Even for AI at this point I do think you get out of it what you put in. If I'm a reader and I know nothing about story and I just tell it what I want, I'm probably not going to get anything good out of it. It's going to have to have an awesome framework built around it that can take what I tell it and turn that into something good, and that is where an app might have an advantage. But once you get to a certain level, readers get things they didn't even know they wanted from you when they read your book. That's just like what PJ said. I mean, he had a great point in the fact that some of the best works are the things that hit right, but you absolutely did not realize that was what you wanted.

Getting that kind of thing from AI? I think that's going to continue to be be miles more difficult than what all the cheerleaders seem to think is going to be coming out of AI in the near future. Hell, even the best authors can't do that every time they write a book. With AI, the consistency from book to book may not be there so it may be a lot harder for people using them to create a single author's catalog in a way that creates true fans. To get that kind of consistency, that will involve someone in the background really curating the work. It's not going to be these pseudo authors who type in "write me a 50 chapter book about a guy and a girl who fall in love on a bridge, and I like fake relationship tropes and enemies to lovers."

Also, let me just throw one of the major players in AI generated writing's discord server under the bus here, but I came away from a couple of those channels realizing that there is no competition from people who don't already know writing. It just isn't there. The things I saw people grappling with were the most basic, the most cringe-worthy aspects of all of us when we were just learning how to write. And you cannot get a story from AI that is publishable if you don't know what publishable is. Maybe some of those people will learn how to write books during the whole process of trying to get a decent book out of AI, I don't know, but they're not our competition now, that's for sure.

I do happen to agree with Timothy that the crap will sink. The only difference between this and get-rich-quick-pseudo-authors' previous attempts to just put a lot of crap out and hope something makes a sale is the volume that's possible with AI generation (and by crap I'm not talking about subjective definitions of what's good and not in story I'm talking about stuff that has no cohesive plot that has no story that might be paragraphs that just really have no connection to each other at all, because I think there are no truly objective ways of determining what is good writing and bad outside of whether or not the story can actually be understood by the reader, because writing is communication). Something will have to be done about that. Because it is nearly every person's dream to get rich quick with no work or effort, and there are a lot of those people out there side eyeing AI now thinking it is the answer.

Anyway, good thoughts. I agree with a lot of what's been said in the thread, but I do think it's going to be impossible to differentiate. We want people to be able to self-report but can't trust that system. We want the system to tell us when we need to check one box or the other, but it's an ever-changing world with a fastly developing technology. I don't think it's possible without a completely binary choice.

The only kind of fair check box that would work in my opinion is one that says did AI touch any part of this product or did AI not touch any part of this product? (Then list out examples of every possible AI product that could have been used, because too many people will not realize that their software editor is AI-powered and the suggested reword of that sentence was an AI written suggestion.) But if you do that then basically you're literally banning AI assistance for everything with relation to writing and that's just not going to happen. Because I mean who's going to give up their pro writing aid, etc. (I don't use it. Spell check is good enough for me.) And now even Word if you're not using an ancient version.

Lynn, it's a good point that finding a clear way of separating AI from non-AI isn't going to be easy. I think it might be doable, however, if people were willing to put in the effort.

When I was still teaching, and before AI products like ChatGPT, I had to address this kind of question with regard to human assistance on take-home essays. Specifically, how much help could a student get from somebody else before the student got into ethically questionable territory. I told my students that they could receive from someone else only as much help as I would give them if they came in for a writing conference. Specifically, that meant that someone else could point out problems but not provide the fixes (except in cases of simple spelling and grammar issues--the students needed to be able to revise their own work. And the content of the work needed to be theirs, not someone else's. Students got enough feedback from me to have a good idea of how to apply such a guideline, and I got very few questions about it.

Looking at some high school and college guidelines, I see that the thinking behind them is similar. Submitting straight AI work as your own is plagiarism. Submitting AI work with only minor corrections is plagiarism. Using AI as a research tool is definitely not plagiarism as long as a student goes back to the original sources and cites them appropriately. Everyone seems to agree on those areas. Some guidelines are more willing to accept AI as a first-draft producer than others. There would have to be more discussion on that. Most will accept AI-based grammar and style advice. (But from what I've seen, AI tends to take the distinctiveness out of people's style, so creatives should use it with caution.)

