Google DeepMind’s Suma is an AI agent that can play 3D video games.

Google DeepMind Introduced a unique Artificial intelligence Wednesday (AI) model that can play 3D video games like humans do. The AI ​​model is called a Scalable Instructable Multiworld Agent or SIMA, and it’s learning how to interact with different gaming environments and perform different tasks. The model is currently in research and is being trained to master more complex movements. Once perfected, the AI ​​model could have a wide range of use cases both online and in the real world, Google says.

In a ___ Blog post, DeepMind’s SIMA team explained that the AI ​​model is not being developed to be a super-intelligent gamer that can beat any game. Instead, the goal is to teach it how to navigate a 3D game in open-world situations and understand how a human would react and interact with it using natural language instructions. will do Google highlights that this is a very difficult task and one that should be focused on.

“It’s an important goal for AI in general, because while big language models have given rise to powerful systems that can learn about the world and make plans, they currently have nothing to do with us. There is no capacity to take action,” the SIMA team said. .

Tasks performed by the AI ​​model SIMA
Photo credit: Google DeepMind

To create the learning environment for the AI ​​model, Google DeepMind partnered with eight game studios and trained SIMA on nine different video games. Some of them include No Man’s Sky by Halo Games, Tears fall By Tuxedo Labs, Goat Simulator 3 And Walheim By Coffee Stan Studios, and more. The AI ​​model was exposed to new interactive worlds in each game and had to learn how to navigate the world, interact with objects, use menus and more.

The company has also created four research environments, including a new environment built with Unity called Construction Lab. In this particular environment, the AI ​​model was tasked with building sculptures from building blocks to test its object manipulation and intuitive understanding of the physical world.

Google highlighted that the current version of SIMA has been tested in 600 basic skills including navigation such as turning left and driving a car, object interaction such as climbing a ladder and preparing a helmet etc. The tasks were largely simple and could be completed in 10 seconds.

According to the company, training an AI model on a diverse set of 3D video games where it follows human instructions and acts accordingly could have a big impact in the future. The company is now focusing on SIMA training with more complex instructions “requiring high-level tactical planning and multiple subtasks, such as finding resources and building camps.” Google indicates that these tasks could translate to the real world and that the company is working on ways to help humans.

---