Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

An Anthropologist LLM to Elicit Users' Moral Preferences through Role-Play

Gianluca De Ninno, Paola Inverardi, Francesca Belotti

TL;DR

The paper addresses how to align AI systems with users' soft ethics by eliciting situated moral preferences through an immersive RPG framework analyzed by a customized LLM (GPT Anthropologist). It demonstrates that narrative, context-rich data combined with an anthropological interpretive lens improves predictive alignment with individual moral decisions in the digital privacy domain. Key contributions include a novel RPG-based elicitation method, the integration of narrative co-construction with LLM profiling, and empirical evidence that anthropologically framed, data-rich inputs outperform decontextualized inputs in predicting user decisions. The approach has practical implications for developing human-centered AI assistants that adapt to users' moral preferences across domains, with potential for scalability and cross-domain application through AI-enabled Game Masters and diary annotation pipelines.

Abstract

This study investigates a novel approach to eliciting users' moral decision-making by combining immersive roleplaying games with LLM analysis capabilities. Building on the distinction introduced by Floridi between hard ethics inspiring and shaping laws-and soft ethics-moral preferences guiding individual behavior within the free space of decisions compliant to laws-we focus on capturing the latter through contextrich, narrative-driven interactions. Grounded in anthropological methods, the role-playing game exposes participants to ethically charged scenarios in the domain of digital privacy. Data collected during the sessions were interpreted by a customized LLM ("GPT Anthropologist"). Evaluation through a cross-validation process shows that both the richness of the data and the interpretive framing significantly enhance the model's ability to predict user behavior. Results show that LLMs can be effectively employed to automate and enhance the understanding of user moral preferences and decision-making process in the early stages of software development.

An Anthropologist LLM to Elicit Users' Moral Preferences through Role-Play

TL;DR

The paper addresses how to align AI systems with users' soft ethics by eliciting situated moral preferences through an immersive RPG framework analyzed by a customized LLM (GPT Anthropologist). It demonstrates that narrative, context-rich data combined with an anthropological interpretive lens improves predictive alignment with individual moral decisions in the digital privacy domain. Key contributions include a novel RPG-based elicitation method, the integration of narrative co-construction with LLM profiling, and empirical evidence that anthropologically framed, data-rich inputs outperform decontextualized inputs in predicting user decisions. The approach has practical implications for developing human-centered AI assistants that adapt to users' moral preferences across domains, with potential for scalability and cross-domain application through AI-enabled Game Masters and diary annotation pipelines.

Abstract

This study investigates a novel approach to eliciting users' moral decision-making by combining immersive roleplaying games with LLM analysis capabilities. Building on the distinction introduced by Floridi between hard ethics inspiring and shaping laws-and soft ethics-moral preferences guiding individual behavior within the free space of decisions compliant to laws-we focus on capturing the latter through contextrich, narrative-driven interactions. Grounded in anthropological methods, the role-playing game exposes participants to ethically charged scenarios in the domain of digital privacy. Data collected during the sessions were interpreted by a customized LLM ("GPT Anthropologist"). Evaluation through a cross-validation process shows that both the richness of the data and the interpretive framing significantly enhance the model's ability to predict user behavior. Results show that LLMs can be effectively employed to automate and enhance the understanding of user moral preferences and decision-making process in the early stages of software development.

Paper Structure

This paper contains 29 sections, 14 figures.

Table of Contents

  1. INTRODUCTION
  2. BACKGROUND
  3. Human Values elicitation: OHV and VSD
  4. Redifining values through narrative anthropology and phenomenological anthropology
  5. Serious Games
  6. METHODOLOGY AND STUDY DESIGN
  7. Game Design: domain, scenarios and rules
  8. Recruiting and engaging audience
  9. Game session and dynamics
  10. Brief questionnaire
  11. Model personalization
  12. Research Protocol Evaluation
  13. NARRATIVES CO-CONSTRUCTION
  14. RPG session and data collection data
  15. Data processing
  16. ...and 14 more sections

Figures (14)

  • Figure 1: RPG scenario 1
  • Figure 2: Sample of Player Sheet
  • Figure 3: The first question of the questionnaire
  • Figure 4: Demographic and Self-Description Table
  • Figure 5: Player's choices and interpretation of game scenarios
  • ...and 9 more figures