The new world model from Google DeepMind allows users to generate and navigate dynamic, interactive environments in real time from a text prompt.
Google DeepMind has announced Genie 3, a groundbreaking world model capable of generating entire interactive worlds from a simple text prompt. This new AI allows users to navigate and alter these dynamic environments in real time at 720p resolution. The company states this is a major step forward in creating simulated worlds for training AI and a new frontier for generative media.
ALSO READ Google CEO to Staff 'Accomplish More' Amid AI Push
Genie 3 operates by auto-regressively generating each frame of the environment, taking user actions and the previously generated world into account to maintain consistency. This allows for real-time navigation at 24 frames per second. A key innovation is "promptable world events," which lets users alter the simulation—such as changing the weather or adding new objects—using text commands, creating dynamic "what if" scenarios.
Google DeepMind positions Genie 3 as a "key stepping stone on the path to AGI," as it provides an unlimited curriculum for training AI agents like robots and autonomous systems. Beyond research, the technology could create new opportunities in education, training simulations, and entertainment, allowing users to explore everything from historical settings to fantastical animated worlds generated on the fly.
ALSO READ Australia to Ban YouTube for Children Under 16
The announcement builds on over a decade of Google DeepMind's research in simulation. Genie 3 follows previous models like Genie 1 and the video-generation model Veo. While earlier models could generate environments or videos, Genie 3 marks the first time the company has combined these capabilities with real-time interactivity, a significant technical breakthrough that maintains consistency for several minutes.
While acknowledging limitations such as a constrained agent action space and short interaction times, Google DeepMind is moving forward cautiously. Genie 3 is currently available as a limited research preview for a small cohort of academics and creators to gather feedback on safety and applications. The company plans to explore making the technology available to more testers in the future, aiming to responsibly develop a tool with vast potential.
Disclaimer: This article was generated with the support of AI and
edited for clarity by the PulseNext team. Except for the headline and featured
image, the content is sourced from a syndicated feed. For details, please refer
to our [Terms & Conditions].
0 Comments