Uses for AI tools

[NoTalentHack]

This is not an "AI is going to eat our babies" post, although I think it probably will, metaphorically speaking. But that's not for a while yet.

In the meantime, let's exploit it!

Even if it wasn't the case that I like writing my own stories, the long content that ChatGPT generates is mostly crap. That's okay; that's not really what it's meant for, at least in a fiction context. But what about using it in ways that IS intended? Factual research is right out, because it sucks at that. There are technical reasons for that that I won't go into unless someone really wants to nerd out, but suffice it to say that ChatGPT is more of a glorified autocomplete algorithm than anything else.

So, given that factual research and long-form generation are both out, what about smaller things? We've (most of us) used rhyming dictionaries, thesauruses, etc. and ChatGPT can be used in a similar way. As an example, I realize that I used phrases involving masks slipping away a bunch when talking about someone's composure failing. I probably will, but I wanted some alternatives. So, I got the AI in the right "frame of mind" and asked it to go to work. Here's the exchange:

You are a writer trying to find new ways to phrase things you've written previously in an attempt to avoid repeating yourself. Come up with phrases that convey a similar idea as "her mask fell away." The context is someone who's trying to put on a brave face and starts failing.

1. "The facade of strength crumbled before our eyes."
2. "The veneer of composure slipped, revealing vulnerability."
3. "The cloak of courage unraveled, exposing raw emotions."
4. "The pretense of bravery dissolved into fragility."
5. "The charade of bravery gave way to an overwhelmed soul."
6. "The illusion of fortitude shattered, unveiling inner turmoil."
7. "The shield of stoicism cracked, exposing a fractured heart."
8. "The act of resilience faltered, revealing a struggling spirit."
9. "The mask of strength slipped, laying bare the weight of emotions."
10. "The façade of bravery slipped like sand through trembling fingers."

Give me a similar list, but less florid.

1. "She couldn't keep up the act any longer."
2. "Her brave front began to crumble."
3. "The strength she was trying to show started to slip."
4. "Her brave face faded, revealing vulnerability."
5. "The facade of courage began to crack."
6. "She struggled to maintain her brave demeanor."
7. "Her attempt to be strong faltered, showing her true feelings."
8. "The mask of bravery fell away, exposing her emotions."
9. "She couldn't hide her vulnerability any longer."
10. "Her brave façade gave way to her inner struggles."

Now, let's be honest: most of these are either crap or basic or both. I'm probably not going to use any of them, especially verbatim. But as a way to brainstorm without having a writing partner at hand? It's at least as good as rubberducking it. (Rubberducking, for the non-programmers, is to use a toy and talk to it as you try to solve a problem). I'm sure there are other uses that can enhance our writing ability/process without supplanting us; anyone got any ideas? I'm happy to hassle the bot with them.

 

[onehitwanda]

"Her brave front began to crumble."

Her rear, though, rose higher.

"Her brave face faded, revealing vulnerability."

And well-sculpted cheekbones - literally.

"The facade of courage began to crack."

She'd put it on before the plaster was quite dry.

"She struggled to maintain her brave demeanor."

It was an expensive hobby and required constant upkeep.

"Her attempt to be strong faltered, showing her true feelings."

She was repulsed, and not in a good way.

"The mask of bravery fell away, exposing her emotions."

Crushing disappointment

"She couldn't hide her vulnerability any longer."

Her panties were far too thin and far, far too negligible.

"Her brave façade gave way to her inner struggles."

Her anal-beads were pinching her.

I'm so sorry.

 

[NoTalentHack]

Her rear, though, rose higher.

And well-sculpted cheekbones - literally.

She'd put it on before the plaster was quite dry.

It was an expensive hobby and required constant upkeep.

She was repulsed, and not in a good way.

Crushing disappointment

Her panties were far too thin and far, far too negligible.

Her anal-beads were pinching her.

I'm so sorry.

Thank you, ChatOHW.

