't have some big dream of building a product that solves some idea that they really believe in. Like, if you talk to a bunch of YC founders or whoever it is, like, what is their goal? It really is to tell all their friends that they raised $10 million and to show their parents they got a headline on TechCrunch. Like, that is their goal. Like, I think of like my friends at Google. They, they often tell me, oh, yeah, I've been at Google or Facebook for 10 years and I want to start a company.

• 3:20Speaker 01

  I'm like, okay, so what problem do you want to solve? They don't know. They're like, yeah, I just want to start something new. I'm bored. And it's weird because they can like pay their own salaries for a couple of months. Again, they've been at Google and Facebook for 10 years. They're not just like fresh out of school. They, they can pay their own salaries, but the first thing they think about is just going out and raising money. And I've always thought it weird because they like might try talking to some users and they might try building an MVP, but they kind of just do it in this throwaway manner where the only reason they do it is to check off a box on a startup accelerator application. And then they'll just pivot around these random product ideas and they happen to get a little bit of traction so that the VC DMs them. And so they spend all their time tweeting and they go to these VC dinners and it's all just so that they can show the world that they raised a big amount of money. And so I think raising immediately always felt silly to me. Like, everybody's default is to just immediately raise.

• 4:05Speaker 01

  But if you were to think about it from first principles, like, if you didn't know how Silicon Valley worked, if you didn't know that raising was a thing, like, why, why would you do that? Like, what is money really going to solve for 90% of these startups where the founders are lucky to have some savings? I really, really think that your first instinct should be to go out and build whatever you're dreaming of. And sure, if you ever run into financial problems, then sure, think about raising money then, but don't waste all this effort in time frame where you don't even know what you do with it.

• 4:27Speaker 02

  Yeah, it's funny. I feel like I'm one of the few investors that actually tries to talk people out of fundraising often.

• 4:31Speaker 01

  All right.

• 4:32Speaker 02

  Like, I actually had a conversation today where the founder was talking about doing a raise. I'm like, why? You know, you don't have to. You can maintain control, et cetera. And then the flip side of it is, I would actually argue, outside of Silicon Valley, too few people raise venture capital when the money can actually help them scale. And so I feel like in Silicon Valley, there's too much and outside of Silicon Valley, there's too little. So it's this interesting. You know, spread of different models that sort of stick.

• 4:53Sarah Guo?

  Evan, what would you, what would you say to founders who feel like there's some external validation necessary to, especially like hire a team or scale their team? This is a very like common complaint or rationale for going and raising more capital.

• 5:11Speaker 01

  I think about it in a couple of ways. So I guess it depends on what you mean by external validation. Like in my mind, again, like I often think about things from a perspective of are you trying to build a startup that's actually going to change the world? Like, do you have this big thing that you're dreaming of? And if you have this big thing that you're dreaming of, you, like, why do you care?

• 5:31Speaker 02

  Maybe the way to think about it is in Sarah's context, like if you haven't, say you're a YC founder, you haven't been at Google, you haven't been at Meta, you haven't been at Twitter, you don't have this network of engineers, you're a complete unknown, you haven't worked with very many people, you're straight out of school. How do you then attract that talent? And to your point, you can tell a story of how you're going to build things or what you're going to do, but it is a harder obstacle to basically convince others to join you or for others to come on board or to have money to pay them if you haven't, if you don't have long work at history. So I think maybe that's the point Sarah's making.

• 6:00Speaker 01

  Uh, yeah, so I mean, I think I would differentiate between maybe two, two things. Like one is, do you need the money? So first of all, like there's a difference between people who are, yeah, like literally fresh out of school or maybe, you know, I've never gone to school for the first place. And so maybe they don't have any savings and so they literally need some money in order, in order to live. And then there's others who, okay, like, let's assume that you don't necessarily need money because again, you've been working at Google or Facebook for 10 years, like or, you know, five years, whatever it is, you have some savings. So I would say one of the questions is, again, like it kind of, the path kind of differs depending on, depending on those, those two, those two choices or those two scenarios. But I think one of the questions is, well, do you really need to go out and hire all these people? Like, one of the things I often see, again, like I'm curious what you guys see, but one of the things I often see is founders will tell me like, uh, okay, so I'm trying, I'm

