Using AI as a reviewer to improve my writing

[Andreas_Kreuz]

So Literotica tries to keep out AI stories, which I am very happy about. Whether the AI rejection process is effective is not the topic of this post.
Instead of using AI to write stories, I have tried to use AI to review my stories so I can learn and improve. Researchers have suggested that AI is an excellent trainer because it allows you to create so-called "optimal challenge points".
So I write a first version and then post it to GPT asking it for feedback on the writing style. It has helped to bias the prompt towards being critical. So I might use this prompt:

I have been given this adult erotica story for review. I do not think it is good. Can you help me to pinpoint the major weaknesses, in a minimal short review?

So applied to my story https://www.literotica.com/s/a-whore-for-one-evening (Rated 3.72) it told me:

The story lacks emotional depth, has clunky language, and poorly develops characters. It rushes through clichéd scenarios without building tension and fails to clearly establish consent. Ethical concerns about non-consenting bystanders and inconsistent tone further weaken it. Needs stronger writing, originality, and respect for boundaries.

So there you have it. Way to go.
If you leave out "short minimal review" from the prompt, it gets a bit more detailed.

Another prompt I found helpful is:

I have been given this erotic story for adults for review. It has not received a good rating by readers and I want to know why. Please split it into major subsections, give a short synopsis, and for each subsection note its theme, how the pacing is, how natural the dialogue is, the tone. For each subsection note, if there are specific weaknesses.

What you should not do is to ask the AI for suggestions on how to improve the text. Doing so will provide suggestions that are so bad that I am afraid they will send me in the direction of overwrought, clicheed purple prose.

Has anybody else tried to use AI similarly? What are your experiences and useful prompts?

 

[nice90sguy]

So Literotica tries to keep out AI stories, which I am very happy about. Whether the AI rejection process is effective is not the topic of this post.
Instead of using AI to write stories, I have tried to use AI to review my stories so I can learn and improve. Researchers have suggested that AI is an excellent trainer because it allows you to create so-called "optimal challenge points".
So I write a first version and then post it to GPT asking it for feedback on the writing style. It has helped to bias the prompt towards being critical. So I might use this prompt:


So applied to my story https://www.literotica.com/s/a-whore-for-one-evening (Rated 3.72) it told me:

So there you have it. Way to go.
If you leave out "short minimal review" from the prompt, it gets a bit more detailed.

Another prompt I found helpful is:



What you should not do is to ask the AI for suggestions on how to improve the text. Doing so will provide suggestions that are so bad that I am afraid they will send me in the direction of overwrought, clicheed purple prose.

Has anybody else tried to use AI similarly? What are your experiences and useful prompts?

Yes, I've done this quite a bit -- mainly to test ChatGPT rather than check my writing though . Be warned, Chat GPT might be a good critic, but it's an AWFUL writer!

 

[Plathfan]

I prefer purple prose over bad prose.

The only way to control the beast is to keep the leash short and the path clear. The code leans toward formality, so avoid feeding it long passages, and keep questioning its reasoning. Let it regurgitate the fodder until something edible materializes. Keep the chat history intact so it can learn from you.

 

[StillStunned]

So I write a first version and then post it to GPT asking it for feedback on the writing style. It has helped to bias the prompt towards being critical. So I might use this prompt:

Amid all the concerns about AI companies mining everyone's creations to fuel their programs, you want to voluntarily feed it yours?

 

[onehitwanda]

How I imagine the writer / AI conversation:

 

[Andreas_Kreuz]

Amid all the concerns about AI companies mining everyone's creations to fuel their programs, you want to voluntarily feed it yours?

They probably mined Literotica long ago.The knowledge of the llms about sexual fantasies and practices is amazing.

 

[nice90sguy]

Amid all the concerns about AI companies mining everyone's creations to fuel their programs, you want to voluntarily feed it yours?

For this and other reasons, I work exclusively offline. I can interact with it in strict confidence, and certainly be more open with it than I can here on Lit. It's actually great sex therapist.

