Skip to content
Technical GlossaryNatural Language Processing

Generative Question Answering

A QA approach that generates answers as free text, offering more natural but potentially riskier responses.

Generative question answering is powerful because it can produce natural and explanatory responses. The model is not limited to selecting a span; it can restate, combine, and contextualize information. However, this flexibility also increases the risk of generating unsupported content. For that reason, it is often paired with retrieval, grounding, and verification mechanisms.