Bias in AI is when a system treats people or groups unfairly, usually because of patterns in its training data or choices made while building it.
In plain words
AI learns from examples made by people, and machine learning can copy the human biases in them. If a group is missing or misrepresented in the data, the system's results can be unfair to that group. Bias can cause harm even when nobody intends it, so teams audit data before training and check predictions for bias before launch.
A quick example
Imagine a hiring tool trained on past hiring decisions. If those decisions favoured one group, the tool can learn to do the same. Auditing the training data and comparing the tool's results across groups can reveal the problem before it is used.
Where you'll meet it
- Hiring, lending and other decisions about people.
- Checking data and model results before a system goes live.
Related
Sources
- DocsTowards a Standard for Identifying and Managing Bias in Artificial Intelligence (SP 1270), NIST · read 29 Sept 2026
- DocsFairness, Google Machine Learning Crash Course · read 29 Sept 2026
Checked by WhatIsAI against the sources above · 29 Sept 2026. Spotted an error? Report it.