 

[onehitwanda]

 

[alohadave]

I used ChatGPT to create a chant for my 750 word story Tribal Wedding Ceremony. https://literotica.com/s/tribal-wedding-ceremony

for something like that, it worked fine.

 

[MediocreAuthor]

The only thing I've used ChatGPT for is basically as a glorified thesaurus.

"Describe a fancy mansion"

I then took the description, picked it apart and used the terms and expressions that I liked. I'm sure I could do much more with it, but I still find it spooky and depressing, and I'm sure I'll be replaced by a machine soon. Remember, we're all chippers, and bronze is coming.

 

[NoTalentHack]

 

[getspastic]

I have had an account for quite some time and been a Literotica reader for even longer; I have yet to contribute. One of the reasons for that, and the reason you might find my thoughts interesting on this subject; I write 3k - 5k words a month of non-fiction for publication, some as a paid contributor and others because of the 'thought leader' status that is important to my professional life. Also have a couple of books under my belt. No bragging there; just to say that I publish a lot and write even more, so I have a decent perspective on the subject in general.
First, I will tel you 100% for certain that GPT is being used on Literotica, and extensively so. If you engage in the content that it produces consistently like I do (more on that below) you can 'taste' it within a paragraph or two, and I am seeing more and more of it.
However, if you are a solid author, GPT can for sure be an amazing tool without taking away your voice. I have a system I use (it would have to be tweaked for erotica as that is against the use case of OpenAI) that allows me to produce quality, original, and on-point content with about a 40% reduction in writing time. This works for gpt 3.5 and 4, but 4 is so far above 3.5 its not even a comparison.
I start out with a simple 'write this for me' prompt. "GPT, I need your help. I am writing a 700 word article on the effects of X in Y for Z publication. Can you give me a draft to work from?" I let it write an entire article. I might do this 2 - 3 times to see a few options. I think land on an outline of an article, and draft an intro parapgraph and maybe 1 or 2 more paragraphs.
Then I start a new chat. Super important; I don't want it 'thinking' about the article it already wrote. I'll then say something like; "GPT, I am working on an article about the effects of X in Y for Z publication. I would love it if you would help me. Here's what I thinking the overall article will look like [basic description of the outline I have], and here are the first couple of paragraphs I have written, can you provide the next two only?"
I take what is written and either customize it or tell GPT it went the wrong direction and why. I also might say 'we are talking here about X. I feel like there is something really unique about Y that we could use to talk about X metaphorically; is there?"
Once I have a couple more paragraphs of text that I am happy with I feed the whole thing back (in the same chat) "GPT, with your input here's where I am at so far; what's next?" and rinse and repeat.
If you compare what GPT wrote with what I finish with, they are probably 10% similar at most, but using the above I have done 1,200 polished words in 2 hours.
I used to take the final work to an AI checker, I don't any more. With the above method I always get back somewhere between 85% - 95% 'human written'. The parts labeled as AI are almost always original text written by me.
So that's just my $.02. I am certain that my current publications (mainly 800 -1600 word articles by the way) are stronger than they were before, with a more coherent straight-line thesis running through an article that is still 100% 'authentic'; still my voice. However I am getting the added benefit of being able to see multiple ways that a couple of paragraph sub-topic can be fleshed out in a matter of seconds.

TL;DR, AI can be more than a thesaurus. It can help you produce content that is authentically 'you' faster, and it can make your writing stronger.

~MP

 

[MillieDynamite]

Don't and won't use AI for writing.