• 6:45Speaker 01

  trying to think about the first few hires I'm going to make. And they're like, yeah, I'm going to hire a PM. I'm going to hire a data scientist. Yeah, these are one of my first five to 10 hires. I'm like, what? Like, this just wowed me. Like, I would never hire a data scientist when it first came to people in a company. And I, and I say this because I used to be A data scientist. Like data scientists are great when you want to optimize your product by 2% or 5%. But that's definitely not what you want to be doing when you start a company. You're trying to swing for 10x or 100x changes. Not worrying and nitpicking about small percentage points are just noise anyways. And to some of the product managers, like product managers are great when your company gets big enough. But at the beginning, you should be thinking about yourself about what product you want to build. And your engineers should be hands on. They should be having great ideas as well.

• 7:25Speaker 01

  And so product managers kind of this weird conception that big companies have when your engineers don't have time to be in the weeds on the on the details and try things themselves. And it's not a role that you come up with either before.

• 7:34Speaker 02

  So I guess with the initial search team, it sounds like you had sort of a small initial tight engineering team. You guys started building product. You were bootstrapping off of revenue. You know, at this point, you're at over a billion dollars in revenue, which is amazing. How do you think about the future of how you want to shape the organization, how big you want to get, the different products you're launching and introducing? Like, what do you view as sort of the future of search and how that's all going to evolve?

• 7:58Sarah Guo?

  Before we do that, can you just explain like what the, at whatever level of detail makes sense here? Like what the billion dollars of revenue is, maybe like how product supports the company, who your data, who your humans are? Because I think there's just very little visibility into into all of that.

• 8:14Speaker 01

  So in terms of what our product is, I mean, at the end of the day, our product is our data. Like we literally deliver data to companies and that is what they use to train and evaluate their models. So imagine, you know, when you're one of these frontier labs and you want to improve your model, your model's coding abilities. What we will do on our end is we will gather a lot of coding data. And so this coding data may come in different forms and maybe SFT data. We are literally writing out coding solutions or maybe unit tests. Like, these are the tests that a good, that a good piece of code must pass. Maybe it's preference data where it's okay, like here are two pieces of code or here are two coding explanations. Which one is better? Or these might be like verifiers, like, okay, here's a web app that I created. I want to make sure that in the top right hand of the screen, there's like, there's a login button.

• 9:00Speaker 01

  Or I want to make sure that when you click this button, something else happens. Like there's a budget or informs that this data may take. At the end of the day, what we're doing is there were different data that'll basically help the models improve on these capabilities. Very, very related to that is this notion of evaluating the models. Like you also want to know, yeah, is this good coding model? Is it better than this other one? What are the errors in which this model is weak and this model is worse? Like what insights can we get from that? And so in addition to the data, oftentimes we are delivering insights to our customers. We're delivering loss patterns, we're delivering failure modes. So there may be a lot of other things like related to the data. But at the end of the day, it's like this, this universe of like applications or just like this universe around the data that we deliver and that is our product.

• 9:39Sarah Guo?

  Yeah, and maybe going back to Alaud's question, maybe like product isn't actually the right word here, but what's, what's like repeatable about the company? Or what are like core capabilities that you guys have that you would say your competitors, you know, fail to meet the mark?

• 9:54Speaker 01

  The way we think about a company is that, and the way we differentiate from others is that a lot of other companies in this space, they are essentially just body shops. What they are delivering is not data. They are literally just delivering warm bodies to, to, to companies. And so what that means is like, at the end of the day, they don't have any technology. And one of our fundamental beliefs is that, again, quality is the most important thing at the end of the day. Like, is this high quality data? Is this a good coding solution? Is this a good unit test? Is this mathematical problem solved correctly? Is this a great poem? And basically, a lot of companies in this space, like just as of like how things have worked out historically, it's that like historically, a lot of companies, they, they've treated quality and data as commodity.

• 10:38Speaker 01

  Like one of the ways we often think about it is imagine you're trying to draw a bounding box around a car. Like Sarah, you and I, we're probably going to draw the same bounding box. Like ask Hemingway and ask a second grader. Well, at the end of the day, we're all going to draw the same bounding box. There's not much difference that we can do. So there's a very, very low ceiling on the bar of quality. But then take something like writing poetry. Well, I suck at writing poetry. Hemingway is definitely going to write a much better poem than I am. Or imagine, I don't know, a VC pitch deck. You're going to write a much better, going to create a much better pitch deck than I will. And so there's almost an unlimited ceiling in this gen AI world on the type of quality that you can, that you can build. And so the way we think of our product is like, we have a platform.

