Whose Memory Is It? Scope-Aware Commit Rules for Long-Term LLM Memory
Persistent memory allows an LLM agent to carry experience across conversations, but it also turns a local reasoning mistake into a durable one.
ProofPaper ↗
Key points
- If memory retains only the resulting sentences, those once-useful possibilities can later return as facts.
- We identify this missing context as discourse ownership: the world, branch, or speaker that licenses a proposition.
- We introduce CASK (Causally Anchored Scoping Keys), a commit rule that preserves this signal so that shared-world facts enter durable memory while provisional content remains available only within its original scope.
- Controlled long-conversation conflicts and tool-agent traces show that this design improves memory admission and prevents provisional content from contaminating later answers while complementing runtime provenance.
Sources (1)
- [1]Whose Memory Is It? Scope-Aware Commit Rules for Long-Term LLM MemoryarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 6, 07:05 PM
Persistent memory allows an LLM agent to carry experience across conversations, but it also turns a local reasoning mistake into a durable one.
If memory retains only the resulting sentences, those once-useful possibilities can later return as facts.
Extractive summary: sentences quoted from the sources.