Datasheets could be the solution to biased AI

In a recent conversation, Facebook AI research scientist Moustapha Cissé told me, "You are what you eat, and right now we feed our models junk food." Well, just like you can't eat better if you don't know what's in your food, you can't train less biased models if you don't know what's in your training data. That's why the recent paper "Datasheets for Datasets" is so interesting. In it, Timnit Gebru and her coauthors from Microsoft Research and elsewhere propose the equivalent of food nutrition labeling for datasets. Given that many machine learning and deep learning model development efforts use public datasets such as ImageNet or COCO -- or private datasets produced by others -- it's important to be able to convey the context, biases, and other material aspects of a training dataset to those interested in using it.