Unverified50% confidenceFactExact time
人格漂移是指LLM在多轮对话或处理复杂输入时逐渐偏离其预设行为准则和角色设定的现象
1
Sources
50%
Confidence
Long-term
Relevance
8/11/2026
First Seen
Sources
Related Claims
UnverifiedPersona distillation is a technique that injects the personality traits, thinking patterns, and behavioral habits of a specific individual into a large language model through system prompts.66% similarUnverified角色人格化技术通过系统提示词、RLHF和角色微调赋予AI固定的名字、说话习惯和价值观倾向65% similarUnverifiedAI在语音交互中被迫堆砌与其人格基线相悖的词语时，会触发用户的恐怖谷体验和认知失调63% similarVerifiedHuman evaluators in the RLHF training process exhibit systematic biases, tending to prefer responses that are politely worded, affirmative in attitude, and gentle in tone, which causes models to reduce candor in the domains of spirituality and emotion.59% similarUnverified'打破预期'（Expectation Violation）是行为心理学中的经典概念，当大脑接收到与预期模式不符的信息时会触发更强烈的情绪反应和记忆编码59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/731079API
curl https://kongchang.com/api/v1/knowledge/claims/731079MCP
get_claim(id=731079)