The problem varies dramatically by field and subfield. ML is by far not the worst affected. There are fields that get papers are published in "reputable" journals that are transparently nonsense, they get a large number of citations, and this happens all the time.
