Framing Effects in Independent-Agent Large Language Models: A Cross-Family Behavioral Analysis
arXiv:2603.19282v2 Announce Type: replace
Abstract: In many real-world applications, large language models (LLMs) operate as independent agents without interaction, thereby limiting coordination. In this setting, we examine how prompt framing influenc…