LQET：人机反馈回路中的来源感知信念更新方法
Provenance-Aware Belief Updating in Human-AI Feedback Loops
AI 导读
论文提出 LQET（Latent Quotient Evidence Training），一种结合稀疏贝叶斯监督与稳定性约束的关系型训练目标，让语言模型依据证据来源而非报告数量进行信念更新。
正文 · 原文
Abstract:In repeated human-AI interaction, a single observation can be quoted, summarized, and returned to a model as multiple reports, making repetition look like independent corroboration. We study whether language models can use supplied report lineage to update according to evidence provenance rather than report count. We introduce Latent Quotient Evidence Training (LQET), a relational objective that combines sparse Bayesian supervision with constraints that encourage stable predictions when reports add no new evidence and consistent updates to the same new evidence across relay histories. On a controlled binary benchmark, LQET reduces new-evidence update RMSE by 41-44\% and log-odds drift after eight relay steps by 89-91\% relative to task-only training across Qwen3-4B and Phi-4-mini, with three paired training seeds per backbone. The gains extend to a closed-loop simulation in which model confidence controls how many relayed reports re-enter the model's context in later rounds. By grounding belief updates in supplied provenance, LQET offers a route toward dynamic human-AI systems in which influence follows new evidence, not the number of times the same evidence returns.
| Comments: | Accepted Oral at the NeurIPS 2026 Workshop on Dynamic Alignment in Human-AI Coupled Systems |
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.13008 [cs.AI] |
| (or arXiv:2610.13008v2 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.13008 arXiv-issued DOI via DataCite |
Submission history
From: Tianyi Huang [view email]
[v1]
Fri, 9 Oct 2026 05:09:01 UTC (1,267 KB)
[v2]
Tue, 27 Oct 2026 01:31:39 UTC (1,641 KB)
来源：arXiv cs.AI · arxiv.org