Advancing Fairness in Generative AI Through Intrinsic and Extrinsic Bias Evaluation and Mitigation

This research focuses on understanding and mitigating bias in generative AI models, specifically by examining both intrinsic and extrinsic biases. The project aims to develop a unified evaluation framework and bias mitigation strategies to promote fairness across real-world applications, such as finance and healthcare. The goal is to ensure generative AI systems do not propagate harmful societal biases, and the research explores bias detection and mitigation across various deployment stages.