The problem is that there's a vast range of values between “using AI to research/hone your arguments” v. “AI writing your comments for you”, and between the rule itself and dang's various remarks on it, where exactly the rule draws the line is about as clear as mud.