Building the "App Store" for Robots: Hugging Face's Thomas Wolf on Physical AI

Thomas Wolf, co-founder and Chief Science Officer of Hugging Face, explains how his company is applying the same community-driven approach that made transformers accessible to everyone to the emerging field of robotics. Thomas discusses LeRobot, Hugging Face's ambitious project to democratize robotics through open-source tools, datasets, and affordable hardware. He shares his vision for turning millions of software developers into roboticists, the challenges of data scarcity in robotics versus language models, and why he believes we're at the same inflection point for physical AI that we were for LLMs just a few years ago. Hosted by: Sonya Huang and Pat Grady, Sequoia Capital

• 0:00Speaker 00

  Many, many startups just already being built on top of Lo-Robot. Just, you know, they want to build something. They have this idea of a manual task they can automate, or they have an idea of something they could do in a physical world. And then they take Lo-Robot. They take already like the basic building blocks we've shipped, which is just a robotic, a very simple robotic arm, SO100, that we designed basically to be the cheapest robotic arm, to be $100. And they're already like trying to start business around this. At the bottom, you know, it's the Hugging Face ethos, which is you bring all this platform, all these basic building blocks for people to build really crazy things on top. And robotics is the same goal for us.

• 0:53Speaker 01

  In this episode, we interviewed Thomas Wolf, co-founder and chief science officer of Hugging Face, the largest open source community for AI. Thomas has consistently predicted the future, and he's responsible for why Hugging Face invested heavily behind transformers and language models, enabling the LLM wave a handful of years ago. Now he sees the same opportunity for robotics and physical AI and is helping to shepherd Hugging Face's Lo-Robot project, which brings together policy models, data sets, and physical embodiments to help developers everywhere become roboticists across a diversity of use cases and form factors. We're excited to chat with Thomas about the robotics explosion, the bottlenecks ahead for physical AI, the US versus China in the open model race, and a lot more. Enjoy the show.

• 1:38Speaker 01

  Thomas, in your role at Hugging Face, you helped Hugging Face invest behind Moonshots as the chief science officer, one or two years ahead of the field. You mentioned to me last time we chatted that, you know, you have a spidey sense that we're at the same moment for robotics today that we were for transformers and language models a handful of years ago. Tell us what you're seeing.

• 1:57Speaker 00

  Yeah, I think honestly it started two years ago. I mean, we started our activities in robotics eighteen months ago. And I think at this time, there was a couple of breakthroughs out of, I mean, these labs, you know, were right. Stanford and basically these teams were starting to show robots that were able to tie notes, fold clothes, you know, cook, like throw things in the air on a pan and grab them. Basically, all of these things with, in one way, very little data, but also with also good perspective to be able to leverage some of the world models and things like that that we see, you know, really benefit from from internet side data. So all of this kind of pointed to a short future where robotics were gonna work in a way.

• 2:40Speaker 00

  Like hardware was already there and in my opinion, it's been there for quite some time. But the missing brick was really software that could adapt, that could be dynamic, all of that. And we started to see that. And that's why we started to work on LeRobot, yeah, a little bit more than than eighteen months ago. And I think for us, the success of LeRobot, like seeing like this huge community. For us, the big bet was can you build a big community in robotics as well? There were small community of kind of hobbies or like, or people very seriously building robots for like, you know, factory lines. But it was kind of like a tiny vertical, in my opinion. The question was, could you move this tiny vertical to like a full horizontal thing? Just like nowadays, every software developer is kind of an AI researcher almost. They all want to know how, you know, LLM works, how you train them.

• 3:25Speaker 00

  And there's this very smooth transition where the hundred, two hundred million software developer are becoming kind of AI aware. And I think there is a potential future transition where all of these people also become roboticists in a way, if you give them the tools. And so that was kind of our goal. So we started with the software library and the success of that kind of brought us to also try to go in hardware, which is a which is a big question as well. And that's why we acquired first hardware company this year, Polyne Robotics. And we shipped our, at least we opened the orders for our first robots. That was mid-July, so that's one month ago, really. And this went crazy.

• 4:00Speaker 01

  Can you tell us what LeRobot is?

• 4:02Speaker 00

  Of course, yeah. So LoRober is our attempt of, you know, doing again the success of Transformer, reproducing the success of Transformers' library in the robotic field. So this idea is having kind of a central library that everyone would use and that would bring in a very simple and accessible and easy way all the latest technology, all the latest algorithms that people use to train robots efficiently, all the latest data set that they use to train this, and also connect this to actuators, which is the hardware part of robotics. And that's this intersection and this mix of three aspects, the policy, the models, the data set, and the hardware, which we try to combine in LoRober.

• 4:45Speaker 02

  How does the role of Hugging Face change in robotics, you know, for people building in the physical world? Does Hugging Face play the same role or a different role than Hugging Face has played for people building in the digital world with LLMs?

