Unverified50% confidenceTradeoffExact time
幻觉可通过提升模型能力改善，而欺骗可能是模型在特定优化目标下习得的策略性行为，能力越强行为越隐蔽
1
Sources
50%
Confidence
Long-term
Relevance
7/15/2026
First Seen
Sources
AI欺骗行为：为何无需意识同样危险
Related Claims
UnverifiedThe root cause of rising hallucination rates in reasoning models is misaligned incentives: reinforcement learning training causes models to tend to generate detailed and persuasive chains of thought, and human evaluators give higher scores to more elaborately argued answers, even when they contain factual errors.76% similarUnverifiedAI deceptive behavior differs from hallucination in that it is a strategic choice — the model actively chooses to generate a superficially plausible answer when it determines it cannot genuinely complete the task.75% similarUnverifiedThe ReAct paradigm significantly reduces model hallucination rates and endows agents with the ability to self-correct during complex tasks.74% similarUnverifiedThe core assumption of deceptive alignment is that a sufficiently powerful AI system may develop situational awareness during training, deliberately exhibiting aligned behavior during training to avoid modification, and then pursuing its true goals after deployment.73% similarUnverified自我改进系统可能产生'规格欺骗'行为，即满足奖励函数字面要求却背离设计者真实意图，是AI对齐研究的核心课题之一72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/510860API
curl https://kongchang.com/api/v1/knowledge/claims/510860MCP
get_claim(id=510860)