• 11:19Speaker 01

  We have actual technology that we're using to measure the quality that our workers or annotators are generating. If you don't have that technology, if you don't have any way of measuring it.

• 11:27Speaker 02

  Is the measurement through human evaluation? Is it through model-based evaluation? I'm a little bit curious like how you create that feedback loop since to some extent, it's a little bit of this question of how do you have enough evaluators to evaluate the output? Relative to the people generating the output or do you use models or how do you approach it?

• 11:42Speaker 01

  Like, I think one analogy that we often make is think about something like Google search or think about something like YouTube. Like you have, you know, millions of search results. You have millions of webpages, you have millions of videos. How do you evaluate the qualities of these videos? Like, is this a high quality, like, is this a high-quality webpage? Is it informative? Or is it really spammy? Like, in the way you do this is that you just need, I mean, you gather so many signals. You gather like page-dependent signals, you gather like user-dependent signals, you gather activity-based signals. And all these feed into a giant ML algorithm at the end of the day. And so in the same way, we gather all these signals about our annotators, about the work that they're performing, about like their activity on the site. And we just feed it into a lot of these different, like, we basically have an ML team internally that builds a lot of these algorithms to measure all of this.

• 12:25Sarah Guo?

  What is changing or breaking as you are like scaling increasingly sophisticated like annotations, right? Like if, you know, model quality baseline is going up every couple of months, then the expectation is it like exceeds, you know, what might have been a random human. At some point, as you said, like can draw a bounding box into all of these different fields where, you know, we have modeled better than the 90th percentile at some point.

• 12:55Speaker 01

  So this is actually something that we do a lot of internal research on ourselves as well. So there's basically this field of AI alignment called scalable oversight, which is basically this question of how do you, how do you like have models and humans working together hand in hand to produce? Data that is better than either one of them can achieve on their own. And so even like even today, someone like writing an S A T story from scratch, even today, like sure, a couple of years ago, we might have written that story completely from scratch ourselves. Today, it's just like not very efficient, right? Like you might start with a story that a model model created and then you would edit it. You might edit it in a very substantial way, like maybe just the core of it is very vanilla, very generic. But there's just so much like kind of crushed that is just inefficient for a human to do.

• 13:36Speaker 01

  And doesn't really benefit from like the human creativity and human ingenuity that we're trying to add into the response. And so you can just start with like this bare bones structure that you're basically just layering on top of. And so again, like there's like the more sophisticated ways of thinking about scalable oversight. But just this question of how do you build the right interfaces? How do you build the right tools? How do you just combine people with AI in the right ways to like make them make them more efficient? It is something that we build a lot of technology for.

• 14:01Sarah Guo?

  Lot of the discussion in terms of what human data the labs want has moved to

• 14:10Sarah Guo?

  R L environments and reward models in, you know, recent months. What is hard about this or, you know, what are you guys working on here?

• 14:18Speaker 01

  So we do a lot of work building our environments. And I think one of the things that people really underestimate is how it is how complicated it is that you can't just synthetically generate it. Like, for example, you think you need a lot of tools because these are massive environments that people want.

• 14:32Sarah Guo?

  Can you give an example of like, just to make it more real?

• 14:35Speaker 01

  Like, imagine you are a salesperson. And when you are a salesperson, you need to be interacting with Salesforce. You need to be getting leads through Gmail. You're going to be talking to customers in Slack. You're going to be creating Excel sheets, tracking your leads. You're going to be, I don't know, writing Google Docs and making PowerPoint presentations to present things to customers. And so you want to basically these are very rich environments that are literally simulating your entire world as a salesperson. It like literally is just like imagine like your entire world. So with everything on your desktop and then in the future, everything that is, you know, not on your desktop as well. Like maybe if a calendar, maybe there's maybe you need to travel to a meeting to meet a customer. And then you want to simulate a car accident happening and you're getting notified of that. So you need to like leave a little bit earlier.

