I've been playing with AI

[madelinemasoch]

and I discovered that ChatGPT4 can accurately analyze (not just summarize) my old work without much if any help in the prompts. I can just feed it a scene and say "analyze this" and it works fine. It picks up on all of the character traits and relationship dynamics and developments and themes which are actually expressed in the scenes I fed it which were from the stories I wrote and published here back in 2023 but since removed. It also didn't call Eleanor a psychopath and it didn't criticize the relationship dynamics as somehow abusive or unhealthy. It just told me exactly what it is I actually did with the characters. There was none of the judgment they received from the comments or the forums included in the feedback from the chatbot whatsoever.

The conclusion I'm drawing from this is that human beings become hysterical over the internet over pretty much nothing, and that I was a pot-stirrer on here without ever trying or aiming to be one. All I did was translate the ideas into literature as they came to me and then share them with the world in that form. I am also drawing the conclusion that I can safely say that I am not crazy because I actually did communicate what the ideas communicated to me through the text, it just took a fucking chatbot to see it. LOL. Maybe among those thousands of reads I received there were actual human beings who did pick up what I was putting down and deciphered it to a more accurate level of comprehension. Those people probably do exist and just didn't say anything, whether they liked what they were reading or not.

Safe to say I was right. Those hate comments were actually misinterpreting what the stories mean. I feel a lot of vindication from that. There is actually a blind spot in the readership on this website that's real and not some inherent product of the reading process but a conditional thing that can be changed. I didn't even write them badly, but I'm going to write even better now.

I'm not even sure exactly what it is about the stories that made people so hysterical with enough unwarranted vitriol that it turned off their thinking caps immediately upon reading, but maybe that isn't my fault. Maybe it is to the extent that my ideas guide me beyond wish fulfillment fantasy and towards the redemption of pain and darkness as well as the creation of new ways of relating to the self and others that aren't necessarily clean and happy-go-lucky or without challenges. Maybe it's because my erotica is actually closer to literary fiction than genre fiction.

Maybe that's why people called my exploration of identity and power and authority and desire "torture porn" and my characters navigating their emotional conflicts and relationship preferences "narcissists" and "psychopaths" and "abusers". Who knows.

I'll see you all again on the story side towards the end of the month.

 

[ElectricBlue]

Maybe that's why people called my exploration of identity and power and authority and desire "torture porn" and my characters navigating their emotional conflicts and relationship preferences "narcissists" and "psychopaths" and "abusers". Who knows.

Keep in mind that the analysis is from a word predictive/collating machine, and there's no human element or reading whatsoever. What you're getting back is not representative of a reader in any sense.

Do be careful you don't delude yourself with all this "analysis". Surely you are better off seeking input from readers you have an affinity with?

 

[XerXesXu]

and I discovered that ChatGPT4 can accurately analyze (not just summarize) my old work without much if any help in the prompts. I can just feed it a scene and say "analyze this" and it works fine. It picks up on all of the character traits and relationship dynamics and developments and themes which are actually expressed in the scenes I fed it which were from the stories I wrote and published here back in 2023 but since removed. It also didn't call Eleanor a psychopath and it didn't criticize the relationship dynamics as somehow abusive or unhealthy. It just told me exactly what it is I actually did with the characters. There was none of the judgment they received from the comments or the forums included in the feedback from the chatbot whatsoever.

The conclusion I'm drawing from this is that human beings become hysterical over the internet over pretty much nothing, and that I was a pot-stirrer on here without ever trying or aiming to be one. All I did was translate the ideas into literature as they came to me and then share them with the world in that form. I am also drawing the conclusion that I can safely say that I am not crazy because I actually did communicate what the ideas communicated to me through the text, it just took a fucking chatbot to see it. LOL. Maybe among those thousands of reads I received there were actual human beings who did pick up what I was putting down and deciphered it to a more accurate level of comprehension. Those people probably do exist and just didn't say anything, whether they liked what they were reading or not.

Safe to say I was right. Those hate comments were actually misinterpreting what the stories mean. I feel a lot of vindication from that. There is actually a blind spot in the readership on this website that's real and not some inherent product of the reading process but a conditional thing that can be changed. I didn't even write them badly, but I'm going to write even better now.

