Google has just taken another big step in the field of artificial intelligence with Project Genie, a web-based experiment developed by Google DeepMind, allowing users to create and explore worlds built by AI in real time.
This project was launched about five months after Google introduced Genie 3, a "multi-functional world" model capable of simulating the environment and reacting like real life.
If Genie was previously only tested within a limited range, Project Genie has opened up the opportunity for a group of paid users to directly step into the worlds created by AI.
Project Genie's operating method is designed to be simple and user-friendly. Users start by entering text suggestions to describe the context, characters, and desired world style.
They can also upload or create images to shape the space appearance. Before participating in the experience, users can preview and fine-tune the scene using Google's AI tools such as Nano Banana Pro and Gemini.
When entering the environment, users can explore the world from the first or third perspective. In this space, the characters are not standing still but are always moving, can walk, fly or drive through continuously expanding landscapes according to the explorer's journey.
Unlike static 3D environments, Project Genie creates new terrain in real time, while memorizing the areas that have been created to maintain consistency when users return.
Due to the requirement for high computing power, each experience session is currently limited to 60 seconds. During this time, users are provided with a dedicated processing "chip", ensuring a private and stable experience.
Access to Project Genie is currently for those who subscribe to Google's AI Ultra package, priced at $250 per month and only applicable to users aged 18 and over in the United States. Google said it will expand the scope of deployment in the near future.
Behind Project Genie is a larger ambition to build world models, an AI system capable of simulating how the environment works and reacts to human actions.
Instead of just solving narrow tasks like playing chess or creating images, these models aim to recreate the diversity, unpredicability and flexibility of the real world.
This technology is expected to be applicable in many fields such as robot training, architectural design, game development, film production, historical simulation and skills training in virtual environments.
Project Genie also allows users to edit or remix existing worlds based on other people's suggestions, creating a community-based creative ecosystem.
In addition, the platform also provides a collection of selected experiences and random worlds to inspire. Users can record videos of their journey of discovery and share them with the community.
However, Google is still cautious when calling Project Genie a prototype. The company admits that the world in the game sometimes does not reach a high level of realism, physics can work abnormally and character control can sometimes be jerkily, lag. Some advanced features that were previously introduced have not been integrated.
However, in the context that many competitors such as Runway or World Labs are pursuing similar goals, Project Genie shows that the world created by AI is no longer a fictional idea.
