Microsoft presented AI model capable of replacing game engine. Meet Muse
Microsoft boasted an AI model that can generate the game's graphics in real-time while reacting to the player's interactions.
Microsoft has been heavily investing in artificial intelligence for several years now. Up until now, it has focused on business and practical uses, but the company also aims to apply this technology in gaming. The latest manifestation of this is the newly revealed model of generative AI.
The model is called Muse and it is designed to generate real-time game graphics, as well as control characters.
- Muse was developed by the Game Intelligence and Teachable AI Experiences groups, which are part of the Microsoft Research department. The people working there do not have much experience in game development, so they enlisted the help of Ninja Theory, the dev of the Hellblade series.
- The choice of Ninja Theory isn't surprising, as the model was trained on the shooter game Bleeding Edge, released by this studio in 2020. Before the game begins, players need to consent to match recording, allowing Microsoft Research to access a vast amount of data to "feed" their AI model.
The solutions used are similar to those from last year's experiment, where AI was employed to generate the first Doom live.
- Muse examined more than a billion images and logged interactions with the controller, which altogether equate to seven years of gameplay (if a single player were responsible for everything), and through this, the model learned to create graphics by predicting what should appear on the screen.
- Muse should be able to maintain the consistency of the generated graphics for several minutes. It's hard to tell how well it turned out, as Microsoft only provided videos in 300x180 resolution, making it practically impossible to spot typical AI distortions.
However, the mere fact that the longest of these materials last for two minutes is impressive.
For now, it's just an experiment
Of course, it's important to keep in mind that this is merely an intriguing experiment and not something that will be implemented in games anytime soon.
Traditional graphic generation will be much more practical in a long time. Especially since, to train this type of AI, you first need to build the entire game and then record years of gameplay. Therefore, there is no question of any savings here, which, after all, is the main argument for using generative artificial intelligence in game development.
However, it must be admitted that as an experiment, this is a significant achievement. Microsoft has made some of the Muse elements available as open source through Azure AI Foundry, so we are curious about what other developers will be able to do with this technology.
