The challenge of continuity in AI-written serialized fiction
A user argues that limited context windows make chatbots lose track of long-running stories. They say Sherpa uses separate agents for writing tasks and a “Narrative World Model” to track story history.
TLDR
A user argues that answering one prompt at a time and working within limited context windows make general chatbots poorly suited to serialized fiction. They say Sherpa uses separate agents for ideas, structure, research, prose and review, while a “Narrative World Model” tracks the story’s history. The user argues that serialized stories need to keep track of who is alive, what characters know and which places exist.
Combined views
2.8K
1 Source, first seen 8h ago