Advancing Artificial Intelligence Capabilities Through Realistic Environment Simulations

Google is building its own ‘world modeling’ AI team for games and robot training

Google has announced the creation of a new “world modeling” artificial intelligence (AI) team, led by Tim Brooks, to develop AI models that can simulate physical environments. This ambitious project aims to advance the development of artificial general intelligence (AGI), allowing for more realistic simulations in video games and robots.

DeepMind, Google’s AI subsidiary, has revealed plans to assemble a new team focused on creating “world models” capable of generating highly realistic virtual environments. Brooks, who joined DeepMind in October 2024, will be at the helm of this initiative. The goal is to build upon existing AI projects such as Gemini AI models, Veo video generator, and Genie – Google’s prior world model for simulating playable 3D environments.

The “world modeling” team aims to create AI models that can simulate real-world scenarios by solving issues related to training at massive scale, curating training data, and exploring the integration of multimodal language models. The ultimate objective is to push the boundaries of AI capabilities and achieve AGI before competitors do.

DeepMind’s new team will work alongside Google’s other AI projects, including Gemini AI models and Veo video generator, which have already demonstrated impressive results in generating realistic 3D environments.

The creation of this world modeling team marks a significant escalation in the battle to develop an artificial general intelligence system that can outdo human capabilities. The move aims to stay ahead of the competition by developing more advanced AI systems.

Google’s push for AGI is not surprising, as OpenAI CEO Sam Altman recently announced breakthroughs in their own AGI research. Autonomous AI agents may start to join workforces this year, according to Altman.

Source: www.theverge.com