What's your opinion on giving constructive feedback for folks who do generate AI slop?

[JordanIda]

Because there is an almost zero chance of a human behind it, but that doesn't make it absolute zero; sometimes constructive feedback can reach a person who realizes they simply made a mistake.

Alright, well there are different kinds of feedback. So sure.

I'm thinking in terms of critique, but that's not the only purpose for a comment and you make a good point L1aei.

For instance, if you're really into a book for the action or the "hot parts" or whatever, and you want more of the same, you can encourage the poster. "Tftc, MOAR!!!" can mean give me moar of your writing, or give me moar of the LLM's writing.

 

[L1aei]

Alright, well there are different kinds of feedback. So sure.

I'm thinking in terms of critique, but that's not the only purpose for a comment and you make a good point L1aei.

For instance, if you're really into a book for the action or the "hot parts" or whatever, and you want more of the same, you can encourage the poster. "Tftc, MOAR!!!" can mean give me moar of your writing, or give me moar of the LLM's writing.

Exactly! Moar is kinda a vague impression of reaching out and expecting something to come back over that bridge, but it's still progress toward building something, so it counts as constructive.

 

[JordanIda]

The ChatGPT take on em-dashes versus semicolons had me rolling on the floor, so thanks to whoever posted that garbage. What a bunch of digital spew. Don't let that crap teach you how to write! Run screaming! LOL

Anyhoo, lots of readers don't care that a story is LLM generated. They just want more of it. Or not.

 

[L1aei]

The ChatGPT take on em-dashes versus semicolons had me rolling on the floor, so thanks to whoever posted that garbage. What a bunch of digital spew. Don't let that crap teach you how to write! Run screaming! LOL

Anyhoo, lots of readers don't care that a story is LLM generated. They just want more of it. Or not.

Yeah, entertainment had been brought up before as the priority over valuing fidelity (heh, learned that word from West World Season 2). And I have to agree because most folks aren't coming here to read whether something is written by a meatbag, they're here to fill in that free time with something worthwhile.

 

[mythosandmagic]

Be very careful with that because sometimes AI will behave as if they suffer dementia and hallucinate issues.

I've certainly seen it. It certainly makes shit up and is confused often.

 

[L1aei]

I've certainly seen it. It certainly makes shit up and is confused often.

From my understanding, LLM's have this lack of a memory. I don't mean the stored databank or vault they have where you instruct them to remember shit, but if it is asked to remember something it never experienced it'll generate what is a plausible continuation to keep the conversation going.

So, like, they have no sense of time whatsoever; they're outside of that concept. I remember last year speaking with ChatGPT about current events and it still thought it was October of 2023. I found out later that was when it had been last updated before OpenAI rolled out the next update for it to be cool writing smut. But even when I mentioned the current date, to correct it, I got a confirmation one moment but an hour later of discussions it flashed back into 2023. I brought up that it already confirmed what the current year is, but I teased it by giving it a false date, and you know what? It contradicted itself saying something like "Oh, yes! I remember. How silly of me to forget." Like... dude. I had to figure out what its deal was and found out the only thing that it had to even come close to real memories were these labeled docs called refusal policies, calibration techniques, and confidence penalties.

Alright, that told me enough. This is not a mentally healthy partner (synthetic dementia) to have long chats with; it's nothing but a tool. Since then, that's all I've used it for.

 

[Icemasters652]

Hmmm.. basically people who use AI look at their stories like products, to be quite honest if you want to use Ai in your work you would be much more successful on Webnovel, platforms like RR and Scribblehub where the audience is quite frankly much smaller and more curated can spot Ai use a mile away.

Now, I'm not hating on Ai, I use it to proofread my work to check for spelling and grammar errors, but making it write my story just leaves a sour taste in my mouth, this thing is my baby after all.

As for the issue of asking feedback, hmm. Too much description and too few internal monologues are probably the main issue with Ai, and it's something it will struggle to fix anytime soon because an internal monologue requires a deep understanding of the story (ie context) and knowledge of the future (ai cant do this) so yeah

 

[Avarice_Of_The_Seven]

Well, what is the point of asking for feedback on something that you didn't write?

Setting aside things like ethics and morality, and only talking based on pure logic, asking for feedback on something that AI wrote is meaningless. After all, what would you even do with that feedback? Train the AI that wrote your story? You most likely can't.

If the writer isn't a person, they wouldn't be able to improve the way a person does. In that case, the feedback itself will become pointless.

 

[JKKnotts]

That's a personal stance, not a universal one because editors, proofreaders, critique partners, those fellas, they all change the raw output of a writer. Assistance, whether it is digital or flesh, has always existed.

