Unverified50% confidenceFactExact time
Large models exhibit hallucination phenomena, fabricating non-existent product terms, calculation formulas, or regulatory rules in an affirmative tone.
1
Sources
50%
Confidence
Long-term
Relevance
9/21/2026
First Seen
Sources
Related Entities
Related Claims
Unverified大模型在生成代码时会产生幻觉，编造出看似合理但实际不存在的函数名、参数签名或API接口79% similarVerified大模型幻觉是指语言模型生成看似合理但实际不正确或不存在的内容79% similarUnverified大模型幻觉是指LLM在生成内容时以高置信度输出看似合理但不符合用户需求的内容77% similarVerifiedLLM产生幻觉的深层机制在于模型被优化为生成'听起来合理'的文本而非'事实准确'的文本，模型没有独立的知识检索模块74% similarUnverified对话式大模型输出的证明文字基于统计模式生成，无法保证每个步骤逻辑严格成立74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/938183API
curl https://kongchang.com/api/v1/knowledge/claims/938183MCP
get_claim(id=938183)