ChatGPT your way past writer's block

[ecorche]

Here is what ChatGPT has to say about when it should be credited:
"You should credit OpenAI's ChatGPT whenever you use it as a source of information or help in your writing. This can be done by adding a note or citation that states "Powered by OpenAI's GPT-3 language model." The exact format and placement of the credit can vary depending on the context and your specific citation style."
So, "whenever you use it as a source of information or help" is a pretty broad brush. Maybe I should start crediting my thesaurus, too. LOL.

 

[ecorche]

I'll be right back. I'm going to start a thread on when and how to credit coffee.

 

[Voboy]

You have a good point with beta readers. They will provide feedback that will change the direction of your story. The difference is that they're doing this based on skill and experience. They should be acknowledged for their time and valuable insights.
I don't view ChatGPT as a beta reader. It's more like a thesaurus. It gives me a bunch of options to consider. Most of them are crap because, unlike beta readers, ChatGPT has no skills or experience to guide its suggestions.

This is false. It has learned, within the limits of its programming; it has been trained. That's what makes it AI.

I've never read a story where the author credited a thesaurus, or a spelling and grammar tool, except in jest.

To me, the difference there (and the reason why I compared AI to beta readers) is that a thesaurus is passive. It does not interact with you. It just sits. In intelligence terms, it is crystallized: final. Unchanging. You go consult it. An AI strikes me as a more interactive experience: it is capable of changing the help it gives you. It is fluid, not crystallized... like a beta reader.

I don't see any distinction at all between AI and beta readers, as I'm sure I've made clear. I'm prepared to accept the one as I accept the other, but I definitely feel that both of them detract from the purity of the writer's mind and abilities. And I think it's only fair to point that out to the reader.

I see it as a spectrum, of sorts: at one end sits a writer with nothing but a piece of paper and a pencil, at the other sits that same writer with an AI onscreen and a beta reader waiting in the wings. We're all somewhere on that spectrum. Unapologetically, I think those closer to the paper-and-pencil writer are more pure about the stories they tell.

 

[ElectricBlue]

as detecting AI generated text can't be easy or simple.

Right now (January 2023), every example I've seen of AI having a go at fiction - both here and elsewhere - has given itself away within a few sentences. The examples I've seen are repetitive, incoherent once a few sentences are in place, and filled with false connectivity, correlating things that don't, in fact, correlate. Each phrase or sentence might be okay on its own, but after a while, the cumulative effect becomes nonsense.

A human reader can easily see that. A bot might be able to parse some kind of a "likelihood rating", if it's smarter than its writing buddy's bit; but how does one tell, for example, if someone has run a story through Grammarly or Word spell check, or can competently edit stuff by themselves? To my knowledge, you can't. Although it's pretty easy to spot where someone might benefit from such tools.

I haven't seen enough of its use in other disciplines where it might do better, but in terms of fiction, it's got a way to go.

 

[AwkwardlySet]

Right now (January 2023), every example I've seen of AI having a go at fiction - both here and elsewhere - has given itself away within a few sentences. The examples I've seen are repetitive, incoherent once a few sentences are in place, and filled with false connectivity, correlating things that don't, in fact, correlate. Each phrase or sentence might be okay on its own, but after a while, the cumulative effect becomes nonsense.

A human reader can easily see that. A bot might be able to parse some kind of a "likelihood rating", if it's smarter than its writing buddy's bit; but how does one tell, for example, if someone has run a story through Grammarly or Word spell check, or can edit stuff by themselves? To my knowledge, you can't. Although it's pretty easy to spot where someone might benefit from such tools.

I haven't seen enough of it's use in other disciplines where it might do better, but in terms of fiction, it's got a way to go.

You have a point, but it is not hard to imagine these AI tools getting better in a year or three. And there is the question of manpower as well. We all know the process of approval for Lit stories for that exact reason. It is also possible that many would use AI to generate text and then maybe do some corrections to the nonsensical stuff. So it would be slightly altered AI story. How do you detect that? No matter how you address this, eventually, It will come down to personal integrity and lines, I think.