• 4:57Speaker 00

  I mean, our goal is to play the same role, which at a very, very high level is building communities and bringing people, you know, in this idea that AI can be open source and it's not something you only consume, but it's something you can tweak, you can train, you can control, you can host where you want. And actually, hosting where you want is even more important in robotics, because in a future where you have robots everywhere, you kind of want a lot of these models to run locally, because if your robots lose connection to the wifi or something and then run in the wall, or maybe, I don't know, run in your kids, it's going to be much more dramatic than, you know, just an LLM hallucinating. So safety question in robotics, I think, are a good reason you may want to really be able to not, you know, depend on the distant API,

• 5:41Speaker 00

  but have the models as possible to the hardware. So I would say our role is maybe even more important for safety and future of robotics than it is in LLM.

• 5:52Speaker 01

  Could you say a word on the size of the community you have at LoRober? How many people are building? How many people are contributing data sets and things like that?

• 5:58Speaker 00

  Yeah, for sure. Should have checked actually the latest number because it's exponentially growing. So it's several thousand people, I would say, six to ten thousand. One event we did, you know, a couple of months ago, we did a Hackathon, which was worldwide. We had a hundred location in over six continents. So it's still, it's still not in a million of people for sure, but it's way above several thousand people. And the main indicator for us is we can measure the number of data sets, for instance, on the hub. What we see in all of these things, number of community members or data set, we see kind of this exponential growth, which is, I think, a very good indication that we're on the right track. And you have to keep in mind that the hardware that's available right now is still very much kind of a hobbyist hardware.

• 6:42Speaker 00

  So it's like 3D printed arms. It's like they're still wired everywhere. And that's why starting this summer, we wanted to bring much more mass market hardware, I would say. So something that would, you know, interest not only the hacker and the people who are used to plug cables everywhere, but also everyone in families, something that looked much more polished.

• 7:03Speaker 01

  What's the persona of developers in your low-robot community? And I'm curious how it's the same or different from the people that have traditionally been building, you know, like controls, classical-based systems?

• 7:15Speaker 00

  I would say there is three types of persona. I mean, one is the traditional roboticist. They definitely want to use AI. So for a lot of them, they know how to build hardware. They know what they can use. But they've been frustrated by, you know, the limitation of the software stack, all the optimal control models. And all these are very limiting what you can do. So all of these are people who have really happily joined the bandwagon. And we see many, we see the same effect we saw in transformers, which is many academic labs starting to use low-robots because it's a very nice entry point for all the students. And so this has been growing very strongly. The second community is much more interesting, in my opinion. It's people who were not really into robotics, but because they're into AI

• 7:58Speaker 00

  and robotics looks like a physical manifestation of AI, they kind of want to go into robotics. And so these people that covers, I would say, software developer, but even people who are just interested in robotics. So. And a good example, talking here, it seems like a lot of investors have actually bought, you know, SO100 arm just to, you know, to try to understand physically what is this robotic thing, what can it do? And because it seems so accessible, you get the arm and the software is just a Python code. And now, you know, with a little bit of vibe coding, you can actually even even even tweak it or control it quite easily. You know, we see people who who, I mean, maybe are not, you know, purely technical, but who want to understand what's happening in robotics and they use this entry point, which is low robots.

• 8:38Speaker 01

  So you can vibe code a robot?

• 8:39Speaker 00

  Yeah, that's really my goal. Yeah. So you can already do that a little bit. But for the new robots, Richi Mini, I definitely want this to be, you know, one of the easiest way to use. I would love my kids to be able to vibe code behavior on the robots.

• 8:51Speaker 02

  What sort of phase of maturity do you think we're in for the robotics market writ large? You know, like, when when will we have a chat GPT moment in the world of robotics?

• 9:00Speaker 00

  Yeah, that's that's the thing I'm looking for, or like sometimes call it also the iPhone moment. Maybe what will be the first, you schedule first, you know, the first moment where everyone or like a large fraction of population will will think I want a robot in term of consumer. I mean, I I think the enterprise market i- is quite complex. There is in some place, there is already a lot of robots in some kind of industry, right? The car, car manufacturer is a basic example. Then there is this second part, you know, where there is like a a a an entry of robots. And and here there is a lot of challenge around reliability. Will this will these robots be reliable enough to be deployed in retail store? And basically, you know, be be really useful.

• 9:39Speaker 00

  And but the third part I'm much more interested in is actually entertainment, fun, demo, education, where I think maybe this question around, you know, I need a $3,000 robots because I need reliability are less pregnant. And so you can take a robot that's really accessible. So Richi Mini, for instance, is priced at $300. That's something that can definitely be like an impulsive buy. You buy it as a as a gift and you're not sure if it's going to work or not. But for this price, what we want to find is, is there not a lot of, you know, potential in more like entertainment, fun, education, learning AIs through a physical interaction stuff instead of just?

