AIGlossaryLab
⌕
← Back
SafetyBeginner2010s

Bias

AI Bias

AI bias is a repeated unfair pattern in an AI system’s results, often caused by its data, design or use.

Explained simply

AI bias is a repeated unfair pattern in an AI system’s results, often caused by its data, design or use.

1Biased data
→
2AI learns pattern
→
3Unequal result
→
4Review and correct

At a glance

Category
Safety
Difficulty
Beginner
Introduced
2010s

Real example

A hiring model trained mainly on past male candidates may unfairly score women lower.

Why it matters

Bias can unfairly affect people in hiring, lending, healthcare and other important decisions.

Timeline

Origins

The ideas behind Bias begin developing.

2010s

Bias becomes a recognised term or technique.

Wider use

Research and practical applications increase.

Modern AI

Bias becomes connected to newer AI systems and products.

Today

Bias remains relevant in the safety area of AI.

Dates for broad concepts may describe the period when the idea emerged or became widely used, rather than one exact invention date.

Learn next

Continue with these connected terms:

AlignmentAI alignment means trying to make an AI behave in ways that match human goals, rules and values.→Explainable AIExplainable AI uses methods that help people understand why an AI produced a particular result.→GuardrailsAI guardrails are rules and controls designed to reduce unsafe, incorrect or unwanted behaviour.→

Simple infographic

A quick visual way to understand Bias.

1Biased data
→
2AI learns pattern
→
3Unequal result
→
4Review and correct
1 of 6

Was this helpful?

Your feedback helps improve this explanation.