04:00
2026-05-29
arxiv.org
large-language-models
Mind Your Tone: Does Tone Alter LLM Performance?
A new study from arXiv (2605.29027) found that tonal variations in prompts cause systematic but model-dependent accuracy shifts in large language models (LLMs) on objective multiple-choice questions. …