Researchers release dataset to expose racial, religious, and gender biases in language models

Reseachers at Amazon and the University of California, Santa Barbara find evidence of biases in language models.