Using large Language models as a road map for establishing core values in a legal vacuum

Abstract

This approach offers a structured method for organizations to develop and articulate their ethical frameworks, particularly in areas where legal guidance is limited or nonexistent. Problem: This study investigates establishing core values in a legal vacuum, where research, design, or implementation of an invention or innovation is feasible but lacks regulations. We leverage Large Language Models (LLMs) to analyze codes of conduct from 1000 organizations (profit and not-for-profit) to identify core values. Metrics such as accuracy, bias, completeness, consistency, and relevance are used to validate the performance of LLMs in this context. From 493 non-profit organizations and companies on the Fortune 500 list, a total of 8646 core values including variations across 89 sectors were found. Using accuracy, bias, completeness, consistency and relevance as metrics for evaluating result from the LMMs, the number of core values is reduced to 362. The research employs a ten-step decision-making process to guide ethical decision-making when clear rules, laws, or regulations are absent. The framework presents how objectivity can be maintained without losing personal values. This research contributes to understanding how core values are established and applied in the absence of formal regulations.