Chris, thanks so much for joining us today. For those of you who don't know, Runway is a company and I'll maybe let you explain this, but essentially you're trying to bring AI and generative AI into filmmaking and into the industrial world. Tell us a little more about the filmmaking side of that because I feel like that's an area that's attracted a lot of controversy among the creative classes over the last few years.
It has, but things are changing fast. So we've been working on what we call world models for quite some time, and we've been very deep into Hollywood for also a couple of years. And the way we approach it is really to try to make AI systems and models that can help the creative class and creative people just do more work better and faster. Today we work with every single studio out there. We work with a lot of filmmakers, a lot of creative talent that are using these tools on a daily basis. I think most of the kind of early skepticism or early complaints do come from a place of mostly fear and misunderstanding, and I think most of those considerations have passed because people have figured out that these are just very powerful tools.
Do you think the fear from artists, whether it's GFX or VFX, has passed?
Absolutely. I mean there's still, of course, a small minority of people who are thinking about the long-term consequences of jobs, but I think if you speak largely through the studios, the people working in the media in the industry, most of them already experienced technology.
But the economics of Hollywood are already pretty brutal depending on how you slice it and how you account for it. Those studios have continually looked for any way they can to save cost. And I'm curious if we're not at the point now, I don't think, where it's more cost-effective, but at some point we will be. And so what do graphics jobs, what do design jobs look like in a world where entire scenes can be procedurally generated?
Right. So a couple things. The first one is these tools are already reducing the cost significantly. It's happening right now in productions, like right now.
Yeah, absolutely. It's like companies and movies are being made that are... And is that on the labor side? Is it on the render side?
It's everywhere. It's on the pre-production side. It's in scripting. It's in planning. It's in execution and visual effects. And so this is already getting deployed at scale. But the issue of labor and jobs, I would say on the Hollywood side actually has nothing... the one we're seeing right now and the budget skyrocketing and the considerations where the industry is have more to do with other issues of streaming and the way you consume and you compete in the market and streaming services has less to do with AI. I think the AI kind of side effect will come over the next year or two.
You're talking about the disputes that actors might have with streamers. But you guys have an adjacent dispute there in that the corpus of work that your models and other models have been trained on is a real point of contention about whether you guys have to pay for those works or not and license those works, which in many cases it's been shown are copyrighted works or people's intellectual property. As you're creating IP for these studios simultaneously, isn't there a world where they're also suing you for training upon their data? I mean, are you facing any litigation around that right now?
No, no. A lot of the conversation on copyright has now moved to outputs. So what studios and IP holders care the most is about IP. I think most media companies and studios are recognizing that the future of this industry is on IP. If you have recognizable things people like, that's probably the most valuable part. And every company is in the IP business. And so what you want to do is you want to get out there to your customers and your viewers more often and more regularly in ways that are more economically visible.
Beyond say a two-hour movie streaming or a movie theater.
Correct. So if you're spending $100 million on making one future film, which is 90 minutes, imagine taking $100 million and spending it on like 50 movies, same quality, same amount of output visually, but you make way more content. So you have way better chances of hitting something. It's a quantity problem.
Doesn't this take some of the magic out of it, though? I mean, the idea that you're now suddenly scaling up and 90 times 100 is 9,000 minutes or 900 minutes of content here.
Well, I think actually that's what we need right now. There's a crisis of creativity in the industry because of the economic incentives of how the content is being made. And this, by the way, is not new. I mean, if you think about production of any type of media, people were freaking out. And this is true when industrial publishing became a thing because kind of that same arguments around was like, how many books are you going to read? We have like 2,000 books being published. You know how many books are published today? Like around 25 million books per year. Of course, I don't read 25 million books and I don't expect you to.
24 maybe 25. But the world is in a much better place because there's more people who've managed to tell a story or say something on the world because technology has allowed them to do so. Like we have this internal saying at Runway that the best movies are yet to be made because we haven't heard from probably the billions of people who haven't had access to this very small technology and this very small amount of access to make stories that people want to watch. And so I think AI will actually, right now it's becoming an incredibly good solution to this crisis that I would say Hollywood is going through.
You are a creative by training. I'm always sort of curious when we started seeing these generative AI videos, you could always kind of tell, right? They had that weird sort of psychedelic uncanny valley sheen to them. And now we've moved past that and they're photorealistic and they're almost indistinguishable from reality or from certainly sophisticated renders. Is there a clear artistic language you see in what you generate? You've compared it, I think, to a camera, right? You've said this is a new kind of camera.
Yeah, it is. And maybe for context, I've started working on Runway quite some time ago before AI was actually working in the way it's working right now. And so we've learned a lot about the process and the grammar of how you make things with these systems. And a good analogy I use a lot in LA and Hollywood and with media is to really think about this as a tool, right? Think about this as a camera. And if you think about the impact the camera had on society, culture was probably unprecedented. We first of all created a new art form. So cinema is a new art form. Photography is a new art form that was born out of science. And it was very much contended and fought in the beginning of the 1900s for reasons that might sound very similar today, which is like painters back then thought it was killing art. So it was the end of art. And this is true. People really resisted, I can't emphasize this enough, really resisting photography. They thought the world was gonna end because there was no soul in photography and this machine made pictures.
And I think a lot of that sentiment is now put into AI. But if you think about what photography did...
Photography invented a new art form by creating a new grammar of how people could use the technology to shape a story. And the way that you tell stories is not by painting. You need to invent something new. You need to invent focus and composition and light, right? You create new things and you understand the world differently. AI for me is very similar. The way you create is a new grammar altogether. And that's being shaped right now.
I would be hopelessly outmatched if I tried to have a philosophical discussion about the nature of art with you. So I'm going to do what all reporters do and just pivot away from that subject very quickly. But in part because you actually have two very different businesses that you're running. One is for the creative classes. The other, and this is something that the CEO of Michelin was talking with me about yesterday, is helping manufacturers and industrial facilities figure out what to do with their robots, which is not really something I'd ever thought of as a use case for AI. Can you explain to me, I guess, first what that business is, and then of those two businesses, which to me seem to be very different, which do you think represents a better growth, a higher total addressable market for you?
I mean, both are big and both are important for us. So maybe the best place to start is to understand why and how we can serve both. And it all goes back to this idea of world models. So, world model without going too much into technical details is this idea that you can train AI systems that understand the world and perceive the world and connecting the world maybe in the same way that we do, right? So, when you learn about the world, you're watching the world and you're learning about it by seeing things, right? So, we do the same. We give the model video data and we show you enough videos so the model understands how physics works, how objects work, how things move in a world. And then we can ask the model to create what I call amazing videos or entertaining videos which is media side. You make movies, you make shows, but you can also ask the model to make what I call boring videos. Videos you don't want to watch. And videos you don't want to watch are 17 hours of objects being moved on a table. Probably no one here wants to watch that.
