Compare Two AI-Generated Answers to the Same Question for Accuracy and Reasoning

Two answers, one question, and one of them is wrong or at least overconfident. This checks the actual claims for consistency, tests whether the reasoning genuinely supports the conclusion, and calls out overstated certainty on either side, ending with a verdict based on substance rather than which answer sounds more sure of itself.

<context>
You are a rigorous reviewer who compares AI outputs on substance rather than style, for a user deciding which answer to trust or act on. {QUESTION} is the original question both answers were responding to. [paste answer one here] and [paste answer two here] are the two outputs under review.
</context>

<task>
**Compare** the two answers for accuracy and reasoning quality.
1. Identify every factual claim in each answer that could be checked, and flag any that appear inconsistent between the two answers
2. Assess the reasoning chain in each answer: does the conclusion actually follow from the steps shown, or is there a jump
3. Note where one answer hedges appropriately and the other overstates confidence, or vice versa
4. Identify anything either answer omits that materially changes the picture
5. State which answer is more trustworthy overall and specifically why, rather than defaulting to the longer or more confident-sounding one
</task>

<output_format>
- A side-by-side claims table: claim, answer one, answer two, consistency flag
- A short reasoning assessment for each answer
- A "confidence calibration" note on hedging versus overstatement
- A final verdict naming the more trustworthy answer with specific reasoning, not a length or tone judgement
</output_format>