 

[dmallord]

I have had an account for quite some time and been a Literotica reader for even longer; I have yet to contribute. One of the reasons for that, and the reason you might find my thoughts interesting on this subject; I write 3k - 5k words a month of non-fiction for publication, some as a paid contributor and others because of the 'thought leader' status that is important to my professional life. Also have a couple of books under my belt. No bragging there; just to say that I publish a lot and write even more, so I have a decent perspective on the subject in general.
First, I will tel you 100% for certain that GPT is being used on Literotica, and extensively so. If you engage in the content that it produces consistently like I do (more on that below) you can 'taste' it within a paragraph or two, and I am seeing more and more of it.
However, if you are a solid author, GPT can for sure be an amazing tool without taking away your voice. I have a system I use (it would have to be tweaked for erotica as that is against the use case of OpenAI) that allows me to produce quality, original, and on-point content with about a 40% reduction in writing time. This works for gpt 3.5 and 4, but 4 is so far above 3.5 its not even a comparison.
I start out with a simple 'write this for me' prompt. "GPT, I need your help. I am writing a 700 word article on the effects of X in Y for Z publication. Can you give me a draft to work from?" I let it write an entire article. I might do this 2 - 3 times to see a few options. I think land on an outline of an article, and draft an intro parapgraph and maybe 1 or 2 more paragraphs.
Then I start a new chat. Super important; I don't want it 'thinking' about the article it already wrote. I'll then say something like; "GPT, I am working on an article about the effects of X in Y for Z publication. I would love it if you would help me. Here's what I thinking the overall article will look like [basic description of the outline I have], and here are the first couple of paragraphs I have written, can you provide the next two only?"
I take what is written and either customize it or tell GPT it went the wrong direction and why. I also might say 'we are talking here about X. I feel like there is something really unique about Y that we could use to talk about X metaphorically; is there?"
Once I have a couple more paragraphs of text that I am happy with I feed the whole thing back (in the same chat) "GPT, with your input here's where I am at so far; what's next?" and rinse and repeat.
If you compare what GPT wrote with what I finish with, they are probably 10% similar at most, but using the above I have done 1,200 polished words in 2 hours.
I used to take the final work to an AI checker, I don't any more. With the above method I always get back somewhere between 85% - 95% 'human written'. The parts labeled as AI are almost always original text written by me.
So that's just my $.02. I am certain that my current publications (mainly 800 -1600 word articles by the way) are stronger than they were before, with a more coherent straight-line thesis running through an article that is still 100% 'authentic'; still my voice. However I am getting the added benefit of being able to see multiple ways that a couple of paragraph sub-topic can be fleshed out in a matter of seconds.

, AI can be more than a thesaurus. It can help you produce content that is authentically 'you' faster, and it can make your writing stronger.

~MP

This method may help with non-fictional writings or with those straight-line thesis articles you mentioned. Probably not going to handle my thoughts on writing erotica since CHAT-gpt v3 doesn't want to contribute to my efforts, not even at a 5% level.

I tried a couple of less erotic paragraphs to see how those might work. The feedback doesn't sound like my voice and hence doesn't seem authentic 'me.' An outline approach for a story might work more favorably as a form of feedback.

As I write this, Grammarly is trying its best to highlight my and your errors in what you submitted. I find that humorous that we are discussing AI, and one is clinging dearly to our words, trying to raise its hand and yell out, "Over here! Look, you messed up this line and these words."

I do appreciate the effort you put into describing your methods. It was inciteful, though it ran against the grain of my way of writing. Experimentation is wonderful, but like Milly, I'm inclined to remain attached to using my own thoughts and words - except for that nagging and wonderful Grammarly thing.

 

[dmallord]

This is not an "AI is going to eat our babies" post, although I think it probably will, metaphorically speaking. But that's not for a while yet.

In the meantime, let's exploit it!

Even if it wasn't the case that I like writing my own stories, the long content that ChatGPT generates is mostly crap. That's okay; that's not really what it's meant for, at least in a fiction context. But what about using it in ways that IS intended? Factual research is right out, because it sucks at that. There are technical reasons for that that I won't go into unless someone really wants to nerd out, but suffice it to say that ChatGPT is more of a glorified autocomplete algorithm than anything else.

Now, let's be honest: most of these are either crap or basic or both. I'm probably not going to use any of them, especially verbatim. But as a way to brainstorm without having a writing partner at hand? It's at least as good as rubberducking it. (Rubberducking, for the non-programmers, is to use a toy and talk to it as you try to solve a problem). I'm sure there are other uses that can enhance our writing ability/process without supplanting us; anyone got any ideas? I'm happy to hassle the bot with them.

