Google DeepMind has announced Genie 4, a new world model designed to generate interactive environments rather than just static videos. Unlike previous models that focused on pixel-level prediction, Genie 4 aims to create worlds that respond to user actions in real-time, incorporating live sound and allowing for in-simulation modifications via natural language. This research preview is slated for integration into Google's Project Genie and the Gemini API, though specific release dates and pricing are not yet available. The development highlights a divergence in AI approaches, with some focusing on high-fidelity video generation as a path to world understanding, while others advocate for abstract representation prediction. AI
IMPACT Sets a new direction for generative AI, moving beyond video to create interactive, simulated worlds for agent training and exploration.
RANK_REASON Frontier-lab model release with system card [lever_c_demoted from frontier_release: ic=1 ai=1.0]
- Gemini API
- Genie 3
- Google AI Studio
- Google DeepMind
- Jürgen Schmidhuber
- Minecraft
- OpenAI
- Project Genie
- Sora
- World Models
- Yann LeCun
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →