Technology
Leave a comment

Forget videos. Google Genie lets users generate playable images resembling video games

Forget videos. Google Genie lets users generate playable images resembling video games


Generative AI developers at Google DeepMind have taken on the world’s highest grossing entertainment industry by introducing something called “Genie.” The new artificial intelligence system from Google researchers can create playable worlds resembling video games from images and text prompts. Announced in a paper Friday, Genie is not available to the public yet.

SEE ALSO:

What was Sora trained on? Creatives demand answers.

Similar to Sora, Genie’s creators call it a “world model,” but unlike Sora, it’s an “actionable-controllable world model.” Trained on over 200,000 hours of publicly available 2D platformers that exist across the internet, Genie can take prompts, sketches, and other images and create virtual worlds for users to play around in — creating assets from scratch and generating pixels based on player actions.

The most impressive thing about Genie is the AI’s understanding of physics through hundreds of hours of unsupervised training. This allows Genie to understand differing layers of game mechanics like player control, actions, and movement. Outside of developing 2D platformers, there could be potential use in the field of robotics to help train robots on how to navigate environments.

While nowhere near the high image quality and polished presentation of Sora, Genie is an impressive generative AI. It also prompts questions about the future job security of video game developers and concerns for game marketplaces like Steam, already overwhelmed with spam games and asset flips.





Source link

Leave a Reply