The article is currently paywalled, but a friend of mine sent me the text. The relevant paragraph, lightly edited (the "she" here is Sorelle):
By understanding the biases inherent in the underlying data, she hopes to eliminate bias in the algorithm. Her system looks for correlations between arbitrary properties – like height or address – and demographic groupings like race or gender. If the correlation is expected to lead to unwanted bias, then it would make sense to normalise the data. It is essentially affirmative action for algorithms, she says.I'm not sure I'd describe our work as affirmative action :), but that's a longer argument.