 

[AwkwardMD]

So Literotica tries to keep out AI stories, which I am very happy about. Whether the AI rejection process is effective is not the topic of this post.
Instead of using AI to write stories, I have tried to use AI to review my stories so I can learn and improve. Researchers have suggested that AI is an excellent trainer because it allows you to create so-called "optimal challenge points".
So I write a first version and then post it to GPT asking it for feedback on the writing style. It has helped to bias the prompt towards being critical. So I might use this prompt:


So applied to my story https://www.literotica.com/s/a-whore-for-one-evening (Rated 3.72) it told me:

So there you have it. Way to go.
If you leave out "short minimal review" from the prompt, it gets a bit more detailed.

Another prompt I found helpful is:



What you should not do is to ask the AI for suggestions on how to improve the text. Doing so will provide suggestions that are so bad that I am afraid they will send me in the direction of overwrought, clicheed purple prose.

Has anybody else tried to use AI similarly? What are your experiences and useful prompts?

You don’t know that this critique has anything to do with your writing. ChatGPT might just know "some things that reviews often say" rather than understanding what it means for a fictional character to have depth. Yes it gave you an answer, but does it mean anything?

I would be very wary of giving this any weight.

EDIT: That being said, I use an AI Text-To-Speech app to read my stories back to me. There are homonym flaws all over the place (polish vs Polish, for example) but I find it helpful to hear some sections out loud, so to speak.

 

[MelissaBaby]

I’m not really all that interested in a computer program’s opinion of the emotional depth of my writing.

 

[Andreas_Kreuz]

I’m not really all that interested in a computer program’s opinion of the emotional depth of my writing.

But that is because you are an exceptional writer. You don't need this. The way you create characters with depth is amazing. I fed some of your text into the AI to understand how you do it.

 

[Andreas_Kreuz]

You don’t know that this critique has anything to do with your writing. ChatGPT might just know "some things that reviews often say" rather than understanding what it means for a fictional character to have depth. Yes it gave you an answer, but does it mean anything?

I would be very wary of giving this any weight.

Yes one has to be wary. And sometimes you are being criticized for aspects that you want. (Like there are sex scenes that are too explicit)
But if it says your writing is choppy, it most likely is. And since human readers also called my style too choppy I used AI to learn by getting feedback from it.
AI is not able to evaluate the overall quality. it can only identify factors, that influence it. Like bad pacing or unnatural or stilted dialogue. Then again, when I write about D/S the dialogue is unnatural and stilted, because that's the way Role-players talk when in their role.

 

[AwkwardMD]

I fed some of your text into the AI to understand how you do it.

Please don’t do this unless you have permission.

 

[AwkwardMD]

There's no reason to ask an AI what makes Melissa's dialog so strong. You can just ask Melissa. You are  already talking to her.

 

[Andreas_Kreuz]

There seems to be the idea, that the AI incorporates your input into it language model. I think GPT-1 from a few years back, did that and it ended in a disaster, because it incorporated the input of trolls. So GPT stopped it and only read from verified sources.
Also researchers have found out, that llms get messed up very quickly if they are being fed their own output. The richness of the answers quickly deteriorates, while hallucination multiplies.
This has also been considered the end of improvement for generative AI. Once they have read everything, there just isn't anything more to read and improve on.

 

[MrPixel]

This concerns me:

... Ethical concerns about non-consenting bystanders ... further weaken it. ...

Since when did a lump of computer code become sentient about ethics? You okay with it imposing its concept of morality on your writing?

 

[AG31]

Keep the chat history intact so it can learn from you.

Technical question. How do you "keep the chat history?"

 

[AwkwardMD]

This concerns me:



Since when did a lump of computer code become sentient about ethics? You okay with it imposing its concept of morality on your writing?

That was one of the aspects that made me think it wasn't actually reviewing anything, just parroting "critic-y talking points"

 

[AwkwardlySet]

So Literotica tries to keep out AI stories, which I am very happy about. Whether the AI rejection process is effective is not the topic of this post.
Instead of using AI to write stories, I have tried to use AI to review my stories so I can learn and improve. Researchers have suggested that AI is an excellent trainer because it allows you to create so-called "optimal challenge points".
So I write a first version and then post it to GPT asking it for feedback on the writing style. It has helped to bias the prompt towards being critical. So I might use this prompt:


So applied to my story https://www.literotica.com/s/a-whore-for-one-evening (Rated 3.72) it told me:

So there you have it. Way to go.
If you leave out "short minimal review" from the prompt, it gets a bit more detailed.

