What does “Algorithmic Bias” mean?
Systematic unfairness in model outputs caused by skewed data, assumptions, or modeling choices.
How Algorithmic Bias Works
Skewed or incomplete data
Model learns the pattern
Different error rates by group
Unfair outcomes
Audit and fix.
Practical Example
A resume-screening model is trained on past hiring decisions at a company that mostly hired men. It learns to rate resumes with typical male patterns higher and can penalize equally qualified women. This kind of case was widely reported in coverage of automated hiring tools.
Why It Matters
Biased models can quietly repeat or amplify unfair patterns in hiring, lending, policing, and health care. Checking results across groups is one of the main ways to catch it.
Common Questions About Algorithmic Bias
Q: Where does algorithmic bias come from?
Common sources are unrepresentative data, labels that reflect past prejudice, and design choices such as which measure the model optimizes.
Q: Can bias be removed completely?
Usually not entirely. Teams measure it, reduce it, and document what remains, because different fairness definitions can conflict with each other.
Q: How is bias detected?
By comparing a model's error rates and outcomes across groups, reviewing the data, and testing with realistic cases.