if you can't think of a way to reliably distinguish AIs from humans, that observation alone should raise great concerns which eclipse "spam comments on forums" or "bad results on google"