Google is establishing a new team focused on developing AI models capable of simulating the physical world, marking a significant step in its ongoing AI research efforts. Tim Brooks, who previously co-led OpenAI’s video generation project Sora, has joined Google DeepMind and will spearhead this initiative. He announced the formation of the new team in a post on X, emphasizing the ambitious goal of creating large generative models that can accurately replicate real-world conditions.

In his announcement, Brooks stated, “DeepMind has ambitious plans to make massive generative models that simulate the world.” The newly formed team will collaborate with existing groups at Google, including those working on Gemini, Veo, and Genie. These teams are already involved in various aspects of AI development—Gemini focuses on image analysis and text generation, Veo is dedicated to video generation, and Genie specializes in creating real-time simulations of games and 3D environments.

The job listings associated with this new team highlight a commitment to addressing “critical new problems” and scaling AI models to unprecedented levels of computational power. One job description noted that scaling training on video and multimodal data is essential for achieving artificial general intelligence (AGI), which refers to AI systems that can perform any intellectual task a human can do. The development of “world models” is seen as crucial for applications in visual reasoning, simulation, planning for embodied agents, and real-time interactive entertainment.

Brooks’ team aims to create tools for “real-time interactive generation” based on their models while exploring how to integrate these with existing multimodal frameworks like Gemini. The pursuit of world models has garnered interest not only from Google but also from various startups and tech companies, including Fei-Fei Lee’s World Labs and Israeli startup Decart. These entities believe that world models could revolutionize interactive media such as video games and films or facilitate realistic simulations for training robots.

However, the rise of AI technologies in creative fields has sparked mixed reactions among industry professionals. A recent investigation by Wired revealed that major game studios like Activision Blizzard are increasingly relying on AI to streamline production processes amid workforce reductions. Furthermore, a 2024 study commissioned by the Animation Guild estimated that over 100,000 jobs in the U.S. film, television, and animation sectors could be affected by AI advancements by 2026.

While some startups in the world modeling domain, such as Odyssey, have committed to collaborating with creative professionals rather than replacing them, it remains to be seen whether Google will adopt a similar approach. The potential impact of these technologies on employment in creative industries raises important questions about the future role of human creators.

Additionally, there are unresolved issues surrounding copyright in relation to world models. Some models may be trained using clips from video game playthroughs without proper licensing, which could expose developers to legal challenges. Google asserts that it has secured permission to utilize YouTube videos for training its models under the platform’s terms of service; however, it has not disclosed which specific videos are being used for this purpose.

As Google embarks on this ambitious project to simulate the physical world through AI, the implications for both technology and creative industries will be significant. The development of advanced world models could open new avenues for interactive media while also prompting critical discussions about ethics, employment, and copyright in an increasingly automated landscape.