The Data Factory: Inside the $100B Race for Post-Training Supremacy, with Labelbox CEO Manu Sharma

Hello and welcome back to The Cognitive Revolution.

Today, my guest is Manu Sharma, founder and CEO of LabelBox, a data factory that supplies frontier training data to all of the top Western AI labs and many enterprises that are pushing the performance frontier with task-specific fine-tuned models. This conversation really couldn't be more timely in the wake of Meta's recent $15 billion deal with Scale AI, which of course saw Scale's former CEO Alex Wang join Meta to lead their superintelligence team. Other frontier model developers have been left scrambling to secure their own training data pipelines, and the market overall is still in the process of realignment. These headline-making developments have really spotlighted just how critical super high-quality training data is to today's frontier capabilities, and also how difficult and expensive it can be to create. Post-training budgets are growing rapidly as labs race to imbue their models with differentiated capabilities. And as Manu explains, Every Western frontier lab is now spending over $1 billion annually on frontier training data. Such tremendous investment is required because modern data work has moved far beyond the simple labeling and preference indication tasks that many are familiar with from years past. And so today, with Manu as our guide, we'll be tracing the evolution of post-training from supervised learning from human examples to reinforcement learning from human feedback to the reinforcement learning from verifiable reward paradigm that's ascendant today, and unpacking how all of that has shifted data creation work away from tools that facilitated the collection of human preferences and reasoning and toward environments, which Manu calls gyms, where models can develop new skills, starting with coding, mathematical reasoning, and computer use through a process of trial, error, and highly automated feedback. The bottom line is that today, when you think about training data creation, you should envision not the data sweatshop of the past, but highly qualified and comfortable domain experts working to create sophisticated reinforcement learning environments and verifiers that teach frontier models to solve complex long-horizon tasks. A quick disclaimer before we begin. LabelBox will be sponsoring the show for at least the next month, and so this does qualify as a sponsored episode. Nevertheless, I can sincerely say that this is a conversation I'd have wanted to have anyway, because whether you're tracking AI progress analytically or trying to achieve superhuman performance in your particular AI application, understanding the best practices for training data creation is essential. And besides, the scale and scope of Labelbox's operation is genuinely remarkable. Their most recent capital raise was $110 million in early 2022, and today Labelbox operates as a vertically integrated data factory. They're scaling their expert network in part by conducting over 2,000 AI-powered interviews per day, and their most in-demand experts are already earning more than $250,000 per year On the platform. I haven't yet had the chance to sign up to earn some of that money myself, but I am looking forward to the AI interview experience. And assuming Manu's right that the system does recognize me as a qualified expert, I'll report back, subject of course to any required NDAs, on my experience as an AI trainer. Now, without further ado, I hope you enjoy this deep dive into the ongoing evolution of the red-hot training data market with Manu Sharma. Founder and CEO of Labelbox.

Manu Sharma, founder and CEO of Labelbox. Welcome to The Cognitive Revolution.

Thank you.

So I'm excited for this conversation. Your world is in chaos. The AI world is going through a bit of a reorganization or realignment right now. And when I say your world, I mean the world of data, data creation, human sources of data. You've been in this business with Labelbox for a number of years, and we'll have a chance to dig into all different facets of that. But obviously the news that has sent various forms of shockwaves, I think, through the industry in the last couple of weeks is Zuck and Meta doing this weird deal with Scale where they're getting Alex to come over and lead the superintelligence team or something along those lines. Then of course we've got people leaving Scale and it seems like generally speaking, it's chaos. So what's your report from the front line? Is this like driving a lot of opportunity for you? Are people confused? What are the takes that you are hearing that you think are interesting? Just super interested in kind of your gonzo report from the front to get started.

Yeah. So this is one of the most exciting times in AI industry, just generally across the board, right? Because when you kind of really look into the innovations and pace of progress, I believe we are experiencing the maximum innovations we've ever seen per day or per week kind of time period. And it's an AGI race. It's a race among a number of companies and groups. And it's really interesting to see there are like these big AI labs that are playing and pushing out awesome capabilities of base models, as well as products and product experiences. But then you have a number of teams that are just emerging with some new ideas and taking a bet on alternative techniques and so forth. One thing, one thing clear underneath all of that is that these teams need essentially 3 components to develop frontier AI systems. One of them, of course, being compute. We've been seeing just incredible investments in CapEx across the board. The second obviously is the AI talent. And so these are obviously the researchers who know all the know-how of crafting these neural networks and the architectures and how to train them and what to train on and so forth. And a third equally important piece is actually data. And we are very much in a regime where a lot of the emphasis is going towards post-training. So kind of like in the AI lens, the way these AI models are trained is you have initially a first phase where you're essentially taking all of the data from the web and from the sources where you can get access to, and you're essentially training a base model that can understand the patterns of all data generated by humans across over these, let's say, 100 years or so. And then after the base model is trained, there's this thing called post-training. And post-training is where you really emphasize some aspects of the AI models that we interact with. It really turns these AI models into assistants. And within post-training, There has been a number of techniques over the last 3 or 4 years that have become emergent or very important. And with your question, it is really awesome to see a lot of limelight thrown at this industry where, how do companies produce this post-training data? And so I think the news about Meta is essentially emphasizing that the data is very important for building these AI capabilities. AI models and AI capabilities. And our industry, generally speaking, has been going through a number of exciting step changes. So when we started LabelBox in 2018, supervised learning was in kind of the main swing, right? Everybody was really putting a lot of— creating a lot of training data, which was essentially for supervised training. So you Have humans around the world who would tag images and videos or some snippets of text, and they'll essentially train these models to mimic that behavior. And around when the transformers came out, we started to see the inflection towards unsupervised learning, where most of the learning was happening was on this sort of large-scale datasets. However, as we emerged through that new paradigm, then RLHF became a prominent way to actually make these models useful. useful in every day-to-day life. And this also required yet new forms of data, very specialized datasets with expertise, because now these models have very strong base capabilities and it needed to be— it needed to learn the ways and the skills, if you will, to interact with knowledge work and knowledge workers. And so RLHF, SFT became sort of a main thing. And now we are yet in another new paradigm of reinforcement learning. So nearly every AI lab that I am aware of is really emphasizing on creating datasets for reinforcement learning. And it's like reappeared back in the days we saw all these amazing innovations of DeepMind with AlphaGo and so forth. And they were really trained on reinforcement learning and it reappeared now, but in a new flavor, new form. So that's where, at a high level, that's the trend we have seen. And all of these things actually require specialized datasets, and they need a lot of that. And I think we are, I would say, still in the very early days of actually making these AI systems more and more capable for everyday tasks that power the economy across the knowledge work, right? And so that's what's going on at the macro level. And these AI players who are building these large flagship AI systems or models, They are very aware of like how important it is to have access to ways to producing a whole bunch of datasets in all of these interesting specialized domains that ultimately really get into these useful business domains where the AI can learn about, let's say, what professionals, let's say software engineers are doing in every day-to-day tasks when they're building companies and software. Or if you have, let's say, domains like healthcare, insurance, all these amazing industries that we have, like how do you really learn from like core workflows, produce datasets to create these models? It's going— it's generally very interesting, exciting times. And it is very common to expect, let's say, in an industry where M&As happen and certain teams make these decisions as they see fit for what the opportunities are. And Labelbox has been behind the scenes providing these datasets to most of these AI labs. And it has shifted the workloads maybe a bit, you know, maybe sometimes quite quickly. Sometimes it takes some time for teams to reorient to the new supply chains and so forth. Yeah. So I think that's the macro kind of context on what's going on. And I certainly expect even more changes in our industry over the next few years. There are new ways, there are new teams that are innovating on How do people create datasets? How can we go help these AI labs to produce these datasets? And yeah, so that's what's going on. But it is generally very interesting and exciting times for people who are really keen in how these datasets are created and how these models are learning.

