Stereotype Detection as a Catalyst for Enhanced Bias Detection: A Multi-Task Learning Approach

2025-07-02

AI Summary

Prior work treats bias detection and stereotype identification as disjoint tasks, limiting their mutual reinforcement and overall performance. Method: We propose a multi-task learning framework that jointly models both tasks using a unified encoder-only architecture and QLoRA-finetuned decoder-only large language models, optimized simultaneously across multiple socially sensitive dimensions on our newly constructed StereoBias dataset. Contribution/Results: Our empirical analysis reveals— for the first time—that stereotypes act as an effective “catalyst” for bias detection, reflecting intrinsic semantic associations beyond mere multi-task training benefits. Joint training improves bias detection F1 by +4.2%; decoder-only models achieve performance on par with encoder-only counterparts; and incorporating stereotype signals enhances robustness in bias identification. This work establishes a novel paradigm and empirical foundation for developing interpretable and fair AI systems.

Technology Category

• Computer Vision: Bias, Fairness & Privacy
• Machine Learning: Ethics, Bias, and Fairness
• Philosophy and Ethics of AI: Bias, Fairness & Equity

Application Category

• Social Networks and Social Media: Fairness and bias in social network and social media analysis
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Bias and stereotypes in language models can cause harm, especially in sensitive areas like content moderation and decision-making. This paper addresses bias and stereotype detection by exploring how jointly learning these tasks enhances model performance. We introduce StereoBias, a unique dataset labeled for bias and stereotype detection across five categories: religion, gender, socio-economic status, race, profession, and others, enabling a deeper study of their relationship. Our experiments compare encoder-only models and fine-tuned decoder-only models using QLoRA. While encoder-only models perform well, decoder-only models also show competitive results. Crucially, joint training on bias and stereotype detection significantly improves bias detection compared to training them separately. Additional experiments with sentiment analysis confirm that the improvements stem from the connection between bias and stereotypes, not multi-task learning alone. These findings highlight the value of leveraging stereotype information to build fairer and more effective AI systems.

Problem

• Detecting bias and stereotypes in language models
• Enhancing bias detection through joint learning
• Exploring relationships between bias and stereotypes

Innovation

• Multi-task learning for bias and stereotype detection
• StereoBias dataset with five labeled categories
• QLoRA fine-tuned decoder-only models