Google is creating its own AI team dedicated to modeling the world for games and robot training

Google DeepMind is creating a modern team of artificial intelligence researchers to develop “world models” that can simulate physical environments. The initiative will be led by Tim Brooks, former co-leader of the Sora project at OpenAI who joined DeepMind in October to work on Google video generation and world simulators.

World models are a relatively modern development in artificial intelligence that can serve a variety of purposes, such as creating real-time interactive media environments for video games and movies and realistic training scenarios for robots and other artificial intelligence systems. This is also part of Google’s efforts to create an artificial general intelligence (AGI) system ahead of its competitors.

“DeepMind has ambitious plans to create massive generative models that simulate the world.” Brooks made the announcement on Monday in Post X. Brooks posted two open job postings for research engineers and scientists to lend a hand refine artificial intelligence “world models” capable of simulating real-world scenarios by solving “mass-scale” training problems, curating training data and examining how integrate them with multimodal language models.

“We believe that scaling initial training using video and multimodal data is on the critical path to artificial general intelligence.” DeepMind stated in the job description. “World models will power many domains, such as visual reasoning and simulation, planning with embodied agents, and real-time interactive entertainment.”

The race to be the first to declare AGI is heating up, so Google’s attention is not surprising. OpenAI CEO Sam Altman recently said that the company had achieved a long-awaited benchmark in the tech industry and that autonomous AI agents could begin to join workers in a meaningful way this year.

The modern DeepMind team will work with existing Google AI projects, including AI flagships Gemini, the Veo video generator, and Genie, Google’s previous global model for simulating playable 3D environments in real time.