The concept of AI-driven game development is a [[link]] largely untested, pie in the sky dream mostly supported by people who don't actually make games, including Diablo fan Elon Musk, but that hasn't stopped PlayStation co-CEO Hermen Hulst from predicting that it will become the norm.
While AI has been shown to [[link]] be very good at plagiarism, there's a wild gulf between stealing voices or mangling art and actually making a functioning, competently designed game with any innovative qualities, but that won't stop 'ideas guys' or executives from pushing a future where publishers are able to cut out the creatives.
Speaking with the BBC for the PlayStation's 30th anniversary, Hulst's prediction attempts to placate both sides—the people extremely worried about AI's impact on creative industries, and the folks selling the dream of an AI future—but it sounds just as hard to swallow as the predictions from executives who've been fully AI-pilled.
"I suspect there will be a dual demand in gaming: one for AI-driven innovative experiences and another for handcrafted, thoughtful content," he told the Beeb.
But aside from the folks who actually have an investment in AI, who's actually going to demand an AI-driven game? Even among the players who don't see AI as an existential threat to the people working in the industry, or who simply don't care, there's no tangible benefit for them. And given the current limits of AI, why would they ever pick a game created without the human touch over one created by teams that have some kind of creative intent? Aside from morbid curiosity, anyway.
There's arguably one advantage, though: specifically, in games that aren't just driven by AI at the development stage, but continue to use AI to evolve and adapt to the whims of their players. We don't even need to guess at what these games would look like, because Oasis already exists. This AI game prototype is literally just Minecraft, but a version that uses AI to adapt to your inputs, and which can do things like spit out a new map based on an image you upload.
It's wonky, performs terribly, and while it serves as an interesting and at times genuinely impressive tech demo, it once again showcases AI's inability to actually create anything new. "Imagine what AI experiences could look like if everyone had the power to create them," AI company Decart teased, right after it just showed the world a worse version of Minecraft.
The future Decart envisions is one where you can [[link]] just say you want something to happen, and the AI engine will make it happen, but it's not going to create these things from scratch: it has to pinch it from somewhere, and for obvious copyright reasons this library would need to be limited to pre-approved assets or data shared by companies or individuals who are down for having their creations used in this fashion.
Oasis also doesn't really give us a convincing answer to the question of how games made like this will be able to provide more than just fleeting diversions. Even big sandboxes full of procedural elements like Minecraft are designed by teams of creative humans who make things with specific goals in mind. So even in games where you set your own objectives, they're fun because very talented people worked long and hard to create bespoke mechanics, art, and toys for you to play around with. While being able to just decide "I want something to happen" might sound like an exciting prospect, there's no game there. It's just a novel sideshow we'll likely get bored of quickly before we go back to more curated experiences crafted by developers.
But none of this is to say that AI does not have the potential to provide a lot of value to games. Hulst follows his prediction up with a more reasonable thought: "Striking the right balance between leveraging AI and preserving the human touch will be crucial." The critical part is the choice of the word "leveraging".
There's obviously a place for AI in game development—a lot of bottlenecks and graft that could be lessened, freeing up developers to invest more time in polish and spending less time crunching. A tool that assists rather than taking over. But there's a vast difference between that and "AI-driven innovative experiences".
The simple fact that plenty of influential people are pushing for it means that we will likely start to see AI-created games cropping up from major publishers eventually, but the prediction that there will be any demand for it, or that these games will actually be able to innovate rather than simply plagiarising, seems incredibly premature.