Gebru, Timnit

was written in a period when NLP practitioners are producing bigger (# of parameters; size of training data) language models (LMs), and pushing the top scores on benchmarks. The paper itself was controversial because it led to Gebru being fired from Google, following disagreements with her managers on conditions (withdraw, or remove Google-affiliated authors) for publishing the paper. A lot changed since mid-2021, when I initially wrote this page....

Meta 📑 Instead of changing the data or learners in multiple ways and then see if fairness improves, postulate that the root causes of bias are the prior decisions that generated the training data. These affect (a) what data was selected, and (b) the labels assigned to the examples. They propose the \(\text{Fair-SMOTE}\) (Fair Synthetic Minority Over Sampling Technique) algorithm which (1) removes biased labels (via situation testing: if the model’s prediction for a data point changes once all of the data points' protected attributes are flipped, then that label is biased and the data point is discarded), and (2) rebalances internal distributions such that based on a protected attribute, examples are equal in both positive and negative classes....