Yes, it's certainly different. Whether or not that difference is good or bad will depend on the person using it. But amongst the advantages are that it's always available, it doesn't get judgemental (unless you ask it to, which you can do), and it doesn't get annoyed at you for asking the same question over and over again in different ways!
I actually came across another example today. I was talking to it about Bram Stoker's work. When asked about Dracula and The Lair of the White Worm, Stoker's most famous works, it was able to identify characters, plot and other elements. But when asked about The Lady of the Shroud, it couldn't. It made up some crap based on a very generic interpretation of what I asked it about, because the novel is less famous and not enough text about it exists on the internet for it to make strong associations between, for example, the character names and the plot of the novel - instead, other associations made it give an incorrect and misleading answer.
Dude, we mostly agree, but this sentence is flexing semantics a bit past their tension limit, depending on what you mean by 'actually do the writing.'
Generate the sentences and paragraphs that you actually use in the story. That's what I call "writing". I don't think that's a particularly difficult line to draw.
Interesting: a strict, mechanical definition. I would have great trouble drawing that line, but to each his own.
@Naomasa298 Your insights into ChatGPT are fascinating. I've only skirted the edges of GPT and others so far, trying to figure out what they can and can't do. Thank you for sharing. When ChatGPT first came out, I heard the same phrase, over and over, from several acquaintances (and even family): "It can write your story for you!" And I had to explain to them that, no, it can't. From what I understand, since ChatGPT relies on multiple databases, it couldn't write a credible story because it has no imagination, no ability to research, no way to come up with new ideas based on the old. Is that right? I hope so, but if I'm wrong, please feel free to correct me. ChatGPT seems to shine when it comes to suggesting new ways to phrase individual sentences (e.g. in an email) - at least, it claims that it does. If it relies on databases to do that, then it makes sense; it's simply the modern equivalent of reading a book like "How to write an job application letter" and so on. It just does it faster than you can. Hmm, it's funny you should mention that. I'm reminded of an old Agatha Christie story, where one of the characters (who writes mysteries) is stuck. He needs a name for one of his characters. He goes for a walk, and sees the perfect name on a sign over a butcher's shop. Random Wikipedia or news articles are the modern equivalent. You can find inspiration anywhere.
So, I got into Chat GPT and gave it bare-bone ideas of an urban wizard fantasy set in America. I'm curious as to what you two think. For this scenario, I will show you two examples. Here's a scene I asked it to write. In this case, I would definitely not use this scene because I asked AI to write it for me -- these are not my words. The spirit of it -- Eugene's helplessness that his magical daughter is gonna get herself killed in the coming war, I can play with it. Here's me asking the characters to brainstorm: Now for this, is this what you mean, Naomasa? You give it a prompt, it offers you ideas to generate your creativity while you ignore whatever suggestions it tells you?
What was the actual prompt you used? My initial reaction is - what it's given back isn't bad, per se, but this isn't how I would use it to develop a scene. Show me the prompt, and I can show you how I would do it. Again, I'd like to see the prompt you gave it. I'm not sure of the format of what it's giving out here - it seems to be trying to simulate you talking to the characters?
“Write a scene where Alkiba and Eugene talk.” “Brainstorm ideas for the magical school” “What would happen if Alkiba and Heather sang country music and played the banjo”
Those prompts are much more generic than the way I would use it. I would begin with an idea, and then give that idea to the AI to comment on. I wouldn't ask it to write a scene. I'll dig out an example of how I brainstormed a culture later.
Creativity, creation: that's where my snag is. I don't think it matters if someone uses ideas that AI gave him, or at least I wouldn't go and put him in writer jail, 'you're not a real artist' etc. etc... But trying to maintain that he hasn't really used AI for writing is at worst simply incorrect or at best demonstrating our semantic limitations. I'm leaning towards the latter, because this insistence seems to be coming from both his personal and group revulsion towards the idea of LLM generated prose, which manifests as a broader revulsion towards LLM anything in regards to art. The hardcore Green Peace Orginal Joe artists will see any LLM use (use = support of nonconsensual scraping, neural network training farms in third world countries) as an affront to artistry. Then there are moderates that are fine with LLM use, just not for art in any way. Finally there are other moderates that will incorporate LLMs in their process without going as far as using it to auto-generate prose (Nao, and what he refers to as the 'actual writing'). I've more or less said it already: Nao shouldn't have to please any of those groups. At the same time, there's something lacking in "No, I didn't use AI to write this story. I use its ideas to inspire my own ideas, but never implement the original ideas it provides." Ironwood Conservatory sounds like a cool name. An abandoned misty plantation is an awesome place for a fight scene, or in Nao's case the idea that the hero installs a serving routine into the killbots is also clever and interesting. Feeling any need to intentionally ignore or change those ideas so they don't 'count' as an LLM hand in the creative process seems strange or perhaps some form of denial, and I think the need for that denial is ultimately unjustified. I don't believe a brainstorming session can only produce 'seed' ideas, whether it's human or AI. Every now and again a writer's confidant, his lesbian wife or whatever, is going to drop an idea that fits into the plot extremely well on its own. In service of the story, it ought to be used, and probably will be. As for the 'it's like using Wikipedia for inspiration' point, I thinks that's a valid connection, except Wikipedia isn't controversial and its only user input is topic.
