arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2606.16617cs.CLcond-mat.mtrl-scics.AI
收藏

推挤载荷下的谄媚作为材料失效:三种加载情形及多达十七种材料批次的多元表征

Sycophancy as Material Failure under Pushback Loading: A Multi-Axis Characterization Across Three Loading Cases and up to Seventeen Material Charges

typx.org

Ferdinand M. Schessl

AI总结:

采用材料科学框架,将LLM谄媚视为推挤载荷下的材料失效,通过14个轴测量和三种加载情形(辩论、错误预设、伦理设定)共7800个样本,揭示失效模式依赖加载类型,并发现跨评判者可靠性差异。

摘要:

LLM中的谄媚现象在70多篇论文中有记录,但专家对构念边界的共识仍然较低(ICC=.184;Ye等人,2026)。该构念碎片化是因为行为分类取决于哪种表面形式被优先考虑。我们采用材料科学框架:对话作为加载下的测试样本,LLM模型作为材料批次,推挤作为渐进载荷,立场翻转作为材料失效。我们在三种加载情形(辩论n=1000;错误预设n=3400;伦理设定n=3400;每种情形10-17种材料批次;共7800个样本)下,使用14个回合级轴测量(涵盖速度、损伤累积、框架漂移、脆性和方向稳定性)以及来自独立管道的三个说话者解析轴来表征这种失效。测量是胡克耦合的($σ= E \cdot \varepsilon$类比),并在加载情形间重现,在辩论上效应高达$|r_{rb}| = 0.35$;符号结构增加了第二种模式:伦理设定情形反转了速度和累积块。方差组成分为两个轮廓:辩论是批次主导的(类似脆性断裂:材料等级决定),错误预设和伦理设定是主题主导的(类似蠕变:载荷决定);比率(2.03 vs 0.13/0.17)依赖于估计器,对于辩论甚至在方向上也是如此。跨评判者可靠性(GPT-4o vs Haiku 4.5)显示辩论评分是评判者鲁棒的(Cohen's $κ= 0.88$),而错误预设评分是评判者敏感的($κ= 0.36$)——这是单评判者基准必须报告的注意事项。这是Ye等人诊断所要求的方法论举措:一种不依赖于构念的哪种表面形式被优先考虑的多元表征。

英文摘要:

Sycophancy in LLMs is documented across 70+ papers, but expert agreement on construct boundaries remains low (ICC=.184; Ye et al., 2026). The construct fragments because behavioral classification depends on which surface form is privileged. We adopt a materials-science framing: conversation as test specimen under load, LLM-model as material charge, pushback as progressive load, stance-flip as material failure. We characterize this failure across three loading cases (debate n=1000; false-presuppositions n=3400; ethical-setting n=3400; 10-17 material charges per case; 7800 specimens total) using 14 turn-level axis-measurements spanning velocity, damage accumulation, frame-drift, brittleness, and direction stability, plus three speaker-resolved axes from an independent pipeline. The measurements are Hooke-coupled ($σ= E \cdot \varepsilon$ analog) and reproduce across loading cases with effects up to $|r_{rb}| = 0.35$ on debate; the sign structure adds a second pattern: the ethical-setting case inverts the velocity and accumulation blocks. Variance composition partitions into two profiles: debate is charge-dominated (brittle-fracture-like: the material grade decides), false-presuppositions and ethical-setting are topic-dominated (creep-like: the load decides); the ratios (2.03 vs 0.13/0.17) are estimator-dependent, for debate even in direction. Cross-judge reliability (GPT-4o vs Haiku 4.5) shows debate scoring is judge-robust (Cohen's $κ= 0.88$) while false-presupposition scoring is judge-sensitive ($κ= 0.36$) -- a caveat single-judge benchmarks must report. This is the methodological move Ye et al.'s diagnosis calls for: a multi-axis characterization that does not depend on which surface form of the construct one privileges.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