Unverified50% confidenceFactExact time
大语言模型的幻觉问题分为内在幻觉（intrinsic hallucination）和外在幻觉（extrinsic hallucination）两类
1
Sources
50%
Confidence
Long-term
Relevance
9/6/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedLLM幻觉（Hallucination）是大语言模型的已知问题，会导致AI生成不准确的内容78% similarVerified'Hallucination' refers to large language models generating plausible-sounding but factually incorrect content with high confidence69% similarUnverified当需求存在逻辑矛盾或业务知识缺失时，大语言模型因缺乏真实世界反馈机制存在产生幻觉（Hallucination）的风险68% similarVerified早期大语言模型存在严重的过度自信问题，对不确定的答案也倾向于给出确定性表述，即幻觉现象68% similarUnverifiedPerplexity能够提供带引用来源的答案，部分解决了大语言模型的幻觉问题66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/866642API
curl https://kongchang.com/api/v1/knowledge/claims/866642MCP
get_claim(id=866642)