Wait, It's All Token Noise? Always Has Been: Interpreting LLM Behavior Using Shapley Value

arXiv:2404.01332v1 Announce Type: cross
Abstract: The emergence of large language models (LLMs) has opened up exciting possibilities for simulating human behavior and cognitive processes, with potential applications in various domains, including marketing research and consumer behavior analysis. However, the validity of utilizing LLMs as stand-ins for human subjects remains uncertain due to glaring divergences that suggest fundamentally different underlying processes at play and the sensitivity of LLM responses to prompt variations. This paper presents a novel approach based on Shapley …

abstract analysis applications arxiv behavior behavior analysis cognitive consumer cs.ai cs.cl cs.lg domains emergence however human language language models large language large language models llm llms marketing noise processes research token type value