Forum: Author Hangout

4. I found that AI is good in suggesting how to rephrase paragraphs and in coming up with plot ideas. More of a writer's block remedy for me than a means to speed up writing.

That is where I spend almost all my time, other than final copy editing.

I watched a demo from a professional editor once, who used AI and then went through the manuscript to point out all the errors that the AI didn't catch and all the ones it introduced. Her prompt did require it to follow Chicago Manual of Style, which cut down some errors, but it still was only hitting about 85% correct.

a flood of slop could overwhelm the system. However, since SOL doesn't monetize content, this may not be as large a problem

Lazeez said ZBookStore too.

You mentioned giving plot credit but even the AI has no idea where its pulling it output from even as it sound familiar to us.

I mean the author should be credited as eg Ron Jon and Grok.

AJ

Agreed Switch. I enjoy writing my stories, even when it's painful, and maybe more so when it is. If I just took some ideas and threw them into a machine and cleaned up the output, I wouldn't call myself a writer. It would be like going from being an automotive designer to the guy in the pit at the five-minute oil change place.

But that's me.

I think You, I, and other SOL authors have the same opinion. We take pleasure in the creation of the story. That is our return for the expended effort. Some of us also sell our stories.

P.S.

To those of you who use AI to write any part of your story:

Do you experience any Pride of Authorship?

How are you viewed by others in regard to the AI generated portion(s) of the story - author or editor?

To those of you who read stories containing AI generated content:

Do you get a sense of duplication between the stories posted by authors using AI?

What are your feelings about the story's originality regarding plot and wording?

To those of you who use AI to write any part of your story:

Do you experience any Pride of Authorship?

How are you viewed by others in regard to the AI generated portion(s) of the story - author or editor?

To those of you who read stories containing AI generated content:

I use AI the exact same way I would use a human editor. Every scene starts and ends with me. Anything suggested by the AI, even general copy editing, must be approved by me. I read every word offered and rejected a fair amount. I am proud of what I've written, and believe a competent editing job has made it better. My strengths are characters and story. I'm not as good at narration and action beats.

Last year, I submitted 3 stories to that Lit site, all written prior to 2023 and they were all rejected for being AI generated. It seems their AI detection tools are not much better than the AI writing tools.

Nowadays, I use it to check my punctuation and spelling before I send it to my editor. I know it doesn't catch everything but it helps cut down on what my editor tells me to fix.

I use PWA, but never use the AI rewrite/suggestion tools. They are pretty clearly labeled. Having it suggest simple grammar changes will not make a story AI-generated, but it has that potential if you want it.

I understand what you're saying and I am sure it is pretty accurate. In my case I had written the stories in question before I had even hesrd of things like Grammarly. I had no editor at the time and spent hours going through the stories finding spelling errors and misplaced commas and periods.
When I look at those stories now, they were pretty bad but they weren't AI. haha

The em-dash is a GREAT AI-watermark

Why?

I use the em-dash and nothing in my stories is touched by AI.

I use the em-dash and nothing in my stories is touched by AI.

Same here. I've been using the em-dash for close to 15 years. The idea that the em-dash is an AI giveaway is one of the great myths surrounding AI in general. I've been working with Gemini in various capacities for almost a year now, personally and professionally. I even use it instead of search most of the time, and I hardly ever see it use an em-dash. Granted, that's a sample size of one, but still...

That's why it's such a great way to tell if AI was used. It's a very good indicator.

And very wrong. There are 31,000 em-dashes in my complete corpus, and I've used them for the 10+ years I've posted to SOL.

Typing them on a Mac is trival, as is en-dash. Opt+hyphen for en-dash, Opt+shift+hyphen for em-dash. It's no more difficult than typing €, ñ, é, å, ü, ö, ß, etc., on a US keyboard once you know the keystrokes.

And I already wrote... YOU are the exception. Part of a TINY minority on this (and any other online story) site.

I have 71 in 167k words

Fact is, the em-dash isn't even on our keyboards anymore.

Either is the … (ellipsis) which, on a Mac, is the Opt+semi-colon.

Anyone who uses the Chicago Manual of Style will use the em-dash. And most (if not all) U.S. publishing companies use it or some variety of it.

