Check an AI summary against the original document
Compares an AI summary line by line with the source to catch errors, distortions, missing points, and things the source never said.
The prompt
Compare an AI-generated summary or answer against the source material it was based on.
The source: [PASTE the original document, transcript, report, contract, or data]
The AI output: [PASTE the summary or answer]
What I'll use the summary for: [DETAILS]
Check:
1. Accuracy - every statement in the AI output is supported by the source. Quote the matching source passage.
2. Distortion - meaning changed through emphasis, softening, exaggeration, or missing qualifiers ("may", "in some cases", "subject to").
3. Additions - anything in the output that is not in the source at all (possible hallucination).
4. Omissions - important points, numbers, risks, exceptions, or decisions in the source that the output left out.
5. Numbers and names - figures, dates, percentages, people, and organizations match exactly.
Return:
1. A table: AI statement | Source evidence (quote with location) | Verdict (accurate / distorted / not in source) | Correction.
2. A list of important omissions, ranked by importance.
3. An overall accuracy rating with one sentence of reasoning.
4. A corrected summary that stays strictly within the source.
Use only the source I pasted; do not fill gaps with outside knowledge. If the source is too long to compare fully, say which parts you checked.
Highlighted [brackets] are blanks to replace with your own details.
Fill in the blanks and use this promptMore Working with AI prompts
What to fill in
- PASTE the summary or answer
- DETAILS
How this prompt is built
This prompt follows the RTF structure (Role, Task, Format). RTF tells the model who to be, what to do and what shape the answer should take. It is the quickest structure that still gives noticeably better results than a bare request.
Related prompts
- Reverse-engineer the prompt behind an outputWorks backwards from text, code, or an image to the prompt that likely produced it.
- List the ways an AI agent could fail and how to prevent eachA failure-mode review for an AI agent: realistic scenarios, how you would notice, the damage, and the safeguard.
- Answer my question, then critique it and argue against itGets an answer, an honest self-critique, the strongest counterargument, and a revised answer you can trust more.
- Turn a one-off AI chat into a reusable prompt pack for my teamExtracts the prompts, fixes, and checks from an AI chat that worked into a step-by-step workflow and prompt pack colleagues can reuse.