The Value of Interpretability

Jayaraman explains that interpretability is not just about accountability or trusting models, but rather a means for experts to interact with the system and ensure their hypotheses are accurate. Making a model interpretable does not automatically make it fair or responsible.