> And attentive readers will know the difference between a text produced by someone who cared about the outcome and one produced by someone who did not.

Likewise, I can also tell the difference between a glass of fresh water and a glass of salt-water... however the process of finding out by testing is not enjoyable, and mismatches lead to waste.

> For me, the perceived value of a text changes once I know

It matters much earlier: The "correct" way to begin reading the text depends on that authorship detail. Usually, reading is an attempt to understand what another human mind was thinking when it generated words. We have a lot of internal circuitry dedicated to that kind of task.

If you go in with that assumption and it's actually LLM-generated, that's just frustration and time lost every time the model you're trying to build crumbles.