Hey, we'll continue our interview in a moment after a word from our sponsors. Let's dig in a little bit more on the kind of nature of post-training. The different techniques and then maybe some of the subtle differences. So a year ago, maybe the prevailing wisdom was like maybe 98%, 99% of compute was going to pretraining. And then this post-training, while the data was really obviously critical to make it work well, it was like a small thing that you could rapidly iterate on. It seems like that has changed, but it's unclear how much it's changed and if everybody's changing in the same way. I'm always very confused by Are we experiencing convergence or divergence in the leading model providers' offerings? There's definitely striking convergence, but there's also maybe some divergence and maybe more divergence to be expected in the future. And I think about this anecdote from the Claude 4 training where they had mistakenly left out, I'm sure you've seen this, that one SysPrompt harmful dataset, and then discovered behaviorally, as I understand it, that the model was following harmful system prompts. They're like, why was that? Ultimately traced it back to, well, we omitted, we had a typo in our config files, and this one dataset that was designed to teach the model not to follow the harmful system prompts was omitted. And so that behaviour was not learned as intended. And then they didn't go back and redo all post-training with the right configuration. Instead, they patched it or made an on-the-fly adjustment later. So from that, I infer that the post-training has now become a significant enough part of the overall compute that a company like Anthropic, as committed as they, I think, mostly credibly claim to be to getting this stuff right, didn't feel like they had the luxury of going back. So what can you tell us, obviously without sharing client secrets, but in a sort of abstracted way from any one strategy? How much compute would you say is now going into post-training? Is it as simple? I'm sure it's not. It used to be pre-train, then some SFT, then a little RLHF, then you're done. Or at least that was how the public understood it, the AI public. I assume now there's more interleaving, more steps, and just more complication. So yeah, tell me more about what you think that kind of post-training has evolved into.

Yeah, so I'm confident that in all of the big leading labs, including many of the hyperscalers, that they are actively researching and innovating across the wide front of end-to-end training recipes. I'm sure there are teams that are pursuing some new ideas on how do we make pre-training more efficient and things like that, the right kinds of data mixtures and so forth. But you have to know that pre-training can take a long time. Let's say if it takes a few months to actually train a very large model and so forth, it is in a release cycle when these companies are shipping these products, that is a given. It's going to take X amount of time, 6 months, let's say, to really get something out the door. And of course, we are seeing that some of these teams are training gigantic models as a base model. And we'll talk about it in a bit. These models are, in many ways, used as a teacher model for student models that are actually going to ship. Now, I would say that the very growing budgets are going into post-training. And I think the intuition that I'll give here is that in many ways, these base models have very strong capabilities about understanding the patterns of data and knowledge and so forth across a wide range of modalities. However, what we are still finding is that these models now have to produce useful work in the form of, you know, maybe you call it agents and in form of like, you know, long horizon tasks. And to solve that problem, these research teams have to use techniques to teach that capability. And so there is sufficient knowledge and capabilities in the base model, but you have to then go out to teach these models how to do XYZ tasks. But just taking a very simple example, like in software engineering, These models obviously are so awesome in terms of what they can do in coding, but there's so much still to be desired when it comes to professional software development at large scale. Like, how can these models develop an entire suite of products with minimal supervision? I would say, I think we get— we were starting to see that, but it's not there yet. And to solve that problem, these research teams have to collect the right kinds of datasets and then do reinforcement learning on that. And so that is where a lot of focus is going across the board for these teams to test new recipes and new techniques and see if their models are getting better in those sort of capabilities. And so they have generally a team that is— or a number of teams that are exploring that frontier. Like, how do we get these models really good in all of these sort of categories? I think The 2 obvious macro ones is coding and math, because the fundamental way these RL systems work, in coding and math, you can verify to a large extent, like, you know, did the model get the solution right or not? Right? So let's take a few examples. Like in coding, let's say I come up with a PR request and did the PR actually solve the stated objective? And does my test pass in the codebase? You know, the prior tests and the new tests. And in RL, essentially, the system will try millions of attempts to solve that problem. And if it gets right, it scores really well, and then the gradient, the weights change in the model, and then they keep going to the next one. And similarly, in math, for large— there's a large number of problems that can be verified by numerical answer. And so the kind of very fast progress we saw in the reasoning models and The coding capabilities in the last, I would say, 9 months or so is because of that. These teams are putting a lot of emphasis on just getting that right. And A, it is a very valuable thing to do because coding could be one of the most lucrative knowledge work in many ways. And some teams have a belief that they can actually, if they solve that, they can actually accelerate their AI research and so forth. So it certainly seems like Anthropic believes that, many others do. And then there is new innovations or new ways to train these systems in non-verifiable reward regime where a lot of our day-to-day work is you don't really know what the right answer may be, and there may be multiple right answers, but how do you model the data where these RL systems can track their progress and are they getting it right? And if they create a candidate solution, how can they score that candidate solution? Effectively. And so we have some really promising techniques right now that is underpinning a lot of these kind of latest capabilities across the board you see across the models. Now, I do think that at a very high level, these models seem to feel like converging in the sense like they interact and they look and feel the same from the outside. However, there is this philosophical principle where to be able to really critique a capability, you really have to be at an expert level to really understand the nuances of quality and really kind of like then have sort of a quality judgment, like, you know, this thing, this model is really good at this thing, but not that thing. And so there is convergence maybe at a very high level when you look at very generic capabilities, but there is actually a lot of divergence when you go into the details. So I code quite a bit. It's awesome for me to be able to get back into coding now with AI assistance. And there are many times where I am working and solving a very tough problem in coding, and you're kind of stuck and you're iterating with these models, and you are able to have an intuition. At least I get an intuition where I've hit the ceiling of this model. I'm in a trajectory where I'm not really going to get this model to solve this problem. And like, I'm able to see that across a variety of models where I have an intuitive judgment, like, you know, model Gemini could solve this problem, maybe because it's like very algorithmic and requires some kind of math domain to truly solve that. But maybe when it comes to really refactoring the software or architecting a distributed system, I might use Claude because it has this amazing property where it is asking a lot of questions or in an internal state and getting more knowledge from reading this file, that file, and ultimately trying to make a better kind of software decision. And from a macro, it looks like convergence, but when you actually go into the details, these models are getting optimized for very different goals, it seems like. Now, I do believe that the ambition among these AI labs is to make generic systems that are just so great across the board. But in reality, what we are seeing is that companies are also starting to optimize experiences where their user bases are. For example, in Anthropic's case, it clearly is that they're very focused. They're very intentional on making— solving coding. And when you look at maybe some other models, maybe they're more optimized for consumer use cases and everyday tasks and so forth. And maybe some other labs would be focused on making a general social kind of assistant, everyday assistant, and that would have different sort of optimizations to have. So that's what it, to me, feels like, where we are at right now. Now, there is a very interesting question. I do think that is, ultimately, when these teams figure out how to make the best coding system, how to make the best researcher for scientific domains, are they able to bring it all together back into a single world model? And I think that remains to be seen.

Broadly, I guess I want to understand better what is the role of human data today. Maybe one way to think about it would be like, if we didn't have this human data, how would things be different? How would things fail? I'll just give you a little bit of a prompt. On the math side, it seems like we've seen from things like the DeepSeek R10 paper that you can get a model to learn at least highly verifiable domains like math encoding pretty well. Even without the human traces that are typically used in the supervised fine-tuning. But that comes with some very important downsides. Like, it starts to do it in strange ways and just generally act very strangely. Language switching being one example, odd behaviour. But I think lots more where that comes from if we detach ourselves from human baseline and just spin the RL centrifuge more and more intensively. So I guess one value of human data is it provides this anchor that gives us some hope for some sort of alignment by default, because we kind of know where the starting point is. Another interesting juxtaposition would be, I think it's GPT-4.5 versus O3 Mini is like the best contrast, where GPT-4.5 has really good trivia knowledge, like blows O3 Mini out of the water on ability to answer random questions about the world. But O3Mini is way better when it comes to these sort of reasoning challenges. So we can decouple reasoning and behaviour more broadly from world knowledge. And then there's the difference again between supervised fine-tuning, where my general understanding has been like, it's really important to record the trace and you're training the model on the actual reasoning process as exemplified by the human. I've had a tonne of value from that. But then in the reinforcement learning, is that reasoning trace so important? So yeah, break down kind of the roles, flavors, and major drivers of impact of human data today, as opposed to if you were just doing the RL thing to the max.

