Google DeepMind’s Open-Endedness Team has developed a new AI model called Genie (Generative Interactive Environments). Genie AI can create new 2D video games from texts, images, and even drawings. However, it is still in the research stage and not available to common people yet.

Genie is special and unique because it has been trained from 200,000 hours of unlabelled video footage from 2D platformer games. Other AI models work by giving instructions, but Genies can learn from watching and observing lots of video games. According to experts, this is a big step towards artificial general intelligence (AGI).

Genie AI works in three stages to create action-controllable video games for gamers. 

Here are the processes of Genie AI to create fun video games:

Video Tokenizer: Genie breaks down complex data into small parts in this stage.

Latent Action Model: It observes how things change from one frame to another in the video. It analyzes important actions in a game, like jumping, running, or interacting with objects in this step.

Dynamics Model: Genie predicts what will happen next in the game according to the current gameplay at this stage.

According to Jaspreet Bindra (the founder of TechWhisperer UK), Genie is a great tool that people can use to create 2D video games, even those who don’t know complex programming languages. He also said that Genie AI is not perfect yet, but it is going in the right direction.

Image Credit: Digit

Keep following The Gaming Newz for the latest updates and news from the gaming world!

The post Google’s Genie AI: Turning Words And Pictures Into Video Games first appeared on The Gaming Newz.