The new modeling team will collaborate with Google's Gemini, Veo, and Genie teams to address critical challenges and enhance model scalability. Gemini is known for tasks like image analysis and text generation, Veo focuses on video generation, and Genie specializes in simulating games and 3D environments. The team will work on developing real-time interactive generation tools and integrating their models with existing multimodal models. This approach is seen as crucial for advancing AI in areas such as visual reasoning, simulation, planning for embodied agents, and interactive entertainment.
Key takeaways:
- Google DeepMind is forming a new team to develop AI models capable of simulating the physical world, led by Tim Brooks.
- The new team will build on Google's Gemini, Veo, and Genie projects to tackle critical new problems and scale models to high levels of compute.
- Gemini focuses on tasks like analyzing images and generating text, Veo is for video generation, and Genie is for simulating games and 3D environments.
- The team aims to develop real-time interactive generation tools and integrate their models with existing multimodal models like Gemini.