Unverified50% confidenceOpinionExact time
自回归解码机制存在两个与事实性相悖的缺陷：所有token在训练时被赋予相同权重，以及一旦生成错误token后续会自洽延伸形成幻觉雪球效应
1
Sources
50%
Confidence
Long-term
Relevance
7/15/2026
First Seen
Sources
Related Claims
UnverifiedThe 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.69% similarUnverified在目标检测任务中，标注矛盾会同时影响分类分支和回归分支两个学习目标68% similarUnverifiedThe 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.68% similarUnverified机制可解释性领域面临的核心挑战是多义性（Polysemanticity）——单个神经元往往同时编码多种语义无关的概念64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/517750API
curl https://kongchang.com/api/v1/knowledge/claims/517750MCP
get_claim(id=517750)