Part of the problem here I think relates to scale.
If I invite a load of friends over to my house for a party, they might be in different rooms having different conversations but they’re all my friends in my house. No one cares who I let in or kick out, certainly not either of the next groups.
Let’s say I’m part of the committee for the local community hall. We let our halls out to clubs. Some of the committee go to some of the clubs. I might not be interested in what it is, but if someone I trust says they are OK, I’m OK.
At the local University they have a lot of spaces, each managed by the respective school. Each school has a slightly different ethos. Some of them might let their space to groups that other schools wouldn’t, but it’s not their call. They share some resources but not decision making.
We’ve got this problem emerging. The decisions made by lemmyworld or other large instances are generally in service to their communities, whereas on smaller or more focused instances the instance level decisions are the same as community level decisions.
I don’t know how tech savvy you are, but I’m assuming since your on lemmy it’s pretty good :)
The way we’ve solved this sort of problem in the office is by using the LLM’s JSON response, and a prompt that essentially keeps a set of JSON objects alongside the actual chat response.
In the DND example, this would be a set character sheets that get returned every response but only changed when the narrative changes them. More expensive, and needing a larger context window, but reasonably effective.