R
AI Glossary Term • Term 169 of 224
Plain English Definition
What does “RLHF (Reinforcement Learning from Human Feedback)” mean?
Core Definition
An alignment methodology that trains reward models on human evaluative rankings to steer generative models toward helpful, harmless, and honest behavior.
Real-World Scenario
Practical Example
Reinforcing an assistant to refuse hazardous chemical synthesis requests while remaining helpful for safe chemistry homework.
Conceptual Cross-References