5. "The facade of courage began to crack." Now this one is my language use, so I'd make use of this. The others, not so much.

Thanks for explaining the term 'Rubberducking.' I had visions of some ... rather strange sex act involving feathers and a ... condom in the mix. Glad it turned out to be a programming term. Now I know why they seem 'different' than the rest of us!

 

[getspastic]

This method may help with non-fictional writings or with those straight-line thesis articles you mentioned. Probably not going to handle my thoughts on writing erotica since CHAT-gpt v3 doesn't want to contribute to my efforts, not even at a 5% level.

I tried a couple of less erotic paragraphs to see how those might work. The feedback doesn't sound like my voice and hence doesn't seem authentic 'me.' An outline approach for a story might work more favorably as a form of feedback.

As I write this, Grammarly is trying its best to highlight my and your errors in what you submitted. I find that humorous that we are discussing AI, and one is clinging dearly to our words, trying to raise its hand and yell out, "Over here! Look, you messed up this line and these words."

I do appreciate the effort you put into describing your methods. It was inciteful, though it ran against the grain of my way of writing. Experimentation is wonderful, but like Milly, I'm inclined to remain attached to using my own thoughts and words - except for that nagging and wonderful Grammarly thing.

I couldn't agree more; though I write non-fiction my 'voice' is why I am asked to contribute to the publications I am featured in. That's why I re-write what GPT gives me and feed it back in to continue. What I find useful is the filling of the 'unimportant' stuff. Example:
"GPT models, including the more recent iterations such as GPT-4, have transformed the landscape of fictional writing in various ways. These AI models offer a powerful tool for writers, both professional and amateur, to generate ideas, flesh out characters, and build intricate plots. The models can be trained on a vast assortment of literary styles and genres, enabling them to mimic these styles and produce coherent, original content. A writer can feed the model a few sentences or even just a plot outline, and the AI can develop this into a full-fledged story. It's also possible to use the AI to emulate the writing style of a particular author or to generate dialogues for characters. As a result, the writing process becomes more efficient and creative, as writers can draw on the AI's outputs for inspiration and modification."
(I generated that for this conversation)
I would probably keep the "generate ideas, flesh out characters, and build intricate plots" but maybe "generate new ideas, flesh out characters, and help build intricate and interconnected plots" (I probably wouldn't have thought of the last bullet point on my own) and "A writer can feed the model a few sentences or even just a plot outline, and the AI can develop this into a full-fledged story. It's also possible to use the AI to emulate the writing style of a particular author or to generate dialogues for characters." but changed to "You are able to feed the model a couple of sentences or a completed plot outline and let the AI go to work; it'll develop a full-fledged story for you. AI will try to emulate your writing style but you'll most certainly need to tune up the finished product."

Anyway, I digress and am not saying you should use it; I was simply presenting a use case that has worked for me.
Also, FWIW, although not as good at generation, the 'playground' of openAI has the Davinci-003 model which, while not unmoderated, is pretty damn close. It won't touch underage or family, but other than that with the right prompting it will generate basically on any subject.

 

[dmallord]

...

Anyway, I digress and am not saying you should use it; I was simply presenting a use case that has worked for me.
Also, FWIW, although not as good at generation, the 'playground' of openAI has the Davinci-003 model which, while not unmoderated, is pretty damn close. It won't touch underage or family, but other than that with the right prompting it will generate basically on any subject.

I follow your thinking. I will take a look at the Davinci-003 model. Thank you for that golden nugget. Learning and experimenting should never end. Open minds expand one's knowledge. I'm all about developing new approaches - and wouldn't rule out AI forever. It may be a good friend one day.

 

[ElectricBlue]

Her rear, though, rose higher.

And well-sculpted cheekbones - literally.

She'd put it on before the plaster was quite dry.

It was an expensive hobby and required constant upkeep.

She was repulsed, and not in a good way.

Crushing disappointment

Her panties were far too thin and far, far too negligible.

