The Fundamental Question of AI in Game Development: Why?
I remember the days when games were widely considered mere kids toys at best or a corrupting influence at worse. I remember how liberating it was for both players and the industry to slowly see the medium of video games start to be seen as an art form alongside movies and literature. Just like the latter, games can deliver straightforward entertainment—or even so-called “trash”, whatever that means—but also impactful experiences enhanced by the interactivity inherent to the medium. I’ve seen games start to more and more frequently tell deep stories and be taken seriously for it. I’ve seen—in particular with the rise of indies and what we now call AA—the industry get energized by wild gameplay experimentations and developers expressing deeply personal feelings through their work.
All of this was done by humans who took their craft seriously and built on top of each other to show that games can be a way to deliver anything from silly fun to deeply affecting introspection, sometimes all at once. Humans with ideas they found fun or interesting. Humans with stories to tell.
Yes, shallow corporate products were always pushed out to market for the sole purpose of making money (I lived through the scourge of movie tie-ins), but as bad as these were and as annoyed as we got with them, at least a bunch of humans got paid to make them, and maybe even built up some skills they could later take to more worthwhile projects.
Today’s corporate shallowness seeks to eliminate the human element altogether to cut on time, expenses, and alongside it, anything and everything that slowly managed to convince all of us that games could be art back in the day. “We have to make games quicker” claims a nameless contingent of CEOs in a cycophantic GamesIndustry.biz article that went out the second the doors closed on the Gamescom business days and so didn’t have a second available to spend interrogating the claims made, such as the fact that “By far the biggest expense when making games is salaries”. As Michael Scally pointed out in response to the article on Bluesky:
How could Team Cherry afford to spend 7 years making Silksong? Hollow Knight sold a lot of copies, yes, but it certainly didn’t hurt that they didn’t have to carve out $31 million a year to pay the boss (EA CEO Andrew Wilson’s reported salary in 2025). It also helped that they never got shut down following the release of their masterpiece by a profit-driven parent company to satisfy the all-devouring needs of investors. From early 2024 to just last month, Microsoft shut down Toys for Bob, Alpha Dog Games, Arkane Austin, Tango Gameworks and The Initiative. In October 2024, Microsoft CEO Satya Nadella got a $30 million raise for a total salary of $79 million. Meanwhile, they killed the talented teams behind Prey and Hi-Fi Rush without a second thought.
In this industry where most of the money flows up to the people actively sucking the soul out of games, these same soulsuckers push for increasingly predatory corner-cutting methods promising it to be the future and asserting to the tune of billions in investments that it is inevitable. I’ve been told that a couple times this Gamescom week. That AI cannot be stopped and so, defeated before we even know it, we should simply adapt to it.
In a gold rush, the people selling shovels make the money, so everyone’s rushing to sell shovels. AI asset generation, AI localization, AI voice acting… and the question I keep asking is “Why?” I don’t believe you can be passionate about games as an art form and at the same time create and sell the tools to dismantle that art form. Why? So Satya Nadella can fire a bunch of people to increase his pay by another few millions and you make a fraction of that while the bubble hasn’t burst? And what’s the end result? A bunch of games with slop content barfed out by algorithms failing to imitate what humans used to make?
One argument I’ve seen made is that such tools can help small studios that lack the resources for large expenses such as voice acting, for example. One of the issues with this is that once you open the door, anybody can walk through it. No one can enforce the use of these tools only under a certain budget, and everyone knows for a fact that the bigger companies are the first to jump on such opportunities. In my career, the worst paying clients have always been the AAAs.
And can you really tell me international players enjoy shoddy localization or that Undertale would have been better if Sans talked with an uncanny android voice? As if one of the reasons indie gaming is so full of charm and creativity isn’t because of those studios’ limited resources and the paths they find around those.
And all this talk of inevitability is based on nothing other than the promises of tech CEOs and investors who need everyone to believe in this inevitability if they want to syphon as much money as possible while the bubble is still inflating. They manufacture the shovels everyone’s trying to sell and we take them at their word. Nevermind the fact that after decades of doomsaying, machine translation still hasn’t gotten past “decent” in most contexts and no-one has really managed to make it work properly for game localization where understanding the context of the text is crucial to proper translation. Nevermind that every actual worker who’s seen AI pushed on them by management has reported on how it actually slows and generally hinders the process. Nevermind some AI models actually getting worse as they reingest their own output after flooding the web with garbage, akin to some dystopian non-human centipede. Nevermind that ChatGPT is seeing such underwhelming results as to shake confidence across the AI industry despite years of development and literal tens of billions in investments from companies like Microsoft (remind me… How many Rs in “strawberry”?).
I don’t see an inescapable future. I see overconfidence, hubris and blind faith.
Yes, AI can be a somewhat useful supplemental tool—especially for solo devs—mostly for early prototyping and idea iteration (and yes, there are major ethical concerns even for this very specific use), but do you think these corporations are forcing it so hard and investing so much all for a mere prototype helper? No, these corporations are not looking for tools, they’re looking for weapons. Weapons to squeeze their workforce ever so tightly by forcing them to incorporate half-baked AI solutions more and more, paying them less and less until they can justify removing workers altogether. We’ve been dealing with this for years—decades even—in localization, with ever-increasing numbers of agencies and clients pushing for machine translation post-editing, which in most cases means reworking bad automated translations from scratch, but for half the pay, until some of those companies started cutting linguists out altogether, product quality be damned. And no, generative AI hasn’t made machine translation any better, its hype has only made it easier to justify for corporations.
All of this because managing humans and nurturing artistic talent is inconvenient when you only exist to serve Capital.
Thankfully, the indie devs and publishers I talked to at Gamescom expressed strong anti-AI and “let’s take our time” sentiments. They’ll work with humans and with the budgets they have, and will take the time they need to produce something they can be proud of, no matter their limitations. Bless these people who have not been infected by the need to fart out new products as quick as possible and jam them with as much “stuff” as non-humanly doable, no matter the quality of that “stuff”.
As much as I’ve been fantasizing for years about localizing West of Loathing, I’d rather they don’t translate it than botch it with AI. Similarly, games survived for decades without voice acting, and I’d prefer they keep working around that limitation rather than start hearing a bunch of generic recycled voices in all my games. And don’t get me started on ChatGPT prompts in background assets and Call of Duty AI art (remember how those tools are supposedly meant for teams with low budgets? Poor little Activision…).
What’s the projected end result of all this? Drive passionate people out of their dream jobs as writers, artists, actors, translators, etc. so that we can enjoy endless ChatGPT rip-offs of better works? So that more international audiences can know the joys of AI translation? So that AAAs can keep working with Uncanny Valley Troy Baker years after his retirement, and then years past his death? Strip it all down until we’re playing games that all look, read, sound and play the same because creativity is expensive and inconvenient?
Yes, they are manufacturing those damn shovels. Yes, there will be people selling those shovels and folks willing to buy them. You don’t have to be among them. Fuck the shovels. Fuck the gold rush. When it’s over, we’ll remember who choked us pushing shovels down our throats.
Take a look at what you’re doing and ask yourself what Jeff Goldblum would say.
