Glossary · Phase 6 · Modern LLM engineering
RLAIF
Reinforcement learning from AI feedback: like RLHF, but an AI model applying written principles does the judging instead of people.
Try it now
The written rules become the specification, so they're easy to read and easy to change.
Learn it properly
Constitutional AI · Training with AI feedback guided by written principles. About 35 minutes.
Other terms from the same lesson
- constitutional AI: Training a model to critique and improve its own answers against a written set of principles.
- false refusal: A model refusing a request that was actually harmless, such as declining to explain how locks work.
- jailbreak: A prompt written to trick a model into ignoring its safety rules.
Never trained an AI?
Teach a small one to write, right in your browser. It takes about 20 seconds and needs no account.