04:00
2026-07-08
arxiv.org
large-language-models
Prompt Robustness Is Task-Dependent: Comparing Objective and Belief-Style Questions in LLM Evaluation
A new study finds that prompt robustness in large language models depends on the type of question asked, with objective questions showing different sensitivity to prompt variations than subjective bel…