If the editor is only editing for style and readability while adding factual mistakes along the way, they could be entirely replaced with ChatGPT in a tight loop with the science writer. Write article -> make style changes -> writer reviews again for correctness.

It’s inexcusable that the quality of science communication is so low.