I find the bait of being flattered – my questions are smart! my writing is good! – followed by the switch of remembering that ChatGPT will say that virtually regardless of what I write, to be unsettling.
It's not really a novel insight, but the article does a nice job of simply expressing why the tone of so many llms is irritating.
Solid point. Sometimes the fake flattery feels more creepy than helpful. I rather have honest direct feedback from AI mas totoo mas useful.