Seriously, how many books are being published based on what was claimed, in big publishing houses, using that " Chicago Manual of Style".

I don't think big publishers bother with it nowadays, if they ever did. As long as a submission has been formatted appropriately and is comprehensible, they don't put any effort into enforcing a style.

AJ

they don't put any effort into enforcing a style

Of course they do.

They won't reject a manuscript if it doesn't follow their standards, but they will change it for publication.

Actually, one way for your manuscript to get rejected is by not following their submission standards. For example, if they say double spaced and you single space it, it probably will immediately go into trash. They have standards for reasons.

Publishers rarely do any editing these days. They expect authors to supply print-ready copy So even books with the same publisher can present a mish-mash of styles.

AJ

So even books with the same publisher can present a mish-mash of styles.

I'd be surprised if that were true. If a freelance journalist submitted an article to the NY Times, I'm sure the Times would insist he followed their style guide (which is like the AP Style Guide). No way would they allow the Oxford comma. In fact, the journalist probably followed it because that's what journalists use. Fiction writers hoping to be published should be smart enough to submit a manuscript that they wrote using the Chicago Manual of Style. If they didn't, and the publisher didn't want to spend the money to edit it, they would probably simply tell the author to fix it.

Fiction writers hoping to be published should be smart enough to submit a manuscript that they wrote using the Chicago Manual of Style.

Why? Publishers don't care and it won't improve the stories if the authors have their own style. Some authors get books trad-published without speech marks, without commas and, I believe in at least one case, without any punctuation at all.

AJ

There are at least three recent cases of books published complete with AI prompts in the text. I daresay there could not have been any competent editor reviewing the work.

Out of interest, do you know whether they were trad-published ie by a major publishing house and available in print, or by a small publishing company, self-published or some other method?

AJ

Not sure. I suspect Amazon (so self-published, more or less).

See: https://futurism.com/fantasy-novel-ai-prompt-copy-style

At least some of these are authors with large fan bases, mind you, so the potential for there to have been a competent editor still applies. They're not first-timers, and they likely would have had the resources to have an editor review the books, if they had chosen to spend them.

we can place all of your stories against the MILLIONS of AI-generated stories that use the em-dash, and then decide again whether it is a good indicator for AI-generated content.

If you're talking about millions of AI generated stories, you're not limiting the comparison to stories on SOL. So don't just use my stories. Use the millions of traditionally published novels. Hell, the AI probably learned by reading those novels.

Hell, the AI probably learned by reading those novels.

That is, quite literally, how the AIs were trained. The tech is an off-the-scale enormous next-word predictor based on billions of sentences it's been trained on and the probability of 'what comes next'. If it's producing an em-dash, it's because there's a very high probability of it appearing there in pre-existing work.

Again, that's literally how the tech works.

AI has also been trained on social media, hence certain sentences in AI-generated stories being readily traceable back to social media posts.

Allegedly, AI developers are desperate for more material to steal because all the millions of books they've stolen so far have left AI still less than serviceable.

AJ

Allegedly, AI developers are desperate for more material to steal because all the millions of books they've stolen so far have left AI still less than serviceable.

When it comes to LLMs, I suspect there's an exponential involved eg they'd have to square the number of stolen books input to double the quality of the AI output.

I wonder whether they've stolen the Chicago manual of Style. How would eg Grok respond if one asked it to edit a passage to make it conform to CMoS?

AJ

And I already wrote... YOU are the exception. Part of a TINY minority on this (and any other online story) site.

Which demonstrates the problem succinctly — false positives are FAR worse than false negatives.

Granted, on THIS site, Lazeez is reasonable and responsive.

Other sites will ban authors (or tag them as AI) because they actually know how to use standard English punctuation, and there will be no recourse.

That's ridiculous.

Scrivener also turns '--' into an em-dash. I use that constantly. Every em-dash in my work is human-generated, and there are a lot of them.

I use the em-dash

I don't. I think it's falling out of use in everyday English and I suspect most readers think it's a hyphen. And among the authors who use em-dashes, quite a few don't know how to use it properly.

But them I'm a fan of the KISS principle.