 

[ElectricBlue]

You have a point, but it is not hard to imagine these AI tools getting better in a year or three. And there is the question of manpower as well. We all know the process of approval for Lit stories for that exact reason. It is also possible that many would use AI to generate text and then maybe do some corrections to the nonsensical stuff. So it would be slightly altered AI story. How do you detect that? No matter how you address this, eventually, It will come down to personal integrity and lines, I think.

Agree this. For me, it's wait and see, as to how good AI gets at writing fiction. When it can ape my style, I might look up and take notice, but in the meantime it doesn't worry me - because I've still got imagination, which silicon doesn't. Also, anything AI writes is based on words that have already been written, but I haven't written my next words yet.

 

[XerXesXu]

I'd also note that at this point, the person doing the most harm to 8letters' reputation is the one purporting to be his defender.

The way these things typically pan out: somebody fucks up and gets a hostile reaction, they leave in a huff, in a few weeks or months the ruckus dies down and people forget what the fuss was about, person comes back - maybe there's a token apology, maybe not - and then mostly folk move on and act like it never happened, unless that person does something to resurrect the issue.

Right now we would normally be moving into the "ruckus dies down" stage. But reviving the topic in this thread, and prodding others to defend their criticisms of 8letters and to keep on restating the facts of the matter, prevents that from happening. Do that for long enough and a rift that should have healed in time can turn into a festering sore.

I would like to believe that isn't the intended purpose.

You do believe that it's not the intended purpose. You'd like to think that what I'm saying wasn't so distressingly true. That's quite different.

 

[XerXesXu]

I'll be right back. I'm going to start a thread on when and how to credit coffee.

That's a great prompt for a story, but I'll take that to your coffee thread, at an appropriate time.

 

[Bramblethorn]

So. I've read the thread, and here are my revised thoughts.

I can grant that some writers feel the need to use this as a tool to improve.

My feeling is that a lot of posters here think of AI the way they'd think of a beta reader: another "brain" off which to bounce ideas during the draft/development phase of some of their stories. I do not (and would not) do this, in the same way I feel no need for a beta reader, but that's the only way I can charitably imagine a self-respecting writer using AI in a work intended for publication under their own name.

So my new thinking is that writers who use AI are no more "wrong" than writers who use human editors... but that ethically, in the same way you'd credit that human editor (or, I hope, beta reader) publicly, a writer who uses AI should credit that AI.

It's not by any means a universal norm that editors do get a credit.

I've done a fair bit of professional editing in technical (non-fiction) publishing; in various roles, I've provided editorial services on dozens of books, tens of thousands of pages worth. In all that time, just one of those books has credited me as an editor, and I was quite surprised to see my name in the front matter. The more usual mindset is that I'm providing a technical service, often through a couple of layers of subcontracting, and that I don't get credited for that any more than the workers who physically print and bind the book.

It's not that the service isn't appreciated - I often get lovely notes from the authors saying nice things about my edits. (Also, I get paid, on time.) It's just not viewed as a role that needs to be credited.

I don't work in professional fiction publishing, but my understanding is that crediting editors is not usually seen as required there either. Authors quite often will use their discretion to thank editors, beta readers, etc. as part of an author's note (or occasionally a dedication) but it doesn't appear in the front matter, and many books don't name the author at all. Overall my vague impression is that these discretionary editor credits are becoming more common but are still a long way from universal. From a quick look at my bookshelf, it's about 50-50 whether the editor gets a mention or not.

(Terminology note: "editor" is a term that can be applied to several very different functions. Some of those do get a credit as standard, e.g. the person who coordinates an anthology or the researcher who goes through a famous person's letters/diaries and compiles them for publication, adding context notes. Contrariwise, the technical editing I do sometimes has titles other than "editor". But for purposes of this discussion, I'm talking about "editors" in the sense that we'd use it on Literotica, somebody who works as a kind of secondary collaborator with the author, offering suggestions for improvement but not taking a lead role and not having final say over what gets published.)

