Is This Your Final Answer? Cross-Contextual Consistency as a Measure of LLM Credibility

Siyang Wu, Yibo Jiang, Bryon Aragam

Large language models (LLMs) are powerful black-box systems, making it difficult to discern whether their answers reflect stable internal beliefs or superficial pattern matching. W…