This post is part of a series entitled, "Just Say Not to AI." Click here to pull up all the posts in this series (in reverse chronological order). For the first post in this series, click here.
Last week Hachette's Run for It imprint (under Orbit) cancelled Shy Girl by Mia Ballard, a horror novel, because they are saying it was written using a LLM (Large Language Model AI).
You can read about this from the NYT (gift link) but I really want you to look at the more nuanced and philosophical takes by Authors and commentators Lincoln Michel here and Emily Hughes.
- What It Means that Hachette Just Canceled an AI Novel: Quick Thoughts on the Canceled (Seemingly) LLM-Authored Novel by Lincoln Michel
- Mia Culpa: On Shy Girl and Artistic Optimism by Emily Hughes
When this news broke Thursday night. my phone was blowing up. Why, because unfortunately, I have been involved with this story since the end of last year.
Please read the 2 pieces I linked above by Michel and Hughes because they give and excellent run down of what happened.
On my end, I need to share my anger frustration, and sadness about all of this because this is a much larger issue for all of us than even the nuanced essays above get at.
I was assigned to read this book for review in Booklist. Actually, I assigned it to myself. Why? Because I was on Edelweiss looking for horror title to review. I was prioritizing lesser known names from TRUSTED presses. And "trusted" was the key word in that sentence.
Run for It is a newer Horror only imprint under Orbit. Orbit and its parent Hachette are publishers I trust. I also have a great relationship with the library marketing teams for both. Since Run for It started, one of those marketers and I have worked together to get Run for It to more actively promote their titles so we can give them the reviews they need to get their author's books into libraries.
It has been going very well.
So, when I saw they were publishing Shy Girl. I was excited. Ballard is a woman of color, she had great success with this book when she self published it-- some libraries even had bought the self published version to was so popular.
I was excited to download the review copy and pass it on to Booklist for them to assign me a due date for the review.
That review was originally due in January 2026 to be published in the March issue, 1 month ahead of the April scheduled release. As I was reading it, I started to get contacted by others in the horror community to see if I knew anything about Shy Girl and the AI allegations. (I will not name names about who I was talking to but they were trusted people). Again, I was literally in the middle of reading it.
Like Hughes states in her excellent essay above, the book wasn't great but I could see which library users would really love it. I was also VERY wary of any random reddit allegations about a woman author of color because, well you have to be always.
But the voices were getting louder and people were offering "proof," in the form of AI detectors saying ti was mostly AI.
I may be one of the only people out there who has never used a LLM. I haven't even tried ChatGPT for fun. When I say I am against AI, I go all the way. Not even curiosity has brought me to it. I don't write this to get a medal. I am writing it because I would have no idea how to check if something was AI or not. And yet, I am a thinking person and I know the LLM's exist because they were fed actual human writing (including my first Horror textbook). And, as more an more actually writing is fed to the LLM's the writing the AI spits out is more and more like the writing of actual humans.
Further, as a reviewer I also understand that the vast majority of books out there are fine. The writing is fine. They are not all great. This is why when I give something a STAR you should understand it is great. It does rise above. But also, you need to understand this part, if the vast majority of book are fine (at best) and these books are the ones getting fed to LLMs, LLMs are going to be writing kinda meh at best. You cannot take uninspiring writing as 100% AI anymore.
This is an all of us problem for which I do not have an answer and it complicates everything in this discussion and those we will inevitably continue to have going forward.
But back to my disappointments, anger and frustration with this specific situation.
After talking to many people with concerns about this book and if it was written by AI, I immediately reached out to my people at Hachette Library Marketing to explain what I was hearing. We have a great relationship, as I said, and they were equally upset. I wrote up my concerns about the allegations, I shared that I did not want to blindly believe random allegations, but that also, I would not risk my reputation or Booklist's without them letting me know that they were standing behind this book.
Again, this was in January. They refused to engage in a conversation with me. We waited at Booklist for a reply. We even were willing to hold the review until April. I checked back in at the beginning of February to say, we have not heard back, but we can still publish it in the April issue if you get back to us before the end of February.
Side note: Along the way, I also kept Library Journal abreast of the situation and they were also holding reviews until I heard back.
The fact that these allegations were flying and no one would address them infuriated me, not matter what the outcome was going to be, that was not okay. I am not the public. I am not an influencer. I am reviewing this for one of the top trade magazines. I even said in my email, I want to give you the chance to reply here, but I am not risking Booklist's 150 year reputation for this one book.
In the end, at the very last minute and with really no explanation, the book was dropped.
This leaves me even more frustrated and upset and even worse, stuck in the middle as well because Run for It has some good books in their lineup. Craig DiLouie is one of their biggest authors and he is a good guy who writes very good novels. I especially loved his upcoming title (a review of which will be in Booklist soon). I don't want to punish the authors. This is not their fault. But also, can I trust Run for It anymore? I don't know.
Still to this day, no one has reached out. No one has given those of us who want to promote horror and get it into people's hands the time of day even as we asked nuanced, honest, and fair questions throughout this ordeal. And yet, I know their publicists will continue to send me and my editors at BookList and LJ emails with ARCs asking for our reviews. My instinct is to say no to every book they send. Why respect those who have no respect for me. And yet, many of the editors and all of the authors played no part here.
There is no easy answer here. Again read Michel's and Hughes' takes above. However, there is a better answer than cancelling the book and silence.
We are all going to have to deal with this issue coming up again and again. We have to work together as a book community to all be on the same side, the side of 100% human created books. But to do this we have to committee to engaging in the hard conversations with each other. That is what I am trying to do today, by linking to Michel's and Hughes' essays and sharing my experience on this issue.
I am not offering solutions, and anyone who thinks they have the answers-- especially those with answers on how to stop AI created books from being presented as human created (even if only parts of the book are that way)-- are wrong and not to be trusted.
I am hoping to have the chance to talk to others about this specific issue and the larger problems that surround LLMs and books. Obviously, if that happens, I will share more here. In the meantime, you can use the link at the top of this post for more from my Just Say No to AI series.
No comments:
Post a Comment