Google DeepMind announces SIMA, a general, game-playing AI agent that can carry out tasks like a human

Google DeepMind introduces SIMA, it's AI agent which could play video games just like humans. Know more about Google’s new AI innovation.

Meet SIMA, your new video game partner who can play just like humans. (Google )

Have you ever thought of playing video games with an AI system? It sounds futuristic right? The kind that we see in Sci-Fi movies. Well, Google DeepMind has taken AI advancement to the next level by introducing SIMA, an AI agent which would be able to learn and adapt to gaming skills from different environments. SIMA stands for Scalable, Instructable, Multiworld Agent and the concept is now being researched. The AI agent has the ability to play games like humans with the help of natural language instruction and image recognition. Know what all Google DeepMind's SIMA is about.

What is SIMA?

According to Google's blog post, SIMA is an AI agent which has the ability to learn gaming skills through natural language instruction and image recognition. It can analyze different types of environments to prepare itself for taking action based on the instructed goal. This AI agent will be trained to learn gaming skills to play just like humans. Therefore, it could act as your video game partner, but it will not act like normal gaming bots who do their own things. Currently, SIMA has learned about 600 basic skills including navigation, object interaction, driving a car, etc. Google said, “This work isn't about achieving high game scores. 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.”

How Google DeepMind is training SIMA

Google will be partnering with game developers to give SIMA the platform to train in different games and settings. So far the company has collaborated with eight game studios to test SIMA's abilities. The AI agent was put to the test in 9 different games including No Man's Sky by Hello Games and Teardown by Tuxedo Labs. Playing different games exposes the AI agent to different gaming setups and a variety of skills to learn. With each interactive world, it gains basic skills such as finding resources, navigation, flying a spaceship, etc.

To help SIMA learn skills and act spontaneously, the company worked with four research environments including a new environment in the Unity engine where the agent was tasked to develop a sculpture to test its understanding. Google also tested SIMA with two human players as one played the game and the other provided instructions to SIMA. This was done to enhance the agent's prediction skills.

Google further highlighted that with advanced AI models, SIMA would be able to perform complex gaming tasks which could perform like humans and help to ensure victory in games.

One more thing! We are now on WhatsApp Channels! Follow us there so you never miss any updates from the world of technology. ‎To follow the HT Tech channel on WhatsApp, click here to join now!