In places like Literotica, where the editor isn't getting paid, crediting them is more common. I'd describe it as a norm here, but not a strong one. I've edited for some authors who credited me and for others who don't. The one who I edit for most regularly doesn't generally credit me (not sure if they ever have - not in any of the five stories I just checked) and I'm comfortable with that; I don't require it.

FWIW, being credited isn't always a positive. Authors don't always accept editors' suggestions - or they make last-minute changes that introduce errors etc. after the editor has already checked the piece - so there's a risk of editors being blamed for problems that were completely out of their hands. A couple of weeks back one of the authors I edit for got some feedback to the effect of "great story but you need better editors, let me assist" from somebody who'd made no effort to find out whether the editors actually were responsible for whatever they saw as "errors".

I recall the movie O Brother Where Art Thou. At the beginning, it credited Ulysses. Does anyone think it would've been plagiarism is the Ulysses credit were absent?

In that case no, both because of the nature of what was taken (a broad story outline, heavily adapted for a different setting, and some character names) and because there was no likelihood of those elements being passed off as the original creation of the Coen brothers. Not everybody knows Ulysses, but enough do that they're going to tell their friends.

(Erotica reading being more solitary a hobby than film-watching, this might be an area where authors here need to be a little more careful than a film-maker who can expect to have their work dissected in public.)

I'd give some leeway for easter eggs, because the whole point of an easter egg is that at least some readers will recognise the source.

You have a point, but it is not hard to imagine these AI tools getting better in a year or three. And there is the question of manpower as well. We all know the process of approval for Lit stories for that exact reason. It is also possible that many would use AI to generate text and then maybe do some corrections to the nonsensical stuff. So it would be slightly altered AI story. How do you detect that? No matter how you address this, eventually, It will come down to personal integrity and lines, I think.

It may be worth distinguishing between the discussion specifically about GPT, and the discussion about potential future AI writing tools. I am not a machine learning expert, but it's part of my job to know a little bit about the basics, and I've read one of the creators' papers on how GPT works.

Based on that, my personal opinion - take it for what that's worth - is that GPT can emulate some but not all of the things required to produce a decent story. As we've discussed here, there are various ways it can assist a writer, but in order to produce a good story, one that can hold reader interest for more than a few pages, the human author is still going to be doing most of the work.

We can argue about whether a story that's 80% human work and 20% uncredited GPT work ought to be allowed here, whether it's ethical, but it's not something that's likely to change the Literotica landscape very much. An author might be able to sneak a mostly-GPT story past moderation, since the deficiencies of GPT are mostly things that don't show up easily on skim reading, but if the readers are going to trash it there's not a lot of motivation.

(Pay sites, OTOH, may have problems, because spamming Amazon with GPT-generated "books" only requires a very small success rate to pay for itself.)

Somebody might invent a tool that works in a different way and is able to generate a satisfying plot with internal consistency, but if and when that happens, its characteristics will probably be quite different to GPT, and there's only so much we can usefully discuss without knowing the specifics.

This is false. It has learned, within the limits of its programming; it has been trained. That's what makes it AI.

Cynically speaking: what makes it "AI" (as opposed to "ML" or "fancy regression") is that somebody decided that that term had a better chance of attracting venture capital.

 

[XerXesXu]

Unapologetically, I think those closer to the paper-and-pencil writer are more pure about the stories they tell.

The old 'I can't believe it's not butter test. Oops ... I did it again - Twice. Oh -Ah-Ah. Something's coming over me, I want to make you laugh. It must be love - Madness - HELP - They're coming to take me away.

 

[Bramblethorn]

You do believe that it's not the intended purpose.

Ah, another fellow who believes himself to be telepathic. In that case there's no need for me to type my reply.