So the expert data continues to be extremely important for all of these capabilities that we are about to see, we have seen in the past. And I think in one way it seems like we kind of underestimate how incredible human intelligence really is and all the world and economy that we have built and all the things we do. This is extremely special and we are trying to slice— in many ways we're trying to emulate it into AI systems. In many ways AI systems are superhuman, but on the other hand, they still cannot do many other everyday-to-day tasks yet. And the way I think of it is that the, let's say, in the domains like mathematics and so forth, we've taken that slice of like, okay, here's an interesting, awesome dataset about the truth of how at least the logic and reasoning work in sort of purely scientific domain. And we can retrain these systems to learn from that and adjust its critique and reasoning from that source. But then with that, we've gotten these state-of-the-art systems today, and they're very incredible. But yet when you look at, when you put them into agent, these products, there, I would say like it is a, there's a baseline established now. There's a reasoning behavior that is basically a foundation for all these amazing kind of capabilities we want to now go build. And so To build these new capabilities, let's say, take an example of a sophisticated AI system. In the future, you'd be able to simply call any company. Let's say you want to change your Wi-Fi router or some— something like that. You know, you previously would call a whole bunch of people, stay online for 40 minutes, and try to resolve some issue. You know, we are very soon going to be interacting with purely AI systems, and these AI systems will be agents where there's a concierge who takes your call, routes you to the right kind of specialized expert on how to solve that problem that you have with Wi-Fi router. And maybe the system has to now interact with the entire database to perhaps provision a replacement and things like that, right? It's a fairly sophisticated system with interacting with you with voice, but then has these capabilities to read the databases and navigate a variety of software stacks inside that company to help you as a user. That system, to really build it reliably, there is just an incredible amount of edge cases that first of all exist because we are now working with tens of millions of users and use kind of ways of how people could request these things. The base model is a starting point to develop these capabilities, but the architect of that sort of agent or that solution would need a very good dataset that represents all of those kind of distribution of the use cases or the interactions that the system would have to go through. And if you were to build that system today with base off-the-shelf model, you're gonna require a lot of orchestration on top of the base models. You would need a lot of software stack and agent frameworks and so forth, and you might still not get very reliable output. And that's just one very narrow example. And there's probably like, I mean, like this infinite number of examples and use cases like this in our economy. And all of these things are gonna require datasets for these systems to be rock solid and superhuman, super reliable. And I don't see any other way to go create these datasets. You can't really invent these datasets from algorithmic synthetic approaches. You can bootstrap it. You can make the process easier by all of these techniques of compute or synthetic approaches. But you'll still need this quality judgment from how the authors or creators of this system want the AI system to interact. How do they want the customer service to be like? What personality it needs to have? These things are a design parameter. And these people who are gonna build these systems have a choice on how the systems needs to act on their behalf of their company. And so all of this requires requires kind of new forms of data, whether it's like the voice interactions, whether it's tool use and things like that. Now, I think there was some time period that we saw, maybe starting last year, that a number of research attempts was to mimic reasoning traces. Like, hey, let's ask expert humans, all the academics in universities and so forth, to teach models about how they think about whatever their scientific domain is and break down those problems in some reasoning steps and so forth. And I think what we have found is that, and I think researchers are finding that it's just really hard to ask humans to really express themselves in that reasoning step by step. And in fact, like there was some research papers going back to process reward models where there was really an emphasis to Teach the model how to think step by step and so forth. I think we have largely moved on to essentially modeling the problem and a solution and how would you grade the solution. And that is a new paradigm. So the reasoning is actually an emergent property of training these models with RL. And I'm sure there are ongoing efforts to continuing to make that much higher quality, and you might still need some sort of traces or audits of traces and so forth. But largely where I see things going is in the kind of framework where, that I described, where take any problem and then solve that problem and express that in a form of a way where you know the solution. And if you don't know the solution, you at least know how The AI system could be graded on how good it is performing in that task, and everything else becomes an emergent property to achieving that goal.

Hey, we'll continue our interview in a moment after a word from our sponsors. So the shift in both data and training— I'll try to summarize back to you and tell me if I'm getting it right— is from a supervised fine-tuning mode where The human is responsible for both the final answer, but also the, how did you get there? Spell out your reasoning, let's think step by step, et cetera. And then the model is learning to imitate that pattern of thinking in hopes that learning the pattern of thinking will get you to the right answer at a high rate of reliability. That's the previous paradigm. The new paradigm is Maybe because we've just found it really difficult to collect that data, maybe for some other reasons. Instead, now we're just focused on give me a really good solution, and then let's develop a rubric for evaluating the AI solutions. And then we train with a reinforcement learning signal, presumably where the AI takes its shot, and then another greater model basically gets a, here's what the AI just did, here's the gold standard example, here's your rubric. Score the new one with the gold standard in mind against this rubric. And then the reward signal given back is like the sum of points on the rubric or whatever. How'd I do on that? What else would you, how would you elaborate that?

That's right, actually. And in most of the domains, you'd not know the right solution, but what you will know from these domain experts in all of that respective industry is that how would they, like, They would know what is good and great or excellent. And that quality judgment has to be somehow expressed in the data for the systems to learn. So a lot of the work is going towards now producing these kinds of datasets. And you might have heard, like there's a— you've had a number of speakers who are from these AI labs and some of them are doing new interesting work. You've talked in the previous episodes about RL environments. Where it's like a— the simple thing, the way I think about it is like a gym. You get the model to go into this gym and practice a whole bunch of tasks, practice a whole bunch of things, and it possesses a new skill. And so think of any domain and you want to really model that in some form of an environment where these models can go and play and start honing that skill. And the challenging part actually is how do you craft these RL environments that are representative of Sometimes generalized, very generalized skills. So let's say one way to categorize things is like games could be much more generic planning games, but then let's say coding environments or maybe other environments are perhaps more specialized on some domains and in certain kind of use cases. And yeah, so that's generally the way to go about it. And now this also has a new sort of challenges, like to be able to build these systems and to make this paradigm really work. Yeah. You've got to have very reliable graders, autograders, where these are specialized models that are essentially able to align with how, let's say, an AI system will grade the progress, and it will be aligned with kind of the experts and the domain experts. And so there is another work stream that goes into making sure that that is rock solid. And so once you get that, then you can really scale up this RL training and performance. in the post-training routine.

One little aside, and maybe you'll have other examples too, but I've always wondered about super long context. That strikes me as one that would be just really hard to— you mentioned it's hard to get people to record their thoughts. I've personally experienced that quite a bit because I've advised a number of companies and coached people on like, okay, you don't need that much, especially if you're fine-tuning for a narrow task. The dataset doesn't necessarily have to be super huge. Obviously, the broader the task and the more diversity of inputs, then the bigger the dataset needs to be. But for a lot of these narrow, very purpose-built work-defined workflows, you don't necessarily need that much data. But yet I've always had a hard time of people actually sitting down and writing out the traces. So I've lived that and can empathize with that. When it comes to long context, I haven't personally ever tried, but I've always wondered if you're talking in the Gemini case, right up to a million tokens. It seems very difficult to just drop a million tokens on someone and say, like, okay, you're responsible for the right answer. So I've always assumed that stuff was more programmatically done and needle in a haystack or 10 needles in a haystack or whatever. And then I guess that's just generalized. But maybe you have more insight. Do you have any insight into how, and especially now, like Gemini, not just can find a needle in a haystack, but it has really impressive command of, in my experience, like 500,000+ tokens. It seems to be not just finding the anomaly, but really grokking what is going on over these long contexts that are obviously unseen and codebases that it's never had any access to. So, yeah, I don't know if you would, but any insight into how that has come online and what role, if any, the human data creation engine has played in that process?

Yeah, totally. So you're right, like when it comes to extremely large context windows, you know, how do you really verify or test, or in other words, like create datasets that these models can express like how good they are in those long context windows? So I think there are 2 broad camps that I've seen. And so one of them is, of course, let's say you can model needle in the haystack programmatically, create a whole bunch of hashes and see if you can actually find a particular hash and Large number of noise and so forth. And so those are, I think, that's that camp of testing, perhaps maybe at a very fundamental level, can it find the information in a vast ocean of information data points? But then there's another camp, which is when you test it in the real world. So we're actually creating a ton of datasets that are testing these capabilities of these models in long context and This would include, let's say, let's take an example of financial analysts. There's a very big industry where one of the jobs is to synthesize a vast amount of information like SEC filings about a company and all of the information they may have released about their company performance and things like that. And these can be, you know, 10 documents. It could be also audio transcripts of earnings calls and things like that. And the financial analyst who is really top of their game would be able to not simply just answer a question about, like, hey, what was— like, tell me a factual thing about where is this in 100 documents, which becomes in many ways a search problem. Rather, they would model the problem where, like, okay, given all of these different modalities, I want to accomplish a task. I want to analyze and create a kind of a model of the company. So that I can forecast what maybe the company's earning potential is and things like that. And that is actually very representative of a real-world task. Somebody may take days to go create those things. And models, while they're very good in many of these examples, largely there's more room for them to actually accomplish those tasks in a reliable way across a variety of domains. And so that is an example where you're really forcing the model to understand and reason across multiple modalities and pieces of information. And what we call it internally is multi-hop capabilities. And there's some strong performance, but again, there's so much to be desired. And so that's one of the ways that these AI labs are casting this long and honing the reasoning. Planning capabilities across wide, large context windows. And if you really, let's say, if you throw into video and other streams of other modalities, they fundamentally take a lot of token space because they have to be tokenized and so forth. In many ways, Gemini particularly is very good in, let's say, video understanding and things like that. Like, hey, where, in which frames something happened and so forth. But when it comes to real-world applications, that is trying to kind of mimic or plug into an actual industry domain experience kind of workflows, these teams still need to improve these capabilities across those kind of vectors or domains, if you will. And so there is a way to model these things and test these capabilities. And the details is in the— devil is in the details where like we can induce failures across all the frontier models on these kind of tasks, and that data becomes very valuable for these AI labs to then use it to hill climb.

