Beyond Value Elicitation: Towards Moral Profiles in Early Requirements Engineering via Role-Playing Games and Anthropologist LLMs
This academic study introduces a novel proof of concept for integrating human moral values into the early stages of Requirements Engineering (RE). Addressing the limitations of traditional methods that rely on predefined value taxonomies, the researchers propose capturing tacit, context-dependent moral orientations through immersive role-playing games (RPGs). These RPG sessions generate rich narrative data reflecting situated decision-making processes. The data is subsequently analyzed by a specialized Large Language Model, referred to as GPT-A, which is grounded in phenomenological and narrative anthropology. This model constructs Individual Anthropological Moral Profiles (IAMPs) for users. Validation results indicate that this approach effectively transforms complex behavioral data into coherent narrative representations of user morality. By preserving the dynamic nature of user preferences and trade-offs, the method enables better contextual interpretation of actions and motivations. This research provides a foundational framework for digital system designers to incorporate nuanced human ethical considerations into software development, moving beyond simple value elicitation to comprehensive moral profiling.
Wire timeline
Beyond Value Elicitation: Towards Moral Profiles in Early Requirements Engineering via Role-Playing Games and Anthropologist LLMs
This academic study introduces a novel proof of concept for integrating human moral values into the early stages of Requirements Engineering (RE). Addressing the limitations of traditional methods that rely on predefined value taxonomies, the researchers propose capturing tacit, context-dependent moral orientations through immersive role-playing games (RPGs). These RPG sessions generate rich narrative data reflecting situated decision-making processes. The data is subsequently analyzed by a specialized Large Language Model, referred to as GPT-A, which is grounded in phenomenological and narrative anthropology. This model constructs Individual Anthropological Moral Profiles (IAMPs) for users. Validation results indicate that this approach effectively transforms complex behavioral data into coherent narrative representations of user morality. By preserving the dynamic nature of user preferences and trade-offs, the method enables better contextual interpretation of actions and motivations. This research provides a foundational framework for digital system designers to incorporate nuanced human ethical considerations into software development, moving beyond simple value elicitation to comprehensive moral profiling.
cs.AI updates on arXiv.org