It's lazy and doesn't include context, so I get why there would be backlash

And it's something I do a lot of... hopefully well received 😉

Basically, as an SME, I assume I have a line of questioning that will get the right answer out of the LLM. It shows my train of thought. The idea, if if the user sees what I'm thinking, it will help them do the same later.

Similarly, what I do on CodeReviews, is rather than say outright: "this is wrong, you should do XYZ", I will use

@coderabbitai

as an objective third-party, and instead say:

@coderabbitai does this follow our PRD for XYZ? I thought we should ...

@coderabbitai should we do this or should we instead define XYZ...

Some people may be skipping that step and that turns people off to reading something generated by AI

I do see how it can be used as a sort of tie breaker in a PR scenario, but I think stuff like design choices should have consensus from a team

Things that may come off as nit picky should just be put in checks, eg. “This file doesn't have a new line at the end”