For me, if I want to go for breadth (quantity), AI is fine. For depth (quality), I have to come up with it or commission someone really skilled at their work.
I take Elder Scrolls II: Daggerfall as an example. The massive scale of its world map was ultimately procedurally-generated (i.e. AI). The locations important to the main quest was hand-crafted and generally of better quality.
I pretty much only world build in TTRPG’s. As a result I use AI a lot in world building, for everything from landmasses, to city layouts, to individual NPC’s. Because by the time I get down to the details of the story, the players choices have shaped the story being told, and I no longer have to world build, I just react as the NPC’s.
deleted by creator
I think I’ll use AI to keep track of the lore accurately, if I were to ever write a story
Unfortunately, in my experience, AI is bad at remembering stuff like that. It’s not a database, it’s just a text generator.
There is a difference between AI in general and LLM’s, a subset of AI. Text generators like ChatGPT are the latter.
Grammar checking my properly written-up Wiki entries from notes of my Obsidian vault. Or finding names and acronyms, but I prefer to do that myself if possible.
To get a scale like that, I would just leave it up to extrapolation of the fundamental traits of your world. If my world was scorched, but I only write about a couple cities, you can imagine everything inbetween is a fiery hellscape too.
It’s your universe, not AI’s nor a professional’s. Try to do as much as you can by yourself, if the world your building fits with you as a person, this should be easier than it sounds.
P.S - Proc-gen is not necesserily AI, but an algorithm. A set of rules to follow with given aguments for variation. This is why games such as No Man’s Sky, that heavily use proc-gen, end up with repeating landscapes or planets. The algorithm was given similar arguments. AI is far more complex however, closer to how a biological brain determines something.
The value of art is in the choices you’ve made and why. It’s important to keep this in mind with AI; how you choose to use it should mesh with your vision. Knowing your own strengths and weaknesses is hard enough, but now we must also know the AI’s, if it can truly help us.
For a criminal organization operating a towpath racket, an LLM once proved interesting in generating a rudimentary business plan for it and speculating various knock on effects of the racket. I had to give it a great deal of context and carefully choose which ideas to use, but if you make its outputs your own thoroughly enough, it can be a useful tool.
