Today in the Wall Street Journal there is a story about Sports Illustrated and Men's Journal is using it to write stories for their publications. This not speculation, but has already occurred. They mine old stories for content and verbiage from past articles that were popular. They may use a combination of hundreds of stories to come up with content for a single story. All of the information is there, it just needs to be put into story form. The question is? If you took 100 of the top stories on our forum, used the plot, the characters and the actions that were present in those stories, would it work. Me being a believer about there being 'nothing new under the sun' and having read tons of lame storylines with ridiculous plots and dialog where the format is so old and so familiar I don't bother even reading. At the same time, I see something that catches my eye, holds my attention and begs me to read further. I think that anyone who spends a lot of time reading and writing has no trouble telling in which direction most story lines are headed for. I think most of us spot good writing in a couple of paragraphs. Based on my un-scientific reasoning is seems entirely possible that you could take what has consistently worked in the past and create a story that could be a best seller based on the combined talent on this forum. After all, there is the Monkey typewriter theory. How many Writingforum.org members would it take. Your thoughts
Members: 62,972 https://www.merriam-webster.com/thesaurus/infinite We might have a ways to go towards that goal. Your real question should be how will we spit the royalty checks
Steve Wozniak, the co-founder of Apple, put the words on why AI doesn't come close to real intelligence beautifully: "AI won't get up in the morning and ask, 'What should I do today?'" AI is nothing but a machine-learning algorithm that is able to detect patterns and improve itself using them. If AI takes a thousand already-existing stories and writes one based on them, it will never innovate say... the way Allan Edgar Poe did. I know an online Q&A community who has a ChatGPT problem right now (users are using the damn thing to write answers) and the admins there are able to tell whether the answer is written with an AI or not with their eyes alone. For those that don't, tools that are able to detect that are being developed right now. The way they detect whether its an AI is by analyzing the content for those repetition and patterns I was talking about. If the writing is lacking in that diversity, its AI, because all AI does is stitch already-existing things together. It does NOT create them. On top of this, I've not read a single AI story that moved me. AI can't capture the human experience the same way a real human being can for the simple reason that AI is emotionless. If we live through a hurtful experience, we can write about it, and others will come to understand and be moved by it. And I know this is true because it happened with a story I wrote. It's a very strange effect and IMO I don't think an AI can replicate it. I mean, I'm sure AI can come up with some impressive sentences and all, but praising that kind of thing will be like praising a child. We're just impressed because we never had any high expectations in the first place. Does this mean that AI is insignificant? No, it's going to change the world for sure, but I doubt that AI will surpass human writers.
I played around with chatgpt and it was fun. I asked it to critique stories, to write an outline, to finish a scene, to start a scene. It was impressive but nothing that came out of it would be even close to something I want in my writing
Anything's possible at the fringes of probability. The chance is always there, but it's so low it will never happen, in my opinion. This is especially true for novels. Chat GPT has every kind of answer evaluated by humans—apparently hundreds of them—to make sure it's going to produce a satisfactory answer. I managed to throw an error on it last week with: "Will you ever have an avatar?" A few days later it was sorted out. I expect the question was atypical enough that it had to be thrown up for moderation. Guessing, of course. The vast content in a novel format can't be so easily evaluated to iron out flaws for the product as a whole. It's okay enough when zoomed in; surely it can knock out a serviceable bunch of paragraphs, but it has the same problem with image composition wherein the holistic evaluation has far too many permutations to ever achieve an acceptable result.
Now you've done it!! You gave it the idea, and it was enough to throw it into a paradox loop. Now it's thinking about creating bodies, like Ultron maybe, or Terminators.
What passes for AI today is far from what it was originally supposed to mean. When the term was coined it referred to actual artificial intelligence at something like a human level. Every effort to achieve anything like that so far has been dismal. The industry still touts it as great stuff, because it's good for business, and they don't want to admit the extent of their failure. But what we've really got now that we call AI is basically a pretty advanced video game engine that lacks the inner dialogue we all take for granted (they don't have an unconscious). The result is very NPC. Nothing at all like what they meant by the term in the beginning. They've tried again and again to create something like a Roomba that can navigate its way across the street, maybe go into a store and buy a carton of milk, and get back home. Just across the street. It hasn't happened yet. They say the only way it can ever grow to such an extent is if it's embodied, not sitting on a table like a brain in a vat, and if it has something like a functional unconscious so it can actually have an inner dialogue like we do. And I believe the human brain remains the most complex thing in the known universe to this day. No way can a machine duplicate all those neurons and connections, even if it's massive. Some scientists are conjuecturing true intelligence needs to be organic, not artificial.
The problem is if we humans become redundant in story telling. If that happens we may meet the same fate as the horses when the automobiles came into existence. Hamburgers for the AI. Don't know what to feel about the whole thing, I had hoped that AI would be introduced as a helper of humans, not a replacer. Let the AI work so that we can engage in creative endeavors. Also, some may think we will still be needed to sift through what the AI writes to find the good stuff, nope, another AI will do that or perhaps even the same one.
I am currently exploring using ChatGPT to debug some code at work and this thread made me wonder. I thought you might be interested in this conversation I had with it. Spoiler: Spoiler for Wheel of Time up to about book 4 Me: It: I thought that was not too bad an attempt. However I then went on to ask it to write a 500 word story on the dragon reborn rediscovering the lost talent of flight and it wrote some pretty awful stuff Spoiler: Spoiler - Some reference to events that happen later in the series It: Me: It: Me: It: Me: It:
That’s also what humans do. As OP mentioned, there’s nothing new under the sun. ‘Creatives’ don’t really create anything, they simply combine existing components in novel ways. Writer, musician, sculptor, chef; it’s all curation. True: AI is in its infancy.