 

[TadOverdon]

I'm not so naive as to think that how some folks respond to this one won't be colored by their opinions of Chomsky's politics.

Be that as it may, he's not wrong.

 

[XerXesXu]

Ah, another fellow who believes himself to be telepathic. In that case there's no need for me to type my reply.

I can read the room. I know what you're thinking. It's not telepathy, it's empathy.

 

[AwkwardlySet]

I'm not so naive as to think that how some folks respond to this one won't be colored by their opinions of Chomsky's politics.

Be that as it may, he's not wrong.

View attachment 2206606

I didn't even know Chomsky was still giving interviews. Thank you for this

 

[XerXesXu]

I didn't even know Chomsky was still giving interviews. Thank you for this

That's telepathy.

 

[Bramblethorn]

I'm not so naive as to think that how some folks respond to this one won't be colored by their opinions of Chomsky's politics.

Be that as it may, he's not wrong.

Yeah, I'm not the guy's biggest fan, but I think he's right on this one, and the mindset he describes extends far beyond AI.

See also this XKCD:

https://imgs.xkcd.com/comics/physicists.png

 

[Bramblethorn]

I can read the room. I know what you're thinking. It's not telepathy, it's empathy.

https://psycnet.apa.org/doiLanding?doi=10.1037/pas0000732

 

[XerXesXu]

https://psycnet.apa.org/doiLanding?doi=10.1037/pas0000732

Hahaha. That's what's written on the screwed-up piece of paper I hold in my hand. "Barrage of Google links incoming." How could I have guessed? Ans = Empathy.

 

[SimonDoom]

I'm not so naive as to think that how some folks respond to this one won't be colored by their opinions of Chomsky's politics.

Be that as it may, he's not wrong.

View attachment 2206606

I'm going to disagree, although tentatively, and it has nothing to do with Chomsky.

While this kind of thing may often be true, this example doesn't prove that it is generally true, and I can think of examples where it wouldn't be true. I'm also skeptical of the "gatekeeping" mentality in general, and I sense there may be some merit to the artist who is criticized in this case regarding his claims about "gatekeeping" activity.

Are people poorer mathematicians because of the existence of calculators and computers that aid the ability to do calculations? I don't know, but I doubt it. It may be that the deemphasis on hand calculations means that there are fewer people who can do calculations in their head the way John von Neumann could, but I rather doubt it at the highest levels. And the ability to do basic calculations faster means that our brains can focus more on math problems at a truly high, creative level.

Consider Chess and Go. It's possible that computers, now that they can beat people at both games (it took them much longer to beat people at Go than at Chess, interestingly enough) have taken some of the fun out of the games, but my impression is that people are better Chess players than ever, in part because of the use of computers in their training. I can imagine computers and AI being used in innumerable ways to "speed up" aspects of authorship that may not necessarily retard the creative process. You might ask a computer to give you 20 different ways to compare sex to a baseball game. You can then sift through them, select one or more, alter them, until you arrive at the one you like best. Has this destroyed the creative process? I'm inclined to think it hasn't. To the extent we can process information faster, it may enhance us by enabling us to reach for more interesting metaphors. We might even expand the language. I'm going to defer judgment until I see more real-world examples where really good authors actually try using these tools so we can see what they come up with.

 

[Voboy]

I'm going to disagree, although tentatively, and it has nothing to do with Chomsky.

While this kind of thing may often be true, this example doesn't prove that it is generally true, and I can think of examples where it wouldn't be true. I'm also skeptical of the "gatekeeping" mentality in general, and I sense there may be some merit to the artist who is criticized in this case regarding his claims about "gatekeeping" activity.

Are people poorer mathematicians because of the existence of calculators and computers that aid the ability to do calculations? I don't know, but I doubt it. It may be that the deemphasis on hand calculations means that there are fewer people who can do calculations in their head the way John von Neumann could, but I rather doubt it at the highest levels. And the ability to do basic calculations faster means that our brains can focus more on math problems at a truly high, creative level.