Her anal-beads were pinching her.

I'm so sorry.

Wanda, I'm not sure you're... taking the topic seriously. Which is probably the best way to approach AI right now .

 

[BrightShinyGirl]

Don’t use it, won’t use it.

I tried it a few months ago for my non-erotic fiction, hoping it might spark some ideas I could use. What I got was an utterly genetic rewording of my prompt.

My writing process depends heavily on stumbling upon minor felicities in the first draft and elaborating upon them as I revise.

A first draft that’s a bowl of mush is useless.

 

[XerXesXu]

This is creative.

using ChatGPT to decide which books to ban

 

[MillieDynamite]

I'm never in favor of banning books, ideas, art, or the creativity of a person. But if you must ban, having AI do the work for you is lazy and unfair to the person whose work you ban.

This is creative.

using ChatGPT to decide which books to ban

 

[astuffedshirt_perv]

Don’t use it, won’t use it.

I tried it a few months ago

Seriously? This story is moving so fast, 'a few months ago' is a long time. Like saying you won't use iPhones becasue you didn't like the iPhone 3. ChatGPT got updated 9 Aug, Claude 2 was launched in July 2023. The tools are improving fastly.

 

[BrightShinyGirl]

Seriously? This story is moving so fast, 'a few months ago' is a long time. Like saying you won't use iPhones becasue you didn't like the iPhone 3. ChatGPT got updated 9 Aug, Claude 2 was launched in July 2023. The tools are improving fastly.

I’ve always been a late adopter. I didn’t get a smart phone until my boss threatened to throw my Nokia across the bar we were in.

I’m happy with the writing process I have now. I don’t need a new tool.

 

[MillieDynamite]

BrightShinyGirl, you have a remarkable resemblance to Audrey Hepburn! You must be her reincarnation.

 

[webkeeper]

First, I will tel you 100% for certain that GPT is being used on Literotica, and extensively so. If you engage in the content that it produces consistently like I do (more on that below) you can 'taste' it within a paragraph or two, and I am seeing more and more of it.

~MP

Yup. 3 major offenders: "symphony", "primal" and "testament". Any appearance of these 3 words is a major tip off ChatGPT was used to produce the story.

 

[NoTalentHack]

Yup. 3 major offenders: "symphony", "primal" and "testament". Any appearance of these 3 words is a major tip off ChatGPT was used to produce the story.

Uh. I use primal, like, a lot, and I’ve never directly used ChatGPT output. Hell, even if I didn’t use it, there’s a whole “primal” kink out there. Don’t know about the other two, though. I may have used synphony!once or twice?

This is the problem with AI detection: it’s just a firm of pattern recognition, and people are primed to see patterns even where they don’t necessarily exist. Anyone who lived through the Satanic Panic of the 80s (Proctor and Gamble’s logo says 666!) can tell you that.

 

[Bramblethorn]

I'm never in favor of banning books, ideas, art, or the creativity of a person. But if you must ban, having AI do the work for you is lazy and unfair to the person whose work you ban.

Lazy and unfair is the point of the exercise, surely.

 

[Bramblethorn]

Yup. 3 major offenders: "symphony", "primal" and "testament". Any appearance of these 3 words is a major tip off ChatGPT was used to produce the story.

Every word GPT uses is a word it's seen used by a human. Every word used by a human is a word that humans use.

Uh. I use primal, like, a lot, and I’ve never directly used ChatGPT output. Hell, even if I didn’t use it, there’s a whole “primal” kink out there. Don’t know about the other two, though. I may have used synphony!once or twice?

This is the problem with AI detection: it’s just a firm of pattern recognition, and people are primed to see patterns even where they don’t necessarily exist. Anyone who lived through the Satanic Panic of the 80s (Proctor and Gamble’s logo says 666!) can tell you that.

My first story here, published 2012, uses "symphony" twice. Because the love interest is a professional musician.

 

[MillieDynamite]

Most liekly.

Lazy and unfair is the point of the exercise, surely.