AI judgement is fake news for authors

I have been increasingly concerned over AI and questions of originality of journalists’ work, authors’ manuscripts, plagiarism.A new manuscript submission as agency made to a publisher was rejected because they ran the author’s text through an AI detector and claimed it was mainly AI generated. The manuscript was an anthology of short stories and true histories the author had written and compiled (about the history of dogs) over more than 10 years. The author claimed that most of the text was written before AI was around. The only editing he has done has been within the confines of MS Word (grammar and spell check). He has “NEVER used AI, ever.”

So I ran portions of the dog book text in Grammarly and Pangram and it came back “42% appears to be AI-generated” and “49% AI-generated,” respectively. Incredible.

So, as a test of these and two other AI detection systems, I ran 15,000 first (unedited) words of a manuscript I wrote in 2018 and was published in 2022 (Elephant Safari) and this was the result in a third AI checker: “75% of your text has signs of AI.” Considering I wrote this thriller on the dining room table in the dark of night without any copy and paste whatsoever, I knew this AI plagiarism was misleading, to say the least.

So I went further back and chose text from a book written in 1990… delivered in Nov. 1990, edited by Victoria Wilson at Knopf and still in print: Marlene Dietrich: By Her Daughter Maria Riva Result? On “JustDone” AI checker: “82% AI content.” This manuscript was handwritten on yellow legal pads.

So, the question we all have to ask is this: If AI memory already contains many of the materials, texts, of published books in AI memory… are they all now considered AI owned/generated? Or is AI actually saying that the material is not new to AI and therefore labels it as plagiarized?

The issue here seems to me to be a definition of “original” – original to whom? If an author sends a Gmail with a manuscript to an editor, Gmail (Google) has the file and their AI can presumably read it. Also, if Google or other AI platforms have scanned a previously published magazine article or a book, I believe the very familiarity of what is in the AI memory will give the result that “AI is familiar with this text” and therefore leads to accuse an author of plagiarism.

As for me, I have no faith whatsoever in these so-called AI detection systems. They provided complete nonsense on something I wrote in 2018 and an author delivered in Nov. 1990! To further illustrate the point, I ran Act 1 of HAMLET: “Most of your text is AI/GPT Generated,” so Shakespeare is also a plagiarist? Teachers, professors, and editors everywhere are relying on these false readings and contributing to fake literature appraisal.

Peter Riva, a former resident of Amenia Union, New York, now lives in Gila, New Mexico.