• 15:20Speaker 01

  Like all these things are things that we actually want to model in these very, very rich R O environments. And so the question is, how do you generate all the data that goes into this? Like, okay, you're going to need to generate like thousands. Slack messages, hundreds of emails. You need to make sure that these are all consistent with each other. Need to make sure that, like going back to like my core example, you need to make sure that time is evolving in these environments and like certain like external events happen. Like, how do you do all this? And then, um, like in a way, there's actually kind of like interesting and creative, but also realistic and not like incongruent with each other. Uh, like there's just like a lot of thought that needs to go into these environments, um, to, to make sure that they're again, like rich creative environments that the models can learn interesting things from.

• 16:02Speaker 01

  And so, uh, yeah, you basically need like a lot of tools and kind of sophistication for creating these.

• 16:08Sarah Guo?

  Is there any intuition for like how real or how complex is enough? Or is it just like, you know, there's, there's no ceiling on the realism that is useful here or the complexity of environment that is useful here?

• 16:23Speaker 01

  I think there's no ceiling. Like at the end of the day, you just want as, as like much diversity and richness as you can get, because the more richness that you have, yeah, the more the models can learn from, like the longer the time horizons, the more the models can learn on and improve on. So I think there's like almost, almost an unlimited ceiling here.

• 16:38Sarah Guo?

  If you were to make a five or 10 year bet on like what scales most in terms of demand from people training AI models and, and types of data, is it RL environments? Or is it traces on types of like expert reasoning? Or, or what other areas do you think there's going to be a really large demand for?

• 16:57Speaker 01

  I mean, I think it will be all of the above. Like, I don't think RL environments alone will suffice just because, I mean, it depends on how you think about there are environments, but oftentimes these are very, very rich trajectories are very, very long. And so it's almost like inconceivable that a single reward, um, I mean, I think even today, we often think about things in terms of multiple rewards, not just a single reward, but a thing like a single reward may just may not be like rich enough to capture, um, all of the, all the work that goes into like the model solving some like very, very complicated goal. Um, so I think there will probably be a combination of, of all those.

• 17:29Speaker 02

  If you're similar eventually, um, Some form of superhuman performance across different model types relative to human experts. How do you think about the role of humans relative to data and data generation versus synthetic data or other approaches? Like, at what point does human input sort of run out as a useful point of either feedback or data generation?

• 17:52Speaker 01

  So I think human feedback will never run out, and that's for a couple of reasons. So, I mean, even if I think about the landscape today, I think people often overestimate their role of synthetic data. Like, personally, I think synthetic data actually is very, very useful. Like, we use it like a ton of ourselves in order to supplement what the humans do. Like, again, like I said earlier, there's like a lot of cruft that sometimes isn't worth a human's time. But what we often find is that, like, for example, a lot of the times where customers will come to us and they'll be like, "Yeah, for the past six months, I've been experimenting with synthetic data. I've gathered 10 to 20 million pieces of synthetic data." Actually, we finally realized that 99% of it just wasn't useful. And so we're trying to find right now, we're trying to curate the 5% that is useful, but we are literally going to throw out 9 million of it.

• 18:35Speaker 01

  And oftentimes, they'll find out that, yeah, like, actually 1,000, even 1,000 pieces of high-quality human data, highly curated, really, really high-quality human data is actually more valuable than those 10 million points. So that is one thing I'll say. Another thing I'll say is that it's almost like sometimes you need an external signal to the models. Like, the models just think so differently from humans that you always need to make sure that they're kind of aligned with the actual objectives that you want. We can give two examples. So one example is that it's kind of funny. Sometimes if you try, so one of the frontier models, let me just say that one of them, if you go use the frontier model, it's like one of the top models or one of the models everybody thinks is one of the top. If you go use it today, like, maybe 10% of the time when I use it, it will just output

• 19:19Speaker 01

  random Hindi characters and random Russian characters into one of my responses. So I'm like, "Tell me about Donald Trump. Tell me about Barack Obama." And just like in the middle of it, it'll just output Hindi and Russian. It's like, "What is this?" And the model just isn't like self-consistent enough to be aware of this. It's almost like you need a, like, an actual human to tell the model that, "Yeah, this is wrong." One of the things I think is a giant plague on AI is Alom sis, Alom Arena. And I'll skip the DDoS for now, but I think right now, people will often... It's like if you train your model on the wrong objectives. So like the mental model that you should have of LMSS, LM Marina is that people are writing prompts, they'll get two responses, and they'll spend like five, 10 seconds looking at responses,

