Google's Project Genie lets you generate your own interactive worlds

Google DeepMind introduced Project Genie, giving access to its world-generating AI, Genie 3, to external users. Project Genie allows users to create and explore simulated environments through AI interaction. Access is initially restricted to users in the US, aged 18+, and subscribed to Google's AI Ultra plan. Exploration involves three modes: World Sketching, exploration, and remixing of generated worlds. World Sketching leverages the Nano Banana Pro model to create an initial image for Genie 3's world generation. Users can define character details, camera perspectives, and methods of interaction within the generated spaces. Genie 3 isn't a game engine; it simulates physics but lacks traditional game mechanics. Generations are time-limited to 60 seconds, with a 24fps, 720p presentation restriction. Users can also interact with prompts for worlds others have generated within Genie. Project Genie offers a glimpse into DeepMind's advanced AI capabilities. This represents a novel, cutting-edge AI research project for subscribers to experience.