Another prompt I found helpful is:



What you should not do is to ask the AI for suggestions on how to improve the text. Doing so will provide suggestions that are so bad that I am afraid they will send me in the direction of overwrought, clicheed purple prose.

Has anybody else tried to use AI similarly? What are your experiences and useful prompts?

I am not the one to automatically dismiss anything regardless of how thin it all sounds. And your idea to use AI to review your writing does seem thin. AI simply isn't there (yet) in my opinion. Whether it will ever get there is debatable I suppose. But either way, some posters made good points about taking for granted this "input" you are getting.
Before even considering taking AI's advice, you should at the very least test its abilities. Ask other authors here or elsewhere for input on your stories and then compare it to the "evaluation" you are getting from ChatGPT. Then evaluate all of the input yourself and see how much ChatGPT made sense in its review. It shouldn't be hard to evaluate if it's just spewing nonsense or giving at least some usable advice.
If you are determined to use AI for such purposes then it would be a rational approach to test it first.

 

[Britva415]

So applied to my story https://www.literotica.com/s/a-whore-for-one-evening (Rated 3.72) it told me:

The story lacks emotional depth, has clunky language, and poorly develops characters. It rushes through clichéd scenarios without building tension and fails to clearly establish consent. Ethical concerns about non-consenting bystanders and inconsistent tone further weaken it. Needs stronger writing, originality, and respect for boundaries.

So there you have it. Way to go.

You can’t take that seriously. “Emotional depth?” AI can’t know this, and it really can’t know any of the other claims it made. All it did was parrot literary criticism it has seen in its training material. It can’t connect the meaning of the criticism to the quality of whatever the reviews were covering.

Don’t take a MadLibs of a review format as actual criticism related in any way to your actual story.

 

[Plathfan]

Technical question. How do you "keep the chat history?"

Just don't delete it, and avoid opening too many chats.

 

[Rob_Royale]

I tried this recently as well. Here's the critique I got.

Don't quit your day job.

 

[Plathfan]

People's approach to AI often mirrors their view of life: bitter nihilists see it as a mistake rather than a gift.

 

[AlexBailey]

ME:
Provide a psychological and factual analysis analysis of the post below from an online forum thread discussion about AI’s ability to review and provide useful insights to an human author’s work.

You can’t take that seriously. “Emotional depth?” AI can’t know this, and it really can’t know any of the other claims it made. All it did was parrot literary criticism it has seen in its training material. It can’t connect the meaning of the criticism to the quality of whatever the reviews were covering.

Don’t take a MadLibs of a review format as actual criticism related in any way to your actual story.

ChatGPT4o:
The forum post critiques the ability of AI to provide meaningful feedback on human-authored work, particularly in terms of “emotional depth” and other qualitative aspects. Here’s a psychological and factual analysis of the post:

Psychological Analysis

1. Skepticism and Distrust:
The author displays a strong skepticism toward AI’s capabilities, reflecting a broader human tendency to distrust technology when it ventures into subjective, human-centered domains like emotions or artistic critique. This may stem from a belief that these areas are inherently human and not replicable by algorithms.

2. Defensiveness of Human Creativity:
The dismissal of AI’s critique as “parroted” suggests a protective attitude toward the uniqueness of human creativity. The author may feel that AI challenges or undermines the value of human interpretation, a common emotional response to advancements in AI.

3. Frustration or Disillusionment:
Terms like “MadLibs” and “parrot” convey frustration, possibly due to unmet expectations or an aversion to seeing AI used in fields they value deeply. This framing implies a belief that AI critiques lack authenticity and depth.


Factual Analysis

1. AI’s Limitations in Understanding Emotional Depth:
The author is correct that AI lacks intrinsic understanding of emotions or subjective experiences. AI models, including advanced ones, are trained on vast amounts of data and can mimic patterns of language and analysis but do not “understand” or “feel” in the human sense. Their feedback is based on probabilistic correlations in text, not genuine comprehension.


2. Mimicry vs. Original Insight:
AI-generated critiques can resemble those of human literary criticism, but they are generated by identifying patterns in previously seen data. The claim that AI “parrots” or assembles critiques like “MadLibs” has merit, as AI often relies on reconstructing phrases or ideas it has encountered during training.


