Menu

MENU SELECTION

MENU SELECTION

MENU SELECTION

Google’s game playing AI: Your next gaming buddy!

Google is working on a DeepMind-based AI that could revolutionize your gaming experience by becoming the ultimate virtual co-op partner. 

Unlike traditional play-to-win opponents or super-bots, this AI, known as Scalable Instructable Multiworld Agent (SIMA), is designed to be a general, instructible game-playing companion.

In a recent research article shared by Google (via @rowancheung), SIMA was introduced as a learning AI capable of following verbal instructions and understanding the virtual world it inhabits. Rather than acting as a rigidly programmed AI opponent, SIMA aims to emulate a more natural and human-like gaming companion.

Google collaborated with eight game studios to test the SIMA model in titles like Valheim, Goat Simulator 3, and No Man’s Sky—open-world games chosen to teach SIMA general gaming skills. The current iteration of SIMA can perform approximately 600 basic skills, including navigation,
object interaction, and menu use.

As a player of No Man’s Sky, I’m particularly intrigued by the potential applications of SIMA. Resource gathering and construction can often feel tedious, but having an AI companion capable of completing tasks like finding oxygen or building a base could significantly reduce the grind. This could free players to focus on exploration and more enjoyable aspects of the game.

Early results from Google’s research show promise, with a trained SIMA agent performing nearly as well in an unseen game as an agent specifically trained for it. This versatility is crucial for real gamers, as an AI companion needs to be capable from the outset of a game to be truly useful.

While the advancements in AI gaming technology are impressive, some skepticism remains. While AI models excel at knowledge-based tasks,
their ability to react and adapt in real-time like a human is still limited. 

However, the potential impact of a well-implemented SIMA-like AI on general gameplay could be substantial, offering a new level of interaction without requiring access to source code or an API.

Google emphasizes that its research is still in the early stages, with ongoing learning from various games expected to make SIMA more versatile and adaptable over time. Could your next co-op companion be an AI—and a genuinely helpful one at that? The future of multiplayer gaming could be shaped by this exciting technology, and I’m eager to see how it develops.

Subscribe
Notify of
guest
0 Comments
Inline Feedbacks
View all comments
Latest News
3 weeks ago
3 weeks ago
3 weeks ago
3 weeks ago

COMING SOON

Trailers & Teasers

Most read

1.
2.
3.
4.
ADVERTISEMENT

Google’s game playing AI: Your next gaming buddy!

Google is working on a DeepMind-based AI that could revolutionize your gaming experience by becoming the ultimate virtual co-op partner. 

Unlike traditional play-to-win opponents or super-bots, this AI, known as Scalable Instructable Multiworld Agent (SIMA), is designed to be a general, instructible game-playing companion.

In a recent research article shared by Google (via @rowancheung), SIMA was introduced as a learning AI capable of following verbal instructions and understanding the virtual world it inhabits. Rather than acting as a rigidly programmed AI opponent, SIMA aims to emulate a more natural and human-like gaming companion.

Google collaborated with eight game studios to test the SIMA model in titles like Valheim, Goat Simulator 3, and No Man’s Sky—open-world games chosen to teach SIMA general gaming skills. The current iteration of SIMA can perform approximately 600 basic skills, including navigation,
object interaction, and menu use.

As a player of No Man’s Sky, I’m particularly intrigued by the potential applications of SIMA. Resource gathering and construction can often feel tedious, but having an AI companion capable of completing tasks like finding oxygen or building a base could significantly reduce the grind. This could free players to focus on exploration and more enjoyable aspects of the game.

Early results from Google’s research show promise, with a trained SIMA agent performing nearly as well in an unseen game as an agent specifically trained for it. This versatility is crucial for real gamers, as an AI companion needs to be capable from the outset of a game to be truly useful.

While the advancements in AI gaming technology are impressive, some skepticism remains. While AI models excel at knowledge-based tasks,
their ability to react and adapt in real-time like a human is still limited. 

However, the potential impact of a well-implemented SIMA-like AI on general gameplay could be substantial, offering a new level of interaction without requiring access to source code or an API.

Google emphasizes that its research is still in the early stages, with ongoing learning from various games expected to make SIMA more versatile and adaptable over time. Could your next co-op companion be an AI—and a genuinely helpful one at that? The future of multiplayer gaming could be shaped by this exciting technology, and I’m eager to see how it develops.

Subscribe
Notify of
guest
0 Comments
Inline Feedbacks
View all comments
Latest News
3 weeks ago
3 weeks ago
3 weeks ago
3 weeks ago

COMING SOON

Trailers & Teasers

Most read

1.
2.
3.
4.