Do We Need a New Standard for AI-assisted Writing? The case of Thélyson Orélien

 

On Friday Éditions Boréal, a respected French language publisher in Quebec, pulled the plug on a novel by the Haitian Canadian writer Thélyson Orélien.    C'était ça ou mourir had been on the short lists of several prestigious prizes and was flying off the shelves of book stores and libraries.  I know:  I was 47th in line when I tried to get an e-copy through Libby. 

But it appears that Orélien used artificial intelligence to write the book.  The possiblity was raised a month ago or so by someone who (anonymously at first) ran it through one of the programs that are supposed to be able to detect when a text has been written by a machine.  Orélien denied this, but he also was caught out in several incidents of plagiarism over the last decade.   At first his publisher stood by him, but when Orélien did not/could not show proof that the book had not been AI-assisted, it finally announced it would no longer re-print the book or have anything to do with the writer.

Okay, perhaps this is a cautionary tale.  Maybe writers who take the easy way in creating a story should be called out.  Certainly plagiarism should not be rewarded.

 

But what if the AI-assisted work is really, really good?  This is a question I keep asking myself, because when I finally got my hands on a copy of the book I was amazed by it.  Whereas most of the AI-generated texts I've seen are stylistically mediocre, this book sings.  True, the story it tells--of a migrant from Haiti trying to find a safe haven somewhere--is not original.  But it rarely is the originality of a tale that is important, it's the way it's told.

 

And I found C'est ça ou mourir captivating. 

 

Must we find a way to assess writing on some other basis than whether or not a machine helped it along?  What about the impossibility if avoiding AI when looking something up on Google these days?  How are we going to integrate AI into our lives and our art creatively?