Unverified50% confidenceFactExact time
OpenAI published the paper "Why Language Models Hallucinate" in 2025, in which it observed that DeepSeek-R1 spells out DEEPSEEK letter by letter before counting, whereas DeepSeek-V3 directly gave 2 or 3 (the correct answer is 1).
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/1/2026
First Seen
Valid until: 1/30/2027
Sources
Related Entities
Related Claims
UnverifiedOpenAI、DeepMind等机构的多项研究都证实了大语言模型谄媚问题的普遍性70% similarUnverifiedSources cited in the article include Kalai et al., "Why Language Models Hallucinate" (arXiv:2509.04664), an official OpenAI article, Ji et al.'s hallucination survey (ACM Computing Surveys), and the SimpleQA data in the GPT-5 system card.69% similarUnverifiedOpenAI explained in a September 2025 paper, "Why Language Models Hallucinate", that most training and benchmarks only score answers as right or wrong and give no credit for "I don't know", which leads models to lean toward guessing.67% similarUnverifiedLeading institutions such as OpenAI, Anthropic, and DeepMind all conduct large-scale red team testing before releasing large language models.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1124160API
curl https://kongchang.com/api/v1/knowledge/claims/1124160MCP
get_claim(id=1124160)