论文进阶

Hallucination

在大模型/AI 语境中，指模型生成的内容缺乏输入、检索证据或可验证事实支持，即看似真实但实为捏造的内容。

English Definition

"In LLM/AI usage, the phenomenon where a model generates content not supported by its input, retrieved evidence, or verifiable facts—i.e., fabricated or false outputs that may appear plausible."

常见错误

·不要把它与心理学中的'幻觉'（perceptual hallucination）混为一谈；在 LLM 语境下特指'生成无依据内容'，常带引号以示隐喻。

真实用例 · 2 条

"the deployment of general-purpose LLMs in healthcare is hindered by their propensity to "hallucinate" facts, where evidential support is lacking."

论文Pipeline extracted — review needed·通用大模型在医疗领域被阻碍于'捏造事实'的倾向

"while RAG significantly reduces hallucinations, it does not eliminate them entirely."

论文Pipeline extracted — review needed·RAG 显著降低但未完全消除幻觉

由 pipeline 自动采集，待人工 review