Google has reportedly expanded its reach beyond its dominance in this world, entrusting its DeepMind analysis laboratory with developing generative models capable of mimicking and simulating the physical world. The mission, likely spearheaded by Tim Brooks, a key leader involved in developing OpenAI’s video generator, Sora, will play a crucial role in the company’s quest for artificial general intelligence, according to reports regarding the newly formed team.
Dr. Brooks, who recently joined DeepMind in October, is spearheading a project to develop groundbreaking generative models that simulate the world. According to his team’s function descriptions, this ambitious undertaking will span multiple domains, including visual reasoning and simulation, planning for embodied agents, and real-time interactive entertainment. If you’re intrigued by these concepts, please consider exploring them further and get back to us if interested in joining the effort.
Artificial intelligence, in its simplest form, often attempts to mimic the true nature of reality. While generative fashion models like Sora can mimic patterns they’ve learned from their training data, they lack true comprehension of the underlying causes behind these phenomena. While the AI may effectively produce a video of someone tossing a baseball, it lacks comprehension of the underlying physical principles governing this action. The world of fashion aims to equip machines with sufficient data to accurately analyze how a motion unfolds and predict its probable outcome?
Meta’s chief AI scientist Yann LeCun characterized global fashions in the following manner during an earlier interview this year: “A world model is your internal representation of how the world behaves…You can think of a sequence of actions you might take, and your world model will enable you to predict what the outcome of that sequence would be on the world.”
Global fashion construction is challenging due to substantial computational requirements and limited access to training data, making it difficult to develop accurate models that can operate in a wide range of scenarios.
DeepMind’s team seems determined to broaden its global impact. We aim to develop “real-time interactive era” tools atop fashion models, potentially exploring ways to integrate them with Google’s massive Gemini language model.
While one potential area that DeepMind may endeavour to address is the realm of video games. The job description for the newly formed team highlights its forthcoming collaboration with both the Veo and Genie groups within Google, signaling a strong potential for synergies and innovation. Google’s Sorbet-like video generator and Genie presents a real-world mannequin capable of simulating 3D environments in real-time, blurring the lines between virtual and physical realities. The online gaming industry has already begun to displace thousands of employees. According to a recent survey, more than 86% of the gaming industry has already integrated generative AI tools into their operations, while approximately 15% of existing gaming jobs may be replaced or significantly impacted by 2026.
Potentially enriching this world might arise from exploring its nuances more thoroughly than creating simulations of it.
