That's definitely begging the question.
If you're prepared to accept that GPT-4 can answer questions just as well as humans can, why do you even need to do prompt engineering?
If you're prepared to accept that GPT-4 can answer questions just as well as humans can, why do you even need to do prompt engineering?
* What's the best way to get to Radio Shack from here?
is not the same as
* What's the easiest way to get to Radio Shack from memory when riding a bicycle from here?
You may be under the impression that annual U.S. deaths from medical errors being in the hundreds of thousands miscommunicates but that is truly your opinion. You are merely jumping to conclusions at places another person might not.
And going on to rely on the LLM to validate your perspective is a lossy process. It may not lose your perspective but it loses someone else's and you don't even seem to notice or care.
The post you replied to was saying that the deaths were caused by miscommunication, but you interpreted it to mean that stating the number of such deaths is somehow a miscommunication itself!