I’m inclined to agree with you, although I think publishers may use AI detection as a first pass or first opinion to help alert them to problems. Some are already doing so, especially since it helps them with author conversations (“why did an AI detector indicate your work may have a high amount of AI-generated writing?”). Whether this is a long-term solution, I can’t say. As you indicate, publishers have to evaluate any work based on their own standards of quality. In the long run, if human authors use AI to assist them in producing high-quality work, and no one can tell, not humans and not detection tools, who do we say is harmed and do publishers really care?