[flagged]
This is discussed right in the article.
> For Kanaris-Sotiriou, the question of adopting the use of gen AI to make games was an easy one to answer. “The foundations that it’s built upon, the idea of using other people’s work without permission to generate artwork [...] are unfair,” he says.
I personally think using AI assistance for the code is much less intrusive than using AI for the art and music -- the code isn't as directly experienced by the player as the art.
In this atmosphere, I think it's easy to perceive an implied rejection of and threat to AI generated code, even if the focus is on art assets, because people aren't being entirely direct and forthright about exactly what it is they're upset about, and that makes for a landmine field.
Edit: not a full explanation, but https://www.mikechambers.com/blog/post/2025-09-24-generative... ; this is subtly different. It's a claim that the model will not create infringing output, but that's not the same as "this model was trained only on content which was licensed for the purpose of AI training".
(there's also a discussion of the idea that the output of a model may not be copyrightable at all, which will cause a whole second set of problems for commercial users)
Adobe Firefly models are trained on a dataset of licensed content, such as Adobe Stock, and public domain content where copyright has expired. Adobe Stock content is covered under a separate license agreement, and Adobe compensates contributors for the use of that content.
We do not mine the web or video hosting sites for content. We only train on content where we have rights or permission to do so.
If you can actually write stories or create art, you can see the “seams” in generative content and it gets to be quite nauseating. The fact it was trained on your own output by a trillion-dollar megacorp via theft while you scrape money for rent is the injury to the former’s insult.
Now, as for "seams" in generated out: insofar those seem are visible to the general public and not only those with artistic talent of their own, the seams are reassuring to artists concerned about tge future commercial value of their talents. But insofar as those seams are only apparent to the artistically trained, that concerns artists because if the buyers of art won't necessarily perceive it.
"Normal" price for a AAA game is more like $60, and Arc is 40.
Sure, indie/2D can be had for less (like Factorio or Silksong), but I would not expect an <$40 price tag for a 3D game like that.
Helldivers 2 which services the same niche goes for the same price.
I bet no one listens to the "AI" voices, they have the game muted and chat on Discord with non AI generated humans...
I can't remember the last time I cared about voice lines in Quake or Unreal Tournament or any other multiplayer shooter.
It's not an RPG or a rich-story genre game, so who cares.
On the other hand, lots of AI-generated VO is very easy to spot, and sounds awful. It stands to reason it could meaningfully take away from even a completely plot-free game. If I were a voice actor, I'd feel insulted that anyone would find it comparable to my work.
Other voices cloned with the same tech are usually much worse. There's something about the nature of Dagoth Ur's voice in particular that makes it work well.
If there are only a few sentences of dialogue, how can you know it's "accurate?" "Plausible" is the more appropriate word here. The AI is filling in the blanks left by a limited sample set.
I'd imagine that's why it "works well" - you don't have a complete reference point to compare to. Ex: in your mind, what does Dagoth Ur sound like when he says "awww what a cute puppy!" Now, what does the AI sound like? Is there any reason to say one interpretation is more correct than the other?
You can make it to Dagoth Ur in about 4 minutes :)
Now I'm cleaning up the whole map as a leisurely pace. Now and then not every week. I might be done by 2028-29.