GLOSSARY · AI SECURITY
Reinforcement learning from human feedback (RLHF)
Reinforcement learning from human feedback (RLHF) is a training technique that tunes an AI model toward outputs humans rate as helpful and safe.
Human reviewers rank candidate responses, a reward model learns those preferences, and the base model is optimized against it. RLHF is a large part of why chat assistants feel cooperative and refuse harmful requests.