SelfCheckGPT: Zero-Resource Black-Box Hallucination Detection for Generative Large Language Models

What changed

Uses consistency across stochastic samples to flag potentially hallucinated sentences, with original BERTScore and MQAG variants for black-box language models.

Scientific scope

Uses consistency across stochastic samples to flag potentially hallucinated sentences, with original BERTScore and MQAG variants for black-box language models. Zero-resource means no external factual database, not zero compute or no auxiliary models. Original experiment uses20 sampled passages and QA/answerability components; consistently false answers may evade consistency checks. WikiBio denotes person biographies, not biological research.

Evidence

The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Zero-resource means no external factual database, not zero compute or no auxiliary models. Original experiment uses20 sampled passages and QA/answerability components; consistently false answers may evade consistency checks. WikiBio denotes person biographies, not biological research.

Contributors

• Source-listed authors and technical contributors · SCIENCE: Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.

• SelfCheckGPT: Zero-Resource Black-Box Hallucination Detection for Generative Large Language Models

Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.

Full scientific record · Browse contribution grades · Request a correction