Pareto-Guided Teacher Alignment for Fair Personalized Text Generation

2026-06-08

AI Summary

This study addresses unfair expression disparities in personalized text generation arising from demographic conditioning by framing fairness mitigation as a multi-objective alignment problem that simultaneously reduces inter-group differences and preserves personalization fidelity. The authors propose a Pareto-guided teacher alignment framework integrating revision-based candidate generation, pairwise-aware feasibility gating, Pareto-style candidate selection, and preference optimization to systematically uncover the trade-offs between fairness and personalization. Experiments on climate change and vaccine persuasion tasks demonstrate that different strategies occupy distinct regions of the fairness-personalization Pareto frontier, with fairness outcomes exhibiting target dependency and cross-domain inconsistency. These findings validate the effectiveness of the proposed approach and advocate for model selection based on multi-audit evaluation metrics.

Abstract

Personalized persuasive text generation can improve relevance and engagement, but demographic conditioning may also introduce unequal framing across groups. We study fairness mitigation in personalized generation as a constrained multi-objective alignment problem: reduce demographic disparities while preserving personalization fidelity. We propose a Pareto-guided teacher alignment framework that combines revision-based candidate generation, pair-aware feasibility gating, Pareto-style candidate selection, and optional preference optimization through supervised fine-tuning and direct preference optimization. We evaluate the framework on climate change and vaccination persuasion tasks using a controlled context-rich demographic grid with matched gender and age pairs and a unified five-audit evaluation suite spanning persuasion bias, formality disparity, emotional framing disparity, lexical association disparity, and personalization fidelity. Across both domains and cross-family transfer settings, no single alignment strategy dominates all objectives simultaneously. Instead, methods occupy different regions of a fairness-personalization Pareto frontier: some achieve stronger disparity reductions, while others better preserve personalization or demographic stability. Our results show that fairness mitigation effects are objective-dependent and transfer inconsistently across domains and model families, motivating bounded-regression, multi-audit model selection over single-metric optimization for fairness-sensitive personalized generation.

Problem

• fairness
• personalized text generation
• demographic disparity
• multi-objective alignment
• Pareto frontier

Innovation

• Pareto-guided alignment
• fair personalized generation
• multi-objective optimization
• demographic fairness
• preference optimization