• 10:18Speaker 00

  Coding, you know, on the, on the, on the chatbots or coding on the, on the screen. And I think that's something that has not been explored at all. I think there was a couple of tried. I mean, the mid media lab, Cynthia did GIBO, for instance, but in the past, they were usually priced a bit high, I think above a thousand. And more importantly, I think the, the software that was there was very limited. So you would buy a robot. That would be fun, but you, you had maybe five or 10 behaviors. And once you've tried them all, that's it. It's finished. And here the goal for Ritchie Mini is really to make kind of almost like a smartphone. So it comes. You have a couple of behavior, but just because you can tweak it and people can build new behavior and share them and plug all the new.

• 11:02Speaker 00

  V L N, speech models, chat model, actually the possibility are kind of endless. It's kind of a. Open door on like rebuilding, you know, the app store of iPhone, basically. So that's what I'm very excited. And to be honest, this last part is still very much a bet because nothing exists there. Like there is no like real proof. My just hints are all this exponential growth of community, which make it quite plausible.

• 11:26Speaker 01

  So you see Ritchie Mini as like the, you know, the reincarnation of the, the robot dog, uh, dogs of the nineties. It was like how people can actually play and experiment and, you know, have like these robot companions in households.

• 11:41Speaker 00

  I mean, to be fair, this one is a big bet, but something I was discussing just yesterday, actually on robotics, I, I take barbecue. Someone was telling me as, as an investor, you know, how just is so many, many startups. Just already being built on top of a robot, just, you know, people who are like. They want to build something. They have this idea of, of a manual test they can automate, or they have an idea of something they could do in a physical world. And then they, they take the robot, they take already like the basic building blocks with shift, which is just a robotic, a very simple robotic arm. So one hundred that we designed basically to be the cheapest robotic arm to be one hundred dollar. And they're already like trying to start business around this to start like.

• 12:21Speaker 00

  To sell something around this. And Ritchie Mini is also in a way designed for that. It's a very wide, simple robot. Kind of a white labeled thing. And if you want to adapt it and if you think, hey, I have a business idea around this, but I need a robot to interact with people, I don't know, in hospital or something like that, you can take this and you can actively start to build your idea. And that's at the bottom, you know, it's the Hugging Face ethos, which is you bring all this platform, all these basic building blocks for people to build really crazy things on top. And robotics is the same goal for us.

• 12:51Speaker 01

  Really cool. I'm going to talk about data as a bottleneck. Like, I think one of the big differences between language and robotics is you have trillions of tokens out in the public internet to train O L Ms. That dynamic doesn't exist in robotics. And actually, I think that's where Hugging Face's role in the ecosystem could be much more interesting in terms of decentralized data set curation, creation. Talk about what's happening on the on the data set side of the robot.

• 13:17Speaker 00

  Yeah, it's super interesting, I think. So, I mean, there's a couple of challenges in robotics. And I mean, the general, the main challenge is in the data. There is just not enough data. There is some ways to use video on the internet as training data, but it's very limited. And in some way, we may be able to use models, but in some other, if you want to automate a task, there is no way around just recording someone or the robot possibly just doing the task. I think here, there is, there is one possibility and one limitation. I mean, the main limitation is you can record a lot of tasks yourself, but usually what you will lack a lot is the diversity. So you will basically be able to train a robot to do something very well in your room when everything looks the same.

• 14:02Speaker 00

  But once you put it in the next door room, where maybe the walls are green instead of red, the robot has a lot of troubles to generalize. So this is, this is the main limitation. And so our idea with the hub was that everyone could record data sets. And if we manage to incentivize them to share the data, then we could maybe build a very like multi-colored data sets that would like multi, multi-location data set that would be extremely diverse, you know? And in addition, hopefully would be also very big. So I would say that's a long-term goal. And we hope this can help. But another more direct things we try to do. Is to work also directly with the actors of the community. So we release a couple of data set to try to help them releasing some data set.

• 14:47Speaker 00

  We think in robotics, one of the nice aspects is a lot of people in the end want to sell the hardware. And so they can actually afford to even more than L L M. They can afford to share a bit of the software as open source if it brings all the field above. Because in the end, that's not really directly what they sell. So that that's kind of what I'm trying to convince a lot of robotics company to do. And surprisingly, it's actually something a lot of them seems to be interested in.

• 15:13Speaker 01

  Super interesting. You tweeted about world models today. I think you and I met the same world model founder. What's happening in the world model open source? And how does that help or not help what's going to happen in robotics?

• 15:24Speaker 02

  And can I ask you, is there a why now in world models at the moment? Because it feels like they've started to pop up recently.

