Abstract. A growing body of research on large language models (LLMs) has identified various biases, primarily in contexts where biases reflect societal pat