Google is forming a new team to build AI models that will simulate the physical world.
Where is AI going? Six leaders share their hopes for AI in the coming year, in The Batch:
– Hanno Basse: Generative AI for Artists
– David Ding: Generated Video With Music, Sound Effects, and Dialogue
– Joseph Gonzalez: General Intelligence
– Albert Gu: More Learning, Less Data
-…— Andrew Ng (@AndrewYNg) January 7, 2025
The team will be part of Google DeepMind and will focus on creating generative models capable of simulating complex physical scenarios. Tim Brooks, who recently joined Google DeepMind from OpenAI, will lead the new initiative.
Brooks was one of the co-leads on OpenAI’s video generator project.
🎧 The first NVIDIA AI Podcast of 2025 is here. 🎧
Listen to Ming-Yu Liu, VP of research at NVIDIA, discuss why WFMs & the new #NVIDIACosmos platform are key to advancing physical #AI in various industries like robotics & autonomous vehicles 🔑https://t.co/LiAU87ApKG pic.twitter.com/tbcunxdF7p
— NVIDIA AI (@NVIDIAAI) January 7, 2025
The new modeling team will collaborate with and build upon work from other specialized divisions within Google, such as the Gemini, Veo, and Genie teams. Gemini is known for its AI models that analyze images and generate text, while Veo focuses on video generation.
Genie specializes in creating AI that can simulate games and 3D environments in real-time.
DeepMind has ambitious plans to make massive generative models that simulate the world. I'm hiring for a new team with this mission. Come build with us!https://t.co/pqvALtAvLs https://t.co/vtwgeXl9Dl
— Tim Brooks (@_tim_brooks) January 6, 2025
According to one of the job descriptions, the team believes that scaling AI training on video and multimodal data is critical to achieving artificial general intelligence (AGI). AGI refers to AI that can accomplish any task a human can perform.
World models are expected to enhance various domains, including visual reasoning, simulation, planning for embodied agents, and real-time interactive entertainment.
Team building AI world models
Brooks’ team will work on developing real-time interactive generation tools and integrating them with existing multimodal models like Gemini.
This effort is part of a more significant trend, with various startups and tech companies exploring the potential of world models. These models could revolutionize interactive media, such as video games and movies, and provide realistic training environments for robots. However, the integration of AI into creative industries is not without controversy.
Some game studios have been accused of using AI to cut costs and increase productivity, leading to significant layoffs. A 2024 report from the Animation Guild estimated that AI could disrupt over 100,000 U.S.-based film, television, and animation jobs by 2026. Copyright concerns related to the training of world models also exist.
Some models are trained on clips of video game playthroughs, which could result in lawsuits if the videos were used without proper licensing. Google, which owns YouTube, maintains that it has permission to train its models using YouTube videos as per the platform’s terms of service. However, the specific videos being used for training have not been disclosed.
As Google embarks on this ambitious project, the AI and creative communities will closely watch the developments from Brooks and his team.
Noah Nguyen is a multi-talented developer who brings a unique perspective to his craft. Initially a creative writing professor, he turned to Dev work for the ability to work remotely. He now lives in Seattle, spending time hiking and drinking craft beer with his fiancee.