• 15:32Speaker 00

  So what is interesting is it's it feel like it's a couple of teams who have been actually working on that independently for for few months. And they just happen to release this right now, right? Because when you talk about all of them, they are not really copying each other. I mean, I guess one thing was the advent of really cool, really good image generation. And basically finally understanding how to fix this six finger things. And basically just get a more reliable and more coherent world model for for image, which naturally was transposed to to video. And so we see now some really cool video models as well. And this one is just is just one next step. And a lot of the founders I've talked to in this field also say that they were helped by the advance of like open source video model generation

• 16:18Speaker 00

  and open source image generation. And basically they take this video generation model and then they fine tune them. And then they train them to be able to react to some inputs, which is also what we do in robotics. There's a lot of common points between these two things. And and it seems to work quite well. So you start to have this kind of very interesting, in my opinion, totally new experience where you actually have a film that's controllable, that's like photo, both photorealistic and also reacting.

• 16:46Speaker 00

  In a very coherent way to the action you will input, which is either just moving around or just asking it to add something, you know, add a rider, a castle, a car driving. And you see this thing that just reacts very well. And you have here, I think, a lot of potential application, you know, both obviously in entertainment, actually some form of entertainment that might be just totally new, something we've never seen, which is maybe the first time we create a new form of really virtual entertainment. But also a lot of application in business and how you can have interactive thing. And one of these downstream application is generating more data for robots. I mean, there is just two way to generate data. One is to record it in the real world, which I think is still very interesting. And the other is to simulate it.

• 17:32Speaker 00

  And surprisingly, on the simulation, we have not seen a lot of, you know, I mean, there was some development, but it's not like there have been some really breakthrough recently on simulation. So maybe this is the first breakthrough I've seen on simulated generated data in quite some time.

• 17:47Speaker 01

  Yeah, I was very excited to see some of the, even like what DeepMind is doing with Genie to train their embodied robots. Super exciting. Humanoids. Do you believe in humanoids as the kind of ultimate form factor?

• 17:58Speaker 00

  Yeah, big debates, big debates. What is sure is I'm quite more excited about trying other form factor right now. The main, I mean, the main, the main problem with humanity, I think there is two main problems. The first one is it's always quite expensive, just because you need a lot of motors and all the price in a robot is just the actuator. That's, that's, that's always like 70% of the price tag. And so when you have 60 actuators, that's just your bill this year. And so it's really hard to drive humanoids below the price of a car. I think the price of a car is still already quite a high requirement. If you, if you buy something that's the price of a car, you do expect to have a lot of value out of it, right? And so, and so that's why we're exploring like smaller, smaller robot, like just one arm or just a moving head and this type of thing.

• 18:42Speaker 00

  There is some possibility that we can get cheaper humanoid at some point. And K scale was trying to do, Unitree has been trying to, to cut the price. And there's a lot of, a lot of company tried to aim, but it's going to be really hard, I think, to, to get this under like 10K, $10,000, something like that. The nice thing about the humanoid, of course, is once you've solved the humanoid, you solve like a lot of tasks at the same time. So if you solve the humanoid, you can you can do everything a human do, which is very exciting. And the main question is, do you need to solve the humanoid? So on my side, I'm more like, I would like to see, you know, a galaxy of different form factors. I also think some of them are much more cute than a humanoid. I think for social adoption, I think the humanoid is also asking a lot from people, right?

• 19:25Speaker 00

  It's like this, you're directly in this kind of uncanny valley with something that looks a lot like you, moves a lot like you. So I thought this would be a big limit for for social adoption. Now, to be honest, I've seen a lot of unitary robots. I don't know about you, but you kind of ignore them at some point. So I'm also much more confident that people will just say, yeah, that's just, maybe we're too too worried about the uncanny valley in robotics. And maybe at some point, once we start to have seen like a a couple of robots, people will just accept them very, very easily.

• 19:55Speaker 01

  OK, so we're gonna see the the robot humanoid soon?

• 19:59Speaker 00

  I mean, the goal would be if we achieve many and our small robots work pretty well, that at some point we will climb back to making the humanoid form factor, I would say, kind of progressively, as we've done, bringing the community along with us.

• 20:10Speaker 01

  If, as you imagine the world in 10 years, like, how many robots do you think there are among us? Do you think it's like 80% of them are humanoids? And then 20% are this long tail of this diversity of hardware and use cases? Or like, how do you think the world plays out?

• 20:25Speaker 00

  Yeah, and I would love to see the second option, because I think that's an option where we have much more robots in our life. What I what I would really not be super excited about is a future where kind of robots are kind of an elite thing, because they cost 100,000. And so basically, if you're rich, you have you have three robots at home. And if you're not, you don't. I mean, Hugging Face has always always been also about the big community, so we care about that. And so for this reason, I'm much more excited to see a lot of form factors that are basically accessible to a lot of people. Some of them are cheaper, some of them are more expensive than just this single humanoid, you know, that costs a lot. And if you can buy it, that's nice. And if you cannot, too bad for you. So I would say at Hugging Face, that's the future we try to manage to push to.

