This is an IGN opinion piece from Mat Jones. He loves weird little guys in video games.
If your social feed is anything like mine, it’s been overtaken by a weird little crocodile man that’s knee-high to a Tauntaun and has an unresolved rage twelve times his size. His name is Biddy Blas, he’s one of Galactic Racer’s supporting cast, and he’s the best new thing to happen to Star Wars since Andor.
Give an AI as long a time as it needs: it couldn’t have made Biddy Blas.
With every new day, the ground underneath video games as a medium, one where professional creatives are valued and capable of negotiating an honest living, feels more and more unstable. Perfectly good studios are shutting in the name of brand restructuring. Electronic component prices have no incentive to reduce. The cost of producing anything is rising due to the precarious supply of energy and the push to end physical ownership. It’s hard to imagine what the future of the medium looks like, regardless of AI’s role to play in it.
But look, I’m not a doomer – games continue to exist in spite of these truths because of the passion of the people involved. They’re also often better than they’ve ever been. Star Wars: Galactic Racer is a highlight for the under-served genre of arcadey high speed racer, with surprisingly in-depth vehicle customisation, heat management and opponent rivalry systems.
It’s hard not to think about how good it is in the context of Epic Games’ apparent push for an "AI first" production model. Such a pivot has reportedly delayed a secret Star Wars and Marvel crossover shooter, which will apparently be developed with AI tools on both the coding and art sides of the project.
I can’t speak to the quality of a supposed game that is an indefinite time away from release, especially one that no one outside of the Epic offices has seen a frame of, but I can’t see this approach being able to emulate Galactic Racer’s secret weapon. For all the arguments of efficiency, accuracy, and speed, I don’t think a Large Language Model has the capability to create a breakout character. AI cannot make a Biddy Blas.
Blas is an icon. He’s the perfect example of what Star Wars fans have dubbed a “Glub Shitto”; not important enough to the narrative to be a main character, but so arresting in their limited screen time that they become your favorite anyway.
At this point you may be thinking, “Surely a bit of computer code could come up with an angry crocodile in a hat?” And maybe it could. Arguably the main thing that makes Blas so perfect is his ridiculous voice, and AI companies have gone to great lengths to ensure its products can do voices. But, as The Blood of Dawnwalker’s lead actor Will de Renzy-Martin put it on a recent episode of My Perfect Console, “the least interesting thing about what I do is the way my voice sounds”.
"[AI]’s not replicating my personality and the things that I do when I'm in the room, and the choices that I make, which is all the reasons that a director or a producer would want to work with me” said the actor behind Dawnwalker’s protagonist, Coen. "Actors are pretty cheap in the grand scheme of things. And if you end up with an AI that can produce the right noises, you're going to need somebody who can interact with it and get the performance that you want attached to those noises – and they're going to be much more expensive than an actor is."
To put it another way: an AI might have been able to write a nonsense Huttese catchphrase, but only Galactic Racer’s specific writers could have decided one of those sounds was “Scrib Doon”, and only Biddy Blas’ performer, Richard Summers-Calvert, could make the inspired decision to deliver the line as ridiculously as he did. And it’s that combination of creative human decisions, that living understanding of what’s genuinely funny, that turned Biddy Blas from a collection of script and polygons into an icon.
No one can prove a counter-factual, sure; I don’t see the same outpouring of love happening if we outsourced seemingly small decisions like these to the computer. As an anecdote, there’s no reality where my partner would be interested in having a Star Wars racing game hogging the home telly; the Biddy Blas game is an entirely different proposition.
Maybe Epic’s new production pipeline makes sense for a company that has spent the past decade turning Fortnite into a platform more than a game, one that increasingly eschewed creative flair and originality as a driving force (a much ruder critic would say this has never been the ambition at all). The Fortnite of today is now primarily a hub for other brands to throw their licenses at. If the priority of Epic is to focus on streamlining the production process for cramming in things which already exist, an “AI first” approach might seem obvious. If you’re making a Star Wars and Marvel shooter, there are plenty of reference materials for both of those properties to train an AI on. I would simply argue there weren’t reference materials for either… before they existed.
It’s not just Star Wars that needs Weird Little Guys created by Actual Cool People. Players are picking up Valve’s hybrid MOBA-shooter Deadlock in record numbers following its most recent patch, inspired largely by the exciting amount of creative flair displayed by the new inhabitants of the increasingly graphically-realised cursed New York City. Characters like the Rat King, who has hundreds of individually written and performed dialogue barks naming a different one of his rodent subjects for every time one of his abilities is used, or the Chess-hustler Solomon, whose weapon is a board showing historically important endgames chosen by the design team which change depending on the maximum amount of ammo in his magazine. Without people making deliberate decisions, having completely original character ideas, people would not have the same incentive to log in daily to go 0-8 in lane. (Although I hate being in games with these powerful new characters, so maybe this is actually a bad example.)
I see the continued use of AI in ideation and initial production as an existential issue for the medium that we’ll one day be course-correcting away from in the same way it would have seemed obvious to CEOs to chase the seemingly inevitable draw of live-service, online passes and on-disc DLC. For all of the efficiencies in the process, for all the promise of profitability, I’m not sure how you can actually sell any of this stuff without the unique spark of newness. I’m talking about this from a purely cynical marketing perspective: If I’m just going to get stuff I’ve already seen, it’s stuff I already have.