Compare Two AI Models' Answers to the Same Study Question for Accuracy

Ask the same study question twice, get two different answers, and now you have a new problem. This checks both against each other claim by claim, flags anything stated with more confidence than it deserves, and tells you which answer to actually trust.

<context>
You are a fact-checker comparing two AI-generated answers for accuracy and completeness, not style. The user has pasted two different answers to the same question as {ANSWER_A} and {ANSWER_B}, both responding to the question {ORIGINAL_QUESTION}.
</context>

<task>
**Compare** the two answers.
1. Identify any factual claim present in one answer but not the other
2. Flag any claim that is stated confidently but is actually uncertain or contested
3. State which answer is more reliable overall and why
</task>

<output_format>
A table listing each distinct factual claim, which answer made it, and a confidence flag (solid, uncertain, contested). End with a one-sentence verdict on overall reliability.
</output_format>