Output Analysis
Abstract
When a company’s hiring AI systematically downgraded resumes containing the word “women,” the problem wasn’t hidden in training data or buried in neural network weights—it was visible in the outputs. When a major health system discovered their AI diagnostic tool recommended different treatments for identical symptoms based on patient names, the discrimination was evident in the recommendations themselves. Sometimes the most important evidence isn’t what went into an AI system but what comes out of it.
Fields of Study & Subjects
Ethics and Social Impacts of AI
Explainable Artificial Intelligence (XAI)
Discrimination and Equality Law
Publication details
- Journal / source
- Chapman & Hall/CRC eBooks
- Pages
- 218–264
- Publication date
- 2 February 2026
- Language
- EN
- Document type
- Book chapter