Google’s new AI gaming companion may ruin some video games

Google’s DeepMind lab has unveiled its new AI gaming companion Sima which can play with gamers, acting similar to human players, an advance that has also raised concerns about the future of the online gaming community.

Sima, which stands for Scalable, Instructable, Multiworld Agent, is designed to play along with gamers, following their instructions to carry out tasks in video games.

It is unlike non-playable characters (NPC), seen widely in video games, and instead acts as a fellow gamer to help accomplish whatever the human player is instructing it to do.

Currently in the research phase, the AI would eventually learn to play any video game, even those set in virtual open-worlds, DeepMind said.

“Sima isn’t trained to win a game; it’s trained to run it and do what it’s told,” the company added.

The AI would work like another human player which could impact the result of the game.

“Sima needs only the images provided by the 3D environment and natural-language instructions given by the user,” DeepMind explained.

While such an AI player could be a supportive companion in multiplayer games with a story mode, Sima may also offer gamers an unfair advantage.

For instance, in games involving monotonous repetitive tasks to gain XP, the AI could be instructed to carry out all such work, freeing the human gamer for better strategising.

Gamers shared their concerns on social media that they may find themselves outperformed by undetectable AI agents when playing online games in the future.

“Once this kind of agent improves up to actual AGI level, then people will no longer be able to compete against it,” one user posted on Reddit.

“Anyone will be able to use an agent to win undetectably as it will eventually be able to pick up, learn and master any new game quicker than any person can and then go on to far surpass our ability,” the user said.

Google worked with several game developers, including Hello Games, Embracer, Tuxedo Labs, and Coffee Stain, to train and test Sima, and have plugged the AI into games such as No Man’s Sky, Goat Simulator 3, and Valheim, to teach it the basics of playing these titles.

From its training to play video games, the AI currently has about 600 basic skills, such as turning left, climbing a ladder, and opening the menu to use a map, DeepMind noted.

“Learning to play even one video game is a technical feat for an AI system, but learning to follow instructions in a variety of game settings could unlock more helpful AI agents for any environment,” the company noted.

Eventually, with more training, Sima could be instructed to do more complex functions in games, the AI lab said.

Researchers hope to translate the capabilities of advanced AI models like Sima into useful, real-world actions.

“We hope that Sima and other agent research can use video games as sandboxes to better understand how AI systems may become more helpful,” they said.