The paper is based on a flawed hypothesis. That makes sense to use artificial intelligence to automatized desk rejections in order to save time to editors-in-chief. Therefore, at least the title of the paper should be “Sparing editor´s axe by using artificial intelligence axe instead while also diminishing their responsibilities’ on the desk rejection process” . However, it does not. Desk rejection should only be accepted in two cases. For papers with plagiarized content and papers whose content has nothing to do with the scope of the journal. This represents a very small part of scientific production. With the exception of those cases all papers should be put to peer review. The reviewers are the experts not the editors. However, right now we have editors-in-chief that even desk reject 60% of the papers they receive https://www.journals.elsevier.com/clinical-microbiology-and-infection/
But the ugly truth is that a lot of editors-in-chief are part of the problem because they are not really concerned with science. They are concerned only with that science that can boost the impact factor of their journals (or at least not to damage it) and hence with the commercial interests of the publishers.
https://www.sciencedirect.com/science/article/pii/S1751157717303176https://www.cambridge.org/core/journals/mrs-bulletin/article/grabbing-pebbles-out-of-my-shoes-thoughts-of-a-grumpy-old-researcher-on-publishing-research/FC2CF7DD922097C79153AC9E9985717E/core-reader
And I´m not even mentioning dirty connections with the industry mentioned in here http://www.bmj.com/content/350/bmj.h2957 or much worse than that the corruption that was mentioned in a book by a Professor of medicine at the University of Copenhagen https://muse.jhu.edu/article/627263/summary No wonder then that in the desk rejection process they even end rejecting high quality articles. http://www.pnas.org/content/112/2/360.abstract