Could you describe maybe the different modalities of interaction or of data collection? And again, I'll give you a little prompt from my own experience. This was actually summer 2022. I spent all summer doing fine-tuning of the text-davinci-002 series of models. They never actually released. We had an early kind of preview of that. product, but they never released it. What I found to work well for me, and there was no multimodality at the time, was for any given task, do 10. Then at some point along the way, I discovered the importance of the reasoning trace, but I didn't know that at first. So I would do 10, fine-tune on just 10, then I would have 100 queued up. I would have the model do the next 100. And then I'd basically do sort of rejection sampling or correction and specifically emphasise the ones that it got wrong, fix those, put those back into the dataset, do another 100, and basically try to get myself into some acceptable success rate through, depending on the task, 2 to n rounds of iteration. But I'd say the qualitative modes were like, initially, I'm totally on my own. I have to do the task. Later, it more is about having the AI do the task and review its outputs and find its flaws and correct. How has that evolved today?

In context of fine-tuning these systems and models, you mean?

Yeah.

Well, actually, another next question I was going to ask is around segmenting the market with respect to the hyperscaler frontier labs clearly have kind of one set of needs where they want to go out and just get all the data they can from all the different experts and be amazing at everything. And then you mentioned financial services, and I also think pharma, and obviously a bunch of different more specialist companies that I assume are more engaged in a fine-tuning exercise. So, yeah, we could add another dimension to the breakdown there as well, if that is like a fault line in how data is collected.

Yeah. So I think what you shared about this iterative creation of a small dataset to Perhaps for eval, and eval essentially is some sort of a holdout set from training data in a way. And that is generally considered to be the best practice, right? And it's really surprising, like how many teams or how many individuals across the industry would get that wrong, because it's sometimes the hardest part in the enterprise kind of space is to really think hard about like what the actual workflows or things that they want to go automate or build is. How do you express that in some sort of a dataset and eval? And it is emerging to be a craft in its own, to kind of, like, really understand that trace or trajectory and express that into datasets. And generally, the best teams or the successful outcomes we see is where you're essentially starting with something and iteratively adding on to that evaluation set. And like anything else, There are known knowns and there are unknown unknowns. And so when you're in that journey, more often than not, we find that teams will uncover completely a new edge case that is never thought about. And we have to go back to the drawing board and say, what if we just did it this way or that way? So like anything else, iterative development is the cornerstone of success across the software industry, right? That's what makes us, like the software team, so agile and successful in all these endeavors. I think that is so true in this new world where rather than writing code, we are producing datasets and eval datasets for these AI systems to be evaluated and trained on. And fine-tuning is really interesting. I don't think that fine-tuning like we used to see before is going to last that in the future. What I mean by that is like in the supervised learning paradigm, like 4 or 5 years ago, Generally speaking, most teams would take a base model and then they would sort of specialize. They would take the data that they have, they're collecting from their sensors or their companies, and they would want to fine-tune that model and make it superhuman on that use case. I think that now in like today in 2025, the fine-tuning at best is helping you make the model efficient in whatever the task you want to do. So what I mean by that is maybe it is that you don't want to— maybe it's very expensive to, let's say, run a very large model at a high frequency, you know, millions of queries per hour on a particular use case. And you can achieve really good state-of-the-art performance on that. But given that it is a fairly narrow task, you can actually produce the data from that large model and then distill it down to a much smaller specialized model. And for that queries per minute, per hour kind of operating parameters, there is a better, smaller model that is just going to be better across the board. It will be achieving the same quality at even a fraction of the cost. And so that to me seems like one obvious reason to do fine-tuning of models nowadays. And then there is probably maybe second, I would say, bucket where Some problems truly are so unique where these base models don't have that capability and you just have to get it, have it. The only way you achieve that is through fine-tuning. But if you look at like last few years, we saw companies like, I believe Google released Med-PaLM model, which was trained on healthcare-only data. And my list I was reading somewhere was that the base Gemini models at 2.5 outperforms a hyper-specialized model that was simply trained on healthcare data. And so what does that say? It is saying that these base models, the reasoning capabilities that it's learning from vast amounts of other data, is actually useful to solving that domain problem. And so maybe the idea that many enterprises or many companies thought that they had very special datasets in their companies and petabytes of data, maybe that dataset actually is not that useful because it is not going to be helping you improve the reasoning capabilities, certainly, of the models, because you do need that broad capabilities for models to learn from and emerge the reasoning behaviors. But perhaps fine-tuning might help you to achieve that particular goal more cost-effectively with the things I just said earlier. And yeah, I think we don't see as much fine-tuning across the board among our customers. Now, what we do see is a lot of context engineering, right? So fine-tuning, you're really changing the weights of the model and so forth, but there's so many problems you can actually now solve by simply context engineering, and that is emerging to be a very specialized craft in its own. So prompt engineering kind of emerged as, like, you're coming up with always a way to model a problem in a prompt, but I think the most effective systems that we interact with today, they're actually very carefully context engineered, right? So that not only includes the prompt, but also all of the context that you might give that model or that query through retrieval mechanisms and things like that. And, you know, that is not an easy thing to do, but it is something that is tractable and becoming very effective.

Just to make sure I can pass the Turing test on that coming back, you're saying across these high-value industries, finance, pharma, whatever, the base models are so good at reasoning that you are better off. I'd be interested to hear kind of the scale of this too. Like, if one of these— you mentioned the financial analyst use case. We could collect a bunch of data. We could imagine fine-tuning that. OpenAI offers that on their platform. Of course, you can do it open source, notably. And I had this question in mind too, like, why doesn't Anthropic offer fine-tuning in any way, in anything similar to the OpenAI way? And why doesn't Google? And it sounds like your suggestion is maybe, and correct my numbers, but maybe you're going to create 10,000 examples of what a great financial analysis looks like. And then at the time, at runtime, you'll choose the best 10 that are relevant to this. Maybe 10,000 is too much. You choose the best 10, choose the best 50, put those into context, and it's about what runtime examples you are using that is driving the end performance. But you can get better, as good or perhaps even better results from that than Actually fine-tuning and changing model weights. Do I have that generally right?

