Eh, what? This is specifically about how the people who frequent this subforum feel about allowing or disallowing AI generated contributions. It’s not about whether AI generated fan fiction can be stopped or not. It’s about finding a consensus as to what is the agreed upon standard for posting creative works here. It’s about establishing a norm or sort of moral code of what goes and what doesn’t. Sure, the question that follows might be about how to go about bringing forth this standard, and the difficulty this might present. But that’s not what I understood this to be about.
Well, I certainly view these very differently. The human imagination is not like an algorithm trying to piece together the data it’s being fed in a way that tries to appear like an original idea. Sure, feeding the human imagination with art that’s already out there in the world can help form it, but there’s also room for stuff like complete randomness, individual quirks and sensibilities, intuition, self-awareness or ethics and morals. The human imagination has the ability to generate completely new ideas and concepts that are not explicitly derived from existing data. Humans can think beyond the constraints of already existing patterns and rules, while AI algorithms operate based on patterns and data they have been trained on, which means they can never have a truly original idea.
Exactly this. It’s not a question of whether we can use AI to create art for us. The question is: Why the hell should we? Being creative is one of the must fundamentally human things we do. Why are so many people so eager to have a “tool” (I’d argue it isn’t really a tool, since tools are meant to help you do something, not take you out of the whole process and do everything for you) that takes creating art from us?
Complete randomness is impossible because the brain consists of atoms, and all atoms behave deterministically. Functional magnetic resonance imaging can accurately predict actions before the person is consciously aware of choosing to perform them. The human mind is far more complex than any current artificial intelligence, but its complexity is finite: 86 billion neurons with up to a quadrillion synapses. Whether it takes hundreds, thousands, or millions of years, both artificial and our own enhanced human intelligence will eventually exceed the capacity of the current human mind in every way by many orders of magnitude, revealing to all what a few have already realized, that the human brain is capable of only so many thoughts, ideas, and expressions. Vastly more than any current artificial intelligence, but also vastly fewer than the human and artificial intelligences of the distant future. We're also constrained by existing patterns and rules; it's just that the set of patterns and rules to which we're limited is vastly larger than the set available to current AIs.
Copyrighted works can legally be used without the rightsholders' permission or knowledge—or even in spite of their vehement disapproval—in any way that meets the legal definition of fair use. Studying them is one such legal use. It's legal when humans do it and it's legal when AI does it, too. George Lucas didn't violate any copyrights when he viewed and read copyrighted works such as Flash Gordon, John Carter of Mars, Dune, and Triumph of the Will and incorporated various elements of them into new creations, and AI doesn't violate any copyrights when it studies existing works and incorporates elements of them into new creations, either.
Artificial intelligence will evolve far beyond large language models. This is the beginning, not the end.
There is a fundamental difference there. He did not literally take them, chop their writing up into little pieces and deterministically reassemble it based on a prompt. He was inspired by it, but the Star Wars screenplay was completely original in that sense.
That's the thing! LLMs are not intelligence. Artificial Intelligence could be vital, but this is not artificial intelligence. Artificial intelligence's meaning in compsci circles is not the same as that in science fiction.
Well, this is the fan fiction section after all, so knock yourself out, I guess.
I’ve observed some of that in creative circles, sure. Visual artist have begun to recognize generated images and their various tells (I’ve seen some refer to it as “AI spew”), growing more and more tired of how they are flooding image searches, stock libraries and basically all social media. But I’m not so sure the general, non-artistically inclined population is anywhere near that right now.
Let me preface what I’m saying by that I’m only conveying my personal understanding of these matters. While it’s true that we might not know everything about how exactly the human mind works right now, we do have an understanding of how the algorithms of AI are programmed. I’m sure it’s an oversimplification to put it that way, but it’s my understanding that AI’s main function is to generate something that fundamentally seems like it could be something a human has created. In terms of text generating AIs, it’s basically taking guesses as to what word might follow another because it often follows that word in human texts, to put it crudely. It has learned from studying all available text how common or likely particular phrases are as a response to a particular prompt. The human imagination, as I understand it, does not work like that. When I create a piece of artwork, my creative process does not include guessing what pixel is more likely to appear next to another pixel. I’m trying to render something I’m seeing in reality or something I’m “seeing with my mind’s eye”. I’ve learned to take into account how light and shadow work, how shapes in three dimensions work etc. The process might include experimentation, including aspects of randomness in my work, with me deciding every time if something works to convey my idea or not. And to me it does not seem like these two approaches are very similar.
It's not legal when AI does it, mainly because AI doesn't actually study anything.
Bingo. AI has no idea about grammar or the structure of language. It doesn't understand anything outside of the positive reinforcement it gets from randomly stringing words together correctly. Midjourney doesn't understand the form of the dog, it just understands pixels in a certain order probably mean dog.
Now, if AI actually experienced positive reinforcement and learned from it, we are truly looking at intelligence, as well as the possibility of consciousness. There is no positive reinforcement in computers today. They do what they're programmed to do, the way they are programmed to do it, which includes learning algorithms. Computers and their programming experience no pleasure and they don't know when they have it right, only that it meets the correct parameters fed to it in their programming. Currently, if an AI program was never asked to perform a task, it wouldn't perform any tasks. It would just wait, no matter how much positive reinforcement it had previously received.
Frank Herbert disagreed, as do others.
Rigorous academic extrapolation is hardly fanfiction. The possibilities of going to the Moon and flying supersonically were mathematically proven before they were accomplished.Well, this is the fan fiction section after all, so knock yourself out, I guess.
According to the law, it is legal.
I have written professionally as a ghostwriter.
According to established law, it is certainly fair use in general, as supported by a wide range of academics, library associations, civil society groups, startups, major US companies, creators, authors, and others who have submitted comments to the US Copyright Office, as well as in the European Union, Japan, Singapore, Israel, and elsewhere around the world.
