TAAFT
Free mode
100% free
Freemium
Free Trial
Deals
Create tool
Glossary
Ethics & Safety
Reinforcement Learning from Human Feedback

Reinforcement Learning from Human Feedback

[ˌriːɪnˈfɔːsmənt ˈlɜːnɪŋ frəm ˈhjuːmən ˈfiːdbæk]
Ethics & Safety
Last updated: December 9, 2024

Definition

Training AI systems using human evaluations of their outputs

Detailed Explanation

Machine learning technique that incorporates human feedback to improve AI model behavior and align it with human preferences

Use Cases

Language model training Content moderation systems Autonomous system behavior refinement

Related Terms

Mechanistic Interpretability

Reverse-engineering neural networks to understand their internal computations and algorithms.

Ethics & Safety

Ethical AI Guidelines

Principles and standards for ensuring AI systems are developed and deployed in ways that respect human rights and values.

Ethics & Safety

Process-Based Supervision

AI safety approach supervising the AI's reasoning process, not just the final outcome.

Ethics & Safety