3. Utility of AI Critiques:
While the post dismisses AI reviews entirely, it overlooks potential utility. AI can provide structural feedback, highlight inconsistencies, or suggest areas for refinement based on patterns observed in similar works. However, interpreting “emotional depth” or subjective quality is indeed outside its capability.


4. AI’s Role in Creative Processes:
The post frames AI as an untrustworthy critic but does not consider its collaborative potential. AI can serve as a brainstorming tool or offer mechanical suggestions, leaving the nuanced interpretation to the human author.


Conclusion

While the forum post raises valid concerns about the limitations of AI in understanding and critiquing subjective aspects like emotional depth, it adopts an absolutist stance that dismisses any value AI might bring. A more balanced perspective would acknowledge these limitations while recognizing that AI can still assist with more technical, surface-level, or mechanical aspects of reviewing human-authored work. The emotional response in the post highlights a tension between the expanding role of AI and the human desire to preserve uniquely human domains like art and critique.

 

[Britva415]

ME:
Provide a psychological and factual analysis analysis of the post below from an online forum thread discussion about AI’s ability to review and provide useful insights to an human author’s work.



ChatGPT4o:
The forum post critiques the ability of AI to provide meaningful feedback on human-authored work, particularly in terms of “emotional depth” and other qualitative aspects. Here’s a psychological and factual analysis of the post:

Psychological Analysis

1. Skepticism and Distrust:
The author displays a strong skepticism toward AI’s capabilities, reflecting a broader human tendency to distrust technology when it ventures into subjective, human-centered domains like emotions or artistic critique. This may stem from a belief that these areas are inherently human and not replicable by algorithms.

2. Defensiveness of Human Creativity:
The dismissal of AI’s critique as “parroted” suggests a protective attitude toward the uniqueness of human creativity. The author may feel that AI challenges or undermines the value of human interpretation, a common emotional response to advancements in AI.

3. Frustration or Disillusionment:
Terms like “MadLibs” and “parrot” convey frustration, possibly due to unmet expectations or an aversion to seeing AI used in fields they value deeply. This framing implies a belief that AI critiques lack authenticity and depth.


Factual Analysis

1. AI’s Limitations in Understanding Emotional Depth:
The author is correct that AI lacks intrinsic understanding of emotions or subjective experiences. AI models, including advanced ones, are trained on vast amounts of data and can mimic patterns of language and analysis but do not “understand” or “feel” in the human sense. Their feedback is based on probabilistic correlations in text, not genuine comprehension.


2. Mimicry vs. Original Insight:
AI-generated critiques can resemble those of human literary criticism, but they are generated by identifying patterns in previously seen data. The claim that AI “parrots” or assembles critiques like “MadLibs” has merit, as AI often relies on reconstructing phrases or ideas it has encountered during training.


3. Utility of AI Critiques:
While the post dismisses AI reviews entirely, it overlooks potential utility. AI can provide structural feedback, highlight inconsistencies, or suggest areas for refinement based on patterns observed in similar works. However, interpreting “emotional depth” or subjective quality is indeed outside its capability.


4. AI’s Role in Creative Processes:
The post frames AI as an untrustworthy critic but does not consider its collaborative potential. AI can serve as a brainstorming tool or offer mechanical suggestions, leaving the nuanced interpretation to the human author.


Conclusion

While the forum post raises valid concerns about the limitations of AI in understanding and critiquing subjective aspects like emotional depth, it adopts an absolutist stance that dismisses any value AI might bring. A more balanced perspective would acknowledge these limitations while recognizing that AI can still assist with more technical, surface-level, or mechanical aspects of reviewing human-authored work. The emotional response in the post highlights a tension between the expanding role of AI and the human desire to preserve uniquely human domains like art and critique.

“emotional depth” though… come on!

 

[AwkwardlySet]

People's approach to AI often mirrors their view of life: bitter nihilists see it as a mistake rather than a gift.

It's actually neither. The emergence of AI was simply unavoidable. It was a natural step that came with the progression of technology and computing power. AI has its benefits but it has plenty of drawbacks once you analyze all its aspects. If you are for once interested in actually engaging in a discussion rather than quoting random authors, there are a lot of aspects of AI usage that can be shown as not beneficial, at least for now, while we are still struggling to set some boundaries.