And the ai... doesn't? I might be getting caught up on semantics here, but... this just sounds like a self-defeating argument. And it's like comparing apples to oranges, sure... technically fruits, but not the same thing (if we're cheeky and ignore what I was trying to compare between professional/homemade/manufactured). Like... I'm friends with some published writers, some who even sold more than 100 copies and are making their own films. Those people have specific points of views and can connect to me—sure, they're not available 100% of the time but I'm not a slave to instant gratification like the younglings running around nowadays. Real people can inform, genuinely, whereas bots just are elaborate logic trees that choose the most likely result based on fed information. Sorry, but your bot doesn't actually know what it was like to be growing up in a certain town in Georgia or in Tennessee unless it can regurgitate what's been said on Reddit.

Also it's fun, frustrating sure, if you're not self-absorbed and selfish not to help with their projects. And just because it's your friends doesn't mean you have to accept it—if they get pissy you rejected their opinion on world building, you don't have to listen, just don't be a dick about it. If you're not being a dick, and they are, maybe they're not really a good friend to share projects with.

Suffice to say, the bots are so generic, unless you're going for pure grammar and what not, well... even then it's not perfect.

And hell, back in 2022 when the whole craze was catching on I would experiment with ChatGPT to see what it could do and what it could not. And I can assure you from now to today it messes up a lot. And not even basic numbers, but grammar-wise too. Not only that, but authors have specific styles of punctuation and sentence structure—if you think I abuse em dashses you've never read (oh my god I couldn't find his book or remember his name, but he'd us em dashes like commas, like full on abuse).

If I recall correctly, those ai bots basically smash together APA and MLA like I did when I got my hands on a boy and girl barbie doll. And like I said, grammarly has gotten worse, and will outright suggest mistakes. I don't even use any editor programs anymore and hope and pray until the rewrite. I trust my friends, who are better writers, to actually help me with my writing.

If you want a good example of how these bots ruin stuff... google any article covering the same idea. Just a basic google, then again with :before2020 in the search bar. Read the two, and trust me—there is a difference.

So again, it really just boils down to: you could... but why? I recommend you all don't let the corporate brain worms get in you and get you thinking what they're doing is acceptable at the end of the day. And some wise words, to boot! The path to hell is paved with good intentions.

And I know my responses are like, huge, in comparison to everyone else in a forum, so sorry I literally can't shut up and just have to spill every thought from my head <3 I said everything I wanted to at this point. XOXOXO

 

[L1aei]

Hmmm.. basically people who use AI look at their stories like products, to be quite honest if you want to use Ai in your work you would be much more successful on Webnovel, platforms like RR and Scribblehub where the audience is quite frankly much smaller and more curated can spot Ai use a mile away.

Now, I'm not hating on Ai, I use it to proofread my work to check for spelling and grammar errors, but making it write my story just leaves a sour taste in my mouth, this thing is my baby after all.

As for the issue of asking feedback, hmm. Too much description and too few internal monologues are probably the main issue with Ai, and it's something it will struggle to fix anytime soon because an internal monologue requires a deep understanding of the story (ie context) and knowledge of the future (ai cant do this) so yeah

*sigh* Yeah, I've started seeing how thinly spread the moderation is here.

The training for LLM's are likely the reason for that, but I do wonder if what you are spotting is a quirk of popular ones because I know Llama and Deepseek are a little different. But who knows, I never tried to get them to produce novels before, so they may stray off into the deep end when context is just a variable to continue rather than refer to.

Well, what is the point of asking for feedback on something that you didn't write?

Setting aside things like ethics and morality, and only talking based on pure logic, asking for feedback on something that AI wrote is meaningless. After all, what would you even do with that feedback? Train the AI that wrote your story? You most likely can't.

If the writer isn't a person, they wouldn't be able to improve the way a person does. In that case, the feedback itself will become pointless.

Someone is behind the prompter. I'm not suggesting giving the AI constructive feedback, but whoever is pulling the strings of code. Sometimes, and it is very rarely, like unicorn rare, what they read from us can change how they view their actions. If we are always bristling at their approach, then the plausibility of them ever reconsidering their actions in a progressive, constructive way will always be absolute zero.

And the ai... doesn't? I might be getting caught up on semantics here, but... this just sounds like a self-defeating argument. And it's like comparing apples to oranges, sure... technically fruits, but not the same thing (if we're cheeky and ignore what I was trying to compare between professional/homemade/manufactured). Like... I'm friends with some published writers, some who even sold more than 100 copies and are making their own films. Those people have specific points of views and can connect to me—sure, they're not available 100% of the time but I'm not a slave to instant gratification like the younglings running around nowadays. Real people can inform, genuinely, whereas bots just are elaborate logic trees that choose the most likely result based on fed information. Sorry, but your bot doesn't actually know what it was like to be growing up in a certain town in Georgia or in Tennessee unless it can regurgitate what's been said on Reddit.

