Absolute Stance Bias
Absolute stance bias refers to systematic errors in machine learning models stemming from skewed training data or algorithmic design, leading to unfair or inaccurate predictions across different groups or contexts. Current research focuses on quantifying and mitigating these biases in various models, including large language models (LLMs), machine translation systems, and image recognition algorithms, often employing techniques like counterfactual fairness, reinforcement learning, and bias-aware evaluation metrics. Understanding and addressing absolute stance bias is crucial for ensuring fairness, reliability, and trustworthiness in AI systems across diverse applications, from healthcare and finance to social media and education.
Papers
Understanding the Interplay of Scale, Data, and Bias in Language Models: A Case Study with BERT
Muhammad Ali, Swetasudha Panda, Qinlan Shen, Michael Wick, Ari Kobren
Exploring Bengali Religious Dialect Biases in Large Language Models with Evaluation Perspectives
Azmine Toushik Wasi, Raima Islam, Mst Rafia Islam, Taki Hasan Rafi, Dong-Kyu Chae
Exploring LGBTQ+ Bias in Generative AI Answers across Different Country and Religious Contexts
Lilla Vicsek, Anna Vancsó, Mike Zajko, Judit Takacs
Images Speak Louder than Words: Understanding and Mitigating Bias in Vision-Language Model from a Causal Mediation Perspective
Zhaotian Weng, Zijun Gao, Jerone Andrews, Jieyu Zhao