Project Genie is a Google DeepMind prototype that lets users create and explore AI-generated virtual worlds in real time, accessible to select Google AI Ultra subscribers in the US. The tool features an easy-to-use interface for generating worlds via text prompts or image uploads, offers quick previews, and supports creative remixing and exploration with standard PC game controls.
Project Genie is an experimental research prototype developed by Google DeepMind that allows users to create, explore, and remix AI-generated virtual worlds in real time. Access to Project Genie is currently limited: you must be a Google AI Ultra subscriber (costing around $200 per month), be located in the United States, and be at least 18 years old. Access is rolling out gradually, even for those who meet the requirements, and is available through Google Labs at labs.google/projectgenie. Once you have access, world generation is free with your subscription, and you don’t need to spend additional AI credits.
The user interface is designed to be simple and beginner-friendly. For those new to Project Genie, it’s recommended to start by clicking the “roll the dice” button, which randomly generates a world and character for you. This feature helps users get a feel for the controls and see what the tool is capable of without needing to write prompts or upload images. It’s important to have a stable internet connection and to stay on the page during world generation, which typically takes 10 to 30 seconds.
When creating your own world, you can either write a text prompt or upload an image. Prompts should describe both the environment (landscape, terrain, atmosphere) and the character (appearance, movement, and abilities). Project Genie uses another AI, Nano Nana, to generate a concept image preview of your world before building the full interactive version. You can approve the preview to proceed or edit it to refine your vision, saving time and ensuring the final world matches your expectations. You can also choose between first-person and third-person views, with third-person recommended for beginners.
For effective prompt writing, be clear about your character’s identity, actions, and how it interacts with the environment. Similarly, describe the environment’s type, style, contents, and behaviors (such as wind or dust clouds). Image uploads should be high-resolution, with the main subject centered if it’s to be used as a character. You can even upload photos of real places or artwork to turn them into explorable worlds, or use AI-generated portraits for unique experiences.
Controls are similar to standard PC games: W, A, S, D keys for movement, spacebar to jump or ascend, and arrow keys to rotate the camera. Each session has a 60-second countdown, and you can regenerate worlds to explore variations. There’s also a gallery of pre-made worlds that can be remixed. The video showcases creative examples, such as controlling a pack of cigarettes in a metro, skiing down a mountain, and simulating body cam footage, highlighting the versatility and realism of Project Genie. The possibilities are limited only by your creativity.