Citation·page:reducing-hallucinations:fn17

Reducing Hallucinations in AI-Generated Wiki Content - Footnote 17

Verdictpartial90%

1 check · 4/3/2026

The claim states 97.9% factual accuracy in human conversations, but the source specifies this accuracy is specifically for conversations about recent topics. The claim mentions a multi-stage verification process, but the source does not explicitly use this term, although it describes the process of grounding on Wikipedia and retaining only grounded facts.

Our claim

entire record

No record data available.

Source evidence

1 src · 1 check

hai.stanford.edu/research/wikichat-stopping-the-hallucination-of-large-language-model-chatbots-by-few-shot-grounding-on-wikipedia

partial90%Haiku 4.5 · 4/3/2026

NoteThe claim states 97.9% factual accuracy in human conversations, but the source specifies this accuracy is specifically for conversations about recent topics. The claim mentions a multi-stage verification process, but the source does not explicitly use this term, although it describes the process of grounding on Wikipedia and retaining only grounded facts.

Case № page:reducing-hallucinations:fn17Filed 4/3/2026Confidence 90%