Yes. Like, a lot of the agents that we are seeing now in the industry does require you to have context, right? So let's just take an example of code. When you're interacting with Cursor, one of the things that makes it so effective is that there is a way for base model to get all the context, like what are the coding files and what are the relevant snippets, what are the Functions that are similar. And so this is one of the directory structures, and all of that information is modeled and sent to the base model and for a coder to achieve, accomplish the goal that they're working on. And that example, you know, you can kind of like, that is a really good framework to think about in any other domain. And so what is the task and how do you take all the right relevant information And provide that as a context for the model to accomplish that. And we are seeing it's very effective. And the technology wave is where the base models are actually gonna get better and better. And you might not need as much, maybe you won't need as like many verbose contexts. Maybe the context efficiency becomes more improved over time. But generally, like, that is the arc of technology where these foundation models are racing to absorb the capabilities that the other app layers are trying to develop with all this context engineering and so forth. It may take some time for these base models to be just so great in everything. But meanwhile, if you want to accomplish a certain task and goal, you're really focused— most effective way to do is to just context, like to engineer that system. And like, there are some cases, like we fine-tune models all the time ourselves. And we do that in scenarios where we have to optimize for cost or, and or Where we have to, we have a very unique opinion on how we want the judgments of cases to be. So I'll give you an example. So in our industry, we have, we work with millions of domain experts around the world. And this could be, these are canonical examples, like, you know, physicists, mathematicians, Olympiad-level software engineers, so forth, right? So across the— 70+ countries. And one of the things we do is, like, how do you assess these individuals and expertise? Like, it would be maybe 10 years ago, if you were to approach that problem, one, you know, the teams would try to solve that by, like, creating tests for, like, hey, you know, here's a physics test, here's a math test, and try to perform on those things. But then you can see, like, it's just not scalable. You have to go manually craft these things. So what we do is we actually have an AI conversational system that's powered by the frontier models. And when these experts interview with us, they're interviewing with an AI, and we know about their resume, we know about their research and so forth, and they're having a conversation for 30 minutes or so about whatever domain there is. And then there's an AI system that assesses how good the interview went. And you're essentially trying to make a judgment, like, is this person going to be a create a producer of datasets for teaching the frontier model new capabilities. And a base model will have its own sort of properties or characteristics of that judgment. And we don't necessarily align with that. We think that that is not good enough. And so for us to improve that judgment based on all the things we see, the outcomes of these experts in the datasets, we have to— kind of close that loop. And one of the ways we do that is by fine-tuning these models, because that kind of capability just simply don't exist, that quality judgment. And that's hopefully some examples on where fine-tuning is helping and the kind of use cases for it. By and large, most of the industry that I see across the enterprise space has moved towards context engineering. And we see some examples of fine-tuning here and there, but it's largely context engineering across the board. And in many ways, actually, it's really good because it is— the technology is enabling enterprises to now intuitively think of the most valuable workflows they want to solve and express that in some sort of like a multi-step agentic trajectories and what perhaps the principles of solving that in an effective way would be and build that agentic sort of architecture or system. It might require retrieval, it might require multiple function tool use and calls. And in many ways, it is a software engineering task with some sort of a domain task mapping or domain task engineering. And that's a lot more tractable for a vast amount of enterprises where they may not have the skill set to be machine learning experts and train the models and go into the nitty-gritty of model training and so forth. And that's, I think, one of the reasons why the AI models and the industry have taken off so quickly is because now most of the world is now finding ways to implement it in a So real quick on fine-tuning, since it may be a mode of attack and decline on some of these even quite highly specialized problems, what models do you see people mostly choosing to fine-tune today? I would say generally 3 models, QWEN, LLaMA, Mistral. I think it's the 3 categories that we see in the open source world. And I don't have much visibility into the private models, but I think it usually comes down to the nature of the business in the company. And so in highly specialized industries, maybe the kind of preference for these companies is to use open source models because of the operational constraints, because they're either in air gap situation or they want to have things on their own servers and so forth. But if you go into different industries, like, you know, let's say high-tech digital natives, companies like Airbnb and Pinterest or so forth, they might be actually using the state-of-the-art models from their cloud providers or OpenAI and so forth.

So on context engineering, then maybe you can coach me on a task that I've been working on for a while. Every episode of the podcast I write, and then now Claude does often a very admirable first draft. of my intro essay, which I'll put at the top. And the basic approach there has been pretty much constant across, I'd say, since Claude 2 wasn't very good at doing it, but I would do the same thing. I would just take the transcript of the current podcast and then a bunch of examples of previous intro essays that originally I was just totally writing freehand. And then I just have a very simple prompt that's like, adopting the tone, style, voice, perspective, structure, et cetera, of the attached example essays, write a new intro essay for the attached podcast transcript. What I've always felt a little weird about there, or wondering if I could do better, is notably for all those examples, I don't have the source because I don't have enough context to provide a bunch of transcripts and their resulting essay. So it's not like I have the inputs and output pairs. Instead, I just have a bunch of outputs that I consider good and then the current input. So how would you advise me to maybe think about refining my context engineering to get better first drafts? And I do always still, for now, edit them. I'm waiting for the day when Claude just nails it, but How would you suggest I improve that approach?

And where do you think currently the system is lacking? Like, you think it's not producing good quality or, or you expect it to do other things that it's not doing yet?

That's a great question. I mean, it obviously writes very nicely. It has no trouble generally following my structure. I feel like it gets me on sort of the cadence. I have this like persona. Which is genuine, it's not like an act, but where I'm both very enthused about the AI upside and legitimately fearful of could this get out of control, et cetera. It usually does a pretty good job of picking up on that and finding that sort of balance point. I'd say the things that it doesn't do super well are, and this is where I have this lingering idea that maybe if I was able to give it more of the inputs for those previous outputs, it might help, but I don't have context to do that. It doesn't seem to, in some cases, pull out what's actually most unique and interesting about the current conversation. It sort of gives me something where it's like, yeah, that's fine, but in terms of how I would really want to tee this up and frame it for people, why is this relevant now? What is most interesting about this? How does this relate to broader understanding? It's that kind of stuff where I'm often like, Not quite doing it. Sometimes I get better results by— often I get better results these days. I do sometimes elaborate on the prompt as well. So I'll just say, in addition to my adopting style, tone, voice, whatever, I'll say, in this case, I want to emphasize, or I thought what was most interesting about this was, try to bring to the fore, whatever. And it can follow those instructions reasonably well. I still don't typically feel like it quite hits the heart of the matter a lot of the time. Obviously that's a pretty subtle thing, right? So it's amazing that they've come this far. I'm certainly not taking it for granted, but yeah, that's maybe the best I can describe impromptu where it's not quite hitting for me.

It might be that you might have to stay persistent in context engineering. Like, hey, what are the examples where it didn't get right or it lacked kind of things or gotchas and so forth? And maybe there's a better way to express that in rules and things like that. But in a use case like this, I would argue that it is a lot better for you to actually continue to context engineer and try different base models, perhaps try different ways to acquire new information or knowledge with the tools and so forth to get a transcript, get, you know, maybe background of the people and things like that. And the reason is that every couple of weeks you're gonna see new models and basic capabilities emerge. Including now that we are seeing assistants emerge, like where they have memory as a feature. Things that you might have liked, not liked in that instance, it's been able to capture automatically as a, in the context in a way. And, you know, maybe things are not great right now, but it will continue to get better as the base models improve. Versus let's say if you were to like architecturally, like fine-tune something right now, you're basically freezing an investment. You have to do it right now and you have to continue doing that as the base models get better and better. And one thing for sure is that the base models are improving very fast across the reasoning and this empathetic aspects of things. Even though we saw that with OpenAI 4.5 model, it is remarkable in many aspects that are not necessarily reasoning. And so maybe that is the element that is missing, that creativity and that kind of like sheer touch of humanness and so forth. And so I think it is a macro question, like, do you really want it in this? This is a very creative domain. Like, I think what makes your, this thing so unique is that you are editing, you're editor-in-chief in this particular workflow. And maybe you actually do want to stay editor-in-chief and you want to have just like, you know, set up correctly from the models to help you do the best work for things that makes you very unique about it. And in these cases, maybe the outcome is not to have fully automated situation, but there will be like, there are cases where completely different domains where it's just mundane work and you just want to rather have it fully solved by something In those scenarios, maybe fine-tuning might be a better kind of a technique to do that, solving that problem.

Yeah, that's interesting. I'm not looking forward to being replaced by AI, but I at least want to know when it's happening. I do expect NotebookLM and similar to compete pretty effectively with me in the not too distant future, if not already, and in an asymmetric way. I do recognize that there's hopefully an appeal of a personality that you get to know over time and Watch a person sort of evolve. At the same time, NotebookLM can take any topic that you want to have a podcast on at the drop of a hat, right? And I certainly can't match that. It'll be interesting to see which relative strengths and weaknesses win out. I often also think too, and I have a friend who really emphasises this point all the time, if I just read the Claude output, I'm not sure anybody would really notice all that much difference or think it's any worse. So I think there's, to some degree, I am really just precious, you know, like I have a sense of what is me and what is me matters to me. And it matters because I'm reading it and I'm putting my name on it and I want it to be genuine to me. But that's a bit of a different question than like, what is quality? What serves the audience best? And I can confidently say that when I edit Claude's output to make my own final thing, that it is more true to me. I cannot confidently say actually at this point that it is truly better serving or better informing the audience. I haven't really run that test. I hope it is, but it's not entirely clear. That also might fall into the category of like, do I really want to know? So far I haven't run that experiment, but it would be an interesting experiment to literally just sometimes read the Claude essay and see if anybody comments that the thing was worse. Interesting to just compare and contrast a little bit. frontier developers, what they're doing with data, and then everybody else. And you can sub-segment that if you want. My general sense is like at the frontier, they want as much data as they can get, as many domains as they can get, highest value, highest quality they can probably get. I've heard numbers of $300 an hour being paid out to physicists or biologists around the world to do this sort of stuff. When it comes to that, everybody else layer, which can still be obviously very sophisticated companies. Obviously they have a more narrow focus. I'm interested in what else is different. Do they, for example, primarily use their own team members to do the data work, or are they still interested in going outside? If you're a pharma company, my naive sense would be like, I'd want my own people to do this stuff as opposed to have you go source talent around the world. But maybe that's wrong. And then also interested in kind of the scale. Do they have their own people do it? Do they still want to go outside? How big of a dataset do you need for something like that?

