Unverified93% confidenceFactExact time
'Hallucination' refers to large language models generating plausible-sounding but factually incorrect content with high confidence
1
Sources
93%
Confidence
Long-term
Relevance
8/2/2026
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79% similarVerified研究表明大语言模型规模越大，幻觉的伪装性往往越强，能生成更连贯、更具说服力的错误内容71% similarUnverified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70% 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% similarUnverified大语言模型在涉及高度专业的化学、生物或网络安全操作时，极易产生听起来可信但存在关键错误的幻觉输出66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/679755API
curl https://kongchang.com/api/v1/knowledge/claims/679755MCP
get_claim(id=679755)