This week, Microsoft unveiled its latest venture into generative AI technology, and it's getting a mixed reaction, you could say.

Called Muse, the tech is a first-of-its-kind AI model that's able to replicate a game based upon video footage and player control inputs.

Microsoft's AI research teams are working closely with Ninja Theory on Muse, and have trained the AI on seven years' worth of footage and player data from Bleeding Edge, an Xbox One online multiplayer action title.

Based on all this data, Muse is able to reproduce the game environment and player actions.

The technology is still in its infancy, but with time, it seems Microsoft is optimistic that Muse can be used as a tool to help aid video game development.

Dr. Michael Cook, an AI researcher and game designer, lays it all out in plain English in a blog post. Aiming to dispel the assumption that Muse will be used to generate original games, Cook says it's more about helping a developer understand how players might react if changes are made to an environment.

"They made a tool that let game developers edit a game level using existing game concepts like adding in a jump pad to a place where there wasn't one before," Cook explains. "They then gave this new level to their model, and asked it to show what it thought the footage of a player playing from this new position would look like."

With all this understood, Phil Spencer, in the above video, talks about a potential use case for Muse in the future, whereby it could be fed footage of old games in order to replicate them on modern hardware, without needing the original software or engine.

Cook calls Spencer's comments "idiotic".

He says that it's difficult to understand what AI models do and don't capture; in the example of Bleeding Edge, the years of footage Muse has been trained on still won't contain everything found in the game, or all possible player actions. In other words, an AI reconstructing an entire game based on video and gameplay data couldn't ever be truly accurate.

"Even if this model was a perfect replication of the original executable software, this is not the be all and end all of game preservation," Cook says. "A generative model of what game footage maybe looked like once might be a nice curio on the side of a real preservation process, but it is always going to be inferior to other ways we approach the problem."

The truth is that it's almost certainly far too early to draw any real conclusions from what Microsoft has presented so far. Muse may eventually have practical uses within game development, but it may take some time before it gets there.

And, based on what Cook and other industry people are saying about its use for preservation, that seems like it might be a stretch regardless.

What do you make of Microsoft's Muse AI tech? Discuss in the comments section below.

Please note that some external links on this page are affiliate links, which means if you click them and make a purchase we may receive a small percentage of the sale. Please read our FTC Disclosure for more information.

[source microsoft.com, via youtube.com, youtube.com, possibilityspace.org, eurogamer.net, bsky.app]