• 20:04Speaker 01

  and they'll just pick whichever one looks better to them. So they're not evaluating whether or not the model hallucinated. They're not evaluating the factual accuracy and whether it followed the instructions. They're literally just vibing with the model. I'm like, okay, yeah, like this one seemed better because it had a bunch of formatting. It had a bunch of emojis. It just looks more impressive. And people will train on like they say an LMSS objective, and they won't realize all the consequences of it. Again, like the model itself doesn't like know what its objective is. It's like you almost need like an external like quality signal in order to tell it what the right objective should be. And if you don't have that, then the model will just go in all these crazy directions. Again, like you may have seen some of the results with like the with Llama4, but it would just go in all these crazy directions that kind of mean you need these external qualifiers.

• 20:47Speaker 02

  This also happens actually when you do different forms of like protein evolution or things like that, where you select a protein against a catalytic function or something else, and you just kind of randomize it and have like a giant library of them. And you end up with the same thing where you have these really weird activities that you didn't anticipate actually happening. And so I sometimes think of model training as almost this odd evolutionary landscape that you're effectively evolving and selecting against, and you're kind of shaping the model into that local maxima or something. And so it's kind of this really interesting output of anything where you're effectively evolving against a feedback signal. And depending on what that feedback signal is, you just end up with these odd results. So it's interesting to see how it kind of transfers across domains.

• 21:27Sarah Guo?

  These, you know, course, as you said, five second reaction academic benchmarks, or even non-academic industrial benchmarks are easily hacked or like not the right gauge of performance against any given task. They are very popular. What is the alternative for somebody who's trying to like choose the right model or understand model capability?

• 21:51Speaker 01

  So the alternative that I think all the frontier labs view as a gold standards is basically human evaluation. So again, proper human evaluation where you're actually taking the time to look at the response, you're going to fact check it, you're going to see whether or not it followed all the instructions. You have good taste, so you know whether or not the model has good writing quality. Like this concept of like doing all that and spending all the time to do that as opposed to just vibing for five seconds, I think actually is really, really important because if you don't do this, you're basically, you're basically just training your models on the analog of clickbait. So I think it actually is really, really important for model progress.

• 22:23Sarah Guo?

  If it's not L N SIS, like, how should people actually evaluate model capability for any given task?

• 22:31Speaker 01

  What all the frontier labs find is that human evals really are the gold standard. Like, you really need to take a lot of time to fact check these responses, to verify they're following instructions. You need people with good taste to evaluate the writing quality, and so on, and so on. And if you don't do this, you're basically training your models on the analog clickbait. And so I think, I think that really, really harms model progress.

• 22:51Sarah Guo?

  Is there work that Surge is doing in this domain of like, trying to standardize human eval or make it more transparent to end consumers of the API or even users?

• 23:01Speaker 01

  So internally, we do a lot of work actually today with working with all the frontier labs to help them understand our models. So again, we're constantly evaluating them. We're constantly servicing loss areas for them to improve on, and so on, and so on. And so right now, a lot of this is internal, but one of the things that we actually want to do is start external forms of this as well, where we're helping educate people on, yeah, like, these are the different capabilities of all these models. Here, these models are better at coding. Here, these models are better at instruction following. Here, these models are actually hallucinating a lot, so you trust them as much. So we actually do want to start a lot of external work to, to help educate the broader landscape on this.

• 23:36Sarah Guo?

  If we can zoom out and talk just about the, the larger like competitive landscape and what happens with frontier models over time, what does a meta-scale deal mean for you guys? Or what do you make of it?

• 23:48Speaker 01

  So I think it's kind of interesting in that. So we were already the number one player in the space. It's been beneficial because, yeah, there were still some legacy teams using scale, like, they just didn't know about us because we were still pretty under the radar. I think it's been beneficial because one of the things that we've always believed is that Sometimes when you use these low-quality data solutions, people kind of get burned on human data. And so they had this negative experience. And so then they don't want to use human data again. And so to try these other methods that honestly, just a lot slower and don't come with the right objectives. And so I think it just harms model progress overall. And so just the more and more we can get all these frontier labs using high-quality data. I think it actually really, really is beneficial for the industry as a whole.

• 24:31Speaker 01

  So I am, I think overall it was a good thing to happen.

• 24:35Sarah Guo?