AJ

I think it's falling out of use in everyday English

We're talking fiction. In other forms of writing, for example, what's within a pair of em-dashes would be inside parentheses. And sometimes the way it's used in fiction, a colon would be used in other forms of writing. And don't forget interrupted speech in dialogue is depicted by an em-dash. (Trailing off speech uses an ellipsis.)

It is definitely not falling out of favor.

It is definitely not falling out of favor.

That is an assertion without evidence, just like mine.

I'm not sure how you'd go about getting a definitive answer. I don't think Google Ngrams can help.

Google seems split. Allegedly em-dash usage has recently mushroomed because of its popularity with AI and tech companies. But other sources say it's declining because eg it's difficult to use.
And because of its popularity with AI and tech companies!

AJ

I'm not sure how you'd go about getting a definitive answer.

Simply because the Chicago Manual of Style still says to use it. And that's what writers of fiction use.

Simply because the Chicago Manual of Style still says to use it. And that's what writers of fiction use.

We'll never agree on that. I've never met a published author who gives Chicago a second thought, even if they've heard of it. Perhaps it's different in America and you're all taught Chicago in school but that seems unlikely to me.

AJ

I've never met a published author who gives Chicago a second thought

From https://www.acrolinx.com/blog/four-editorial-style-guides-that-every-writer-needs-to-know/#:~:text=The%20Chicago%20Manual%20of%20Style&text=It's%20the%20standard%20for%20book,well%20as%20style%20and%20usage.

Currently on its 17th edition, The Chicago Manual of Style is beloved by writers, editors, and publishers. It's the standard for book publishing in fiction and nonfiction and is often used in the arts and humanities for academic papers. It has a lot of instruction on the publishing process, such as preparing a manuscript, proofreading, formatting, and citation, as well as style and usage.

From https://thewritelife.com/writing-style-guide/#:~:text=CMOS%20is%20a%20set%20of,two%20variations%20of%20source%20citation.

Who uses Chicago Manual of Style?

Chicago is the preferred style of print publishers in both fiction and nonfiction

From https://rightyourwriting.com/2020/07/09/what-style-guide-should-i-use/

The mainstay of the trade publishing industry, CMOS is my first suggestion to most clients. Not only does it cover most grammar questions (new) authors have, it also overviews things like how to handle text messages in novels. If you're not sure how you should write something, CMOS likely has an opinion on it.

And the list goes on and on and on.

Chicago Manual of Style

There may be a bit of Cross-Atlantic crossed wires going on here. British publishers do not reference the CMoS - why would they? American is a different language from English.

There is a British equivalent - the New Oxford Style Manual, but a lot of British publishers have in-house style guides too.

British publishers have in-house style guides too.

I imagine all major trad-publishers, British and American, have an in-house style. But that's for print layout etc, the double-spacing that SB mentioned above, for example. A good agent will usually have software to coerce stories into that format. I believe Amazon and other on-line publishers have a pre-processor that does something similar.

AJ

None of those links, as far as I can tell, are from published authors. They're from 'writing experts' and such.

Show me a Stephen King or a James Patterson, respected authors who are passionate about educating beginners in their craft, telling writers they need to follow Chicago. They don't, because it wouldn't be true.

AJ

Maybe that's one of my chapters that AI had never touched got flagged in an AI testing site - I love the em-dash!

I make extensive use of the emdash, endash, and other less-frequently used marks (including dieresis), so I wonder what will happen there.

including dieresis

I wouldn't know when to use or how to make.

I wouldn't know when to use or how to make.

Mostly out of use in English, but looks like an umlaut.

In English, this ö has a diaeresis and is used to inform the reader that the two letters are not a diphthong, e.g., coöperate, naïve (one of the few still regularly used).

At least on a Mac, you make it the same way you add an umlaut — Option-u + char.

Mostly out of use in English, but looks like an umlaut.

It should appear in some botanical Latin names (which allegedly are mostly Greek) eg Drosera scorpioïdes, but most people omit and mispronounce them.

AJ

AI is, in fact, spectacularly bad at determining if something was AI-generated. Figuring out what does work is a seriously hard problem, one that is leading to PhD-level work and to significant business opportunities.

AI is, in fact, spectacularly bad at determining if something was AI-generated.

