Engine failure
Can game makers put Genie back in the bottle?
The SuperJoost Playlist is a weekly take on gaming, tech, and entertainment by business professor and author Joost van Dreunen.
Middleware is sexy as hell, is what I always say.
Often, the unsung toolset that makes it possible to build the universes in which we play is the software engine, the foundational codebase that makes up the interactive world. New features expand creatives’ ability to dream up novel experiences. It is fundamental, yet the least understood category. That’s the lesson I take from this week’s stock drop.
On Friday, Google’s release of Project Genie, a prompt-based AI solution that generates explorable 3D environments, triggered the sell-off of several prominent game makers’ shares.
That’s entirely too soon, as I don’t expect people to start spinning up their virtual universes quite yet for a few reasons.
First, established publishers have been at this for a while, and even their worlds lose their appeal. It’s no secret that Ubisoft, which has been making all manner of more-or-less historically accurate environments, has been struggling financially. Here’s a legacy game maker with strong IP and more French people with art degrees than you can throw a baguette at, yet it’s having a terrible time innovating. That’s too bad, of course, and evidence of why the idea of instant open-world games is going to be difficult to realize.
Second, creating compelling game worlds is harder than it seems—much like how AI can’t simply prompt its way to the next great American novel. Developers like Rockstar spend such attention to detail that their games, like Red Dead Redemption II and Grand Theft Auto 5, hold up extremely well, even a decade later. Part of the secret sauce is building proprietary engine software. It is this combination of in-house tools and incredible talent that makes it possible.
And even then, it’s not a given. Ubisoft employs 17,000 people who do this for a living, and even they struggle to consistently generate universes that captivate players. The creative vision required isn’t something you can commoditize through better prompts.
Third, Genie is currently only available to Google AI Ultra in the US, which costs you $250 a month. While I’m enticed by the idea of creating my own private playable environments, I’m going to forfeit the opportunity and have professionals make me something instead. For that kind of money, I can buy three copies of GTA6 and a sandwich.
Navigating these AI-generated worlds also happens at a measly 24 frames per second. That makes for great prototyping and clever demos. But a fully immersive interactive experience it is not.
The immediate counter-argument is, of course, that prices will come down, and that in a year or five, from now, we’ll all be quietly snickering at home in our own custom adventures.
I think it’s unlikely.
People enjoy agency but only for the fun parts. Ask yourself how much time you’re willing to spend deciding what to watch on streaming video. Between a dozen providers and hundreds of shows, audiences are often exasperated within minutes of having to navigate, decide, and commit.
Remember when sneaker companies made it possible for people to custom-design shoes on the internet? Yeah, that went nowhere. People like it when professional creatives make something marvelous for them. You are the wizards. Show me some magic!
The real challenge, however, is economic.
