Source record

Published by arXiv cs.HC (T1 source). The original is at https://arxiv.org/abs/2610.12666.

Pipeline notes

The summary and note below are generated by the signal pipeline — they are Beyond Desk’s reading, not quotations from the source.

iRAG is a human-in-the-loop decision-support system combining Retrieval-Augmented Generation with Hybrid Decision-Making. Its knowledge base starts empty and is populated through human decisions stored as retrievable precedents. The system progresses from silent observation to evidence-grounded contestation to autonomous handling with abstention. Experiments show reduced decision error under 40% knowledge drift, with a stability–plasticity trade-off.

ArXiv paper with a concrete RAG architecture for evolving knowledge bases and quantified drift results. Worth flagging because the authors openly note human usability is still unevaluated.

Cited by

No citations on record.