# What is RLAIF? Definition and meaning

Uses an AI model, rather than humans, to produce the preference labels used for alignment.

## Definition

**RLAIF**: Uses an AI model, rather than humans, to produce the preference labels used for alignment.

In short: Reinforcement learning from AI feedback.

## When it is used

Scaling alignment data when human labelling is too slow or costly.

## Learn more

The full guide to **[RLAIF](/tools/rlaif)** explains how it works, when to use it and how it compares with the alternatives, with examples. The [glossary](/glossary) lists every term on the site from A to Z.

Previous term: [ResNet](/glossary/resnet) · Next term: [RLHF](/glossary/rlhf)
