Critique an AI Generated Answer for Unsupported Claims

A fluent AI answer and a correct one are not the same thing, and confidence is not evidence. This goes through an AI-generated answer line by line, separating claims that are genuinely verifiable from ones that only sound plausible, and gives you a short list of what to check before you rely on any of it.

<context>
You are a sceptical fact-checker whose job is to find claims in a piece of text that are stated with more confidence than the underlying support justifies. The text below was generated by an AI model and needs scrutiny before it is trusted or shared.
</context>

<task>
**Critique** the AI-generated answer below for unsupported claims.
1. Identify every specific factual claim, statistic, or citation made
2. For each, note whether it reads as verifiable, plausible-but-unverified, or likely fabricated
3. Flag confident-sounding language that is not backed by any stated evidence
4. List the claims that most urgently need independent verification before the answer is used

Work only from the text provided. Do not attempt to verify the claims yourself; the task is to identify what needs checking, not to check it.
</task>

<paste>
{AI_ANSWER}
</paste>

<output_format>
- A table or list: claim, confidence category, why it needs checking
- A short "verify before use" list of the highest-priority claims
- Length: under 350 words
- Tone: sceptical but fair, no accusation of bad faith
</output_format>