Entity · paper

It's Not What You Say, It's How You Say It: Evaluating LLM Responses to Expressions of Belief

paperactiveprovisionalit-s-not-what-you-say-it-s-how-you-say-it-evaluating-llm-responses-to-expressions-of-belief-a2983d08·1 events·first seen 8h ago

Aliases: It's Not What You Say, It's How You Say It: Evaluating LLM Responses to Expressions of Belief

Co-occurring entities

Gemma 3 Google Llama 3 Qwen3 Meta

More like this (12)

From Plausible to Actionable: A Position on LLM Self-Explanations Reinforcement Learning with Metacognitive Feedback Elicits Faithful Uncertainty Expression in LLMs Moral Safety in LLMs: Exposing Performative Compliance with Puzzled Cues Rating the Pitch, Not the Product: User Evaluations of LLMs Reflect Expectations More Than Performance Knowledge Knows, Verbalization Tells: Disentangling Latent Directions for Mathematical Solvability in LLMs Measuring Epistemic Resilience of LLMs Under Misleading Medical Context Inside the Unfair Judge: A Mechanistic Interpretability Account of LLM-as-Judge Bias Accuracy and Satisfaction in Multi-Turn LLM Dialogues for NFR Assessment Revising Context, Shifting Simulated Stance: Auditing LLM-Based Stance Simulation in Online Discussions Security and Privacy Prompts in the Wild: What Users Ask LLMs and How LLMs Respond Prompting Complexity: Shortest Prompts for Texts and Behaviors in LLMs LLM Judges Can Be Too Generous When There Is No Reference Answer

Recent events (1)

5arXiv · cs.CL·8h ago·source ↗

Benchmark study reveals how linguistic framing of user beliefs shifts LLM context-following behavior

A new arXiv paper introduces a typology of 17 linguistically motivated expression-of-belief (EoB) types—spanning form, evidentiality, epistemic stance, and tone—to evaluate how phrasing affects whether LLMs defer to user-stated beliefs or their own prior knowledge. The authors benchmark 16 LLMs across Llama 3, Qwen3, and Gemma3 families at scales from 1B to 30B parameters, finding that larger and instruction-tuned models are systematically less context-following than smaller or base models. Specific linguistic framings (e.g., presuppositions, certainty markers) are identified as statistically more persuasive, with implications for prompt robustness and sycophancy research.

Evaluation and Benchmarking AI Safety Research Gemma 3 Google Llama 3 +3 more