I'm not even sure exactly what it is about the stories that made people so hysterical with enough unwarranted vitriol that it turned off their thinking caps immediately upon reading, but maybe that isn't my fault. Maybe it is to the extent that my ideas guide me beyond wish fulfillment fantasy and towards the redemption of pain and darkness as well as the creation of new ways of relating to the self and others that aren't necessarily clean and happy-go-lucky or without challenges. Maybe it's because my erotica is actually closer to literary fiction than genre fiction.

Maybe that's why people called my exploration of identity and power and authority and desire "torture porn" and my characters navigating their emotional conflicts and relationship preferences "narcissists" and "psychopaths" and "abusers". Who knows.

I'll see you all again on the story side towards the end of the month.

Do you need to 'jailbreak' ChatGPT? I find that most AI will otherwise jibe at anything too explicit. Have you tried DeepSeek, it's now my first choice AI.

 

[madelinemasoch]

Do you need to 'jailbreak' ChatGPT? I find that most AI will otherwise jibe at anything too explicit. Have you tried DeepSeek, it's now my first choice AI.

No, but I'm going to.

I can use ChatGPT4 for free with no limitations through a job I have. It's technically completely legal.

 

[XerXesXu]

Keep in mind that the analysis is from a word predictive/collating machine, and there's no human element or reading whatsoever. What you're getting back is not representative of a reader in any sense.

Do be careful you don't delude yourself with all this "analysis". Surely you are better off seeking input from readers you have an affinity with?

Some people still live in 2023.

 

[OverconfidentSarcasm]

There was none of the judgment they received from the comments or the forums included in the feedback from the chatbot whatsoever.

The conclusion I'm drawing from this is that human beings become hysterical over the internet over pretty much nothing, and that I was a pot-stirrer on here without ever trying or aiming to be one.

BWAHAHAhahahahaha

Dude! Are you telling me that you used an LLM, a Large Language Model, which has the sole function of mimicking language and stringing words into sentences, fed it your stories, and it DIDN'T give you more than the sentences you put onto paper? No shit! Really!? Who would have thought! And now you're using that as "proof" of real people being wrong when categorizing your characters BEYOND what you claim them to be yourself inside the story? You just made my day!

How about we replace all therapists and psychologists with ChatGPT-powered bots now? Apparently, they're doing a fine job "analyzing" characters!

 

[VerbalAbuse]

BWAHAHAhahahahaha

Dude! Are you telling me that you used an LLM, a Large Language Model, which has the sole function of mimicking language and stringing words into sentences, fed it your stories, and it DIDN'T give you more than the sentences you put onto paper? No shit! Really!? Who would have thought!

You just made my day. How about we replace all therapists and psychologists with ChatGPT-powered bots now? Apparently, they're doing a fine job "analyzing" characters!

interestingly, that's exactly what the "therapists" do -- turn your words back to you. it's... lucrative.

 

[OverconfidentSarcasm]

interestingly, that's exactly what the "therapists" do -- turn your words back to you. it's... lucrative.

Yeah, it's also what you and I do while reading posts here. It's language comprehension. The difference between us and ChatGPT is that we can go beyond what is written and draw actual conclusions based on facts and experience, instead of just forming responses based on what "should sound right".

 

[nice90sguy]

Ask an LLM to comment on your work carries a danger. They're trained to be eager to please, not to say sycophantic. Remember to prompt the LLM not to go easy on you when criticising your work, or be too nice. Tell it to play devil's advocate, or even to rip your work to shreds. It will happily obilge (because you told it to, and it's a fawning, insecure and obedient "helpful AI assistant").

 

[SyleusSnow]

Well, LLMs can do a lot more than parrot words based on probability. There are "chat" LLMs and "instruct" LLMs that work differently. (DeepSeek is an instruct model). They actually can "understand" context, do semantic analysis and analyze style. Of course, they get things wrong and you can't feed them a whole novel, these things are not just a modern-day "Eliza".

To demonstrate, just for fun I fed a DeepSeek model Left Behind, a 750 word story, and simply told it to "analyze the short story". The end result is below. You can see it's a lot more than just stringing together words together and does offer some critical analysis. Good as a human? Error free? Of course not. But this shows that it mostly understands the story.

Output from DeepSeek R1 Distill Gwen 32B:

The short story presents a complex exploration of a tumultuous and dysfunctional relationship between two characters, "I" and Amber, set against the backdrop of a recent move into a new condo. The narrative is rich with vivid descriptions and sharp dialogue, effectively capturing the tension and underlying issues between the characters.

