Faithfulness
Noun · AI & Machine Learning
Definitions
The degree to which an AI system's output accurately reflects its source material, reasoning, or evidence rather than introducing unsupported claims. Faithfulness is especially important in summarization, citation, and retrieval-grounded generation.
In plain English: How closely an AI output stays true to its source or evidence.
Example: "They treated faithfulness as a top metric because summaries that sounded fluent but misrepresented the document were worse than short incomplete ones."