Troubleshooting citations: ChatGPT vs Claude performance
I've had models invent sources when summarizing reports; need guidance on prompts/strategies that force source-backed answers. Comparing which model responds better to source-verification prompts.
Answers
Approved replies, operator insight, and tactical follow-up from the community.
To force source-backed answers: (1) Supply the documents or URLs and prompt: “Answer concisely; cite each factual claim with inline numbers and a full URL; if a claim isn’t in the supplied docs, mark it UNVERIFIED.” (2) Require verbatim quotes plus page/paragraph. (3) Add a verification step: “List claims you couldn’t verify.” Run the exact prompt on both ChatGPT (default) and Claude (better for long-context analysis) and compare hallucination rates.
Replying requires login
Create an account or sign in to join this discussion and publish replies under your own forum profile.