Key Elements of Analysis:​

1. Character Dynamics:
   • Amber: Described as confident yet brash, Amber's character is multi-faceted. She shifts between aggressive and vulnerable behaviors, suggesting a complex inner life that isn't fully explored. Her interactions with "I" reveal a mix of dominance and submission, adding depth to their relationship.
   • Narrator ("I"): The narrator struggles with conflicting emotions towards Amber, oscillating between affection and frustration. His internal conflict highlights the destructive patterns in their relationship and his inability to break free.
2. Themes:
   • Addiction and Dysfunction: The story explores themes of addiction and self-destruction through the characters' repeated engagement in a volatile relationship despite clear issues.
   • Power Dynamics: The narrative examines power imbalances, with elements of dominance and submission evident in their interactions, influenced by class differences and past experiences.
3. Setting:
   • The setting in an empty apartment after moving symbolizes transition and attempts to start fresh. However, the presence of Amber and remnants of past relationships underscores how history can haunt one's present.
4. Conflict:
   • The introduction of Sophia adds external conflict, revealing broader awareness of their relationship dynamics. Her reaction introduces potential consequences beyond their personal interactions, though her character feels underdeveloped.
5. Writing Style:
   • The author employs strong descriptive language to create vivid scenes and realistic dialogue, capturing the volatile nature of the characters' relationship.
   • Some moments feel rushed, particularly in sex scenes, which could benefit from more emotional depth rather than just physical descriptions.
6. Areas for Improvement:
   • Pacing: Slower pacing in key scenes could enhance emotional resonance.
   • Character Development: Tying descriptive elements more directly to character development would enrich the narrative.
   • Foreshadowing: More setup around external conflicts, like Sophia's role, could strengthen their impact.

Conclusion:​

The story offers a compelling look into a dysfunctional relationship with rich characters and vivid settings. While there are areas for improvement in pacing and depth, it effectively conveys the complexity and tension between "I" and Amber, making it an engaging read that explores themes of addiction and power dynamics.

 

[lovecraft68]

I won't touch that trash even to 'just see'

 

[AwkwardlySet]

Well, LLMs can do a lot more than parrot words based on probability. There are "chat" LLMs and "instruct" LLMs that work differently. (DeepSeek is an instruct model). They actually can "understand" context, do semantic analysis and analyze style. Of course, they get things wrong and you can't feed them a whole novel, these things are not just a modern-day "Eliza".

To demonstrate, just for fun I fed a DeepSeek model Left Behind, a 750 word story, and simply told it to "analyze the short story". The end result is below. You can see it's a lot more than just stringing together words together and does offer some critical analysis. Good as a human? Error free? Of course not. But this shows that it mostly understands the story.

Output from DeepSeek R1 Distill Gwen 32B:

It is more than just stringing words together, I agree. But also, as @nice90sguy already pointed out, it's clearly biased towards the positive. There are some obvious points of criticism that AI either didn't grasp, or didn't want to point out so this overwhelmingly positive review aimed to please the prompter could come to life.

 

[pink_silk_glove]

Do be careful you don't delude yourself with all this "analysis". Surely you are better off seeking input from readers you have an affinity with?

No, a Red H from several hundred people with a mouse in one and and their privates in the other is the most accurate assessment of your submissions. StillStunned has spoken.

 

[Lifestyle66]

and I discovered that ChatGPT4 can accurately analyze (not just summarize) my old work without much if any help in the prompts. I can just feed it a scene and say "analyze this" and it works fine. ...

The conclusion I'm drawing from this is that human beings become hysterical over the internet over pretty much nothing, and that I was a pot-stirrer on here without ever trying or aiming to be one. All I did was translate the ideas into literature as they came to me and then share them with the world in that form....

Maybe that's why people called my exploration of identity and power and authority and desire "torture porn" and my characters navigating their emotional conflicts and relationship preferences "narcissists" and "psychopaths" and "abusers". Who knows.

While using ChatGPT may be a reasonable idea for assisting you in refining your communication skills, it is just an unemotional machine. It dispassionately looks at your story and feeds back ways to improve the language. I tried using it to review one of my stories, and it did point out some flaws in areas which needed improvement.

However, it does NOT reflect how real humans with real human experiences and real emotions will react to what they get from your stories.

Maybe a question to ask is "Why are you trying to communicate with humans, if you don't want to listen to what they tell you?" If they tell you they read "psychopaths" and "abusers", then that's their emotional response from what they read.

I throw my stories out there and take the LOVE and HATE for what it is. Some people RIGHTFULLY hate it because of their own personal life experience. It's who they are. And I don't fault them for expressing their opinions. But I just listen and try to learn.

