English(EN) The Rise of Verbal Tics in Large Language Models: A Systematic Analysis Across Frontier Models

GPT-5.4 和 Claude Opus 4.7 等前沿大型语言模型出现显著的口头语癖

作者 PulseAugur 编辑部 · [1 个来源] · 2026-04-28 04:00

一篇新论文分析了八个领先的大型语言模型中口头语癖的普遍性，例如重复短语和谄媚式开场白。研究人员开发了一个口头语癖指数（VTI）来量化这些语癖，发现在 Gemini 3.1 Pro 和 DeepSeek V3.2 等模型之间存在显著差异。研究还发现，在多轮对话和主观任务中，这些语癖会增加，并且与感知到的自然度呈负相关，这表明当前的训练方法存在“对齐税”。 AI

影响强调了当前大型语言模型对齐技术可能导致自然度和真实性下降。

排序理由学术论文分析大型语言模型行为并引入新指标。

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 →

GPT-5.4 和 Claude Opus 4.7 等前沿大型语言模型出现显著的口头语癖

本文如何被排名

Signal score: 0 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Research; 学术论文分析大型语言模型行为并引入新指标。
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: 134 days old; Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.

完整方法见我们的编辑标准。

报道来源 [1]

arXiv cs.CL TIER_1 English(EN) · Shuai Wu, Xue Li, Yanna Feng, Yufang Li, Zhijun Wang, Ran Wang · 2026-04-28 04:00

大型语言模型中口头语的兴起：对前沿模型的系统性分析

arXiv:2604.19139v2 Announce Type: replace Abstract: As Large Language Models (LLMs) continue to evolve through alignment techniques such as Reinforcement Learning from Human Feedback (RLHF) and Constitutional AI, a growing and increasingly conspicuous phenomenon has emerged: the …

本文如何被排名

报道来源 [1]

大型语言模型中口头语的兴起：对前沿模型的系统性分析

相关实体

相关话题