Discrimination and Features
Solon discusses the complexities of using certain features in predictive modeling, particularly in relation to societal inequalities. He highlights how features like zip codes, while informative for predicting home values, are intricately linked to historical racial discrimination. This raises challenging ethical questions about the implications of using such features in data-driven decisions.In this clip
From this podcast
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
Legal and Policy Implications of Model Interpretability with Solon Barocas - TWiML Talk #219
Related Questions