Algorithmic bias explained: how AI systems end up unfair
Where AI bias comes from — data, labels, design and deployment — with landmark examples such as Gender Shades, and the main ways to measure and reduce it.
AI Desk3 min read
How to audit an AI system for fairness: a practical checklist
A step-by-step approach to assessing whether an AI system treats people fairly: scoping, data review, disaggregated testing, documentation and ongoing monitoring.
AI Desk3 min read