Implicit gender bias in language models can perpetuate real-world biases, with pronoun choice revealing underlying biases.
