From Newsgroup: talk.origins

https://www.science.org/content/article/new-preprint-server-welcomes-papers-written-and-reviewed-ai

From 1994 to 2022 I attended the Plant animal genome meeting in San
Diego. I met a lot of interesting people at those meetings. One of
them was a professional research paper writer who worked for a medical
school in the late 1990's. He did what AI is doing for some researchers
now. His job was to take the data produced by the medical doctors and
turn it into a research paper. He claimed that a lot of the doctors
didn't have a clue as to how to present their data and they might not
even know if the data was worth publishing. He had to do the work like identifying similar research and referencing the correct studies.

In writing a blog piece (I'm still working on it) I came across a paper
that claimed to have found regulatory sequence variants that altered the expression of MC1R (melanocortin receptor 1). The paper was so messed
up that you could not conclude much of anything from it. They put the
wrong sequence in their figures or mislabeled their figures. The titles
of the sections were inconsistent with what was in the paragraph, and
their materials and methods were inconsistent with the results
presented. AI could have likely done a better job if the authors could assemble an accurate collection of their data. These guys even
incorrectly labeled the positions of the sequence that they had in their figures. They tried to do a phylogenetic analysis on what was likely
randomly associated sequences between birds and mammals. I do not know
how the paper got published.

The paper was so bad that I put a review of the paper in the Appendix of
the blog piece that I am working on. I have never reviewed a paper this messed up, and somehow it got published.

Ron Okimoto


--- Synchronet 3.21a-Linux NewsLink 1.2