Algorithmic Bias

noun

say it: “al-guh-RITH-mik BY-us” /ˌælɡəˈrɪðmɪk ˈbaɪəs/

Unfair results from a computer system, such as a model that treats some groups of people worse than others, often learned from biased data.

What is algorithmic bias?

Algorithmic bias is when a model's decisions or mistakes fall unfairly on some groups of people, such as a screen that passes over qualified applicants from one district. It usually comes from the data: past decisions with old habits in them, groups with few examples, or proxies for a sensitive feature. Leaving the sensitive column out doesn't remove it.

In a sentence

  • “Algorithmic bias crept in through a column that stood in for the neighborhood.”
  • “The audit found algorithmic bias: qualified applicants from one district were invited half as often.”

Level: Journeyman · More AI Basics terms at this level

Updated