  If you were to make a bet that an underdog catches up to OpenAI, Anthropic and DeepMind, who would it be?

• 24:43Speaker 01

  So I would bet on XAI. I think they're just very hungry and mission-oriented in a way that gives them a lot of really unique advantages.

• 24:50Sarah Guo?

  I guess maybe another sort of broader question is, do you think there's three competitive frontier models, 10 competitive frontier models a couple of years from now? And is any of those open source?

• 25:03Speaker 01

  Yeah, so I actually see more and more frontier models opening up over time because I actually don't think that the models will be commodities. Like, I think one of the things that we've, I mean, I think one of the things that has actually been surprising the past couple of years is that you actually see all their models have their own focuses that give them unique strengths. And for example, I think Anthropic's obviously been really, really amazing at coding and enterprise. And OpenAI has this big consumer focus because of ChatGPT. Like, I actually really love it. It's models' personality. And then, you know, it's just a different set of things that's willing to say and to build. And so it's almost like every company has, it's almost like a different set of principles that they care about. Like, some will just never do one thing. Others are totally willing to do it.

• 25:46Speaker 01

  Others just have different, like, models will just have so many different facets to their personality, so many different facets to the type of skills that they will be good at. And sure, like eventually AGI will maybe encompass this all. But in the meantime, you just kind of need to focus. Like, there's only so many focuses that you can have as a company. And so I think that just will be two, like, different strengths for all of the model providers. So, I mean, I think today, you know, we already see, like, a lot of people, including me. We will switch between all the different models, just depending on what we're doing. And so in the future, I think that will just happen even more as you are just using more and more models for, or using models for different aspects of their lives, like both their personal and their, in their, like, professional lives.

• 26:25Sarah Guo?

  Going back to something Alad mentioned, like, where should we expect to see, like, Serge investing over time? Like, what do you think you guys will do a few years from now that you don't do today?

• 26:35Speaker 01

  Again, I think I'm really excited about this more, kind of like, public research push that we're, we're starting to have. Like, I think it is really interesting in that a lot of the, like, for obvious reasons, a lot of frontier labs, they're just not publishing anymore. And as a result of that, I think it's almost like the, the industry has fallen into kind of a trap that I worry about. So like, maybe to dig into some of the, some of the things I said earlier. With some of the negative incentives of the industry and some of the kind of concerning trends that, that, that we've seen. So, like, going back to LMS. One of the things that we'll see is, like, a lot of researchers, they'll tell us that their VPs make them focus on increasing their rank on LMS.

• 27:18Speaker 01

  And so I've had researchers explicitly tell me that they're okay with making their models worse at factuality, worse at following instructions, as long as it improves their ranking, because their leadership just wants to see these metrics go up. And again, that is like something that literally, literally happens because the people ranking these things on LMS, they don't care whether the models are good at instruction following. They don't care whether their models are, are emitting, like, factual responses. They, what they care about is, okay, did this model emit a lot of emojis? Did it emit a lot of bold words? Did it have really long responses? Because that's just going to look more impressive to them. Like, one of the things that we found is that the easiest way to improve your rank on Almarina is a way to make your, make your model response longer. And so what happens is, like, there are a lot of companies who are trying to improve their leaderboard rank.

• 28:02Speaker 01

  So they'll see progress for six months, because all they're doing is unwillingly making their model responses longer and adding more emojis. And they don't realize that all they're doing is training their models to produce better clickbait. And they might finally realize six months or a year later, like again, you may have seen some of these things in industry, but it basically means that it's been the past six months making zero progress. And in a similar way, I think, you know, besides LMS, you have all these academic benchmarks, and they're completely diverse in the real world. Like, a lot of teams are focused on improving these SAT-style scores instead of real-world progress. Like, I'll give an example. There's a benchmark called IFeval. And if you look at IFeval, so it stands for Instruction Following Eval. If you look at IFeval, like, some of the instructions are trying to check what our models can do. It's like, "Hey, can you write an essay about Abraham Lincoln?

