Discussion AI Navigation System - Global AI or Individual AI or Hybrid AI Navigation?

The Unity AI & Navigation section seems only for NavMesh so ill ask about my AI navigation system here.

I was talking to an AI about AI navigation. I brought up a Global AI design and an Individual AI design where either examine and determine how to navigate the 3d environment. The AI said that Global AI is more efficient but less flexible or less unique and the individual AIs in this system well seem robotic or like a hive mind; while the Individual AI system is more precise or unique in how AIs will navigate the environment but that it is more costly to process the more AI there are. Then i said, what about a Global AI that examines and determines a lot of the general things about the environment, and then the individual AIs can also examine and determine some of the finer details of the environment? The idea is to have the best of both design. Now my experience with AI navigation is very limited. Iv mostly just been working with individual AIs doing their own navigation. I probably wont have much AI so the individual AI would do. But I like the idea of a Global AI establishing in general how to navigate the environment, and with individuals AIs still being able to make their own decisions on how to navigate the environment, like say an AI hates grass, so they will avoid grass as much as possible. Well i don't know if i would need a hybrid system to include that individuality even with just the Global AI system - who knows. This is sort of a continuation discussion I had with the AI but now im hoping that humans can chime in. Im in the very early stages of trying to implement the AI navigation system and could use some advice or ideas. I am not using Unity's NavMesh system or other navigation systems like A*. They seem a bit complicated or technical and im not sure if id like them. But personally id just like to build my own system from scratch. If anyone has any advice or ideas on it please let me know.

I do have an OpenSpace feature, where the AI would know how open the space they are in is or how enclosed it is, which would effect certain AIs in how they react such like as if they were claustrophobic and so on. The AI suggested a way to also examine and determine narrow spaces, like a path through rocks, or a cave entrance, or a gap under a fallen tree, etc. that an AI can fit and go through. It was interesting. I think it was separate from an AI navigation but I assume most of that would already be included by default/design in an AI navigation system. But I thought id bring that up in case anyone had any input or ideas on this Narrow Path idea the AI suggested.

The game I am prototyping this navigation system with involves fast AI navigation that should not know exactly what path to take through a maze just because they have a target in mind but at the same time should be able to make quick decisions as they discover new areas or paths to eventually get to their target. And there can be quite a lot of them going after a target at a time for different reasons. I have an Individual AI system at the moment and it works fine for the most part, but its like a really haphazard system that I put together that somehow works for now. I just built it on the go and well, im having some spaghetti code issue with it.

 

Make your post more readable, structure it better and no big blocks of text.
Make it smaller there is probably no need for such a long post, your post should be half of what is written.

I wanted to help you but this was just to much for me.

 

Its clearly a discussion if you look at the title. Its not a question asking for an answer from you otherwise iv would have kept it simple and short. Its mainly an open discussion on AI navigation. I already have the system implemented but wanted other people's ideas or experience on implementing such a system if any.