Measuring What Matters: Connecting AI Ethics Evaluations to System Attributes, Hazards, and Harms
This research paper critiques current AI ethics evaluations for failing to account for system attributes and hazards, proposing a holistic framework to align...