DeepMind launches Genie 2 to create interactive 3D worlds from images and text

DeepMind has introduced Genie 2, a new AI model that creates interactive 3D worlds from a single image and text description. This successor to the earlier Genie model can simulate real-time scenes, allowing user interactions like jumping and swimming. Genie 2 can generate diverse environments and maintain consistency in scenes, unlike previous models that struggled with memory and detail. It is designed primarily as a research tool for creating interactive experiences and evaluating AI agents. DeepMind's advancements in world modeling come amid growing interest in generative AI. The company has recently expanded its team, hiring experts from other AI organizations to enhance its capabilities in video generation and simulation technologies.