Plural Voices, Single Agent: Towards Inclusive Ai In Multi-User Domestic Spaces

Date

10/2025

AI Overview

Why it matters

Domestic artificial intelligence agents face ethical, autonomy, and inclusion challenges. They frequently overlook the needs of vulnerable populations such as children, older adults, and neurodivergent individuals in multi-user households.

Study at a glance

Question: How can a single domestic AI agent equitably balance safety, autonomy, and conflicting needs across diverse household members?
Setting and data: 36 participants from 3 families, including 8 children, 6 older adults, 4 neurodivergent individuals, and 18 neurotypical users; 100,000+ synthetic entries; public conversation datasets.
Method: Participatory co-design workshops, prototype development, user testing with interviews and surveys, and automated LLM-based benchmark evaluations.
AI tools: Agora-4B, AgoraNest, Phi3:14b, Qwen-3 4B, Kimi-VL-A3B-Thinking, i2vgen-xl

Key findings

The single-agent Agora-4B outperformed multi-agent baselines in compliance (76% versus 70%), fairness (90% versus 85%), and safety-violation rates (0% versus 7%).
Users reported high overall satisfaction with the prototype app, awarding an average score of 4.2 out of 5 across 36 participants.
In synthetic benchmarks, the model achieved an 82% fairness mediation rate compared to 61% for multi-model baselines, while lowering safety violations from 4.2% to 1.5%.

Summary drafted with AI from the full paper and may contain errors. See the full paper for details.

Domestic AI agents faces ethical, autonomy, and inclusion challenges, particularly for overlooked groups like children, elderly, and Neurodivergent users. We present the Plural Voices Model (PVM), a novel single-agent framework that dynamically negotiates multi-user needs through real-time value alignment, leveraging diverse public datasets on mental health, eldercare, education, and moral reasoning. Using human+synthetic curriculum design with fairness-aware scenarios and ethical enhancements, PVM identifies core values, conflicts, and accessibility requirements to inform inclusive principles. Our privacy-focused prototype features adaptive safety scaffolds, tailored interactions (e.g., step-by-step guidance for Neurodivergent users, simple wording for children), and equitable conflict resolution. In preliminary evaluations, PVM outperforms multi-agent baselines in compliance (76% vs. 70%), fairness (90% vs. 85%), safety-violation rate (0% vs. 7%), and latency. Design innovations, including video guidance, autonomy sliders, family hubs, and adaptive safety dashboards, demonstrate new directions for ethical and inclusive domestic AI, for building user-centered agentic systems in plural domestic contexts. Our Codes and Model are been open sourced, available for reproduction: https://github.com/zade90/Agora