Google's Project Genie lets you generate your own interactive worlds (www.engadget.com)

Google's DeepMind has launched Project Genie, which allows users to generate their own interactive virtual worlds. Based on the recently introduced Genie 3 model, this AI system not only creates stunning images but also responds dynamically as users navigate through the simulated environments. Initially designed as a tool for training AI agents, Genie 3 is now accessible to the public, albeit for subscribers of Google's $250 per month AI Ultra plan within the US who are at least 18 years old. Project Genie offers three modes: World Sketching, exploration, and remixing. In World Sketching, users interact with the model by generating a source image through the Nano Banana Pro model. They can customize their character, select camera perspectives, and adjust various aspects before exploring the AI-generated world. However, it's important to note that despite its impressive visuals, Genie 3 lacks traditional game mechanics, presenting outputs limited to 60 seconds in 24 frames per second at 720p. This development is significant for the AI/ML community, marking a shift towards more interactive and user-driven AI applications, while fostering innovation in how AI can enhance creative expression.