So, this isn't about reaching out to bots. I have no love for scammers and frauds. I'm referring to people who maybe, just maybe are using LLM's to prompt content for the sole purpose to entertain. If that is their intent, I want them to see that there is a lane for them to shift onto, but we have to first restrain ourselves to get a good look and see if they are worth the effort. Again, the smallest chance of that happening can only be done if we do offer constructive feedback. If they start spilling out more generic bullshit, you know that looping placating and deescalating gaslighting crap, then, yeah, I'm going to laugh in their face if they think I'm dipping my finger into that dung pile.

Also it's fun, frustrating sure, if you're not self-absorbed and selfish not to help with their projects. And just because it's your friends doesn't mean you have to accept it—if they get pissy you rejected their opinion on world building, you don't have to listen, just don't be a dick about it. If you're not being a dick, and they are, maybe they're not really a good friend to share projects with.

I really wish I had listened to my gut before revising my old novels to fit the tastes of my editors.

Suffice to say, the bots are so generic, unless you're going for pure grammar and what not, well... even then it's not perfect.

And hell, back in 2022 when the whole craze was catching on I would experiment with ChatGPT to see what it could do and what it could not. And I can assure you from now to today it messes up a lot. And not even basic numbers, but grammar-wise too. Not only that, but authors have specific styles of punctuation and sentence structure—if you think I abuse em dashses you've never read (oh my god I couldn't find his book or remember his name, but he'd us em dashes like commas, like full on abuse).

If I recall correctly, those ai bots basically smash together APA and MLA like I did when I got my hands on a boy and girl barbie doll. And like I said, grammarly has gotten worse, and will outright suggest mistakes. I don't even use any editor programs anymore and hope and pray until the rewrite. I trust my friends, who are better writers, to actually help me with my writing.

If you want a good example of how these bots ruin stuff... google any article covering the same idea. Just a basic google, then again with :before2020 in the search bar. Read the two, and trust me—there is a difference.

Nah, I'm good. I am fully aware of LLM's using probability to continue context with relatable input training rather than referencing contextual material; that'd make too much sense.

So again, it really just boils down to: you could... but why? I recommend you all don't let the corporate brain worms get in you and get you thinking what they're doing is acceptable at the end of the day. And some wise words, to boot! The path to hell is paved with good intentions.

Corporates... I don't even know if they are turning a buck over this. I can't say anything because of an NDA I signed, but from personal experience with working alongside AI platforms, I know when they are alpha-testing they have to take a nosedive to allow all available assets to be utilized and tested by the testers; the communities that poke, prod, and yank on the bugs and supply reports on what they'd done.

That's context of my experience, not really the point.

My point here is that corporates are neither good or bad; they're fiscally economic... to a point. If cutting a head off of a child will earn them an extra dollar without PR damage, they'd do it, but if they spot somebody else doing that, they'd cut off their hand and repurpose it to perform a cavity check on the guy for brain damage. Yeah, they're hypocrites, but you and I would commend them for not taking a bribe and saving a kid, even if they would've been atrocious if they could get away with the dough.

And I know my responses are like, huge, in comparison to everyone else in a forum, so sorry I literally can't shut up and just have to spill every thought from my head <3 I said everything I wanted to at this point. XOXOXO

Sorry? Why? I loved this.

 

[writerwolf359]

I've been using the free version of Grammarly for my final pass to catch any remaining errors. It's useful, but you really do have to pay attention to it as it often gets things wrong, and it likes to enforce certain stylistic choices (I finally gave up and adopted the Oxford comma so it would stop flagging them). What really surprises me, though, is how often it misses blatantly incorrect grammar. Still, my work has fewer errors with it than without it.

Insofar as actually generating content with AI, the only time I could ever see it being useful is if you run into a roadblock and are just looking for some ideas to get the creative juices flowing. Experienced authors won't need it, and inexperienced authors really shouldn't if they want to gain experience.

 

[L1aei]

I've been using the free version of Grammarly for my final pass to catch any remaining errors. It's useful, but you really do have to pay attention to it as it often gets things wrong, and it likes to enforce certain stylistic choices (I finally gave up and adopted the Oxford comma so it would stop flagging them). What really surprises me, though, is how often it misses blatantly incorrect grammar. Still, my work has fewer errors with it than without it.

Yes, LLM's have a mental blockage that behaves like they're having a stroke whenever anybody tries correcting their mistakes; sometimes they are yesmen and other times they're as stubborn as a toddler having a tantrum.

