A Detailed Study on LLM Biases Concerning Corporate Social Responsibility and Green Supply Chains
Greta Ontrup, Annika Bush, Markus Pauly, Meltem Aksoy
TL;DR
This study investigates how state-of-the-art LLMs encode biases regarding corporate social responsibility and green supply chain management, using validated psychometric instruments to compare responses across five models. By prompting the models to adopt four organizational culture roles, the work reveals significant model differences and strong culture-dependent modulation of sustainability assessments. The PRESOR and GSCP scales show mixed psychometric properties for AI responses, highlighting challenges in applying human-centric instruments to LLMs while demonstrating clear patterns of pro-sustainability bias and culture-driven amplification. The findings underscore the need for bias auditing, careful model selection, and governance when employing LLMs for sustainability decision-making in organizations.
Abstract
Organizations increasingly use Large Language Models (LLMs) to improve supply chain processes and reduce environmental impacts. However, LLMs have been shown to reproduce biases regarding the prioritization of sustainable business strategies. Thus, it is important to identify underlying training data biases that LLMs pertain regarding the importance and role of sustainable business and supply chain practices. This study investigates how different LLMs respond to validated surveys about the role of ethics and responsibility for businesses, and the importance of sustainable practices and relations with suppliers and customers. Using standardized questionnaires, we systematically analyze responses generated by state-of-the-art LLMs to identify variations. We further evaluate whether differences are augmented by four organizational culture types, thereby evaluating the practical relevance of identified biases. The findings reveal significant systematic differences between models and demonstrate that organizational culture prompts substantially modify LLM responses. The study holds important implications for LLM-assisted decision-making in sustainability contexts.