How do you know?

Are there like rules of thumb or guidance that you can give people? How much does that kind of cost? In a way, this could be like your intro sales pitch or like orientation of a new customer. But yeah, I want to get the lay of the land both at that top tier, but I think I understand that better. And then especially at that kind of everybody else tier.

Yeah. So in Frontier AI Labs, we work almost with everybody at this point. And there is, I would say, like an insatiable appetite to get access to new novel specialized datasets that are in the regime of kind of reinforcement learning and is helping them teach models longer horizon tasks, do that reliably, do that across the board, across a variety of domains, right? Not just including mathematics or coding, but rather all the other things as well. So that is certainly the case. It's actually getting more in demand over time. And so the budgets across the frontier AI labs on data is actually increasing. To give you a sense, like each of the frontier AI labs is probably spending over a billion dollars a year on data. And so that is like fundamentally increasing across the board. And these frontier AI labs, like they're trying all these different ways to produce these datasets. I think some teams try to hire contractors directly. And so there's an emergence of these staffing agencies that we are seeing in the world. They're all, all they do is just simply help them hire domain experts very quickly in bulk. And so staffing agency models have become very exciting for investors nowadays in Silicon Valley, just because like now they're actually doing things in the AI space, but more often than not, The AI labs do ultimately require or want really high-quality data delivered to them very quickly. And to produce the data, you have to operate a data factory, as they say. So LabelBox is essentially a data factory. We are fully verticalized, so we have a very vast network of domain experts. And we use awesome ways to actually screen and vet these experts, but then That is just one part of the story. You have to actually go build tools and technology to then produce these datasets. And actually, like today, the state of the art is that most of the datasets are essentially hybrid datasets where we have to use all of these amazing techniques with AI systems and synthetic dataset approaches to infuse it with the human experts to produce these novel datasets. We looked at our network and looked at the kind of earnings across the board. Guess how— what is a yearly earning for a top quartile or like a top AI trainer? What would you guess what they're making per year?

Am I assuming they're working a full-time 2,000 hours?

You can assume that.

I mean, the $300 an hour that I quoted seems high, but $100 an hour doesn't seem High. So maybe I'll say if they're doing a full 2,000, that would be $200,000. And then I'll discount because maybe they can't get that many hours. I'll put it, um, put them at $140,000.

Yeah, that's a good guess. And so our top contributors are earning well north of $250,000 a year. And that, you know, there's a power law. And so you kind of see it sort of asymptote in around maybe $40,000, $50,000 a year for other domains in other countries and so forth. But the best, most highly specialized individuals are are earning over $250,000 a year. And we actually expect it to increase as the AI frontier goes deeper into agents and business workflows and so forth. And this is like pretty amazing to see. Just 5 years ago, the average hourly rates for kind of the data stats was much lower than that. We were operating in a very different regime and domain at the time. And now you have essentially These domain experts who are using maybe part-time, they're working on these kind of AI training tasks, and some of them are really changing their lifestyle and to be able to do this basically full-time and have the freedom to do work whenever they want and so forth. So that is the state of the play with human data in frontier, in expanding the frontier. And these are extremely complex tasks, and they're not necessarily anymore tasks that a person would do for 5 minutes or so. These individuals are actually creating RL environments. And so these are endeavors to develop solvers and verifiers for these environments and/or play in these environments to produce an instance of what a good activity would look like and so forth. Now, we actually also have a very interesting, very big business in the enterprise space. So we are We're one of the leaders in building these software tools and platform for producing the training data. And so we have our customers, like some of the best or biggest pharmaceutical companies or robotic medical imaging companies and so forth. And so in those cases, they are developing models where they do have to work with their own experts. So medical coding is a really great example. It's something I think will be solved with AI. Excited to see, you know, their companies doing that. And in these scenarios, you have to have individuals who are really good experts in that domain. All right. And it is such a nuanced situation where like the insurance codes, medical codes, particularly in the United States, it's an industry in its own. And so you have to tap into those individuals from that industry or train new experts to do that task really well. And I'm sure companies that are starting to do really well and build these AI systems, they are finding all these leverage points. They understand that nuance of how it's done and that they're probably trying to train a larger workforce to actually go do that in a kind of factory sense. And in these scenarios, these teams have to develop the entire data factory themselves, if you will. They have to build the technology and the tools. Then they have to figure out who these experts are going to be, how they're going to work in the factory to produce these datasets. So LabelBox is operating a kind of large-scale data factory, primarily for frontier labs and for enterprises. What we do is we actually offer them a technology platform to do it themselves. And in those scenarios, they are using their domain experts to do it, you know, run the data factory themselves.

Do you have to give them a sort of Palantir-style forward-deployed expert to help them develop the process? I assume that they need that help in most cases.

In some cases, yes. In some other cases, no. It really is a fact-based— it really depends on sophistication of customers. So if you're, let's say, super high-tech digital startup, AI startup, application startup, perhaps you want to have full control and architect and operate the system yourself. But maybe in some other industry, you're getting more help from our company.

Yeah, that makes sense. One other kind of random market segment. I don't know if this really exists or not, but I just did an episode not long ago on the concept of sovereign AI, and there's a lot of different facets of what that might mean, whatever. But one big one is countries around the world with different languages, different cultures, different value systems, whatever they might want to make an investment in whatever they can make an investment in. It's not always entirely clear what they should do to get models to be more fluent in their language, more familiar with their values, respond in more culturally sensitive and appropriate ways, just to have more local knowledge, right? Whatever the— probably all the dimensions at once they'd like to improve. I wonder if there's any business there for you. And whether there is or isn't, a theory that I've had in the past is that if you are Brazil, for example, or you are maybe even India, although India is big enough that maybe they want to have their own national champion. But I've said maybe what you should do by default is just make an investment in data collection and then just go give that data to the frontier developers and be like, hey, You can be better in Brazil. We've done the data collection work as the Brazilian government. Here's a big data dump, please use it, and then you'll be better for us in Brazil. And that way they don't have to worry about, oh my God, what are we going to do about data centres? And what are we going to do about frontier researchers? And how are we going to afford the $100 million to compete with what Zuckerberg is reportedly offering people? Although I'm a little sceptical of that number, except in the most rare cases. But yeah, any, is there any like sovereign AI demand coming your way? And if you were going to advise the president of Brazil or Mexico, or for that matter, India, what would you tell them they should do on the data front?

Yeah, so I believe there is tremendous opportunity in the government to essentially, I think what would look like a leapfrog or overhaul of entire ways of doing things to this new way of making it super efficient. And I think the different countries have a different sort of system. Like in some countries, they, the government, our government is providing certain services and they are actually owning the full stack of those services themselves. And in some countries it's more privatized model where they're essentially creating the environment and creating a spec. Hey, we want private companies to be able to do XYZ things. And so in scenarios where like some governments are developing Or have been in the business of offering services to their citizens, which I think is most of the governments really, right? So some shape or form, there is a lot of opportunity to rethink like what would be an AI data experience here? And it probably across by all dimensions would mean they would be able to render those services a much better experience at a fraction of cost and get out of that legacy system. And finally, we have a technology that allows them to do that. I mean, many of these governments are probably running on legacy Fortran and COBOL and things like that, which I believe, like I was learning somewhere from Doge team, that they uncovered this massive amount of things that are just run on legacy systems and so forth. So there's tremendous opportunities, and I think there's an opportunity for rethinking experiences in a pure digital native AI-first manner. And think about India where so many citizens could benefit from basic knowledge understanding of, okay, how does this thing work? What are the amenities? What are the kind of services available to me and all these different towns and villages and so forth? So there is that. Now, the investments really, I would argue that the government should be focusing on the outcomes, like what the experiences they want, and then mapping it back to how do they achieve that in the most efficient way. And whether it's letting private companies to develop those solutions and implement them. And in some other areas where they're based on a particular use case, they're looking at like, how do we re-architect or develop those services? And in many cases, it will become a question of like, how do you get those 3 components together, compute, data, and talent to go build those capabilities? And you would need certainly 3 of them to making that effective.