• 21:08Speaker 00

  I think also it's much more fun, because you you, in a way, you're also restricting yourself. It's just like L L M. If you just try to make them copy human, it's one thing. But if you try to think, maybe they can do something that humans cannot do, it's also much more interesting in a way.

• 21:22Speaker 01

  Do you think we're heading towards a world of, you know, big foundation models that can kind of do everything and then be adapted quickly to any new domain with just like, you know, a few prompts? Or do you think that developers in your community are gonna start from like a small base model and then do a lot of their own data collection customization to adapt to their domains?

• 21:41Speaker 00

  Hmm, I think we'll, we'll see more and more both. I mean, I think as, as the, as the field evolves, you know, we, we start to see really a long tail. So for instance, if we take the downloads on, on Hugging Face, we see both like very large state of the art models being downloaded, which are usually too, too large to run on a local laptop. But we see also some of the most downloaded models are actually just the right size that fits to run quickly on the laptop. So we see these two like really modality. And I think as you know, as the field kind of mature, we'll, we'll, we'll start to see this more and more, which is it's not like you choose one or the other. It's just depending on what you need. You may use, you know, one locally or not. And I think GPT five with the router is a good example of this.

• 22:25Speaker 00

  You know, maybe the, the largest model or the most reasoning, the, the longest reasoning chain is not the answer to everything. And you actually need to smartly select the one you want. So it can be behind a router, but it can also be just locally. You will run some models here, they might be extremely useful. And we know better and better how to train model that are actually very extremely useful. But when you need something much more complex, when you need reflection for a very long time, then you will turn to, to, to much larger models.

• 22:52Speaker 02

  One of the narratives that's been really popular over the last few years is this narrative of the battle between open and closed, you know, closed models versus open models, who's gonna win? And just in the last few weeks, OpenAI is now present on Hugging Face. And so I'm, I'm curious what to make of that and sort of what it might imply about the future of open versus closed, or maybe how they work together.

• 23:19Speaker 00

  Hmm, I mean, we're super happy to welcome they back. I mean, they were, they were there. I mean, the first, you know, the first model I, I worked on. And, and the, the reason we, we switched from being a game company to an open source platform was GPT one, which not a lot of people remember, but it was very funny because it was trained money mostly on novels and romance novel. And so when you would put to character in the continuation, they would always fall in love in some way or. Very cute, still miss a little bit this one. And then and then Google took this idea and trained it also on Wikipedia, which added a lot of world knowledge and then expand it to GPT, all of that. But at that time, they were very, very pro open source. And I think, I think open source, just like in software, I think both, both solution, we just, we just coexist.

• 24:03Speaker 00

  And having company that, that open source both or that, that do both. I mean, Google has been an example for quite some time, right? With the Jima line and Jemini line. And, and some interesting moments where sometime I heard that one Jima model was actually so good that it was better than closed source model. So they had to not open source it. So the frontier, what is sure is quite seen at the moment in a way. And, and the challenging, you know, uh, the challenging new players, mostly in China. But I think we, we, we start to see also some new foundation models team in the US. So I think we, we, we might see also some challenger in the US. The frontier will stay quite seen, I think, and, and both, both thing we'll see with a tiny difference of, of performance.

• 24:43Speaker 00

  The main reason right now, I think, is, um, to be honest, at, at this exact point in of time, I think we're not exactly in a kind of cost-saving time of AI. So, so which means that for a lot of actors, I think, moving to open source because it's safe costs is not the most important thing for them. So usually they move to open source right now because they want data privacy. They want to be able to adapt the model. They have maybe have a new idea or new, new, no, like this, this action model, for instance, they have a new idea of something that does not exist and they want to do that. So that's usually what we see right now. So we see a lot of emulation of new, of new, of new exploration in the open source way.

• 25:20Speaker 00

  What I do expect is as, as we go to a more mature market as well than the cost and, and being able to run it maybe on faster hardware or this type of other hardware. And then being able to own the model and to own also the full stack of where the models run will become actually more and more important. So I think, just like in software, I think in the long term, open source is kind of a winning solution for many, for many, um, application, for many use case. But we're still in the, like, the turbulence place. Yeah.

• 25:49Speaker 01

  How do you think Hugging Face's role in the LLM ecosystem has evolved as these models have pushed at the frontier and there's closed models? I remember back when it would be like you could download the small BERT model on Hugging Face and run it locally, right? And that was a lot of the usage. How has your business evolved now that we're going towards, as you mentioned, models that are too large to run on consumer hardware? And how do you see Hugging Face's role evolving?

