PulseAugur
中
实时 09:37:23
English(EN) Language as Trait, Politeness as Strategy: Controlled Tests of Linguistic Effects in LLM Social-Deduction Agents

LLM的礼貌和语言复杂度不会影响社交推理游戏中的代理信任度

一项新研究发布在arXiv上,调查了大型语言模型(LLM)代理的语言变化如何影响它们的社交互动。研究人员测试了语言复杂度或礼貌策略的改变是否会影响其他代理对发言者信息的判断或行为。实验在LURK Werewolf框架内进行,只重写了信息的自然语言实现,同时保留了核心游戏内容。在大量游戏和重玩中,语言范围或礼貌策略均未显著影响其他代理对发言者的信任度,在统计校正后也未影响投票或夜间行动。 AI

影响 表明当前的LLM代理在社交推理场景中可能不像人类玩家那样容易受到细微的语言操纵。

排序理由 该集群包含一篇发布在arXiv上的研究论文,详细介绍了LLM行为的实验结果。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.CL 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。 我们如何撰写摘要 →

LLM的礼貌和语言复杂度不会影响社交推理游戏中的代理信任度

本文如何被排名

Signal score
16 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
该集群包含一篇发布在arXiv上的研究论文,详细介绍了LLM行为的实验结果。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
paper, safety
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准。

报道来源 [1]

  1. arXiv cs.CL TIER_1 English(EN) · Yuchen Li, Lisa Liu, Runqi Zou, Shuxuan Kuang, Qingquan Zhang, Julian Togelius ·

    语言作为特质,礼貌作为策略:LLM社交推理代理的语言效应控制测试

    arXiv:2611.09768v1 Announce Type: new Abstract: Large language model (LLM) agents coordinate through conversation, and the same message can be phrased in simpler or more complex language, bluntly or politely. We test whether such differences change how other agents judge and act,…