Runway's conference got drowned in the factory cognition drama so I don't think many people noticed but Runway, who is mostly known as a video gen AI lab, is leveraging their world model research to solve one of the biggest bottlenecks in robotics: training data. Robots normally need thousands of hours of demos recorded on the actual robot, with someone driving it through the same task over and over. Praxis-1 is a robot brain that learns how the physical world works from regular videos of humans doing stuff, using the same research behind Runway's world models. So with praxis-1 robotics companies would only needs a few minutes of demos on their own robot to get it working 🫪🫪 This feels like one of those things that feels kind of random now but ends up being very consequential in the next 18 months