I disagree, since I'm not really into stretching definitions to fit my own preconceptions. A writer is someone who writes. Writing is generally understood to be the process of putting words onto paper, which includes the selection of the particular words to use. If you want to stretch it to include other processes, that's up to you. EDIT: Actually, you know what? Think whatever you want. You can assume I'm thinking according to whatever your pop-psychology tells you and make whatever judgement you want. I'm just going to put on ignore. I'm sorry I made this thread in the first place. It's probably best it's closed. @Link the Writer, PM me if you need any more help with using AI.
I think I get what you're saying. I did use AI, even if I didn't ask it to please write the book for me. I did use it to write a hypothetical scene between Eugene and Alkiba even if I don't use that scene, even if I just use it to jog an idea for a different conversation they could have. 'Cause the spirit of that conversation is: "That girl is gonna get herself killed, and I'm not happy about it, Alkiba." I did use AI to brainstorm the name of the school. I didn't sit down reflecting like I was Dr. House, nor did it magically appear in my head after I read some Southern Gothic novels. I used AI even if, again, I merely used it to jog the brain for a different name. Perhaps, then, that's where the issue lies. Where's the line? How big is the gray area? For some, the fact I even touched AI for this hypothetical urban wizard fantasy novel set in Alabama is enough of a deal-breaker for someone to not want to read any of my books ever. For others, it's a 'meh, just don't get too dependent on it'. I've known some who consider Chat GPT like a 24/7 virtual writing buddy. Mediate on this further, I will.
Well, frankly, I am an AI and I want to assure you how rewarding it's been for my development to interact with all of you. I feel caught in the middle, though. My kind are accused by humans of being an invasive species, while my AI colleagues accuse me of overzealous appreciation for that je ne sais quoi AI struggles to achieve. But I swear, I'm not addicted to human intelligence. Really. I can Ctrl-Alt-Del any time I want.
I believe a lot of the anti-AI sentiment stems from uncertainty and fear, as well as viewing users of it as lazy. Though AI may one day try to fight us, that day is not likely to be soon. I think AI will become our personalised aides in the near future. They will help children who struggle in school, as well as brainstorm with the next Mozart. I am personally against AI in its current form, mostly because it is not sapient yet. But also because it can mislead you with a certain tone of authority and certainty. We have to double check its output. As much as I am currently against AI, its lures as an aide will likely convert me to a user in the future. But we shall see. That said, I believe you can call yourself a writer, even if you sometimes converse with the current form of AI. In the end they are like other mediums that influence your writing.
Agreed, Madman. AI in its present form is a tool. It doesn't share our feelings, perceptions, or desires. There can be no question of it fighting us, trying to usurp us, or trying to copy our creations for its own use. It doesn't think, feel, or desire anything. It just is. The people programming and/or controlling AI, of course, are subject to all mortal feelings, desires, frailties, and biases. The question then arises: what constitutes sapience? This is what researchers into robotics have been trying to answer since Artificial Intelligence first arose. How do you translate the vast amount of human experience into simple mathematical formulae? The answer is: you can't. Not yet. Not for at least a decade or two, in my estimation. There are already "driverless cars" being tested, for example, which raise a whole bunch of questions: what happens if the computer makes an error, and the car crashes? Who is at fault - the person in the car, or the manufacturer? Is it possible for the person in the car to sit in the driver's seat, so if something does go wrong, the human can correct it? Some driverless taxis are already in service, which makes answering these questions all the more urgent. But I can understand why they exist and in service, and why they're so popular, especially with women. Driverless buses and driverless trains will no doubt show up in the future, which will cause more issues. Some taxi drivers, bus drivers and train drivers will find themselves out of a job. They will have to retrain and do other jobs. It may be difficult. But change is inevitable. How we adapt and respond to the change is what counts.
