I know next to nothing about AI, so please correct me if I’m wrong, but it seems like the thought process of a dungeon master is a difficult starting point, since they’re balancing out multiple levels of considerations. They’re simulating a world, but also trying to shape a story (plus modelling the players & writing decent prose). The data would seem to be simpler to understand if you’re dealing with a pure simulationist DM, or player who’s 100% roleplayer (or munchkin), as the chains of reasoning would be focused on maximizing a single clear metric.
I ask because if so, some of those options might be also be easier to produce than a true AI Dungeon run.
I know next to nothing about AI, so please correct me if I’m wrong, but it seems like the thought process of a dungeon master is a difficult starting point, since they’re balancing out multiple levels of considerations. They’re simulating a world, but also trying to shape a story (plus modelling the players & writing decent prose). The data would seem to be simpler to understand if you’re dealing with a pure simulationist DM, or player who’s 100% roleplayer (or munchkin), as the chains of reasoning would be focused on maximizing a single clear metric.
I ask because if so, some of those options might be also be easier to produce than a true AI Dungeon run.