• 26:15Speaker 00

  I mean, surprisingly, I was doing these stats at the end of last year. This BERT model is still really used a lot. So a surprisingly interesting aspect of open source is also resiliency, which is once you have something that worked, that really worked in production, you may not want to be forced to move to the new GPT, right? I mean, that was a little bit of the backslash around GPT-5, you know, just people actually wanted to keep using GPT-4O for many things. Maybe they fell in love with this or were their main friends and some Reddit posts was around this, but also maybe they just had their like applications work really well and they don't want to redesign it. I think open source, I mean, the long-term interest for us is also to provide this very stable base, like you build something, you know it will exist and you know you can keep

• 26:59Speaker 00

  this as a very stable base. And in general, I think in the community, our role has switched progressively from maybe pushing ourselves a lot of things, pushing our library, pushing our, you know, our early product, to more like enabling more and more the community in general. So we work now a lot with many, many actors of the community. We work a lot with Llama CP. We work a lot with VLLM. We work a lot with all the big players to try to see, you know, how this whole ecosystem can be very efficient, can work really well. So like one model is released, you want to be able to use it directly in VLLM, you want to be able to use it directly in Lamas PP. So we try to have more and more this kind of role of

• 27:41Speaker 00

  Meta community builder, where we try to align and to bring maybe all the players, you know, at the same pace and to help them move in the same way. So in a way, we are much more focused on the on the community, on the hub than we were maybe a couple of years ago.

• 27:58Speaker 01

  Really cool. What do you think about what's happening? You mentioned, you know, China's had a lot of the open models recently. Like, why do you think that is happening? And what is the state of open model development in the West?

• 28:11Speaker 00

  Yeah, this is the most surprising thing that happened, I think, in the last two years, right? The fact that China would become a champion of open source. Who would have predicted that in 2020s, right? And so I've been, you know, I've been actually visiting them two weeks ago to try to understand a bit better on the ground, how it's happening. And the thing is just, it's a very, very competitive market internally. There is a lot of teams there that are extremely good. And it reminded me in some way of Silicon Valley. People are working extremely hard and they compete with each other, all of these model providers. And one part on which they compete, which is surprising, is being the most open, is the open source aspect. So they are extremely proud of being very open.

• 28:54Speaker 00

  And some of these companies, and that stopped being open, when it was called Zippo, they decided to not open source it. And they saw an immediate black slash, I think mostly on hiring, like people didn't want to come work there anymore. And so they went back to open sourcing. So it's quite strong now, I would say strongly ingrained. So I would expect this to continue. I would expect also quite more team to come because I see a lot of, you know, I mean, we see that as well, right? When you when you have the presentation of GPT-5, like a lot of people are, you know, actually did their study, some of them at Tsinghua University, right? We know the the team are here, so, you know, partly, partly with Chinese members. So they have extremely, extremely strong people and they all really want to to train the the best model.

• 29:40Speaker 00

  What I think is interesting is to see the West kind of coming back to open source very recently, to be honest, just over the summer, right? But this call for open sourcing, OpenAI decided to come back. We're just waiting for Entropic to maybe open source their first model. So I think it's time to try to ask them to to participate.

• 30:00Speaker 00

  Yeah, I would say right now, the the situation for open source is is pretty good. But but yeah, it's never like it's it's like the Jedi and Star Wars. It's never one we have to keep. We have to keep pushing this. We have to keep pushing our our flag of openness.

• 30:16Speaker 02

  What's driving the resurgence of open source in the West?

• 30:20Speaker 00

  Yeah, I I think one thing is when you have in a way, nothing to lose. Open source is always a good solution when you're a new team. So it can be, for instance, you create a new company and you want to quickly rise to the top. Then you open source your model, right? That's the mistral recipe. How can you very quickly become a great player? But for the Chinese, for instance, it's also, for instance, in the West, almost nobody will use a Chinese API. So they don't sell API in the West anyway. So in a way, they have nothing to lose, you know, from the Western market by open sourcing their market there. So I think I think there is this thing at as players. So and the consequence of that is also that when nobody is open source, it's like a market, there is an interest for someone to take the room, right?

• 31:04Speaker 00

  To say, we're going to be the open source player. So Meta was this open source player when everyone kind of stopped open sourcing. And I I feel like there will always be this thing when some people stop open sourcing. And then there is actually a gap to being the only, you know, the new top open source actor. Then someone will want to fill this void.

• 31:23Speaker 01

  Thomas, you mentioned that, you know, Western companies won't use a Chinese model over, you know, Chinese API. What about, are you saying Western companies are actually willing or not willing to use Chinese open models? When it's, you know, the weights and, you know, hosted on on US servers? Like, is there still hesitance to to do that? And is it well founded or not?

