Classify AI bias examples by source category to select better mitigations.
sort-buckets source-map score-chips Place each concern into its primary source bucket. Systemic bias Statistical bias Human bias Past staffing left Spanish-language tickets waiting longer, and the model learned wait time as difficulty. The training set has only 4% examples from customers using screen readers. Labelers mark assertive customer messages as hostile more often when names sound foreign. A policy requires in-person branch history, excluding applicants from areas where branches were closed. A proxy feature captures college prestige because historical hires came from a narrow referral network. Reviewers accept high model scores without reading the low-confidence warning. The target variable measures…
Sign up free — one personalized lesson every day, matched to your role and goals.
Already have an account? Sign in