Technology Category
Philosophy and Ethics of AI: Morality & Value-based AIHumans and AI: Learning Human Values and PreferencesMachine Learning: Large Multimodal Models (LMMs)
Pluralistic Alignment for Healthcare: A Role-Driven Framework
🤖 AI Summary
Large language models (LLMs) struggle to align with diverse ethical and cultural values in high-stakes domains such as healthcare. Method: This paper introduces EthosAgents—the first role-driven, multi-value alignment framework tailored for medical applications. It enables lightweight, generalizable alignment by explicitly modeling user/clinician roles alongside their cultural backgrounds, contextual constraints, and value preferences, thereby supporting personalized, norm-aware response generation. Innovatively extending modular pluralism, the framework implements a joint role–value modeling mechanism compatible with multimodal and multi-scale LLMs via plug-and-play integration. Results: Evaluated across seven open- and closed-source models, EthosAgents significantly improves multi-value alignment performance, demonstrating strong generalizability, effectiveness, and cultural sensitivity in medical AI deployment.
Technology Category
Application Category
📝 Abstract
As large language models are increasingly deployed in sensitive domains such as healthcare, ensuring their outputs reflect the diverse values and perspectives held across populations is critical. However, existing alignment approaches, including pluralistic paradigms like Modular Pluralism, often fall short in the health domain, where personal, cultural, and situational factors shape pluralism. Motivated by the aforementioned healthcare challenges, we propose a first lightweight, generalizable, pluralistic alignment approach, EthosAgents, designed to simulate diverse perspectives and values. We empirically show that it advances the pluralistic alignment for all three modes across seven varying-sized open and closed models. Our findings reveal that health-related pluralism demands adaptable and normatively aware approaches, offering insights into how these models can better respect diversity in other high-stakes domains.
Problem
Research questions and friction points this paper is trying to address.
Aligning language models with diverse healthcare values
Addressing personal cultural situational factors in health
Ensuring model outputs reflect population diversity
Innovation
Methods, ideas, or system contributions that make the work stand out.
Role-driven framework simulating diverse perspectives
Lightweight generalizable pluralistic alignment approach
Adaptable normative awareness for healthcare diversity
🔎 Similar Papers
No similar papers found.