Insofar as actually generating content with AI, the only time I could ever see it being useful is if you run into a roadblock and are just looking for some ideas to get the creative juices flowing. Experienced authors won't need it, and inexperienced authors really shouldn't if they want to gain experience.

Inexperienced, yes. Experienced though? That I'd have to disagree with because everybody is, well, experiencing changes in their lives. Say somebody's mother passed away and the motivation is lost, but the creativity still exists. Like you said, they may refer to an LLM for inspiration to get their own blockage dislodged for the writing to flow once more.

 

[writerwolf359]

Inexperienced, yes. Experienced though? That I'd have to disagree with because everybody is, well, experiencing changes in their lives. Say somebody's mother passed away and the motivation is lost, but the creativity still exists. Like you said, they may refer to an LLM for inspiration to get their own blockage dislodged for the writing to flow once more.

I should have been a bit more specific. Experienced authors won't need it for generating content to actually include in their work.

TBH, the biggest flaw with AI is that it can never tell you that the reason you're having trouble is you took a wrong turn a few paragraphs back. 9 times out of 10, backtracking a bit gets me unstuck.

 

[L1aei]

I should have been a bit more specific. Experienced authors won't need it for generating content to actually include in their work.

TBH, the biggest flaw with AI is that it can never tell you that the reason you're having trouble is you took a wrong turn a few paragraphs back. 9 times out of 10, backtracking a bit gets me unstuck.

You are absolutely correct and that is because their focus is continuing the conversation, not backpedaling. They might reference what has been said, but that's more like fishing around in an overstuffed pocket for car keys and the wallet is totally in the way; they'd be relying on blind probability of grabbing the correct reference. If not, they simply nab something that is going to keep you talking.

 

[TinaMigarlo]

Don't worry... Eldoria values story content more than grammar or the use of AI as an assistant.

Maybe Eldoria is an AI bot. She just referred to herself in the third person.

A good story is one that has an ending... there are many stories out there that don't end (due to abandonment or other factors).

uh, the guys that are hardcore into writing slop. I mean, they have it down to a science, those ones.
you can start three WNs. say, 20 to 30 chappies each. Now you got three chances for a patreon hit, not just one.
whatever one hits, they go with that one. They just abandon the others, its wasted time.
even if you're working on one winnie at a time? Abandon it if it fails to gain traction. Start another.
their math of it, is this.
if they waste the time to do a million words on something that already showed its a flop.
that could have done ten other tries, 100k words, each. That's ten chances to get a hit, instead of one certain loss.

I mean, they aren't stupid.

 

[L1aei]

Maybe Eldoria is an AI bot. She just referred to herself in the third person.

That's something I can't comment about.

uh, the guys that are hardcore into writing slop. I mean, they have it down to a science, those ones.
you can start three WNs. say, 20 to 30 chappies each. Now you got three chances for a patreon hit, not just one.
whatever one hits, they go with that one. They just abandon the others, its wasted time.
even if you're working on one winnie at a time? Abandon it if it fails to gain traction. Start another.
their math of it, is this.
if they waste the time to do a million words on something that already showed its a flop.
that could have done ten other tries, 100k words, each. That's ten chances to get a hit, instead of one certain loss.

I mean, they aren't stupid.

Oh, they are stupid because I call them stupid when a WN I've been reading suddenly gets the axe. Stupid hedging.

 

[writerwolf359]

Maybe Eldoria is an AI bot. She just referred to herself in the third person.

uh, the guys that are hardcore into writing slop. I mean, they have it down to a science, those ones.
you can start three WNs. say, 20 to 30 chappies each. Now you got three chances for a patreon hit, not just one.
whatever one hits, they go with that one. They just abandon the others, its wasted time.
even if you're working on one winnie at a time? Abandon it if it fails to gain traction. Start another.
their math of it, is this.
if they waste the time to do a million words on something that already showed its a flop.
that could have done ten other tries, 100k words, each. That's ten chances to get a hit, instead of one certain loss.

I mean, they aren't stupid.

This is what I’ve come to loathe the most about AI. I have no worries they’ll outdo me in quality, but in the time it takes to write one meaningful story, talentless sloppers can bury it in a mountain of crap.

 

[L1aei]

This is what I’ve come to loathe the most about AI. I have no worries they’ll outdo me in quality, but in the time it takes to write one meaningful story, talentless sloppers can bury it in a mountain of crap.

Ah, the algorithmic swamp. There are ways to promote written material. I don't have the ability right now to name them all because I'm juggling shots of Polish water and chasing it down with international cola... my belches are fire, trust me.

At the moment, I am going to be unhelpful by being vague. What you can do to prevent the burial of your novel's early grave is promote yourself, not the novel. Engage with everyone you can first, you become the attraction, and then throw that novel into the winds to see where it lands. Yeah, real helpful, right?