Google DeepMind Launches Genie 4 for Interactive World Simulation

Google DeepMind has announced the launch of Genie 4, a real-time simulation model for interactive 3D worlds featuring sound and talking characters. Gemini models assist in background planning to maintain simulation consistency, enhancing the interactive experience. However, Genie 4 remains an early research model, with longer sessions leading to visual inconsistencies. This development marks a significant step in agent design, combining new technologies with concepts predating large language models, thereby opening new avenues in the field of artificial intelligence.