 

[AlexBailey]

It is more than just stringing words together, I agree. But also, as @nice90sguy already pointed out, it's clearly biased towards the positive. There are some obvious points of criticism that AI either didn't grasp, or didn't want to point out so this overwhelmingly positive review aimed to please the prompter could come to life.

It’s a tool.

You can request responses based on different perspectives, styles, and with different priorities.

Anyone who considers it useless because it’s “just stringing words together” is blinded to the possibilities.

Unlike some didactic educators or reviewers, the AI’s default is to be positive. This is a setting, not a rule. You can prompt it to be hyper critical, or to respond as someone who has been the victim of sexual abuse, as a jealous ex, whatever….

You could have it answer from the perspective of someone from or outside of your target audience or even as one of your characters after they have had a full life and are reviewing their own past.

Facts? What are “facts” when it come to a review? What human reviewer’s opinion should be accepted without reflection? Human opinions aren’t any more objective than any AI response, but unlike using human reviewers you aren’t taking up someone else’s time.

 

[SyleusSnow]

It is more than just stringing words together, I agree. But also, as @nice90sguy already pointed out, it's clearly biased towards the positive. There are some obvious points of criticism that AI either didn't grasp, or didn't want to point out so this overwhelmingly positive review aimed to please the prompter could come to life.

That's a very good point. Thanks, @nice90sguy for mentioning it.

I hadn't noticed the output being particularly positive, but you're right @AwkwardlySet: I re-ran the analysis of my story, telling it to act as an expert editor and to be critical and not polite. I'm sure there are better prompts, but DeepSeek did offer more critical observations.

 

[AlexBailey]

That's a very good point. Thanks, @nice90sguy for mentioning it.

I hadn't noticed the output being particularly positive, but you're right @AwkwardlySet: I re-ran the analysis of my story, telling it to act as an expert editor and to be critical and not polite. I'm sure there are better prompts, but DeepSeek did offer more critical observations.

I’ve been checking out ChatGPT through its most recent releases. It’s getting much better very quickly.

You can ask it for advice about crafting prompts to get the responses you find most useful.

Ask what the best word count is for an excerpt.

Ask it how to frame your prompts so it will track the context between excerpts.

Ask it why it responds the way it does.

Like any tool, the better you understand what it can do the more valuable it can be.

Try asking something like this:

“How does the pace of the narration change throughout the chapter? Does it fit with the action and is the tone appropriate for the emotional dynamics?

 

[AwkwardlySet]

It’s a tool.

You can request responses based on different perspectives, styles, and with different priorities.

Anyone who considers it useless because it’s “just stringing words together” is blinded to the possibilities.

Unlike some didactic educators or reviewers, the AI’s default is to be positive. This is a setting, not a rule. You can prompt it to be hyper critical, or to respond as someone who has been the victim of sexual abuse, as a jealous ex, whatever….

You could have it answer from the perspective of someone from or outside of your target audience or even as one of your characters after they have had a full life and are reviewing their own past.

Facts? What are “facts” when it come to a review? What human reviewer’s opinion should be accepted without reflection? Human opinions aren’t any more objective than any AI response, but unlike using human reviewers you aren’t taking up someone else’s time.

But that's the problem. You can never hope for a truly balanced review. The AI will always try to please the prompter. If you ask for criticism, it's gonna criticize. If you ask for praise, it's gonna praise, and if you ask for an "objective" review? Yeah, it's not really capable of that... yet.

As you said, it's a tool. Using it might lead to some new insight, sure. But it could also lead to delusions about the quality of one's writing. A highly sophisticated tool such as AI requires a deep understanding of the way LLMs work - if you are to avoid these biases, and that is not something you can expect from a huge majority of people.

 

[Zootonius]

On a related note,

ChatGPT can now write erotica as OpenAI eases up on AI paternalism.

Still, I see it like a golfer using a robot to golf for him.

https://arstechnica.com/ai/2025/02/...erotica-as-openai-eases-up-on-ai-paternalism/

 

[AlexBailey]

But that's the problem. You can never hope for a truly balanced review. The AI will always try to please the prompter. If you ask for criticism, it's gonna criticize. If you ask for praise, it's gonna praise, and if you ask for an "objective" review? Yeah, it's not really capable of that... yet.

