Google Allows Users to Create Their Own Interactive Worlds with Project Genie
Google DeepMind has launched Project Genie to publicly experience its world model called Genie 3. Adult users in the US will be able to create and explore AI-generated virtual environments in exchange for a specific subscription plan.
World Model Goes Public
Tech giant Google is taking another significant step in artificial intelligence research. The company's AI research unit DeepMind is making its 'world model' Genie 3, introduced this summer, available for users outside Google to experience under the name Project Genie. This system allows users to create their own interactive virtual worlds and navigate within them.
How Does Project Genie Work?
Users who want to try the project must subscribe to Google's $250 monthly AI Ultra plan, reside in the USA, and be over 18 years old. Project Genie offers users three different interaction modes: World Sketch Creation, Exploration, and Remix.
In the first stage, Google's Nano Banana Pro model creates the source image that Genie 3 will later use to create a world to be explored. At this stage, the user can define their character, determine the camera angle (first-person, third-person, or isometric), and choose how they want to explore the world. Before the model begins its creation, the user can have Nano Banana Pro generate a sketch of the environment it will see and make corrections accordingly. It is also possible to write one's own prompts for worlds created by others with Genie.
Not a Game Engine, But a Simulation Tool
Experts emphasize that Genie 3 is not a game engine. Although its outputs may look game-like and it can simulate physical interactions, it does not include traditional game mechanics. The created worlds are limited to 60 seconds, and presentations run at 24 frames per second and 720p resolution. This system is fundamentally positioned as a tool for training AI agents.
A New Phase in AI Research
The launch of Project Genie indicates the rapid advancement of AI's creative and simulative capabilities. Such world models could be used in the future for developing deeper research agents as well as in areas like accelerating scientific simulations. Google's move appears to be opening a new front in the cloud and AI infrastructure competition. Industry observers note that these developments once again bring to the fore the question of what the new frontiers of AI might be.
Although Project Genie is currently offered with limited access, it is considered a concrete indicator of both the interest in AI-generated interactive environments and how rapidly technical capabilities in this field are progressing.