I do find it odd that people who are "anti-AI" will quite happily use an AI detector tool that is built on AI platforms.

I've also found that Quillbot is spectacularly bad at detecting AI. ZeroGPT is better, but still not great.

I ran an experiment and generated a 'scene' for a story. I fed the original Gemini-generated scene, edited version that I'd worked on "a little" and "a lot", and then a complete rewrite of the same scene into both QB and 0GPT.

Quillbot told me all of them were "100% human". 0GPT did better, but didn't label any of them as 100% AI. On the edited versions, quite a bit of the text that it flagged as AI was actually bits I'd edited.

Again, a sample size of one, but still...

To be honest, it doesn't really fill me with any confidence in the AI detectors at all.

When writing AHL, I used Gemini for feedback, brainstorming and occasionally word choice or sentence structure problems. I would feed the scene I'd written to the AI and ask for it's general impression and feedback. I'd then use that feedback to make changes to the scene, feed it back and make more changes until I was happy with what I'd produced. The feedback was generally constructive (as per my instructions to the AI at the start of the task), and it was 'useful' (by which I mean the suggestions were genuine areas to be improved and the suggestions on how to improve it were pretty good) about 70% of the time, there or thereabouts.

In the editing process, I did the same thing, fed in the story scene by scene, but gave it very different instructions at the start. Its responses were therefore more those of an "editor" than a "collaborator," picking grammar issues, spellings, weak word choice and poor sentence structure.

Exactly my process.

To those of you who use AI to write any part of your story:

Do you experience any Pride of Authorship?

Your story "A Healing Love" does not meet the criteria specified in the first line for as you said, "It's very much my own work."

Will you create a tag that we can use to self report?

I think he was talking about 'AI polished' and 'Mix AI/Human'.

I'd like 'organic', to signify AI-free. Given the negative feelings some readers have towards AI, I can imagine it's a tag readers might search for.

AJ

I didn't see it when I posted my story. I'll look again and fix it, thanks.

I have a question....

What happens if someone has a... weird... way of writing that could be mistaken for AI but the individual may be autistic/dropped several times as a child/whatever, and their stories get flagged as AI, but they're not? What is the procedure, how do they even go about challenging the accusation? Is there a process? How would you even go about investigating?

I have to admit, I've never knowingly read an AI story, however, I've recently heard a song that I really liked and when I went looking for more songs from the band, I found out it was totally AI generated. I honestly, couldn't tell. Possibly musicians could easily do so. But me? Not a clue.

If Terry Pratchett was just starting to release novels today, would he be accused of AI use, because his work is like nothing else out there? (There is stuff like it out there now, because people try to emulate his style).

I know you have all said "There are tools", but for a mere pleb like myself, you may as well be talking in a foreign language for all the comprehension I have.

AI is getting better and doing so scarily fast. It wasn't long ago that you could easily spot AI generated images because the people in them had fifteen fingers each. Now it's getting really, really hard. Will the same issue arise for AI generated text? Personally, I think that it's going to go the same way as AI images.

Obviously, if the story was written in the 90's, then it's going to be a mute point, and if the author has a body of work stretching back decades, then there is going to be a personal style involved that can also be used. But what about 'new' writers?

I have to admit, that the contrarian in me is also having intrusive thoughts, demanding that I deliberately write and submit a story written like it was written by an AI (Which is problematic because I don't actually know at this time what an AI written story looks like, but let's not let facts get in the way of a good rant) because, well I've no idea why really. I think I am basically, fundamentally, annoying.

What's the point of going to extremes to write like an AI?

For some individuals, there is always a point, even if no one else sees the...err... point. Isn't that one of the things that makes us...us?

I see and understand your point. I was just...curious.

The whole thing about AI text that it's all so similar and predictable that it becomes tedious and boring to read.

I have the same opinion of some writers who churn out the same 'book' with just a few name changes. Clive Cussler springs to mind in that regard and there are a few other 'mainstream' authors I have read over the years who are similar. Early warning signs (for me) are authors who have a many, many, books in a series. Repetition inevitably sinks in, partly (I think) due to pressure from publishing companies to keep writing 'What works' leading (possibly) to a fear from the writer to try something new.

