Tacit Knowledge and Hermeneutics: Critical Reflections on Automated Text Interpretation with Large Language Models

Abstract

Overview This study explores the interplay between tacit knowledge and hermeneutical interpretation within the domain of large language models (LLMs). In this study, we examine the question of whether automated systems can approximate interpretive practices that depend on implicit understanding, situated experience, and contextual embedding. To this end, we draw on philosophical and psychological accounts of tacit knowledge and traditions of hermeneutics. Utilizing a comparative experimental design, multiple LLMs are prompted to interpret texts from the domains of philosophy, theology, literature, and conversation. We find considerable variability and constrained reproducibility in the interpretive outcomes. These findings suggest that the core dimensions of tacit knowledge, particularly embodied inference and context-sensitive understanding, remain inaccessible to current generative models. The paper posits that hermeneutics cannot be reduced to formal pattern recognition and calls for refinedtheoretical and empirical criteria for evaluating machine-based interpretation. Content Python Sourcecode Text corpora Different Prompts