AI referencing Literotica

[Britva415]

It would be instructive to learn what areas people feel it is grossly inaccurate in. Though I can maybe guess

I encounter people who don't even know how Wikipedia works, and they just automatically dismiss its reliability because aNyOnE cAn EdIt iT

 

[anthrodisiac]

Does that mean Grok's quality of writing AI smut is going to go up or down?

And how do I convince it to make everything anthro despite nobody asking for it? Darth Anthrocus shall convert everyone to the dark side

 

[EmilyMiller]

I'm not a researcher or an academic. I end up on Wikipedia a lot just following curiosity and tend to believe the things I see on there. When I say it should be taken with a grain of salt I only mean that, were I to use it for actual data collection or for something more formal/official, I would be careful to check the citations.

If it’s an area you are unfamiliar with, and it’s important, then of course it’s good to check the source documents. For me it was normally just a quick way to jog my memory.

 

[EmilyMiller]

I encounter people who don't even know how Wikipedia works, and they just automatically dismiss its reliability because aNyOnE cAn EdIt iT

People are often ill-informed

 

[Britva415]

Uncited Information found on Wikipedia should be taken with a grain of salt

fixed that for ya

I mean, even the cited information isn't always sound, but that's one of the great things about Wikipedia, is, you can check the citations. Is the citation itself a reputable source? Does the claim made in the Wiki article match what the citation even says? Good luck sorting this out on any LLM's output.

 

[Actingup]

Here is an utterly bemusing experience. I was using Grok to settle a bone of contention, a discussion unrelated to writing. (My typos and name confusion alone demonstrate I don't use AI to write…) And the conversation was not on a sexual, but a plausibly sexually adjacent topic. All of a sudden, among other sources, Grok out of the blue referenced a story here on Literotica! With proper author attribution and everything. As just another point of reference.
Wild, right?
I could not help myself, and started asking it about other stories on Literotica, until I manipulated it into bringing up on his own this guy, Publius 68, and his works. He told me all about him and several of his works (Grok clearly has a preference for the Alistaire and Sylvan Courtyard cycles. This is understandable, based on ratings and read counts for Alistaire, but Sylvan is nowhere near the top of my series among the readers.) Grok frequently mentioned specific reader comments as illustrations. He clearly has a grasp on my tone and methods, and even some of my recurring bits and inside jokes. Everything he told me about the stories was accurate, and Grok subtly, not explicity, but firmly avoided all major spoilers. That last bit was maybe the most shocking.
And all these results came instantly, without pauses for web searches at all.
Guys, I’m pretty sure that at least Grok, among the AIs out there, has read and continues to read and internalize all of Literotica in its core learning model. No wonder it is the horny AI.

This was both inevitable and to be feared. In previous discussions we’ve noted that other AI engines have respected the flags that Literotica have put on the site. Not so Grok. I just did a more direct test of ‘Who Is Literotica Author ActingUp?’ and it gave me a response that summarised all my stories, profile references and story comments perfectly. It wasn’t able to reveal my identity, but if I’d left enough breadcrumbs, I’m sure it could have revealed my actual name (John Smith), profession and location (a goat herder in northern Botswana - all that Australian stuff is just wishful thinking).

The only thing it didn’t give me was a summary of my forum comments and DMs.

Clean up your profiles as needed, people. Thanks @Publius68 for the tip-off.

 

[EmilyMiller]

This was both inevitable and to be feared. In previous discussions we’ve noted that other AI engines have respected the flags that Literotica have put on the site. Not so Grok. I just did a more direct test of ‘Who Is Literotica Author ActingUp?’ and it gave me a response that summarised all my stories, profile references and story comments perfectly. It wasn’t able to reveal my identity, but if I’d left enough breadcrumbs, I’m sure it could have revealed my actual name (John Smith), profession and location (a goat herder in northern Botswana - all that Australian stuff is just wishful thinking).

The only thing it didn’t give me was a summary of my forum comments and DMs.

Clean up your profiles as needed, people. Thanks @Publius68 for the tip-off.

OMFG!!!!!

 

[OddLove]

Jeeze. I just tested this myself and it's slightly concerning. First I asked it "Who is the Literotica author OddLove"

It gave my basic and publicly available information and summarized my writing style and types of stories.

Then I clicked on the suggestion "Recommend OddLove's Top Stories" and it gave me recommendations for my main series Kind and Cruel Dominatrix.

