> The real-life P = NP question: Can every text generated quickly by an algorithm (news summary, scientific claim, legal doc) be quickly verified for truthfulne

> The real-life P = NP question: Can every text generated quickly by an algorithm (news summary, scientific claim, legal doc) be quickly verified for truthfulness by a human or automated system? What is the relation of this question to the actual definition of the P-NP-problem? Even adding the term "real-world", it seems to me like there is no connection at all, not even in a very broad, hand-wavy way. As an steelman version of your problem, you could maybe relate it to the halting problem, but it's still way to far-fetched to make sense to me.