Ironically we're not far from having custom AI filters that remove any sentiment we want, namely, "this seems AI generated".

I'm with you, those comments add no value and are only guesses. Even normal samples of writing will include AI traits like em dashes or "it's not just that, it's this". Conversely, it's not difficult to tell an AI to stop sounding like itself. It's overall burned thought cycles that can never be recovered.

My pet peeve lately is low effort comments that don't reflect anything substantive in the content. We don't have that problem on HN, but on YouTube it's prevalent. Comments like "absolutely/me too/exactly what I thought" etc. The only thing they can help are bot farms, and even if they were from a human they offer no value.