However, the model is not without its limitations, as the worlds it creates only last for up to a minute, with most lasting 10-20 seconds. Despite this, DeepMind is positioning Genie 2 as a research and creative tool for prototyping interactive experiences and evaluating AI agents. The company believes that Genie 2, despite being in the early stages, will be a crucial component in developing future AI agents.
Key takeaways:
- DeepMind, Google’s AI research organization, has introduced Genie 2, a model capable of generating a wide variety of playable 3D worlds from a single image and text description.
- Genie 2 can simulate object interactions, animations, lighting, physics, reflections, and the behavior of non-player characters, and can generate consistent worlds with different perspectives.
- Despite its impressive capabilities, games created with Genie 2 would erase progress every minute, so DeepMind is positioning it as a research and creative tool for prototyping interactive experiences and evaluating AI agents.
- DeepMind believes that Genie 2, while still in early stages, will be a key component in developing future AI agents and Google is investing heavily in world models as the next big thing in AI.