RLHF

What it means

Reinforcement Learning from Human Feedback: humans rate the model's answers, and the model learns to prefer responses people find helpful.

An everyday example

A chef reading diner reviews every night and adjusting recipes until the dishes people love keep showing up.