Expired50% confidenceBenchmarkExact time
Multiple benchmark tests in 2023 (such as the HaHackathon dataset) showed that top LLMs still have significantly lower accuracy than ordinary human annotators on sarcasm recognition tasks
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/9/2026
First Seen
Valid until: 10/7/2026(expired)
Sources
Why Can't AI Understand British Political Farce? The Cultural Comprehension Dilemma of Algorithms
hackernewshackernews7/8/2026
Related Claims
Unverified在事实性评估任务上模型评判者准确率仍显著低于人类专家，但在风格、连贯性等主观维度已接近人类评审员间一致性水平72% similarUnverifiedMeituan LongCat launched the Lohosearch search agent benchmark, where the best score among 11 frontier models tested was only 34.74%, far below the approximately 90% human performance.70% similarUnverified研究显示，在专业技术领域，LLM 的幻觉率显著高于通用知识领域，因为训练数据中领域专家的纠错反馈相对稀少67% similarUnverified在标准行人检测数据集（如MOTChallenge）上训练的模型，在俯视视角下的检测精度（mAP）通常下降40%-60%67% similarUnverified2022年Chen等人在ACL发表的工作发现GPT系列模型在'解释为何好笑'任务上的表现（约62%人类满意度）显著低于'识别是否好笑'任务（约78%）66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/381002API
curl https://kongchang.com/api/v1/knowledge/claims/381002MCP
get_claim(id=381002)