Measuring Normative and Descriptive Biases in Language Models Using Census Data

Open in new window