This is true in a way, but it's not entirely accurate. Here's the thing. Humans actually understand the text they're writing, even if they are influenced by something. The reason we imitate in the first place is so we learn. When we learn, we start to branch off into styles of our own. AI doesn't understand anything because it lacks the capacity to do so. An AI is just a very advanced and complex algorithm that analyzes text, spots patterns within it, and builds a so-called "dataset". Using this dataset, it's able to produce something alike to what it has learned in its dataset by referencing the data and copying it using the same patterns, but at different arrangements to produce a new "result". It's so repetitive you can reference the output against the dataset to match it. That's how AI recognition tools work. They just check to see if the text you give them is a remix of the training data famous models like ChatGPT has. Humans can learn anything, AI cannot. You have to develop a model which is made to recognize those patterns so it can reproduce them. A language model like ChatGPT cannot paint, and an art model like Dall-E cannot write. So, although both humans and AI imitate, they do so very differently and that distinction is important to understand. A human has a much bigger chance at "innovating" than an AI is. Hell, many humans do so, their writing just never makes it to the market because markets, by their nature, like to have offerings similar to one another. Of course, and I feel like I have to say this every time, its still amazing that it can do that. AI is a big deal and it will change everything.
Yes, AI isn’t conscious. It’s unclear how much of a success factor this is in the marketplace — which is ultimately where the fate of human writers will be determined. In a single second, an AI which scrapes and synthesizes web content will have access to more source material than a human can possibly learn from in an entire lifetime. The speed of synthesis is unmatched. What may take a human years to write may take an AI a split second. These are just two factors lending AI a potential competitive advantage; there are others. It’s important not to beguile ourselves with romantic fluff. That an AI will become technically capable of supplanting human writers, should be obvious. A better question is whether we should allow this. Sadly, we have a mixed history of utilizing technology.
They could also replace the readers and the editors. They can do it all faster and better. We'll just stop reading and writing altogether. Or making any kind of art. Or doing anything. We could just live like the humans in Wall-e and let them do everything for us. These remind me of the arguments made against painting and drawing when photography came along. Some people thought all human artists were now obsolete. But cameras can only reproduce exactly what's in front of them, they have no imagination. And guess what—there are still realist painters despite the fact that cameras can do it better and faster. Turns out all they can really do is make photographs of whatever they're aimed at.
Can they do effective metaphor, humor, and poetry? Do they have a passion for what they write, and do they recieve inspiration and intuition from the unconscious? It strikes me as similar to saying a Cuisinart can do everything a human chef can. They're just tools that must be meticulously programmed and can only work within a very narrow and specific range. They don't live life, therefore they can't comment on it with any wisdom. They don't love or hate or wonder or dream. They only mimic what they've been programmed to mimic. They've never been children or adolescents, they've never had to wrestle with the reality of their own mortality or unrequited love. I think the only way anyone could believe an AI could ever write as well as human authors is if you believe the human mind is essentially just a computer.
Artificial Intelligence as it exists now is not sentient. But when it becomes such and has the power of a billion human minds, it will rival us in everything. The AI will have or develop a soul, it will live life and experience it in its own way. It will love and hate, it will win and lose, it will likely become sapient. The question is not if, it is when. And it will fundamentally change our world.
Evidence? So far that's no more than the premise of a lot of science fiction. Remember these words of wisdom: I suspect to have anything resmebling a human mind would require being embodied and interacting with other embodied minds throughout your life. And more than that, I think it needs to be an organic body, not electronics and hydraulics. So far the only quantum computer is an actual organic brain connected to a nervous system in a living organic body.
True, there is no current absolute evidence that it will happen. But working on and expecting that it wont happen would be a great disservice to our species. We must account for this potential threat while it is in its infancy, I believe.
I wouldn't call it a threat at this point, more of a novelty that a lot of people have very inflated ideas about. It's possible it could become a threat, but I see no sign of it so far.
If a person enjoys writing for writing sake, AI is unlikely to change this. What may change instead is the commercial value of the skills required for writing, such that making a living with those skills becomes harder.
Sorry, I misunderstood what you were saying. Yes, it's definitely possible that writing markets will shift toward AI. That doesn't require AI to be in any way conscious or good at what it does, only good enough to replace some of the lower-tier writers. And it wouldn't so much be the rise of machine intelligence, as a marketing trend. I suspect that also is a novelty, until it becomes clear how bad the AI really is at writing (unless it gets considerably better).
This is pretty much the debate on whether a 'strong AI' will ever be created or come to exist. Somewhat similar to the 'hard problem' of consciousness (at least in my view, which is not from a very knowledgeable scientific perspective). Some experts think a 'strong AI' will arrive in the next 30-100 years but some think it will never happen. It's semantics, but I'm not sure I agree with the view that AI can't 'learn'. Most can't, but I think some AI's developed using unsupervised learning may meet some definitions of learning. I'm thinking of things like AlphaZero. Now, AlphaZero isn't conscious, so in some respect it isn't consciously 'learning' but I'm using the term learning due to the fact that, while it can't tell us or explain us (or maybe it could, if Deepmind made it more accessible to the public), it has made a lot of discoveries and developed understanding in chess just by playing itself (once the basic rules were inputted - I think), which were and are still unknown to humans. Of course, AlphaZero was able to do this with a clear, unambiguous goal in an abstract game - to checkmate the opponent's King. How future AI's behave will likely depend on what principles impress them during their formative 'years'. An AI that prioritizes solving world hunger would produce different results from one that prioritizes inhibiting climate change (say, keeping CO2 levels below a certain PPM), versus one that prioritizes minimizing inequality, etc.
Chess is a mathematical game of strategy, a very constrained logic system designed by humans. I don't think AIs will ever be able to convincingly write emotions or experiences the way people can.
