AI is coming for the games industry again, this time with grand claims about generating whole interactive 3D environments on demand.
Google DeepMind and Fei-Fei Li’s $1bn start-up, World Labs, are pushing “world models”, systems built to navigate and recreate the physical world in software.
DeepMind, co-lead of Genie 3, Shlomi Fruchter said: “Creating software and games in particular is changing a lot, and I expect it to change, maybe entirely, over the next few years. This will empower creators and developers to build things faster, better, and in ways that weren’t done before. I don’t think it will replace the existing experience, but we will see more types of experiences that are not available today.”
AI outfits, including Elon Musk’s xAI and Nvidia, want similar tech inside robots and autonomous vehicles, though the quickest wins look aimed at games.
That matters because gaming is still a serious money machine, with Newzoo expecting almost $190bn in revenues this year, and studios always hunting for ways to cut production pain.
Generative AI is already used for visual assets, from landscapes to characters, and some studios are treating it as part of the standard toolchain.
In May, Epic Games and Disney introduced an AI-powered Darth Vader in Fortnite, built with Google and ElevenLabs, as an interactive non-player character.
GDEV Holding, chief executive Alexander Vaschenko, said AI has quadrupled development speed on titles such as Aliens vs Zombies: Invasion. “Based on my professional experience, I firmly believe that both the video game and film industries will soon be unable to function without AI."
World Labs launched its Marble world model last month, while Runway, which works with games studios, launched its first world model in December, and the pitch is faster creation from text prompts.
World Labs, founder Fei-Fei Li, told The Financial Times the tech will hit engines such as Unity and Epic’s Unreal. “This is all up for disruption. Simulation gaming engines are due for improvements.”
The sales pitch goes further, with AI experts claiming players will build worlds themselves, and studios will need less expensive software and fewer specialist skills to generate content.
Mohamed bin Zayed University for Artificial Intelligence, president Eric Xing said: “Now a gamer in front of this world model can put themselves into a virtual world. That makes the game industry very different from today, because producing a personalised game is now a straightforward process.”
Critics are not buying the utopia, warning that more AI means fewer artists, more layoffs and visuals buried under “slop” that looks cheap because it is.
Six European video games unions condemned the growing use of AI this month, saying the tools were “being forced upon us, even though they degrade our working conditions.”
Optimists say it could ease costs, lift creativity and reduce burnout, which sounds lovely in an industry where triple-A games can take years and cost more than $1bn.
DeepMind, associate producer Alexandre Moufarek said: “Often, that’s the time that’s missing at the end of the production. Christmas is coming, and you need to release the game, and you don’t have time to polish the things that you wanted or debug things correctly. The more we put those models in the hands of creatives, I’m sure we are going to discover new ways of working that we haven’t even anticipated yet.”
