Signals 4 · free daily AI digest

Not One Belief: Answer Payloads, Persistence, and Dissociated Confidence Signals in Conversational Question Answering

Paper recorded by Signals 4 on 2026-10-26 in cs.CL. Abstract reproduced from arXiv; link to the original below.

Published 2026-10-26 on arXiv · recorded by Signals 4 on 2026-10-27

Category: cs.CL · 自然语言处理 · first seen 2026-10-27

Abstract

Conversational QA systems should fix errors when users push back and signal which answers to trust. We split each follow-up into its speech act (re-ask, agree, or challenge) and its payload (an answer string, correct or not), and measure four confidence signals on the same transcripts: verbalized confidence, length-normalized likelihood, a frozen hidden-state probe, and P(True). Answers change whe

Read on arXiv →

#142 most recent of 660 cs.CL papers we have recorded · ↑ newer: MiRL: Error-Targeted Reinforcement Learning for Correcting Terminal Ag · ↓ older: Adapting LLM-SC to Living Sacred Heritage: Domain-Specific Extraction,
Cite this page: Not One Belief: Answer Payloads, Persistence, and Dissociated Confidence Signals in Conversational Question Answering: the #142 most recent of 660 cs.CL papers we have recorded (as of 2026-10-26). Source: Signals 4 (Signals API) — https://data.jiangzhang.ca/signals4/t/papers/not-one-belief-answer-payloads-persistence-and-dissociated-confidence-signals-in.html
Free to quote with attribution to “Signals 4 (Signals API)”. Machine-readable: papers.json
Related: More cs.CL papers · arXiv signals · All papers · Today in AI
Get 4 AI signals a day by email — free.
Subscribe free → See all plans →
Get 4 AI signals a day by email — free
All models · All repos · By company · Daily editions