The calculator argument is a good one, and it's one of the parallels that began making me think differently about the use of AI in writing. But I think that all but the most complex math problems have far fewer variables than even a mundane creative writing exercise, once you factor in punctuation, word choice, plot moves, etc.

Consider Chess and Go. It's possible that computers, now that they can beat people at both games (it took them much longer to beat people at Go than at Chess, interestingly enough) have taken some of the fun out of the games, but my impression is that people are better Chess players than ever, in part because of the use of computers in their training. I can imagine computers and AI being used in innumerable ways to "speed up" aspects of authorship that may not necessarily retard the creative process. You might ask a computer to give you 20 different ways to compare sex to a baseball game. You can then sift through them, select one or more, alter them, until you arrive at the one you like best. Has this destroyed the creative process? I'm inclined to think it hasn't. To the extent we can process information faster, it may enhance us by enabling us to reach for more interesting metaphors. We might even expand the language. I'm going to defer judgment until I see more real-world examples where really good authors actually try using these tools so we can see what they come up with.

Creative writing, though, is not a game with discrete rules, and one does not train for it in the same way that a chessmaster does.

I am skeptical that AI would help you become a better writer unless you were very shoddy to begin with. Like, worse than the AI.

 

[SimonDoom]

I am skeptical that AI would help you become a better writer unless you were very shoddy to begin with. Like, worse than the AI.

Maybe. This is why my answer is only tentative. We won't know until we give it some time and see what happens. I'm very skeptical of the "AI will make us lazier authors" argument. I'm sure it will make us change the way we write, but it may simply re-channel how we use our creativity. I think people with a true artistic bent WANT to be creative and are not looking for shortcuts.

 

[alohadave]

I'm not so naive as to think that how some folks respond to this one won't be colored by their opinions of Chomsky's politics.

Be that as it may, he's not wrong.

View attachment 2206606

Meh. Making art does not necessarily need to be hard work. It sounds like "I had to bust my ass to do this, and so will you", or paying your dues.

Just like every single time new technology has made photography easier to do, the current guard decried it as ruining the art. And yet, every time technology has made it easier, the art has expanded and grown. You could say that most of the pictures being taken today are worthless, but we are more connected to the world because the effort required is so minimal.

People expressing themselves via the tools available will always make themselves seen.

Is ChatGPT the same? Maybe not. But it is an exciting new tool that has more people talking about and experimenting with writing. And that's never a bad thing.

Creative writing, though, is not a game with discrete rules, and one does not train for it in the same way that a chessmaster does.

It does have rules of style, they are taught in school to teach people how to do it. Practice writing is a thing that some people do.

 

[Voboy]

It does have rules of style, they are taught in school to teach people how to do it. Practice writing is a thing that some people do.

...which are often broken, with dazzling results. Just ask James Joyce.

My point was that chess and go are both finite: they are bounded systems. The number of possible combinations of moves is knowable, though VAST. Creative writing is not; it has nearly infinite variations.

I think a human writer would have very little to gain from having this current generation of AI "teach" them. I think they would move beyond the skills of the AI very, very rapidly. Now, naturally, AI will improve... but, as you point out, so will the state of the art in creative writing in general.

There may come a day when humans write worse than computers, but if that happens, we're in worse trouble than worrying about mere plagiarism.

 

[XerXesXu]

There may come a day when humans write worse than computers, but if that happens, we're in worse trouble than worrying about mere plagiarism.

There's a premise for a Sci-Fi story.

 

[Bramblethorn]

Meh. Making art does not necessarily need to be hard work. It sounds like "I had to bust my ass to do this, and so will you", or paying your dues.

There's no virtue in hard work for the sake of hard work. If I can find an honest way to get my job done more easily without compromising quality, I'll embrace it.

But some good things do take work. The assumption that everything in life must have a cheat code, that makes suckers and thieves out of people.