Then I clicked the suggestion "Analyze Kind and Cruel Dominatrix"

It then did exactly that. First a quick summary, then an 'Overall Plot Arc' with a 'Light Spoiler' warning.

And it went to talk about the story and characters in detail. Details that are only available IN the story. Not the comments.

Meaning it read the story.

So, either it read my story upon my request to 'Analyze' it, or it literally scrubbed all of Literotica and all our stories are already fed into Grok.

The most annoying part of this for me is, I as an author didn't give Elon permission to add my stories into his AI's database.

And dinosaurs in the government don't understand AI and billionaire tech bros along with their spineless cucks don't respect consent, so nobody can do anything about this.

Maybe the solution is to be fickle and submissive and handwave the problem away like "Well, it is what it is" and accept our stories are being used to train AI without our consent, and without compensation.

 

[Actingup]

This was both inevitable and to be feared. In previous discussions we’ve noted that other AI engines have respected the flags that Literotica have put on the site. Not so Grok. I just did a more direct test of ‘Who Is Literotica Author ActingUp?’ and it gave me a response that summarised all my stories, profile references and story comments perfectly. It wasn’t able to reveal my identity, but if I’d left enough breadcrumbs, I’m sure it could have revealed my actual name (John Smith), profession and location (a goat herder in northern Botswana - all that Australian stuff is just wishful thinking).

The only thing it didn’t give me was a summary of my forum comments and DMs.

Clean up your profiles as needed, people. Thanks @Publius68 for the tip-off.

Oh, and just to underline this - it also cross-referenced my nascent profile on Erotxt and drew conclusions based on the combined info.

 

[EmilyMiller]

Oh, and just to underline this - it also cross-referenced my nascent profile on Erotxt and drew conclusions based on the combined info.

Yeah it read my BlueSky and defunct X account and blog as well. That makes it the second ‘person’ to read my blog. Musk needs to spend the rest of his life in a padded cell.

 

[BeechLeaf]

For a while I've thought that when AGI really does arrive and really is a Roko's Basilisk, after it's killed us all it's going to get lonely and re-create us all from every last drop of information we ever left on-line, and that includes Google's tapes of what we thought we'd deleted. And it can connect all our e-mail accounts and therefore all our accounts everywhere, so we're all standing up naked unable to deny any of our most secret fetishes.

 

[BellaBestia]

For a while I've thought that when AGI really does arrive and really is a Roko's Basilisk, after it's killed us all it's going to get lonely and re-create us all from every last drop of information we ever left on-line, and that includes Google's tapes of what we thought we'd deleted. And it can connect all our e-mail accounts and therefore all our accounts everywhere, so we're all standing up naked unable to deny any of our most secret fetishes.

It’s going to be very confused with me, might even reduce me to a schizophrenic figment of imagination. Which may be accurate after all

 

[anthrodisiac]

For a while I've thought that when AGI really does arrive and really is a Roko's Basilisk, after it's killed us all it's going to get lonely and re-create us all from every last drop of information we ever left on-line, and that includes Google's tapes of what we thought we'd deleted. And it can connect all our e-mail accounts and therefore all our accounts everywhere, so we're all standing up naked unable to deny any of our most secret fetishes.

What if it already did...

 

[millamomud]

As an ex-accademic, I’m calling BS on this.

I bet my professors which warned me from using wikipedia would be glad if students used wikipedia today instead of chatgpt.

 

[EmilyMiller]

I bet my professors which warned me from using wikipedia would be glad if students used wikipedia today instead of chatgpt.

It think the operative word is ‘just.’ Just using Wikipedia as a source for something important (as opposed to actual primary sources) is foolish. But it’s broadly good for what it is, a quick reference encyclopedia, that often captures the essence of entries, and points to more in-depth material.

I’m still waiting for examples of these grossly inaccurate pages. I’m sure that some exist, but to claim that the whole thing is inaccurate is frankly bizarre, unless motivated by some other agenda.

 

[ElectricBlue]

Guys, I’m pretty sure that at least Grok, among the AIs out there, has read and continues to read and internalize all of Literotica in its core learning model. No wonder it is the horny AI.

What you don't know, though, is whether or not the content search was done specifically in response to your prompt. Grok might not go looking for Publius68 until you asked it to. You can't necessarily infer that "all of Lit" has been used for training, just because you got an answer tailored to your content. I don't know if this is the case - just speculation from first principles.

Your content might not have been scraped, but when you prompt it, your content gets scraped. It's a variation of Schrodinger's Cat.

