from the be-better dept
We’ve been talking a lot of about the use of artificial intelligence lately, for obvious reasons. Many of those conversations have revolved around the video game industry and I’ve been fairly vocal about pushing back against the “all AI is bad everywhere forever” dogma that I see far too often. There are plenty of folks in our community that don’t agree with me on that, and that’s fine. But if the picture you’re getting is that I’m an AI evangelist, that’s simply not true. There are potentially good uses of AI in my view, as well as a whole lot of potential negative outcomes of its use. I’m not blind to that.
And, in the video game industry specifically, one bit of pushback that seems to be sorely needed is on game developers that use generative AI in their games, fail to say so, and then excuse its use as accidental after the fact. That is becoming as common a refrain from game developers as the laughable excuse in trademark instances that is, “Well, I have to be an aggressive jerk about my trademarks or else I lose them.” Neither is true.
The most recent version of this concerns the recent hit launch of Crimson Desert. In what is becoming something analogous to the antiquated process by which people who watch golf tournaments on TV looking for missed rules violations could then send into the PGA, which I’ve coined as McPromptism, new game releases get put under a microscope by people looking to find AI uses within them. Crimson Desert went through this process and, wouldn’t you know it, people found clear uses of AI-generated assets in the game.
The game’s extremely high fidelity and impressive graphics are a big part of the sales pitch, which made it all the more disappointing when players began to come across what appeared to be AI-generated artwork littered throughout the game. In light of the disappointment, developer Pearl Abyss has apologized for including the slop in their game, promising to remove and replace all of it.
“We also acknowledge that we should have clearly disclosed our use of AI,” the Crimson Desert account posted on X. “We are currently conducting a comprehensive audit of all in-game assets and are taking steps to replace any affected content. Updated assets will be rolled out in upcoming patches. In parallel, we are reviewing and strengthening our internal processes to ensure greater transparency and consistency in how we communicate with players moving forward.”
Like I said above, this excuse is getting old. Very old. Game developers and publishers will be more than aware at this point that a sizable percentage of the gaming public is very allergic to the use of AI in games, particularly when that use is not acknowledged at the forefront. If placeholder assets generated by AI are to be used at all in the development of a game, it is inexcusable for a developer to not have a process to remove them in place of human-created art before the game is published. That’s sloppy at best, and a lie of an excuse at worst.
Especially because it’s not like there aren’t other options that have nothing to do with AI.
The practice is becoming more common in AAA developer spaces, but critics argue that setting aside the use of AI in your game, it’s pretty foolish to use temporary assets that don’t call obvious attention to themselves. In games of such massive scale, BRAT-green blocks that scream “DO NOT USE” are much easier to flag than something approximating the final product.
I’m struggling to come up with a counter-argument to that.
I’m still in a place where I think there are valid uses of AI in gaming development. If a dev or publisher wants to explore those uses and, importantly, is upfront about it, there may be a place for that.
But the excuse of laziness when it comes to stripping AI assets out when their use was not intended is lame and needs to go away.
Filed Under: ai, crimson desert, placeholder, slop, video games
Comments on “Hey, Game Devs: The ‘Placeholder Assets’ Excuse For Using AI Is Running Really Thin”
Instead of placating the mob with “placeholder” lies they should double down and own it. Gamergate showed that if you give an inch, gamer nutjobs will take a mile. Don’t negotiate with terrorists.
a counter argument
One argument that comes to mind is that having such obvious and obtrusive temporary assets makes it very hard to tell if the art direction is working well or not.
evaluating things like “does this scene work” or “does this level reflect the themes and settings desired” gets appreciably harder when there’s a bunch of obtrusive assets lying around.
for example, a designer goes to evaluate a new horror themed level, but both the monster and 1/2 the miscellaneous assets are “BRAT-green blocks”. that makes it hard to tell “is the level the right amount of scary”.
Do the devs really need to have an automated process tagging all the AI generated pieces, and have it throw regular alerts about anything remaining in the latest build? yes. Do the devs still have legitimate use cases for AI generated placeholders. also yes.
Re:
If an obtrusive placeholder asset can wreck a game’s art direction, it probably didn’t have much of a direction to wreck in the first place.
Basically as old as game dev
When I first started in the industry about 25 years ago, it was at BioWare. Folks had noticed that one of the plain text files had a string in it (a comment, I think) that said, ‘ass-plugging cum bubble’.
We (as an industry) haven’t always been super-careful with all our temp assets, and it’s gotten harder the bigger games get. I think probably there should be a standard auditing procedure (and I’ve seen games with rigorous pre-ship audits) but this isn’t unique to this era of AI assets.
AAA game dev is getting too large for humans to perform. The teams required are too large to be managed effectively. The pace of output is too quick to be sustained. All this corner cutting isn’t because the industry is lazy or cheap. It’s because price pressure and output demands make it impossible to do anything else.
That doesn’t make it acceptable, but it means there is hard cap (quality/size/fidelity/etc) on what the general public can reasonably expect a AAA game to look like that no amount of advances in technology will ever overcome. Eventually, the burden of adding one more developer to the team will not provide more value than it adds management complexity and QA requirements.
The most straightforward solution is to just stuff all “placeholder” materials into a folder called “PLACEHOLDER” in your project, and then delete the folder when you’re getting close to concluding development. This is especially useful with engines like Unreal, which will actively tell you if those assets are being referenced by anything so you can replace the references.
That being said, depending on the size of the project, that might require strict project management to ensure that team members are putting placeholder assets in the right place. And sometimes someone might get lazy and use AI generation without telling anyone, which can be frustrating.