We spoke with Princeton researcher Aylin Caliskan, co-author of a headline-grabbing paper published in Science magazine earlier this month. Her paper details how learning machines can sometimes learn all too well, picking up our biases as well as our brilliance.