Algorithmic Bias (Bias in AI)

Algorithmic bias, or bias in AI, is a repeatable pattern in which a system produces worse or unfair results for certain people or groups. It can originate in the data, in how labels were assigned, in the features chosen, or in how the system is deployed.

Why it matters

Biased systems can disadvantage people in hiring, lending, moderation, and search, and expose organizations to legal and reputational harm. Because models copy patterns in training data, historical inequities can be reproduced at scale.

Example

A résumé screener trained on past hiring decisions may learn to favor profiles resembling previous hires. If that history was skewed, the model repeats the skew while appearing neutral.

Common mistakes

• Assuming that deleting a sensitive column removes bias; other features can act as proxies.
• Measuring only overall accuracy instead of performance for each group.
• Treating fairness as a one-time check rather than ongoing monitoring.

Related

• Training data
• AI alignment
• Data cleaning
• Computer vision

This page is general information written by our editorial team and updated when documentation changes. Servers and providers differ, so test changes safely first. See our editorial policy and disclaimer.