analiza aiqbench
This study identifies stable, latent dimensions in how users phrase advice-seeking queries that are distinct from the topic itself, revealing specific articulation styles that cause models to fail at asking clarifying questions. By treating input variation as a measurable structure rather than noise, the research provides a new framework for stratifying benchmarks and evaluating model robustness across different user interaction patterns.
How You Ask Shapes What You Get: A Theory-Seeded Measurement of Articulation in Advice-Seeking LLM Conversations
arXiv:2608.29591v1 Announce Type: new Abstract: Users articulate the same advice-seeking request in different ways: some specify detailed constraints, others gesture at a vague need. Prior work treats this variation as noise to be averaged away; we instead treat it as a stable, measurable structure in the input distribution. We ask whether articulation (how people ask) forms latent dimensions separable from topic (what they ask about), and whether it is associated with how language models respond. We extract interpretable features from 16,447 advice-seeking prompts pooled from public chat corpora (WildChat, LMSYS, and ShareChat) and recover a small set of latent articulation factors that replicate across train/test splits and across corpora. Because this structure is largely separable from topic, the populations it defines cut across topics and stay invisible to topic- or task-based evaluation. The factors define a handful of recurring articulation styles, one of which stands out: a long-form but information-poor style, roughly one in six prompts in the largest corpus, where models return shorter, vaguer answers and do not ask for clarification even though under-specification is exactly the condition that warrants it. The contrast holds within every topic group and length quintile, and is not under-specification alone -- a second, equally under-specified style does draw clarifying questions. Two independent human annotators reproduce this contrast. We argue that benchmarks should stratify on articulation, and we offer the extracted structure as a measurement instrument for doing so.
Abstract:Users articulate the same advice-seeking request in different ways: some specify detailed constraints, others gesture at a vague need. Prior work treats this variation as noise to be averaged away; we instead treat it as a stable, measurable structure in the input distribution. We ask whether articulation (how people ask) forms latent dimensions separable from topic (what they ask about), and whether it is associated with how language models respond. We extract interpretable features from 16,447 advice-seeking prompts pooled from public chat corpora (WildChat, LMSYS, and ShareChat) and recover a small set of latent articulation factors that replicate across train/test splits and across corpora. Because this structure is largely separable from topic, the populations it defines cut across topics and stay invisible to topic- or task-based evaluation. The factors define a handful of recurring articulation styles, one of which stands out: a long-form but information-poor style, roughly one in six prompts in the largest corpus, where models return shorter, vaguer answers and do not ask for clarification even though under-specification is exactly the condition that warrants it. The contrast holds within every topic group and length quintile, and is not under-specification alone -- a second, equally under-specified style does draw clarifying questions. Two independent human annotators reproduce this contrast. We argue that benchmarks should stratify on articulation, and we offer the extracted structure as a measurement instrument for doing so.
Computation and Language (cs.CL)
arXiv:2608.29591 [cs.CL]
(or arXiv:2608.29591v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2608.29591 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 325 cuvinte
Subiecte conexe
State of the Field: AI for Epistemics and Coordination
LessWrong · 2 oct. 2026
The Continual Constitutional Convention: How language models revise the documents that govern them
AI Alignment Forum · 13 oct. 2026
Self-Modeling Interventions Modulate Emergent Misalignment
LessWrong · 5 oct. 2026
Hint-based CoT faithfulness evals still mostly work on Claude
LessWrong · 30 iul. 2026