Term

Definition

Example

Bias (in general)

Favoring one person or group unfairly over another.

Choosing a friend’s idea even if another is better.

AI Bias

When an AI system gives unfair or inaccurate results because the data it learned from was unbalanced or incomplete.

A face recognition system that works better for one gender or skin tone than others.

Diversity

Including a wide range of people, examples, or ideas in data or design.

Using images of people from different ages, backgrounds, and regions.

Fairness in AI

Ensuring AI treats everyone equally and gives accurate results for all groups.

A language app that works for all accents.

Data Quality

The accuracy, completeness, and balance of the information used to train AI.

Having equal examples of boys and girls in a dataset.