Yes, but abstracting all that, the answer given by the better one is not useful either, and taking it at face value can be dangerous.

An "infection"? Of what? Viral? Bacterial? Of a gunshot wound? In your lungs? The model should recognize bad questions and be a lot more nuanced when replying. (For the record, ChatGPT is NOT being nuanced, it's simply parroting propaganda).