Google DeepMind is forming a new research team focused on developing AI models capable of simulating real-world environments. Led by Tim Brooks, a former OpenAI executive, and video generation specialist, this initiative aims to build upon Google’s previous advancements in large-scale generative AI models such as Gemini, Veo, and Genie.
In a post on X, Brooks emphasized the ambitious mission to create “massive generative models” and invited experts to join the team.
The new team will focus on scaling AI models for real-time, interactive generation of complex 3D environments and multimodal simulations. With applications ranging from video games and robotics training to advanced visual reasoning tools, this project aligns with efforts within the AI community to achieve artificial general intelligence, a system capable of performing any intellectual task a human can.
Why It Matters: World modeling technology has the potential to revolutionize industries such as gaming, entertainment, and robotics. However, it also raises concerns about job disruption, ethical training practices, and intellectual property rights. The formation of this team signals Google’s increasing investment in world models, competing with other AI frontrunners like Odyssey and World Labs.
- Leadership and Vision: Tim Brooks, who previously co-led OpenAI’s video generation project, Sora, will head the new DeepMind team dedicated to building world-simulating AI models.
- Building on Existing Models: The team will expand on Google’s Gemini (multimodal AI), Veo (video generation), and Genie (3D simulation) to enhance interactive, real-time generation capabilities. Genie’s latest iteration, Genie 2, is capable of creating fully playable 3D worlds.
- Applications of World Models: These models could be used for visual simulations in games, interactive media, training robots, and real-time decision-making for embodied AI agents.
- Ethical and Creative Tensions: While some startups like Odyssey pledge to collaborate with creators, others fear that increased automation could lead to job losses in creative industries. A 2024 report from the Animation Guild predicts significant job displacement by 2026 due to AI adoption.
- Legal Concerns Over Training Data: Google’s approach to training its models using YouTube videos has raised questions about copyright compliance, though the company claims its use aligns with YouTube’s terms of service.
Google is building its own ‘world modeling’ AI team for games and robot training – The Verge
