Google is embarking on a bold initiative to create generative AI models capable of simulating the physical world, signaling a significant stride toward artificial general intelligence (AGI). The company is expanding its DeepMind research division, led by Tim Brooks, to tackle the ambitious task of developing “world models.” These models are designed to revolutionize fields like visual reasoning, planning for embodied agents, simulation, and interactive entertainment. The research aims to address critical challenges and scale AI models to unprecedented computational capabilities.
This initiative builds upon the groundwork laid by Google’s existing AI teams, including Gemini, Veo, and Genie. Gemini, Google’s flagship AI model, specializes in tasks such as text production and image analysis, while Veo focuses on video generation. The company sees the integration of video and multimodal data as pivotal to achieving AGI an advanced form of AI that can perform any task a human can. Job postings for the project emphasize the critical importance of scaling AI training across diverse data formats to achieve these objectives.
The development of world models underscores Google’s commitment to pushing the boundaries of AI technology, with potential applications across numerous domains. By simulating the physical world, these models could transform industries ranging from interactive entertainment to robotics and beyond. This move solidifies Google’s position at the forefront of the race toward achieving AGI and unlocking new possibilities in AI-driven innovation.