Sociodemographic biases in language models can have harmful effects, necessitating research into quantifying and mitigating bias.