Another name has just popped into my head, Jack Campbell, writer of The Lost Fleet series. I stopped reading his work because, again, it was just the same idea repeated with different names.

And another thing...

inflicting such an atrocity on readers should be punished.

There is already a mechanism in place for that, it's called scor

:user has been terminated :

I know you're making a joke, but it still needs addressing.

I thank you for all that you do for this site, and it wasn't so much as a joke, as a tacit acknowledgement that I cherish my access level on the site on the proviso that I don't talk about that which I'm not to talk about. 😂

But I appreciate your point on the potential damage being done before ****** become visible. I hadn't thought about that.

If there are stories that you suspect are AI generated, but aren't tagged, report them to me and I'll check them.

That presents a moral dilemma. There are some stories I believe are mostly AI-generated, with the usual florid language, superficial characters, continuity errors between scenes etc, but they're amongst the most highly rated stories on the site. If labelling them as AI-generated causes the authors to remove them, I think that would be deleterious to the site.

Is there, perhaps, a softer way? Every story uploaded gets checked for AI content, your gizmos calculate a probability, and every story gets that probability published as a tag eg AI:48%. If the authors disagree, at least management isn't directly responsible because it's a case of 'the computer says'.

AJ

Can you give an example of those supposedly "most highly rated stories on the site", despite having useless prose and continuity errors?

I said 'florid', not 'useless'. Florid language has its uses, for example in literary fiction where stories don't have plots, or poetry, where the poets can't think of rhymes :-)

I'm not going to publicly name stories, as that might create the sort of friction that I'd like to protect management from.

AJ

I'm not going to publicly name stories, as that might create the sort of friction that I'd like to protect management from.

One clearly AI-generated story, which now bears the Ai tag, has provoked its author to blog that their stories are not AI-generated.

And that's a true story ;-)

AJ

I think that would be deleterious to the site

Given that AI detection (from what I understand from this and similar threads) works on 'flowery' verbosage, then would AI flag AJ for using deleterious (Which I had to go look up, as I have never seen it before) instead of the more 'common' detrimental?

Given that AI detection (from what I understand from this and similar threads) works on 'flowery' verbosage, then would AI flag AJ for using deleterious (Which I had to go look up, as I have never seen it before) instead of the more 'common' detrimental?

71 SOL stories contain 'deleterious'.

I can confirm that you are not an AI because your stories get "its" and "it's" wrong far too often. Unless you're a very smart AI ;-)

AJ

😂

I like to be consistent...

I can confirm that you are not an AI because your stories get "its" and "it's" wrong far too often.

In your story 'The Night Is Dark, Full of Terrors & Is Darkest Before the Storm', after Dearglil Ardatlamash takes possession of Peggy's body, she refers to Peggy as It (capital letter, therefore a name). When Dearglil uses the possessive form, I reckon it should be "It's" not "Its". But I'm open to counter-arguments.

AJ

My understanding is that it's is always the contraction of it is. The possessive is its.

Why?

English is weird.

It follows the pattern of pronouns…

His, hers, its, ours, yours, theirs

'Mine' is the outlier.

I was responding to your OTHER post. Edited mine to show that.

I swear the authors of the 19thC were paid by the word but had to pay the publishers for every period they used.

Most of the best authors of the time were women, and they hated periods.... (boom-tish!)

How about a "Not AI" tag? Just think of the fun you could have when some one uses it on an AI story.

Is authors having fun the reason why they sometimes tag works of fiction as true stories?

AJ

Here is my situation, which I've been open about - I have one story that was AI generated and I edited. I labeled it was AI generated as soon as I was aware the tag existed (I may have posted before the tag's creation). My main work is AI edited, where I read every word of the edits and have final say on what stays in. The rest of my stories have never been touched by AI.

But if I understand your new policy correctly, I as an author will be tagged with an AI Scarlet Letter, and people will be led to assume that all my works posted here are at least AI assisted.

Thank you for the clarification.

For me its an editing tool. Memory and not being speciafially designe for creatinve writing yet is the biggest draw back.

Nails it!

~ JBB

Thanks. Fix that and the the ethical AI bullshit and it will be a much better tool. Won't replace a human for the spark of an original idea.