Ive noticed its kindof endemic in poetry to put unhelpful line breaks which have more to do with visual consistency than any consideration of meaning. Imagine trying to read some of this stuff out loud to an audience. You can easily tell which poets considered how their poetry would be read when they were writing this way, not just how it felt coming out (which, to be fair, is quite important). Tends to be one of the marks of a better poet, but such consideration is neither required nor sufficient for good poetry it has to be pointed out.
Wetness is an illusion. An artificial category imposed on what we call reality by our minds. Which are also illusory. Everything is an illusion. It's all as meaningful and purposeful as the patterns dust makes in a whirlwind. That's all we are- dust in the wind. Everything is dust in the wind.
chatgpt favored a few articles in its training data that used "its not x – its y" more than others. it was sheer accident and coincidence to begin with, but is now being cemented into the language by chatgpt's relative ubiquity and a feedback mechanism where new training data contains this artifact, increasing its favorability in subsequent models. I'm not saying that the phrase didn't exist before chatgpt. it’s not a seismic shift in language patterns – It's a feedback mechanism. The same feedback mechanism causes it to prefer "it's not" vs "it isn't" despite there being no grammatical distinction between them. "It's not" was presumably slightly more popular among the training data it (or rather, its trainers) happened to favor during initial training. The same feedback mechanism causes it to write metaphor like a bored, not-particularly-bright college student in a poetry class they're taking just for the credit.
goddamn am i sick and fucking tired of slam journalism. if this clown wasnt set to shame in restraint for three hours at noon for the entire city to jeer and throw muck at i dont want to hear anything about a fucking pillory.
until the ai carches on and starts intentionally making mistakes to pass our new turing test better than we can