Like in 'The Matrix', but for real: Google's new technology looks like science fiction

Google DeepMind has unveiled Genie 4, an AI system that creates photorealistic interactive 3D spaces in real time with voice dialogues and physics. Users can navigate and change the environment using text commands. The article also mentions competition with the Israeli startup Decart.

Google DeepMind's research lab has introduced Genie 4, a new artificial intelligence system capable of creating photorealistic three-dimensional spaces in real time with voice dialogues and simulation of natural laws. Unlike conventional video generators, users can move inside the simulation using a keyboard, change the environment with text commands, and interact with characters. The system is based on a hybrid architecture combining Google's language models (Gemini Omni, Gemini 3.8 Flash, Gemini 4 Argon) and a world model. It can simulate physics, including aerodynamics and collisions, for approximately two minutes. The presentation took place amid competition, including with the Israeli startup Decart, which is in negotiations to sell to Amazon for $7 billion. Developments by Runway, AMD, and Rivian are also mentioned. DeepMind acknowledges that Genie 4 remains a research project and faces visual distortions during long simulations. Access to the system will be opened to Google AI Ultra subscribers and select developers in the coming weeks.