Google is moving up a gear in artificial intelligence research with the creation of a new team within Google DeepMind. Led by Tim Brooks, formerly of OpenAI, this team aims to develop generative models capable of simulating the real world with unprecedented precision and scale.
Tim Brooks, who joined Google DeepMind last October, made this initiative official in a post on X (formerly Twitter). According to him, the objective is clear: to create massive models capable of generating realistic and interactive simulations of the physical world.
This new team will build on existing work carried out by the teams Gemini, I seeet Genie from Google:
- Gemini : Google’s series of flagship models, capable of handling varied tasks such as image analysis and text generation.
- I see : A video generation model developed internally to push the limits of multimedia content.
- Genie : A technology dedicated to real-time 3D environments, which has recently demonstrated its potential in generating playable interactive worlds.
Towards a new stage: artificial general intelligence
The mission of this team goes beyond simple academic research. The job descriptions shared by Brooks highlight a forward-looking ambitionartificial general intelligence (AGI)an AI capable of performing all the tasks that a human could accomplish.
Among the priorities:
- Develop tools for real-time interactive generation from generative models.
- Integrate these tools with multimodal models like Gemini.
- Harness supercomputing capabilities to address complex problems, ranging from simulation for robotics to intelligent planning and interactive entertainment.
These efforts aim to put Google ahead in the global AGI race, a goal shared by many tech players.
A thriving ecosystem
Google is not alone in this area. Startups like Odyssey, Decart or even World Labs (founded by Fei-Fei Lee, an influential researcher in the field of AI) are also developing similar models.
The potential of “world models” is vast:
- Entertainment : Creation of interactive films and ultra-immersive video games.
- Simulation : Realistic environments for training robots or intelligent agents.
- Scientific research : Advanced visualization and planning for complex tasks.
An innovation not without controversy
Like any cutting-edge technology, world simulation models raise ethical and social questions:
- Impact on jobs : There are concerns about the growing use of AI in creative industries like animation and video games. A 2024 report, commissioned by the American union Animation Guildestimates that more than 100,000 jobs could be affected by 2026. Some studios, like Activision Blizzard, have already integrated AI to compensate for massive layoffs.
- Copyright : Many models are trained on data from videos, sometimes without explicit permission. While Google uses YouTube content in accordance with its terms and conditions, the details of its use remain unclear, which could expose the company to litigation.
- Collaboration with creators : Some startups, like Odyssey, promise to work hand in hand with creative professionals to avoid replacing them. Will Google follow this example?
A long-term vision
With this new team, Google DeepMind aims to combine fundamental research and concrete applications. The promise of interactive simulations and advanced planning models could transform industries like robotics, entertainment and training.
However, the road to AGI remains strewn with pitfalls: ethical issues, social impacts and fierce competition will play a determining role in the success of these technologies.