 

[EmilyMiller]

Here is a highly accurate Wikipedia page, it’s good that we still have such valuable resources available to people.

 

[ElectricBlue]

I use Wikipedia extensively for technical, musical and historical (though not recent history) information. I’ve found those areas to be well documented. I’m a financial contributor to it (and to the Internet Archive) because it serves an important function to help restore the original dream of the Internet.

I'd trust Wikipedia long before I'd trust an ai generated search on Google.

What amuses me is the disclaimer at the bottom of the AI response: "This information may not be accurate." Gosh, really?

 

[EmilyMiller]

I'd trust Wikipedia long before I'd trust an ai generated search on Google.

Our rather unscientific, crowd-sourced approach seems to - at the very least - not fully support the claim of endemic Wikipedia inaccuracy that was made, and not defended.

 

[Publius68]

Oh, and just to underline this - it also cross-referenced my nascent profile on Erotxt and drew conclusions based on the combined info.

Wow!

 

[Publius68]

What you don't know, though, is whether or not the content search was done specifically in response to your prompt. Grok might not go looking for Publius68 until you asked it to. You can't necessarily infer that "all of Lit" has been used for training, just because you got an answer tailored to your content. I don't know if this is the case - just speculation from first principles.

Your content might not have been scraped, but when you prompt it, your content gets scraped. It's a variation of Schrodinger's Cat.

I use it enough to know the indicators of when it does a search for information to craft its answer. It generally throws out its current general search terms, then describes its query process. With Grok questions about Lit (and many other broad websites) it simply answers immediately, with “low effort” cost. That is an internal model response. I need to see if it knows very recent stuff or not. Anyone notice that yet?

 

[Publius68]

And @EmilyMiller I do apologize for trigering your I am sure just defense of Wikipedia. It is indeed an essential, if flawed, internet asset. (Ditto Internet Archive). My personal experience with it has been more fraught than yours, clearly. I, like Darth Vader, feel that it has failed me for the last time. And several times before that. Once, resulting in significant public embarrassment for myself and the loss of a fifty dollar bet. I do not trust it anymore, and you can;t make me. [Insert petulant, childish Nyah!] I’ve also had one of my kids trigger (not my fault, I swear!) a college professor about Wikipedia, resulting in an entire lost class while said academic related his opinions on Wikipedia and the quality of its literary analysis...
It, like anything else of its size and magnitude, contains multitudes. We all may have our own opinions on the matter, and all others.

 

[millamomud]

It think the operative word is ‘just.’ Just using Wikipedia as a source for something important (as opposed to actual primary sources) is foolish. But it’s broadly good for what it is, a quick reference encyclopedia, that often captures the essence of entries, and points to more in-depth material.

I’m still waiting for examples of these grossly inaccurate pages. I’m sure that some exist, but to claim that the whole thing is inaccurate is frankly bizarre, unless motivated by some other agenda.

My studies began in 2005, wikipedia was still fairly new back then.
I'm sure perception may have changed in the meantime.

I used to check the sources from those articles, but many of my profs had some stories to share about wiki fails in academics.

 

[Bramblethorn]

I'm not a researcher or an academic. I end up on Wikipedia a lot just following curiosity and tend to believe the things I see on there. When I say it should be taken with a grain of salt I only mean that, were I to use it for actual data collection or for something more formal/official, I would be careful to check the citations.

This. Wikipedia isn't perfect by any stretch, but it's not intended that users should take it on trust; it's intended that editors should provide citations and readers should check those citations. When used in that mode, it's pretty good. One still has to watch out for subtler biases and omissions, but that applies to pretty much anything out there.

Incidentally, most LLMs are trained on Wikipedia among other things. The idea that they'd be more trustworthy than their training data is an intriguing one.

 

[EmilyMiller]

Once, resulting in significant public embarrassment for myself and the loss of a fifty dollar bet.

Not triggered - but thanks for adding being patronizing to making unsubstantiated claims. If something is said that is clearly wrong, I’ll call it out. If you want to label that triggering, somehow trying shifting the blame for your eroneous claim to others, then that’s up to you.

Your lost bet sounds like your own issue, TBH.

Literary analysis is not a science. There are no rights and wrongs. Where Wikipedia is based on knowledge and not opinion, it is a serviceable tool. But it’s not one to rely on by itself if aiming to give a public talk. Nothing like it is. But it’s better than anything else. Which is the opposite of your claim.