Skip to content
Longterm Wiki
Index
Citation·page:reducing-hallucinations:fn4

Reducing Hallucinations in AI-Generated Wiki Content - Footnote 4

Verdictpartial90%
1 check · 4/3/2026

The claim states hallucination rates of 2-18% for RAG-based systems, but the source states 2%-18% vs 39% for nonparametric memory LLMs (CIS Chatbot/Google chatbot) had fewer hallucinations than the parametric memory LLM (conventional chatbot). The claim states 19-40% for conventional approaches, but the source states approximately 40%.

Our claim

entire record

No record data available.

Source evidence

1 src · 1 check
pmc.ncbi.nlm.nih.gov/articles/PMC12425422/resource
partial90%Haiku 4.5 · 4/3/2026

NoteThe claim states hallucination rates of 2-18% for RAG-based systems, but the source states 2%-18% vs 39% for nonparametric memory LLMs (CIS Chatbot/Google chatbot) had fewer hallucinations than the parametric memory LLM (conventional chatbot). The claim states 19-40% for conventional approaches, but the source states approximately 40%.

Case № page:reducing-hallucinations:fn4Filed 4/3/2026Confidence 90%