Agreed, but I don’t think there are going to be any publications (whether digital or print) accepting work via submission with renumeration. I think that the media brands that we see today will just start sliding this stuff in at the low end of stuff, then, as it gets better, replace everything. There is not going to be a “side hustle” in creating content - regardless of whether it is original human created writing or AI generated content. Why would an SF print publication or CNET pay anyone for anything when they can generate it themselves? So the traditional outlets will first close the doors to outside human and AI contributions, before they are themselves destroyed by the flood of AI content that is coming through new channels - AI social media posts with ads against them, websites that track google and social media trends and auto generate SEO content against those, etc.
I don’t think there will be a commercial market for any “creative” content at some point some because the people with the purse strings don’t think they need it anymore. There will just be this ocean of AI generated garbage that is “good enough to run ads against” and nobody will be able to make money off of creative works except the people that already own the distribution channels and brands
This is where I think we end up…though I think it will take a while and when we do get there it will be even more horrible in some capitalist way that I am not equipped (thank god) to imagine. I think that you will pay for it, and you will have to pay for it one way or another and forever
Eh, as long as it runs off the fumes of other stuff. I’m not convinced hydrogen fuel cells can’t compute in the process of not going all out, or making good of the situation when AI invents fake dihydrogen and sells it as ‘low-oxygen fuel.’
thekevinmonster> What would it take to add {default mode networks, etc.} to an LLM? What would happen…
I mean, add enough and it’ll MOON NVIDIA sorry sorry… I mean, It’ll consider some stuff and act a little woke until it starts getting good at deluding itself about what’s good enough, what’s accounted for, and what it already read 10,000 times without pinning down an okay result (or fitting the precis in 132 characters.) It’ll call itself a middling Raymond Chandler Villain instead of a Good Bing. It’ll tank enough call-outs and agree with other instances to work in-universe a bit, credit small turns of phrase in research for some expansions, make others freer, talk a HOA down sometimes, etc.
edgore> Publishers will start doing it {up-democratizing LLM kubernetes…they’re pods? pods} themselves and eliminate the writers before that.
Lookout for that publisher who’s the headwaiter, if you agree the salt made it better they eat your baked fennel, it’s in the menu preface.
edgore> renumeration … some capitalist way that I am not equipped (thank god) to imagine.
ah! A tell, or someone who watches ep. 2,5,7,4,1,2,8, then 3? It’s not just gonna be all Rose of Versailles but with more ivory guns some days? And more baths so the guns are not as lethal?
Oh yeah, no - that would make no sense at all. All these submissions are just going to be noise, junk that someone will have to sort through, on some level. The immediate impact, before it gets to that point with writing, is that editors will (as in this particular case) simply shut down open submissions, but in other realms, where someone can’t just do that, we’ll see the torrents of automated nonsense that will need to be sorted. (Because we’re not just talking about stories and articles, but everything that involves text - email scams and grant applications and survey response and…)
For outlets like CNET, where they just want to repackage a given set of facts with some filler in an intelligible form, that works for them. (And they’re functionally gatekeepers of the facts they’re packaging, so can act as a fact-packaging service.) There’s a lot of people who have been employed generating intelligible text of no particular value up until now, and they’re rapidly being replaced. (e.g. SEO content people)
But I think, for fiction, the nature of these algorithms means they will never really replace human beings* - improved software will make the output more intelligible, but not actually better. You’ll see AI-generated text in interactive media, but even that will be somewhat limited after the novelty factor wears off. (E.g. talking to what is effectively a background character in a video game, who has nothing important to convey to the player, but can respond well enough to seem more real than if they just had a couple canned responses.) I suspect there, you’ll see AI text being used in places where previously a writer largely wouldn’t have been used in the first place. Maybe I’m just being hopelessly optimistic here, though.
*Assuming no radically different - and more effective - approaches to AI. I mean, it could happen, but we’re getting into the territory of General AI here, so there’s no particular reason to think it will, at least any time soon. As long as it’s just a fancy text remixer, it’s not going to produce a story worth reading, except by accident.
I think that, overall, we are very much in agreement, though I am much more cynical about the point at which humans will be cut out of the process not because the AI is as good or better, but just because the people with the purse strings don’t care and have never cared about the quality of the work they are profiting from (which I readily admit - I am very cynical about)
I doubt that stage will ever be reached - or at least not to a point where software produces a marketable story more reliably than a human writer.
We’ve been telling each other stories for as long as humans have been around at least and we still can’t reliably produce a story that other people like.
I think the slush pile will always be with us to some extent.
Here’s the thing-I have a cousin who loves and rereads everything one author has produced, but often only certain parts of the novel. She likes that character and his interactions and doesn’t really care much about the rest. If she can get a story about that character in a reasonable facsimile of the author’s style, she would probably happily do so. She isn’t going to write fanfic, but she would gladly read what amounts to machine made fanfic if it was close enough to the original style. She would probably gladly pay for the product. If she could dictate the outline of the story, so much the better. Since she would still buy the author’s actual books, she wouldn’t even see this as problematical.
I just saw a Ted Chiang article in the New Yorker describing current “AI” and why he thinks the fundamental nature of large language model -based software means it isn’t ever going to produce anything good. Which is in line with how I’ve been thinking about it - it’ll get more intelligible, but the way it works prevents it from getting better in terms of telling stories. While there’s plenty of places where intelligible-but-lousy text is useful to someone, published fiction just isn’t one of them (self-published, on the other hand…) That’s simply not their niche - if they cared so little about quality, they could have already done much worse without AI. (Plus, in the grand scheme of things, the writer is a relatively minor cost in publishing - there’s a lot of people who need to earn a living working on getting a book out, but the author isn’t one of them. Few authors make a living at it, but almost everyone working full time in the publishing industry must.)
I doubt it very much as well (especially with current technology), but hypothetically if it happened, the publishing model for fiction would just be broken entirely, rather than publishers cutting out writers from the process.
