AI Glossary
What is AI Bias?
When an AI system gives unfair or one-sided results for certain groups or situations, usually because of patterns in the data it learned from or in how it was built.
AI learns from human-created data, and that data reflects human biases. A model may, for example, assume certain jobs belong to men or women, describe some regions or communities unfairly, or work better for English speakers than for others.
Where it matters most: hiring and HR decisions, lending and credit, customer service, healthcare, and any content about people or groups.
How to reduce the risk in your work:
- Never let AI make decisions about people on its own. Use it to assist, and keep a human responsible.
- Read AI-written job descriptions, reviews and customer messages for stereotypes or exclusionary language.
- Ask the AI directly: "Is there any biased or exclusionary language in this text?"
- Use diverse examples when you give the AI examples to follow.
Related: Training Data.