So no shortcuts, in other words, like you don't think they could just come to you and say, hey, we want to go like my silver data on a silver platter for the frontier developers. Where does that break down in your mind?

There are obviously many industries and many players in our market. They would say like, no, no, let's go invest in the fundamentals and we'll figure out the use cases later, but you're going to have to get your data. in the right place and things like that. And I am not so sure that actually helps these teams achieve the outcomes or like the entities to achieve the outcome. Rather, let's walk backwards because now finally we have a technology that can be wielded and molded to achieving the goal. And actually the hardest part is coming up with that intention and the goal and the kind of outlines of what a great experience or service would look like. And funding that. And then once you have that clarity, all the other things are actually fairly tractable. Like, okay, we can help them produce the right datasets. We can help them build the end-to-end systems and evaluate those things and so forth. And I think more often than not, like projects fail because they didn't have that end kind of goal in sight, rather than they were like going into operating in a world where they were acquiring a whole bunch of set of technologies, seeing what it can do. Rather than walking backwards from the customer experience.

Yeah, I think that's a great reminder and something I also, in my very ad hoc, you know, AI consulting for businesses, always emphasize. What problem are you trying to solve? Because there's so many amazing new technologies that you can experiment with and get lost in and have tons of fun with and spend a lot of money on. And yet, if you don't have a clear idea of what it is you are trying to accomplish, it's probably not going to go super well for you. Sometimes people are like, we need to build a platform first and then we can build application. We need to build our AI platform. And I'm often like, I would do a spike. Let's get one thing working first. We'll learn a lot of lessons that way. We'll learn what data we do and don't have. There's all sorts of problems that we can identify in the course of one spike. Probably all that gets thrown away. And then at some point, maybe you mature into a more robust platform. But if you try to build that without specific problems in mind, I think you do set yourself up for a lot of frustration. So I totally agree with that analysis. I do give the frontier developers a lot of credit because I think they have taken a number of steps to close these gaps over time on their own. One really notable one was when GPT-4o dropped, the tokenization of non-Roman alphabet languages got dramatically better, which brought cost down, speed, a lot more To fit in context windows. So they've like definitely prioritized this to a degree, but there's still a way in which it feels like if you're in India or you're Brazil, whatever, you are a second tier user because just the things are so English and so American centric in many ways that if you could just do a giant data dump, maybe first of all, I imagine the frontier developers would take it and then it might lead to more parity between the American experience and the sort of Brazilian or the Indian experience of using the models off the shelf. But yeah, who knows?

I don't know. I think it's really interesting because I think it'll be really interesting to see how it all emerges. So before Labelbox, I worked at Planet Labs. It's a company that is now public and it has, it operates about 400+ satellites in low Earth orbit and scans the entire Earth every day. And I was involved in developing capabilities where we would use computer vision to extract insights from this dataset. So for example, The government of Brazil would be very interested in understanding what is the level of deforestation that they're seeing every year. And back then, there was simply no other alternative way to really understand that on a daily or monthly or weekly basis. But as a company, we could develop that insight because we would scan the entire Earth every day. We'd know exactly, we would basically get an image of the entire Brazil every day at a certain resolution. And we could apply these algorithms and essentially feed that as an insight to the government and to make the right decisions, policy decisions. And that's an example where, you know, there's only certain companies who can do that and share that insight. And it wouldn't make sense for Brazil to actually operate foreign satellites to just do solving that problem. And so there's categories of capabilities or use cases where you have to get the private companies that are best in what they do, bring those capabilities, and then work on integration. With other systems to make whatever policy decisions or services that it gives to citizens. And there's some cases where the governments are very uniquely positioned to curate or cater an intended experience to their citizens. I mean, what could be an example? Maybe experience with tax could be an interesting one. Like, hey, can the governments offer a much more intuitive way To file taxes. Or maybe there's some other, you know, for countries where there is a centralized health service, can it make it super AI-driven where citizens can query and understand or get basic healthcare services conversationally perhaps, and so forth? And so it really becomes case by case. What are the capabilities that the government can invest in and completely rethink? And in some, in some other areas where it will have to really rely on a private industry to offer a bunch of these solutions. I do think that there, we seem to be going in a direction where there are going to be these foundation model companies or models that are developed privately perhaps, or maybe the open source versions, and they will be used as a base model in the sovereign context. And they're essentially then going to either develop applications on top of it that is very uniquely for the sovereign use cases, or In some cases, they will have to fine-tune it or develop custom models on top of it because government has the scale and intent to be able to produce some specialized datasets or gather some specialized datasets that companies might not be able to do so.

Different topic. Why has nobody offered to pay me to install software on my computer to watch me use my computer all day? It seems like one of the obvious relative deficiencies right now of the frontier models, and OpenAI's operator is getting decent, certainly. But you go watch the AI Village and you've got a lot of stumbling around in various nooks and crannies of UIs. It just seems like this race for data acquisition has not extended to let's just go Watch people use their computers, record that, and then bring that kind of behavioral data into the fold. And I've never heard an answer to this question that has satisfied me. So what's your take on why nobody's made me that offer?

It's coming. And I think the reason I have a very clear visibility into this kind of all— we power a lot of these computer use agents right now. Labs are building them. So right now, I think we are seeing first versions of truly useful computer use kind of agents or computer use capabilities, where it's just really the first innings of it, where kind of like from a movie, perhaps like inspiration from Her, where these systems are becoming an AI companion. They can listen to you. They can observe what you're seeing every day from maybe your camera on your phone. But then if you're doing work on a computer, you could just turn that on and it will know everything you're seeing on the screen. And I would say that these models are very poor in that capability right now. So if you were to turn on, let's say, a live AI model that can understand a lot of these kind of things on your screen and so forth, the understanding piece is really good, particularly some areas of UI or text and so forth. But when you're really doing some interesting or like domain work, they often fail. And they will fail not necessarily like in a very unintuitive way. It's like they will fail maybe 3 minutes into the kind of session where maybe the model did not understand or had the kind of memory of what was said or what the intended goals were in the first minute. And so there is failures that are happening across the board, whether it's understanding of a screen at a time, what is it? Like, for example, most models don't really get a lot of these shapes of geometry and so forth. As an example, have a very poor spatial understanding. And they also fail in the time horizon. And as part of the rollout of these AI systems, I would be not surprised at all that the freemium kind of model would be that they will be able to use the sessions and the datasets to then perhaps train and improve the models. But again, there's a lot of work happening through companies like LabelBox where we are actually producing these sessions and these datasets at a very large scale. Across different languages and domain expertise. And I guess in a way you can say like the capabilities are being cooked. It's not rolled out yet, but as part of the product experiences, the companies are going to be able to take the data from the users and actually improve these systems and make it more reliable. And I'm sure like they will make a choice on what privacy, and some users will want to intentionally allow their sessions to be used by the companies to improve their models.

Do you think I have any expertise that would allow me to be a data contributor via LabelBox? I wouldn't aim for 2,000 hours a year, but I wonder if I even have, because I'm kind of a highly generalist jack of all trades sort of person. Is that profile even useful anymore? Absolutely.

Like the way I would, there's a lot of talk in human data where we take examples of doctors, lawyers, that kind of thing. I actually think it's not that actually helpful. It's helpful to just maybe like for new audience to really understand like we're looking for this domain expertise. But actually the way we think about it internally at AbleBox is we are simply looking for individuals who have like high agency and IQ across the board. And that fundamentally correlates with being able to do very generalized tasks even in certain domains. So you would be able to learn completely new tasks, maybe even coding, or maybe in some research where you're able to do long horizon tasks far better than any of these models can do right now. And that means that you can actually produce the training data or signal for these models to learn from. So for sure, absolutely. I think I'll let you in.

All right. I'm looking forward to experiencing the AI interview as well, because I've heard about a lot of those things getting at least experimented with. It sounds like you've scaled yours. And actually, while we're on that, any lessons from the AI interviewing at scale that you would highlight for people?

