Clues that AI or plagiarism is being used to write reviews?

[Gerry Steen]

When I read a review that could be copied, pasted and used almost for any book, I suspect that there has been a use of AI. The author's name and the title of the book has been used but there is no proof from the review that the reviewer actually read the book. For example there might not be any mention of specific and verifiable content from the book.

Another thing that makes me suspicious is if I read reviews of a person who manages to write a review almost everyday. Unless the person stays at home all day and does nothing else but read books all day, this is almost impossible. I believe that it is easier for them to read multiple reviews from other reviewers and then write a review using their information. If someone is doing this, their reviews are probably not well-written and probably breach guidelines. Their reviewer score probably under 50 %. If all of the above apply, I am questioning if they are really reading the books. I am glad that OBC now limits people to selecting only one book at a time. This can limit but does not stop this kind of cheating.

What are your thoughts?

[Shirley Ann Riddern Labzentis]

Yes, Gerry, I have noticed a lot of AI reviews, and certain reviews have two or three reviews in a matter of two days. I've often thought that it is impossible unless they have no life. They need more staff here and have people who just go through reviews and posts for any AI detection.

[Moonbroch+Asterock]

This is a very late reply, but I wanted to say that you've provided some very sound advice. I always try to give the benefit of the doubt when it comes to summarizing or details of the book that hints whether or not they really did read the book. Especially when there is such a thin line between alluding to specific plot points that show they read the book and outright spoiling the whole book. I can also see how providing a basic and vague review can be suspicious because it can be copied and pasted from other review. Honestly, the workaround should be in the other parts of the review such as the likes/dislikes and the star rating.

[Tiffany Dowell]

I'm a stay-at-home mom living out in an unincorporated community, retired, financially stable, house paid off so nope, I don't need to work for anything. It's not that I don't have a life, I've lived it, and now I am living the rest that other people are working to reach. I could easily do multiple children's book reviews per day, considering that most children's books are under fifty pages long. Right now, I'm reading a fifty-three-chapter novel. There is NO WAY someone is reading more than one and writing a good review per day. I get up at 7 and generally go to bed at 2 am, and it takes me a few days normally. I can easily see some taking a few weeks if you account for doing the REAL work.

[Melissa Katherine]

When I was editing reviews, I figured out a few ways that help look for plagiarism and AI. If you copy and paste a person's review in a Word document, then go to "editor," you can check for similarity. Several times this brought up links to Goodreads where the reviewer had copy and pasted someone's Goodreads review. Also, I found the websites GPTZero and Phrasly. Copy and paste the review into one of these sites, and they'll give you a percentage of how much of the work was written by AI. Lastly, I asked ChatGPT to write a review on a book that I kept getting to edit reviews of just to see what it told me. Most of the time, ChatGPT will provide a name for the main character, but it's not actually the name, so that's a huge clue (a simple search on the book's Amazon or Goodreads page will usually tell you what the book is about and what the main character's name is). ChatGPT also can be vague and broad in its description and would focus on character development and the author's writing style.