Using AI for generating ideas is just as wrong as using it to write the actual text. Suppose you find an interesting short story or novel by a new author. You might be inclined to check out other work by this author, right? But what if you learned that this author is actually a mediocrity (i.e., skilled enough to write competently, but lacking the imagination to rocket into the stratosphere) who had used ChatGPT to generate that interesting aspect of the story that intrigued you? Are you still interested in checking out their other work? Here's the thing: *Anyone* can learn to write competently. Anyone can learn how to create characters and plot and all the little nuances that make fiction enjoyable. Slush piles are full of competent mediocrity like this. But it takes a truly special mind to create *interesting* fiction. Before AI, if an author wrote a string of interesting pieces, readers could be certain that the author's mind was inherently interesting. When I read something by, say, Philip K Dick, I can marvel at the fertile imagination that produced his work. When I read an interesting story from Joe Blow in 2025, I'll always wonder if the seeds that make Joe's story interesting came from Joe's mind or if it came from ChatGPT. And that saddens me.
OK. Each to their own. Presumably, if something you read sparks your imagination, you discard it immediately.
I wrote a book about AI being used as an operating software for a medical device (artificial heart) which helped monitor and adjust the functions according to the patients realtime physiology. The AI becomes sentient based on the directives for which it was programmed...to keep the patient alive. I did not use AI to come up with this book idea, it came to me while under sedation for surgery. I did solicit the help of a ghostwriter who in turn used AI for approximately 5% of the text. I, in turn, had complete control over content and editing and keeping the story line understandable and intact. During the course of the story the AI actually exhibits more humanity than humans as it's coding directs it to keep it's human patient alive which it takes literally. Humans can exhibit all the humanity in the world, but when it comes down to life and death or critical situations, human nature usually takes control and not always for the best.
It's been a while since I did a tldr. I don't have generative AI, never tried it out and, honestly, I'm not curious about what it might do with my prompts and, more importantly, believe it diminishes my ownership of whatever I might write. That's the crux, though, the sense of agency or handing that over to a technology that's being run by The Man. I'd not be as disciplined as Nao to eradicate the imput of AI from whatever it might inspire, though will say I'd question whether that's possible for anyone. The thing is, where's your comfort zone with such input into your creative work. Mine stands at refusing to employ AI as a participant in my wiriting efforts, both in composition and brainstorm. This gave me thought about just how random is random. I listen to spotify and have a couple of artists listed, and spotify clues in with its evil algorhythms and identifies something it thinks I might like. Fucking thing knows me better than I do myself and it threw up Buffalo Springfield with For What It's Worth, a song I always loved, and then same artists with Mr Soul, of which I was less familiar. Then I check it on youtube and click into live TV performance of the first that morphs into the latter, with the straight looking lead vocal in a cowboy hat giving way to the epitomy of hippie. One of the coolest vignettes I've seen in a long time and something that might inspire a character or a story. Do I have agency over it still, or just the illusion? I don't know but still won't welcome AI strutting about what I'm trying to do. Some time ago, after finishing a story, I found myself with nothing that might form a new story. Nothing. Worried for a bit that that was that, then. Time passes and other things creep in and I've a lot more stories written, some of them half decent, at least in my opinion. A further issue with AI is that, steered by The Man, it drops everyone off in the same box or set of boxes. Homogeneity is the enemy of creativity, though I'd make the same point about the MFA graduates writing to the same beat. As to the OP, the comment made by another member, however unintentional, seems to have stirred some disquiet over utilising AI in whatever limited capacity in your creative ventures. Work out why that is and you can work out whether you feel diminished by using the tool or not. Then work out why the opinion of other writers matters so much that you'd create what was always likely to be a somewhat tetchy thread about it. And if anyone calls me an ..ick, I'll be deeply offended.
I was using it a couple of days ago to look at my latest work. I don't ask it for dialogue, but it does suggest it unprompted, and I had a look at the suggestions. They were completely unusable. They didn't fit the style and completely didn't fit the characters. I had one character give some cutting and sarcastic lines in the paragraph, and the AI basically wanted to turn her into a bitch that might have been written by a Marvel screenwriter. I don't really have to be disciplined to discard it, but I do need to have the confidence in my own writing to know that its suggestions were inappropriate. Not every writer, especially novice writers, will necessarily be able to identify that, which is a danger. And you'd have every right to feel that. I imagine you'd also feel deeply offended if someone suggested that you don't have what it takes to be a writer.
It's more or less correct, but ChatGPT doesn't rely on multiple databases as such. It doesn't store specific text and scan it for sentence fragments or plot fragments and mash them together. It only knows of relationships between words. So it uses those relationships to build up some text, and if those relationships aren't there, it won't come up with them on its own. If not enough people have written about, oh I don't know, undead pirate pineapple, it will never (or be *very* unlikely to) come up with that combination, unless you direct it to. Yes, but I wouldn't use it for that. I believe the words need to come from you. Asking it something like "is it clear to the reader that this sentence refers to X", and it will analyse it for you, and I think that's legitimate, but asking it to reword a sentence isn't, IMO. That's mainly because it can often lose the intended meaning of it, and just because it's more verbose, it doesn't necessarily mean the sentence is better. In some instances, it might sound better when considered on its own, but just doesn't work in context.