Yeah, so we probably are, I don't know about the others, we believe we are one of the largest rolled out production AI interview right now. We are conducting well north of 2,000 sessions or interviews a day. Right now, probably 50 people are interviewing right now with our Zahra AI interview. And yeah, there's a lot of lessons. So first of all, we were very surprised how much people loved it or enjoyed interacting with an AI about their experience and so forth. And in many ways, when you look at the average rating, when we ask our interviewers, hey, how did the interview go? They are like 4.6, 4.7 average satisfaction score across the board. And that is because they can do the interview. at their time, whenever they feel they're in the moment to have that conversation. And B, it's like incredibly patient in a sense. You're able to just dive into a variety of topics and so forth. And in many ways, it asks you— if you, let's say, have a PhD experience, it will ask you about your research paper, about a very nuanced thing that a human recruiter may not be able to even get to that detail in a first 30-minute session with you, right? And so people are just like, love about sharing their experiences and especially if they can share aspects of the greatness of their work in that fidelity in a very condensed manner. So that's one. Then the people are using it to practice their real-world skills. Like, so we just rolled out like practice sessions of our AI interviews where simply people are Using it as a way to do interview prep, perhaps for other things. And we rolled out with an intention like, hey, our goal is not to go into that sort of category or market. Our contributors asked for, hey, this is so great. I want to be able to just do practice runs for other things in my real world, jobs and so forth. And so that's that. I also find there's really clever ways people try to cheat the system. I've certainly seen people literally just put an iPhone with a ChatGPT Advanced voice and have AIs talk to each other and things like that. And so we just see this human ingenuity to gaming a system and so forth. So that's just, I take it in a positive way. So we have to understand, we have to build technology and systems to stay ahead of that, right? So if we have that truly-like system, and that means we might not, we have to be really good at assessment and finding really solid, good players with right intent. And that is always a pursuit because human ingenuity is just so vast and you just get surprised by ways people employ all these different tools and techniques.

Yeah. Interesting. I'm looking forward to seeing what that experience is like and how helpful the AI cheating assistant is.

Another thing I got very surprised about actually with the AI interviews is just how good it is in not only going deeper into like your context, right? Your resume and experiences or papers and things like that. But how good it is to having conversations in all these different languages, it actually is able to assess how good you are, like how natively fluent you are, let's say, in a particular language and so forth. So we are able to assess all of those things, your different language skills, your domain expert skills. I was honestly, right? Shocked to see how effective these are.

OpenAI voice for that or something else?

We use multiple providers behind the scene. And then we have, as I told you earlier, we have part of our system is fine-tuned capabilities where especially around assessing the conversation and scoring or grading these things. These are all a data engine that we have built. So the more interviews we get, And the more signals we get about how good these human experts are in producing the data, we are using that insight to improve the grading system, if that makes sense. And actually the entire format of interviews and so forth. Cool.

I'm looking forward to checking it out. Maybe going back just to the topic we started with for a second. What is your expectation for the future of the industry? Is this a one-off where Zuck is just doing something crazy, or should we expect this to be like the first domino of what could be a number of kind of frontier developer and data factory company pairings? What would you expect there? And I understand you've got a lot of skin in the game, so I'm not sure exactly how much you want to speculate on that, but—

Right now, one way to think about it is like we are slicing kind of aspects of human intelligence and emulating that in AI. And the question then is like, how many slices are left in the vast human knowledge? And I would argue that there's probably infinite slices out there because we don't really understand how human brain works yet very effectively. And so far we haven't seen systems that work like human brain yet. I think we are getting closer and so forth. And time and time again across multiple paradigm shifts, we have seen from supervised learning I mean, when I started my education, I used to train models with 3 neural networks, 3 neurons and 3 layers in MATLAB and Simulink. And just in a decade or so, we've hopped multiple paradigms, and time and time again, we have come back to data being a very critical, essential ingredient to making these capable, highly capable models. Now the frontier is where we are trying to teach these models very sophisticated knowledge work. And if you look at it, like, they are very capable in certain ways, but they're not so capable at all in actual tasks in knowledge work yet. And the question is, what will it take to go do that? And I think in the RL-informed RL paradigm, we are now at least looking how do we go create or emulate variety of these different domain tasks and in RL, express that in RL environments or datasets and have these models learn. So I think there's a long road ahead in terms of making these AI systems more capable, more integrated in our day-to-day lives, being more reliable in long horizon tasks. And all of this will require data essentially one way or the other. And I think there's some really exciting things that are happening in synthetic data domains. I think that is going to be always a pursuit because of just how, like how much effective or the potential there could be. But then you still have to have some sort of a human grounding judgment to ground the synthetic data, if that makes sense, because these AI systems have to be actually useful and interact with humans at the end of the day. And so we are in very exciting times, but also I believe the data is going to continue to be the primary mode where humans are going to supervise these AI models open-ended in the sense like they're Like, we want to be at the driving seat, and we want to be able to command or manage millions of these AIs to accomplish a task. Managing these millions of AIs to accomplish a task will be imparted or exchanged through some data. And LabelBox is positioned to become a— continue to become a bigger and bigger data infrastructure provider for all of those use cases that are invented right now but are yet to be invented in the future.

That's a perfect place to end. But I have one more question, which is, where does that leave you? This is maybe the hardest one of all, but if you peer into your AI crystal ball, where does that leave you on the question of a plateau around a sort of AGI level versus a prospect of a sort of takeoff from an AGI level into a superintelligence level? And I know all the arguments in all the different directions, but Some of what you've said around, I mean, obviously just positive transfer in general and potentially the sort of human-level coder that can be run in, if it's like as good as the best human-level coders and it can be run in super parallel, when I imagine that, I'm like, geez, once you get to that level, you might take off pretty quick. You might really find all these new architectural innovations at a faster rate than we have been. And the whole thing could get crazy. But then some of the other things that you've said, when I focus my attention on those and I'm like, there's all these million slices and you got to have data for each one, then that leaves me feeling more like maybe we hit a plateau around the best humans, which, by the way, I think I would prefer for safety reasons, for time to adjust. There'll be time for superintelligence, is my view. But what is your expectation? Do you think we'll have any sort of extended period of Kind of human expert level, but not like fast takeoff from there?

So I think in a way both things are actually true. I think we are already in an accelerated takeoff by all means. And we get adjusted with every day to day. Like, hey, why are all these AIs not so great right now yet? It hasn't done my work yet. But 5 years ago you asked me, man, this looks like some crazy capability from a very far future. And so in many ways, right now we are in fast progress. I think the future is going to be very much more synergistic with hundreds of— think about it, there are 30 million developers now or something like that. And I think we're probably onboarding billions of coders over the next few years with this AI assistance. And so that's really interesting and exciting. That has its own properties to it. So there's that. The question, the fundamental question that I think which kind of also tease this out in the philosophy realm is how can these AIs ascertain quality judgment? And there's a lot of things about the world that humans are able to figure out and able to put a quality judgment. And in many, many ways or scenarios, we're not able to necessarily express it, why a certain thing is so good or bad, but we know when we see it. And it's this visceral thought or emotion that appears. from that experience of reality and human mind. And these are the concepts explored very heavily in books like Zen and the Art of Motorcycle Maintenance, which sort of brings a lot of these really interesting ideas about quality and so forth from other philosophers and so forth. I think the question comes down to that, how that quality judgment will be imparted into AIs. And so far, The way we are doing it is with humans teaching it in all these meta ways. These are ways that we're rubrics or grading, things like that. And I'm sure a year, 2 years from now will be very different techniques. But I think that's an open question. I'm very curious about that. I think a lot about that, obviously. And that will be really key for kind of a, I would say, like what you described, the sort of even faster progress. If you somehow can magically Figure that out across vast knowledge of vast space of human knowledge, what makes things excellent, good, bad. And it is gonna be key for making these AI systems really progress very quickly.

Yeah. Taste in, in maybe— that's why it's a domain to watch. This has been excellent. Is there any other closing thought or anything we didn't touch on that you wanted to leave people with?

No, thank you for having me. It was an honor to be here.

My pleasure. That's very kind. So thank you very much as well. Manish Sharma, founder and CEO of Labelbox. Thank you for being part of the Cognitive Revolution.

Thank you.

If you're finding value in the show, we'd appreciate it if you'd take a moment to share it with friends, post online, write a review on Apple Podcasts or Spotify, or just leave us a comment on YouTube. Of course, we always welcome your feedback, guest and topic suggestions, and sponsorship inquiries, either via our website cognitiverevolution.ai, or by DMing me on your favorite social network. The Cognitive Revolution is part of the Turpentine Network, a network of podcasts which is now part of a16z, where experts talk technology, business, economics, geopolitics, culture, and more. We're produced by AI Podcasting. If you're looking for podcast production help for everything from the moment you stop recording to the moment your audience starts listening, check them out and see my endorsement at aipodcast.ing. And thank you to everyone who listens for being part of the cognitive revolution.