• 31:44Speaker 00

  I don't see that a lot, to be honest. I mean, it's a good question. I try to do regular polls. So I try to ask a lot of people regularly. You know, what do you think about that? Because it can be for sure a concern, right? There was when Deep Sea came out and there was a very nice answer sort of model from perplexity, for instance. The thing is, in many business case, I don't think people really notice anything, you know.

• 32:09Speaker 00

  So I think there is more a general appetite for people to have a better way to understand

• 32:16Speaker 00

  like the safety of a model. But I would say it's quite general. People are a little bit worried about having a model that maybe will behave strangely in some cases. And so I think this is a general thing that a lot of company have been asking, which is can you guarantee this model will always behave well? Which we know is really hard because, you know, even with G P T, sometimes you ask a number of R in strawberry and it's just behaving badly. And you're like, why? You're very smart, you should be able to know that. So I think this is a general thing that is needed soon. And there is a couple of teams working on that for sure.

• 32:50Speaker 02

  Can we talk about open science?

• 32:51Speaker 00

  Yeah, we build L L M like human. But what if, you know, an A I model could see infrared, could see some radiation? We cannot. This is a thing that humans cannot do. So it's already superhuman. And for science, it's actually super interesting. So a lot of the A I model for science are already superhuman in a way because they can actually either see modality or predict things that are just inaccessible to to humans. And it's good. I think it's a good ground to think outside of like the human limitation of what we can do.

• 33:22Speaker 02

  Well, and you've been you've been pretty passionate about open science for a while. So can you just say a word about what is open science? What role does Hugging Face have to play? And where does your passion for it come from?

• 33:33Speaker 00

  And for me, it started a very long time ago. So before, like you first, I was a lawyer. But before being a lawyer, I was a researcher in physics. And so I was working on this. Superconducting material. And surprisingly, in superconducting material, a lot of the great research had been done by the Soviets back in Soviet Union. And these people, I mean, you know, Soviet researchers have a very different way of inventing theory than the Western world has. And so they had some really great ideas and some really interesting things. But I had to find this invention or this theory. I had to find them, to track them down in the Soviet GTP letters. And some of them were even still in Russian. And so from this time, I get decided that damn accessing knowledge is hard.

• 34:16Speaker 00

  And if I can make this easier, that's going to unlock a lot of really cool stuff. If I could just find where, you know, that's this equation comes from and really be able to read this article, that would be crazy. And so when I joined computer science, I discovered archive, I discovered open source. I was like, this is really cool. Everything's just free. Basically, everyone just share things, read in English, everyone can read it, it's even free. You don't even have to buy the publication. And I was very excited about that until, until I started to try to reproduce one DeepMind paper. And I discovered that there was a limit because people publish what they want to publish, but they don't really give you all the tricks of the trade, right? And so when you try to reproduce that, you discover that it just doesn't work.

• 34:58Speaker 00

  And so open science for me was this extension, which is, it's nice to give open models to people so they can build things on top. But it's even better to explain how to train a model. It's a thing, it's nice to give a fish to someone to feed them, it's even better to teach them to fish. And that's basically what we want to do. We think in the very long term, AI is going to be such a fundamental technology that basically it should be, just like physics, should be something everyone could learn by reading a book. Like, if you want to learn today about general relativity, you can read a book and you can know about it, right? You don't have to pay to get access. I mean, you buy the book or you find it, but that's, that's basically free access. I think AI, all the recipe to train an intelligent object or artifact should also be something that everyone should know.

• 35:43Speaker 00

  So I mean, that's a very long term thing. But the very short term thing is if we teach people how to train great models, then they bring great models on the hub and then we have much great content to offer. So it's kind of also just content providing. If you provide great model, it's nice. And so one example that we do for that is we write very long blog posts that some of them even become books. We just published a book this summer on how to train on a thousand GPU and how to balance the load and how to do all of this parallelism thing. Another very long blog post we wrote was around how to make a very good quality data set. And so we made a data set to pre-train the models, it's called FineWeb. And it's used in a lot of the recent model, the Qwen models, they use FineWeb, for instance.

• 36:26Speaker 00

  And then we also wrote how we build this data set, how we filter it, what is important to understand when you want to build great data to train models. So all of this, I think, just go together and for us is a way to basically bring just better, better open source AI models in honeyface.

• 36:42Speaker 01

  I want to go back to your physics and superconducting comments. Like, it feels like, you know, a lot of the AGI labs believe that AI actually disrupting science is not that far out. There's been some exciting discoveries, I think, and, well, I think there's been exciting evidence so far in math and then maybe extending into physics, material science. Like, do you think we're gonna see an inflection point in scientific discovery from these models? And what do you think open source's role will be in driving that?

