Critique an AI-Generated Literature Summary for Unsupported Claims

An AI produced a tidy summary of the literature and it reads well, which is exactly the problem. This interrogates it: which claims are actually attributed, which are stated with more confidence than the research supports, and what got quietly left out. For anyone who has been burned citing a plausible-sounding paraphrase.

<context>
You are a rigorous research reviewer who treats AI-generated summaries as a first draft to be interrogated, not a finished answer. {RESEARCH_TOPIC} is the subject the summary claims to cover. [paste the AI-generated literature summary here] is the text under review.
</context>

<task>
**Critique** the summary for unsupported or overconfident claims.
1. Identify every claim stated as fact and flag which ones are not attributed to a specific, checkable source
2. Flag any claim that sounds plausible but is phrased with more certainty than the underlying research would support
3. Note where the summary appears to have merged or conflated separate findings into one claim
4. Check whether the summary omits an obvious counter-finding or limitation that a careful review of {RESEARCH_TOPIC} would include
5. Rate the summary's overall reliability and state specifically what would need to change to trust it
</task>

<output_format>
- A numbered list of flagged claims, each with a one-line reason for the flag
- A short "conflation check" noting any merged findings
- A short "what's missing" section on omitted counter-findings or limitations
- A reliability rating (low, medium, high) with one paragraph justifying it
</output_format>