PRIDE -- Parameter-Efficient Reduction of Identity Discrimination for Equality in LLMs

2025-07-18

AI Summary

Large language models (LLMs) frequently reproduce gender and sexual orientation biases present in training data, exacerbating discrimination against LGBTQIA+ individuals. To address this, we propose PRIDE—a community-informed, parameter-efficient fine-tuning framework that jointly optimizes fairness on the QueerNews corpus and WinoQueer benchmark. PRIDE integrates low-rank adaptation (LoRA, adding <0.1% parameters) with soft prompt tuning (10 learnable tokens). Evaluated across multiple open-source LLMs, PRIDE reduces bias scores by up to 50 points versus baselines and increases neutral output proportion from near 0% to 36%, markedly improving inclusivity. Its key contributions are: (i) the first integration of structured LGBTQIA+ community feedback into lightweight fine-tuning; and (ii) a reproducible, auditable fairness optimization paradigm designed for sustainability and transparency in bias mitigation.

Technology Category

• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy
• Philosophy and Ethics of AI: Bias, Fairness & Equity
• Machine Learning: Ethics, Bias, and Fairness

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Large language models for search
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

Large Language Models (LLMs) frequently reproduce the gender- and sexual-identity prejudices embedded in their training corpora, leading to outputs that marginalize LGBTQIA+ users. Hence, reducing such biases is of great importance. To achieve this, we evaluate two parameter-efficient fine-tuning (PEFT) techniques - Low-Rank Adaptation (LoRA) and soft-prompt tuning - as lightweight alternatives to full-model fine-tuning for mitigating such biases. Using the WinoQueer benchmark, we quantify bias in three open-source LLMs and observe baseline bias scores reaching up to 98 (out of 100) across a range of queer identities defined by gender and/or sexual orientation, where 50 would indicate neutrality. Fine-tuning with LoRA (< 0.1% additional parameters) on a curated QueerNews corpus reduces those scores by up to 50 points and raises neutrality from virtually 0% to as much as 36%. Soft-prompt tuning (10 virtual tokens) delivers only marginal improvements. These findings show that LoRA can deliver meaningful fairness gains with minimal computation. We advocate broader adoption of community-informed PEFT, the creation of larger queer-authored corpora, and richer evaluation suites beyond WinoQueer, coupled with ongoing audits to keep LLMs inclusive.

Problem

• Reduce gender and sexual-identity biases in LLMs
• Evaluate parameter-efficient fine-tuning for bias mitigation
• Improve fairness in LLMs with minimal computational cost

Innovation

• Uses LoRA for bias reduction
• Applies soft-prompt tuning marginally
• Evaluates with WinoQueer benchmark