Human feedback
People comparing answers so the model learns which are better.
Learn it through a story
- How a Chatbot Goes to School (Jugnupur Tales Vol 2, Ch 5)
Short on time? Read the 5-minute version.
Related terms
- Grounding Tying an AI's answer to real sources, so it can be checked.
- Guardrails Protections that help an AI refuse harmful requests.
- Hallucination When an AI says something made up as if it were true.
- Image generation AI that creates new pictures from a text description.
- Knowledge cutoff The date after which a model has no training information.
- Language model A program that has learned patterns in language and predicts words.