Mitigation of Gender and Ethnicity Bias in AI-Generated Stories through Model Explanations

2025/09/03 by Martha O. Dimgba, Sharon Oba, Dimgba, Martha O. +5
Social Sciences · #Artificial Intelligence (cs.AI) #Computation and Language (cs.CL) #Ethics and Social Impacts of AI #FOS: Computer and information sciences

paper · pdf · doi:10.48550/arxiv.2509.04515

openalex publication_date 2025/09/03 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Language models have been shown to propagate social bias through their output, particularly in the representation of gender and ethnicity. This paper investigates gender and ethnicity biases in AI-generated occupational stories. Representation biases are measured before and after applying our proposed mitigation strategy, Bias Analysis and Mitigation through Explanation (BAME), revealing improvements in demographic representation ranging from 2% to 20%. BAME leverages model-generated explanations to inform targeted prompt engineering, effectively reducing biases without modifying model parameters. By analyzing stories generated across 25 occupational groups, three large language models (Claude 3.5 Sonnet, Llama 3.1 70B Instruct, and GPT-4 Turbo), and multiple demographic dimensions, we identify persistent patterns of overrepresentation and underrepresentation linked to training data stereotypes. Our findings demonstrate that guiding models with their own internal reasoning mechanisms can significantly enhance demographic parity, thereby contributing to the development of more transparent generative AI systems.

Citations

• Addressing bias in generative AI: Challenges and research opportunities in information management
• LLMs are Biased Teachers: Evaluating LLM Bias in Personalized Education
• Unboxing Occupational Bias: Grounded Debiasing of LLMs with U.S. Labor Data
• The Life Cycle of Large Language Models: A Review of Biases in Education
• When AI Eats Itself: On the Caveats of AI Autophagy
• The Pursuit of Fairness in Artificial Intelligence Models: A Survey
• AXOLOTL: Fairness through Assisted Self-Debiasing of Large Language Model Outputs
• Self-Debiasing Large Language Models: Zero-Shot Recognition and Reduction of Stereotypes
• Escalation Risks from Language Models in Military and Diplomatic Decision-Making
• "Kelly is a Warm Person, Joseph is a Role Model": Gender Biases in LLM-Generated Reference Letters
• Fairness and Bias in Algorithmic Hiring: a Multidisciplinary Survey
• Fairness and Bias in Algorithmic Hiring: A Multidisciplinary Survey
• Gender Bias in Multimodal Models: A Transnational Feminist Approach Considering Geographical Region and Culture
• Bias and Fairness in Large Language Models: A Survey
• Bias and Fairness in Large Language Models: A Survey
• BAD: BiAs Detection for Large Language Models in the context of candidate screening
• Multimodal Composite Association Score: Measuring Gender Bias in Generative Multimodal Models
• Self-Refine: Iterative Refinement with Self-Feedback
• Perspectives on the Social Impacts of Reinforcement Learning with Human Feedback
• DALL-Eval: Probing the Reasoning Skills and Social Biases of Text-to-Image Generation Models
• Ethical and social risks of harm from Language Models
• On the Dangers of Stochastic Parrots
• Bias Out-of-the-Box: An Empirical Analysis of Intersectional Occupational Biases in Popular Generative Language Models
• Debiasing Pre-trained Contextualised Embeddings
• Language (Technology) is Power: A Critical Survey of "Bias" in NLP
• Algorithmic Fairness
• Bias in Data-driven AI Systems -- An Introductory Survey
• A Survey on Bias and Fairness in Machine Learning
• Mitigating Gender Bias in Natural Language Processing: Literature Review
• Fairness in representation: quantifying stereotyping as a representational harm
• Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word\n Embeddings
• On the relation between accuracy and fairness in binary classification

Related

• More Women, Same Stereotypes: Unpacking the Gender Bias Paradox in Large Language Models
• Generating Clues for Gender based Occupation De-biasing in Text
• More of the Same: Persistent Representational Harms Under Increased Representation
• Evaluating Large Language Models through Gender and Racial Stereotypes
• Unveiling Gender Bias in Terms of Profession Across LLMs: Analyzing and Addressing Sociological Implications