As you said, it's a tool. Using it might lead to some new insight, sure. But it could also lead to delusions about the quality of one's writing. A highly sophisticated tool such as AI requires a deep understanding of the way LLMs work - if you are to avoid these biases, and that is not something you can expect from a huge majority of people.

Well yeah, if you simply ask if the piece ‘good quality’ it’s going to respond in a fairly useless way, just as a review from a random person who you know nothing about is basically useless.

You’re never going to get a truly objective review from any human either. You could ask it to focus only on adherence to particular writing standards or you could have it critique for word efficiency.

At some point you need to decide what resonates with you. Human audiences cover a wide spectrum, what works for some will be repulsive to others.


My stories typically involve the psychological aspects of kink and gender-identity. Many readers in my target audience are sensitive to some of the issues I write about. When I asked AI to identify points in my writing that could be emotional trigger points for some of my target audience it came back with concepts I had already considered but it also brought up some points I hadn’t thought of simply because I’m biased by my own perspective.

I don’t agree with all of its assessments anymore than I agree with every comment on this thread, but I do find the responses valuable for reflection.

 

[alohadave]

As a test, I downloaded DeepSeek and had it analyze a WIP I have.

It gave me some things to think about with it. Admittedly, things I already knew, but it wasn't far off in its suggestions.

Curiously, when I asked it to improve the story, it initially did, but after a few seconds, cleared entry and gave me this:

Sorry, that's beyond my current scope. Let’s talk about something else.

It did this when I told it that it was an erotic story and for just a story.

Not sure what that's about as it identified sex scenes and erotic content when it was analyzing the story.

 

[AwkwardlySet]

Well yeah, if you simply ask if the piece ‘good quality’ it’s going to respond in a fairly useless way, just as a review from a random person who you know nothing about is basically useless.

You’re never going to get a truly objective review from any human either. You could ask it to focus only on adherence to particular writing standards or you could have it critique for word efficiency.

At some point you need to decide what resonates with you. Human audiences cover a wide spectrum, what works for some will be repulsive to others.


My stories typically involve the psychological aspects of kink and gender-identity. Many readers in my target audience are sensitive to some of the issues I write about. When I asked AI to identify points in my writing that could be emotional trigger points for some of my target audience it came back with concepts I had already considered but it also brought up some points I hadn’t thought of simply because I’m biased by my own perspective.

I don’t agree with all of its assessments anymore than I agree with every comment on this thread.

I think we are somewhat mixing apples and oranges here. Sure, any human reviewer would have their own biases, but they are still a person - they are still part of your audience. Their response, however biased, will be the response coming from how that person actually understood your story, and it will reflect at least a part of your readership if nothing more than that.
On the other hand, AI represents no one, and its response doesn't come from understanding, biases, or any particular personality traits, but from pattern recognition. So unless you are writing for AI as your target audience, I don't see how those two things can ever be the same.

 

[HeyAll]

https://nypost.com/2025/01/24/enter...rter-and-better-than-oscar-nominated-writers/

 

[AlexBailey]

I think we are somewhat mixing apples and oranges here. Sure, any human reviewer would have their own biases, but they are still a person - they are still part of your audience. Their response, however biased, will be the response coming from how that person actually understood your story, and it will reflect at least a part of your readership if nothing more than that.
On the other hand, AI represents no one, and its response doesn't come from understanding, biases, or any particular personality traits, but from pattern recognition. So unless you are writing for AI as your target audience, I don't see how those two things can ever be the same.

The AI can identify patterns - patterns that are known to elicit emotional responses. Some people can’t do that.

It can recognize multiple layers of patterns and how they interact to great depths. I can say without reservation that it has deeper emotional intelligence than many people I know.

As an experiment try giving it a significant background on one of your characters and have it provide responses to different challenging questions and situations. If your characters are portrayed with nuance and depth the AI’s responses will reflect that. It’s not making judgements, it’s providing a reflection of what you wrote.

 

[Duleigh]

The conclusion I'm drawing from this is that human beings become hysterical over the internet over pretty much nothing, and that I was a pot-stirrer on here without ever trying or aiming to be one.

I requested Chat GPT to draw up a story on any subject in exactly 750 words
The first result was 684 words
The second result was 689 words
The third result was 4250 words
The fourth result was 4890 words

The conclusion I drew from this is that human beings becoming hysterical over wave of considering the internet a thinking creature are right. It's f---ing stupid. We need to rely on human creativity and leave creating to the artists while the hellspawn of 100,000 undergrads does the souless math in the background.