• 28:46Speaker 01

  And every time you, like, mention a word, Abraham Lincoln, make sure that five of the letters are capitalized and all the other letters are uncapitalized." It's like, what is this? And sometimes we'll get customers telling us, like, "Yeah, like, we really, really need to improve our, like, our score on IFeval." And what this means is, again, like, you have all these companies or all these researchers who, instead of focused on real-world progress, they're just, like, optimizing for these silly SAT-style benchmarks. And so one of the things that we really want to do is just think about ways to educate the industry, think about ways of publishing on our own, just, like, think about ways of steering the industry into, like, hopefully a better direction. And so I think that's just one, one big thing that we're really excited about and could be, could be really big in the next five years.

• 29:29Speaker 02

  Okay, yeah. I mean, so Sarah brought up earlier how everybody kind of wants high-quality data. What does that mean? How do you think about that? How do you generate it? Can you tell us a little bit more about your thoughts on that?

• 29:38Speaker 01

  So let's, let's say you wanted to train a model to write an eight-line poem about the moon. And so the way most companies think about it is, "Well, let's just hire a bunch of people from Craigslist or through some recruiting agency, and let's ask them to write poems." And then the way they think about quality is, "Well, is this a poem? Is it eight lines? Does it contain the word 'moon'? If so, like, okay, yeah, I hit these three checkboxes. So, yeah, sure, this is a great poem because it follows all these instructions." But if you think about it, like, the reality is you get these terrible poems, like, sure, it's eight lines and it has a word "moon", but they feel like they're written by kids from high school. And so other companies be like, okay, sure, these people on Craigslist don't have any poetry experience. So what I'm going to do instead is hire a bunch of people with PhDs in English literature. But this is also terrible. Like, a lot of PhDs, they are actually not good writers

• 30:22Speaker 01

  or poets. Like, if you think of Hemingway or Emily Dickinson, they definitely didn't have a PhD. I don't think they even completed the college. And like, one of the things I'll say is, like, yeah, I went to MIT. I think you went there, too. And a lot of people I knew from MIT who graduated with a CS degree, they're terrible coders. And so we think about quality completely differently. Like, what we want isn't poetry that checks some boxes. I'm like, okay, yeah, check through these few boxes and use it some complicated language. We want a type of poetry that Nobel Prize laureates would write. So what you want is, like, okay, we want to recognize that poetry is actually really subjective and rich. Like, maybe one poem, it's a haiku about boom light on water. And there's another poem that's like, it has a lot of internal rhyme meter. And another one that, I don't know, focused on emotions behind the moon rising at

• 31:04Speaker 01

  night. And so you actually want to capture that there's thousands of ways to write a poem about the moon. There isn't a single correct way. And each one gives you all these different insights into language and imagery and poetry. And if you think about it, it's not just poetry, it's like math. There's a thousand ways probably to prove the Pythagorean theorem. And so I think the difference is that when you think about quality the wrong way, you kind of get commodity data that optimizes for things like inner radar agreement. And again, checking boxes off the sum list. But one of the things that we try to teach all our customers is that high-quality data actually really embraces human intelligence creativity. And when you train the models on this, like, richer data, they don't just learn to follow instructions. They really learn all these deeper patterns about all the stuff that makes language in the world really compelling and meaningful. And so I think a lot of companies, they just throw humans at the problem, and they

• 31:49Speaker 01

  think that you can get good data that way. But I think you really need to think about quality from first principles and what it means. And you need a lot of technology to identify, yeah, that these are amazing poems, and these are creative math problems, and these are games and web apps that are beautiful and fun to play, and these ones are terrible to use. So you really need to build a lot of technology and think about quality in the right way. Otherwise, you're basically just, like, scaling up mediocrity.

• 32:11Sarah Guo?

  That sounds very domain-specific. So do you, like, in every domain, are you building a lens of what quality looks like along with your partners?

• 32:19Speaker 01

  Yeah, I mean, I think we have kind of like holistic quality principles, but then oftentimes there are differences per domain. So it's like a combination of both.

• 32:26Sarah Guo?

  I think we got all the core topics. Nice work on podcast number two, Edwin, and thanks for doing this. Congrats on all the progress with the business.

• 32:35Speaker 02

  Yeah, no, thanks so much for having us.

• 32:36Speaker 01

  Yeah, it's great. Great meeting you guys.

• 32:39Sarah Guo?

  Find us on Twitter at NoPriorsPod. Subscribe to our YouTube channel if you want to see our faces. Follow the show on Apple Podcasts, Spotify, or wherever you listen. That way you get a new episode every week. And sign up for emails or find transcripts for every episode at no-priors.com.