Ethics & SafetyFeatured

Bias in AI

Definition

Systematic and unfair discrimination in AI outputs resulting from biased training data or algorithms.

In-Depth Explanation

AI bias can reflect and amplify societal prejudices related to race, gender, age, and other characteristics. Sources include biased training data, label bias, and algorithmic design choices. Addressing bias requires diverse datasets, careful evaluation, and ongoing monitoring. It remains a major challenge for responsible AI development.

Real-World Example

A hiring AI trained on historical data might discriminate against women if past hiring was biased toward men.

0 views0 found helpful