AI technology company Runway has introduced its first physics-aware world model, GWM-1, aimed at simulating reality to train agents and power various applications such as video, robotics, and avatars. Unlike traditional AI systems, a world model like GWM-1 learns an internal simulation of how the world functions, enabling it to reason, plan, and act without the need for extensive real-life training scenarios.
Runway’s CTO, Anastasis Germanidis, emphasized the importance of establishing a robust video model as a foundation for a comprehensive world model. The company’s GWM-1 is positioned as a more versatile model compared to competitors like Google’s Genie-3, offering the ability to create simulations for training agents across diverse domains, including robotics and life sciences.
Additionally, Runway has introduced specialized versions of the world model, including GWM-Worlds, GWM-Robotics, and GWM-Avatars. GWM-Worlds, an interactive app associated with the model, allows users to generate a world based on prompts or image references, incorporating elements like geometry, physics, and lighting. This tool, running at 24 fps and 720p resolution, has implications for gaming and holds promise in educating agents on navigating and interacting in physical environments.
Source: TechCrunch
Leave a Reply