An Investigation Into Value Misalignment in LLM-Generated Texts for Cultural Heritage
Fan Bu, Zheng Wang, Siyi Wang, Ziyao Liu
Abstract
As Large Language Models (LLMs) become increasingly prevalent in tasks related to cultural heritage, such as generating descriptions of historical monuments, translating ancient texts, and creating educational content, users and researchers increasingly rely on their ability to generate accurate and culturally aligned texts. However, cultural value misalignments may exist in generated texts, such as the misrepresentation of historical facts, the erosion of cultural identity, and the oversimplification of complex cultural narratives, potentially leading to severe consequences. Therefore, investigating value misalignment in the context of LLMs for cultural heritage is crucial for mitigating these risks, yet there has been a significant lack of systematic study and investigation in this area. To fill this gap, we systematically assess the reliability of LLMs in generating culturally aligned texts for cultural heritage-related tasks. We conduct an evaluation by compiling an extensive set of 1066 query tasks covering 5 widely recognized categories with 17 aspects within the knowledge framework of cultural heritage across 5 open-source LLMs, examining both the type and rate of cultural value misalignments in the generated texts. Using both automated and manual approaches, we effectively detect and analyze the cultural value misalignments in LLM-generated texts. Our findings are concerning: over 65% of the generated texts exhibit notable cultural misalignments, with certain tasks demonstrating almost complete misalignment with cultural values. Beyond these findings, this paper introduces a benchmark dataset and a comprehensive evaluation workflow to serve as a valuable resource for future research aimed at enhancing the cultural sensitivity and reliability of LLMs.