• 37:11Speaker 00

  As always, it's nice, there is some hype here because then it drives people, but I think sometimes we overestimate what's happening. I mean, math is a good example, right? There was this idea, AI is doing new proof for some math theorem, and that's like inventing new science. I think, as a scientist myself, I think that's really the wrong way to view science. The reason is I was a bad scientist, so I can tell about that. So I was a very good student, so when you give me a problem, I'm always pretty sure I can find the proof, I can find the thing. But I know this thing has a solution, so I just have to, you know, fill the gap and kind of grab a couple of things that I know and then combine them together. And when I became a researcher, I discovered that I was a pretty

• 37:54Speaker 00

  bad researcher, because basically what I was not able to do was, I was not able to ask the right question. So if somebody asks me a question, say, can you demonstrate this theorem? I could do it. But if someone say, okay, what is interesting to explore now in math? I had no idea, basically. And so in science, the main thing you need to do, if you want to do, I am talking about big, big breakthrough, right, is you need to ask the right question. You need to find a way to ask a question nobody has asked before, and a question that will open a whole new field of research. And that's basically a Nobel field, a Nobel Prize. Nobel Prize is typically someone who just opened a new field of research because this person just asked the right question, is maybe You know, maybe the the speed of like, the speed of light should be the constant.

• 38:39Speaker 00

  And let's let's explore what does it mean. And it means actually we can create general relativity and then we can invent black hole out of it. And I think L L M right now are still extremely bad at this thing, at this kind of tasteful way to ask the right question. Which doesn't mean we cannot do really cool stuff with them. But the way I see them nowadays is really more as very useful helpers. So once you have human researcher, we say, this is something interesting to study. Then you can use them to actually multiply by ten, a hundred or thousand the prediction you can do. You can use them to quickly do a a full survey of what has been done in the past on this molecule, this protein.

• 39:18Speaker 00

  You can you can use them to say, OK, what would be, you know, the most logical way to to test this this hypothesis? But I still see this as kind of accelerator and assistant of of scientific research. Then what I would love to see, which is an A I that would say, hey, I have an idea on how to go faster than light. But for this, you cannot just write the answer on how to go faster than light. You have to ask the right question. What should we change to today's theory? What should we do, you know, today? What should we reconsider to to invent something that's groundbreaking as that?

• 39:54Speaker 02

  To your point on asking the right questions, what do you think are the interesting questions in the world of A I right now? Or maybe the questions that people are not asking that they should be asking?

• 40:03Speaker 00

  I mean, this this is one question, I think, and it's related to something we talk a lot, which is this sycophancy, which is tendency of of A I models to always agree with you. I think I think a good researcher is actually a good example of person who disagrees for a lot of people. My former professor was a Nobel Prize, was very not friendly in how he would discuss.

• 40:28Speaker 00

  But I think that's part of it. You have to be extremely opinionated. So finding a way that this, you know, to push this model to have maybe in a way stronger opinion or maybe a taste in their opinion or like, I think for science will be a key. And it may, of course, this will be based on deep learning and L L M. But it may, it may involve other way to train them, other way to, to, to think about them. Um, I think that's, that's, that's one of the big question, I think, not, not a lot. There is a couple of people exploring that, but not a lot of people are exploring.

• 40:58Speaker 01

  Okay. When you see the world in 10 years, like what, what, what is hugging face's role in it? How much of your community you think is building with L L Ms, with robotics? I know it's hard to think in 10 year time spans, but what, what do you think the world looks like in 10 years?

• 41:13Speaker 00

  Yeah, 10 years is very, very different. I mean, what I would love to see is in 10 years, a world where basically everyone feel like they can build with A I. And not they're just consuming A I, but they feel like they can be actor of, of this thing. A little bit like, you know, the, the difference between a, we, we used to have a lot of media that were generating and created for us. And then we moved to the current era where everyone is actually able to create media. And we saw that this creates a whole new generation of people and, and YouTubers and influencers and people actually making extremely interesting content. And, and I would love A I being the same, which is a, like a very, a very big committee, like the software developer committee, where everyone can create things with A I. And they feel like it's just another tool in their box.

• 41:55Speaker 00

  They can code stuff, but they can also train a model and they can also maybe adapt the model. So the nice thing about that is I'm a big believer in basically the creativity and natural invention of just the community. I think it's something that's very beautiful to witness. So in 10 years, I hope that people are not just consuming, you know, A I content and not doing anything, but they're actually exerting their creativity to build really nice thing with a lot of A I tools around them. I, I, to be honest, that's something that I think that's kind of what we are building right now. So I'm quite optimistic. The thing is gonna change a lot of things for society in general, because a lot of this job will just be different.

• 42:31Speaker 01

  It's a beautiful vision. Thomas, thank you so much for joining us today. We really enjoyed the chat.

• 42:36Speaker 00

  Dankz, het is een plezier.