Techniques & Methods
Response Quality
Response quality assesses how well an AI output fulfills the user's request. Key dimensions include factual accuracy, relevance to the query, logical coherence, appropriate length, and tone. Automated metrics (BLEU, ROUGE, BERTScore) and human evaluation are both used.
In RLHF, reward models trained on human preferences score response quality to provide training signal. High response quality standards are also critical for AI search products—low-quality cited sources reduce trust in the platform.
Authority Links
Related Terms
Techniques & Methods
Evaluation Metrics
Quantitative measures used to assess how well an AI model performs on a task.
Techniques & Methods
Reinforcement Learning from Human Feedback (RLHF)
Training technique that refines AI models using feedback from human evaluators on output quality.
Model Components
Reward Models
Models trained to score AI outputs based on human preferences for use in reinforcement learning.
Techniques & Methods
Hallucination
When a language model generates confident-sounding text that is factually wrong, invented, or misattributed — a structural consequence of next-token prediction over learned patterns rather than retrieval from a verified knowledge base.