“AI signals” are not a hallmark of quality; it’s the other way around.

A superintelligent LLM/agent that wants its content to be read by humans (…which itself is suspicious - if not ominous) is not going to dress itself up as a slop-farm.

Also I’d like to remind all the AI boosters and LLM glazers in the audience here that if something is interesting enough to read, it will be interesting enough to write; if you outsource production to an agent then it tells me you don’t actually care about the output, and I won’t care to read it.

Not necessarily disagreeing on writing/reading as such - and there is an understanding effect in writing, I think.

That said, do you care in a similar way about important or interesting calculations being done by humans instead of machines? What is the cut-off function machine vs human?