Explore the hidden biases in Natural Language Processing (NLP) with Dr. Dirk Hovy in this 33-minute conference talk from the Alan Turing Institute's workshop on "Bridging disciplines in analysing text as social and cultural data." Delve into the methodological challenges that cut across research disciplines, examining various types of biases including selection bias, annotation bias, model bias, and design bias. Learn about the MACE (Multi-Annotator Competence Estimation) approach and its implications for NLP. Discover how American English influences NLP models and consider strategies for moving forward in addressing these biases. Gain insights into the interdisciplinary synergies required to tackle the complex issues at the intersection of NLP, machine learning, humanities, and social sciences.