Bias and Power in Natural Language Processing

Explore the critical intersection of bias, power, and Natural Language Processing (NLP) in this thought-provoking lecture presented by Arjun Subramonian from UCLA. Delve into the complex nature of bias in NLP models and their impact on marginalized communities. Examine the sociotechnical aspects of bias, its sources throughout the NLP lifecycle, and the specific context of gender exclusivity. Analyze various bias metrics and interventions, while critically assessing their limitations. Gain insights into the connection between bias, harms, and the underlying power structures in which NLP operates. Discover the importance of adopting more intersectional and reflexive approaches to NLP development and implementation. Learn from Subramonian's expertise in inclusive graph machine learning, ethics in AI, and the integration of queer perspectives in computer science research.

Bias and Power in NLP