Ever before artificial intelligence (AI) tools became widely used, businesses have been experimenting and testing the technology in various fields. There was never an AI chatbot that allowed us to play games, even though we had AI bots for creating material, creating photographs, editing videos, and other tasks. However, that might change now.
Google unveils SIMA (Scalable, Instructable, Multiworld Agent)
Google recently announced that it is developing SIMA, an AI agent that is being taught to play video games like a skilled human player. The AI agent is being taught in a variety of gaming scenarios to reach human-level proficiency. SIMA is designed to work as a cooperative colleague with human players, in contrast to normal AI that could interfere with or overpower human gameplay.
SIMA which is still in the research phase can now master any video game, including open-world adventures and linear storylines. Eventually, SIMA will be able to play any video game, including open-world games with no clear way to end the game. Consider it more of an additional player who gets along nicely with your group, even if it's not meant to replace the current game AI. It combines picture recognition and comprehension of three-dimensional environments with training in plain language.
To train and test SIMA, Google collaborated with eight-game developers: Coffee Stain, Tuxedo Labs, Hello Games, and Embracer. To teach SIMA the fundamentals of playing games like No Man's Sky, Teardown, Valheim, and Goat Simulator 3, researchers inserted the AI agent into the titles.
Additionally, Google stated in the publication that SIMA is a big advancement, moving the focus of game-playing AI from specific games to a more general approach. This is the first time an AI agent can understand different game environments and do tasks using commands in natural language, much like a human player would.
SIMA simply needs screen visuals and directions in normal language to function thanks to a straightforward interface. Because of its accessibility, it can communicate with different virtual environments without requiring specific APIs or access to the game's source code.
Among its 600 talents, the present version of SIMA excels at basic activities like object interaction and navigation. Google wants to improve SIMA's ability to handle more challenging jobs, such as multi-step goals and strategic planning.
ALSO READ: SpaceX targets greater distance for Starship in 3rd test flight milestone; Deets inside