Humans can produce noise, but humans using LLMs can produce orders of magnitude more of it.
But your stance is reasonable. If it improves the product, who cares who/what produced it. I personally find reviewing machine-generated code and the process of code review with a machine much more exhausting than interactions with humans, but you may feel otherwise.