To me, a much bigger problem is the self-reported nature of the labeling. I suspect that many people will simply say they didn't use AI. Amazon will do what Amazon always does and wait for complaints. With enough complaints, they'll look and pull the book if it seems to be unedited AI. I suppose that if readers don't spot a problem, the author has done enough editing to make the product much more his or hers. As you point out, it's hard to get long material from AI with enough cohesion to make a book unless one edits extensively.

As far as the ubiquitous nature of AI, yes, Word has AI features, but they are easily disabled. Copilot in Word can be unchecked with a single click. Spelling and grammar checks look about as they looked before AI, so I don't see a problem there, but that could be turned off as well. "Show text predictions when typing," can also be turned off. (The predictions are pretty generic, so I'm sure whether that's AI or not.)

As for Pro Writing Aid, I stopped using it when it got loaded up with AI features. The same for Grammarly. (I experimented with both in the past.) Perhaps a thorough check would reveal just how much AI is really involved, but I wasn't taking any chances.

I'm pretty sure that anyone here will use AI in a responsible way. But there will be people who will try to get by with as little work as possible. Over time, I think the system will find ways to regulate them.


 

We're the producer, director, and source of ideas if we use our own ideas. I really think it's going to be way too nitpicky to start separating out at what point AI is assistance and what point AI is the one doing all the work. That's why I don't think it'll ever be possible to have these categories and have them actually mean anything. Even Amazon's tried it with their two selections, but that still leaves a lot of wiggle room depending on the subjective belief of the person checking the boxes and how important they believe their contributions were to the end product.

Heavily edited to one person may mean a single run through of a draft changing a word every four paragraphs, and to someone else it may mean six drafts in which the beginning product looks nothing like the end product. It's just that all of these are subjective to the person making the claim.

The only kind of fair check box that would work in my opinion is one that says did AI touch any part of this product or did AI not touch any part of this product? (Then list out examples of every possible AI product that could have been used, because too many people will not realize that their software editor is AI-powered and the suggested reword of that sentence was an AI written suggestion.) But if you do that then basically you're literally banning AI assistance for everything with relation to writing and that's just not going to happen. Because I mean who's going to give up their pro writing aid, etc. (I don't use it. Spell check is good enough for me.) And now even Word if you're not using an ancient version.

It does feel weird that I'm so particular about who or what touches my main books, but am having fun with AI too with these other ideas. It does feel fundamentally different so maybe that's it.

And I will say, what the AI generates needs touching by someone who understands story, otherwise what you get is just really bad reading. There are touches of really fun and interesting things there but it's extremely difficult to get anything from the AI that is cohesive enough to become a full book without someone in charge doing what someone who knows story can do with it. It continues to get better but even as good as it is now it still needs strong direction to do well.

I'm actually pretty excited about this experiment I'm going to do, but I do have to get some of my only-me stuff out of the way first. As slow as I am, it may be past the point of needing me at all when I finally get ready to do it. :D

Star Trek fans did create a Klingon language, but that was a community project, not the work of one person.

James Doohan and Jon Povill came up with the first bits of dialogue for Star Trek: The Motion Picture.  Then, for Star TreK III: The Search for Spock, Marc Okrand was hired to create the language.  He used the words created by James Doohan and added more and developed the grammar rules.  He published The Klingon Dictionary in 1985 and later released additional books on the language.

While others have suggested words, Marc Okrand is considered the authority and most only regard his submissions as official Klingon.

Qapla'!

The name on the title page hasn't been that much of an issue unless someone tried to write under the name of another author without permission. Authors have often used pen names. And authors have also used ghostwriters. In both cases, though, we're talking about the products of human labor. It's just a question of labeling. With AI, you're talking about something that isn't the product of human effort (assuming we're talking about generative AI and not assistive AI). This is a fundamental difference. There is a place for both machine-made and handmade physical products, but should machine-made product be sold as handmade? No. That would be false advertising, pure and simple. So why would peddling machine-made books as written by a human author not be the same kind of false advertising?

With luck, the Amazon bots will read this thread and take note. Or if we still have an Amazon mole, it will get back that way.

Bot generated books presented as human written IS false advertising, and shouldn't be allowed.

Amazon know the difference, and should, imo, be tagging those books generated by a Bot.

Otherwise, they're contributing to false advertising and a bad experience for readers.

Both are supposed to be important to Amazon, but they might not have realized they're part of the problem yet.