Tyranny of Objectives
The conversation explores the shift from post hoc interpretability methods to a proactive approach that incorporates fairness objectives during model training. By focusing on how features develop over time, insights into model behavior can be gleaned, allowing for a deeper understanding of performance across various predictions. This evolution emphasizes the need to optimize for interpretability rather than relying on retrospective explanations.In this clip
From this podcast
Machine Learning Street Talk (MLST)
#92 - SARA HOOKER - Fairness, Interpretability, Language Models
Related Questions
What is the challenge around explainability in AI as discussed in the episode Evaluating Model Explainability Methods with Sara Hooker - TWiML Talk #189 and the clip Interpretability in AI?
What is the challenge around explainability in AI as discussed in the episode Evaluating Model Explainability Methods with Sara Hooker - TWiML Talk #189 and the clip Interpretability in AI?
What are the key topics in AI interpretability as discussed in the episode Evaluating Model Explainability Methods with Sara Hooker - TWiML Talk #189 and the clip Interpretability in AI?