Bingo. AI is shockingly good at building simple things, helping with direct questions about items. It cannot replace humans in its current state.

At this point it's CEO bluster just like the blockchain, where the suits are talking about technology like they personally handcrafted it while the actual engineers are sitting in the back of the room thinking "uh, there's no way it can do that".

I think we're going to see a couple hilarious cycles of some shit thinks they can replace humans with AI, fail spectacularly, and then quietly go back.

You do realize it isn't staying the same, right?

There is no status quo with AI.

It's within literal months that leaps are occurring that defy most expert expectations and predictions.

While yes, creative writing is not part of the target of where models are improving right now (and there are IMO clear mistakes being made with foundational models contributing to that poor performance), we're probably less than one dev cycle from the best AI outperforming an above average video game writer with institutional integration of the models.

And really, people thinking this is going to put writers out of business are missing the true value add for publishers.

You'll see the same amount of writers as before. What will change is the amount of writing.

Being able to have a core writing team do the normal work they do of writing out main and side quests and then feeding all that writing into a model spitting out side NPC dialogue fitting in with the events taking place allows developers to make their world come alive in ways previously only accessible to the largest budgets in the industry like RDR2.

This also allows games that are successful to transition into more of a live service product without needing to have a massive audience.

For most live service games, you need as many people as possible playing to justify dedicating resources to continued development, or you need a subscription fee. But niche products with a dedicated fan base which aren't overly popular are too small to justify continuous content development.

With AI that equation changes. More games have the opportunity to keep players engaged longer for continuing adventures when a smaller team can use generative systems to flesh out the product.

Everyone praises No Man's Sky for their continued development with a team of about a dozen putting more and more content out, but the other side of the coin is that they can only successfully deliver updates that feel weighty because they are leveraging procgen to extend their efforts.

Imagine the next version of FF online where not only is there a core main story everyone experiences, but there are also individualized stories woven into it that are shaped around your interactions. Where every NPC can be spoken to and any one of them might lead to your next individualized adventure. A world that feels at once epic and shared with millions of other players while also personal and unique just for you.

Even if the individual writing wasn't as planned out as world event scenario writing from lead writers, I'd sure as hell prefer to spend $16/mo on a world with little repetition and endless adventures than a world that only has a hundred hours of story every year and is mostly running the same things over and over in between waiting for small bursts of content updates.

AI makes perfect sense for any live service provider, and Square Enix has one of the most successful live service products to date. Of course they are going to be investing into it as it rapidly improves.

There are some fundamental obstacles to that. I don't want, for instance, that a game AI does that which I tell it to do. I want to be surprised and presented with situations I haven't considered. However, LLMs replicate language and symbol patterns according to how they are trained. Their tendency is to be cliche, because cliche is the most expected outcome of any narrative situation.

There is also the matter that ultimately LLMs do not have a real understanding and opinions about the world and themes. They can give us description of trees, diffusion models can get us a picture of a tree, but they don't know what a tree is. They don't have the experiential and emotional ability to make their own mind of what a tree is and represents, they can only use and remix our words. For them to say something unique about trees, they are basically randomly trying stuff until something sticks, without no real basis of their own. We do not have true generalized AI to have this level of understanding and introspection.

I suppose that sufficiently advanced and thorough modelling might give them the appearance of these qualities... but at that point, why not just have the developers write these worlds and characters? Sure that content is much more limited than the potentially infinite LLM responses, but as you wring eternal content from an LLM, most likely you are going to end up leaving the scope of any parameters back into cliches and nonsense.

To be fair though, that depends on the type of game we are talking about. I doubt that a LLM's driven Baldur's Gate would be anywhere as good as the real thing by a long margin. But I suppose it could work for a game like Animal Crossing, where we don't mind the little characters constantly rambling catchphrases and nonsense.