AI: Novel, Personalized, On Demand Media and its Potential Impact [+]

J.Quondam

CR 1/8
What do you think? Assuming this is a thing that can and will occur, what impacts do you think it will have on culture, fandom and individuals? And would you employ such a service?

I think it would (will) find a happy medium in the culture. Given the proliferation of forms of easy entertainment available nowadays, however, I suppose it could end up diluting entertainment media to some degree.

I don't think that, movies, TV series, novels, team sports, video gaming, etc, are going to disappear. That's just because part of what makes "culture" culture is the shared experience of such things. Some people will strongly gravitate to bespoke fiction, and others will be repelled by it, and everyone else will float in the middle somewhere. Eventually AI-gen would just settle into its "natural" place, wherever that is... probably be a similar place as fanfic or fantasy football or some other niche?

Also, as independent (open source, free, etc) generators, I doubt such AIs would ever get terribly big simply due to regulation and/or control by the companies that create them. So I can imagine established brands engineering custom AIs to give fans bespoke fiction within that universe. For example, Disney might (try to) sell a subscription to their AI that would churn out Star Wars graphic novels or animated shorts featuring characters (existing or generated) within the Star Wars universe, but who are constrained by the AIs guardrails to operate inoffensively and solely within official canon. On the other hand, I suspect that free/open AI projects would likely remain pretty small scale, rough around the edges, and serving a fragmented niche. Branded or not, though, solely AI-gen media just doesn't strike me as something that would change the entertainment media landscape too significantly, tbh.

As far as how it might impact the volume of new works by human creators, it's not clear to me that AI would change much of anything at all, except maybe in the tools used by those creators. I mean, a fundamental part of being human is that people like to create, so they would continue to do so. Of course, people also need to pay the bills, so imo examining that particular aspect of AI on culture is more difficult in a hypothetical world mostly free of economic concerns.


Would I use such a thing? I doubt it. Most of what I do consume is recommended by others, not what I'm "in the mood for" right now. HOWEVER, my exception: computer games. I would be definitely interested in AI-generation (eg, imagery, plotting, rulings, and maybe audio) in solo CRPGs, especially relatively lightweight turn-based games (eg roguelikes) that I could tune to my my own thematic and mechanical preferences. Some of this sort of thing has already long existed to a degree, but I think AI promises to make those types of games MOAR!! in ways that would likely appeal to me. But that's another pretty niche thing, too.
 

log in or register to remove this ad


Umbran

Mod Squad
Staff member
Supporter
What I want to talk about here is (again, speculative) a world in which anyone can summon by way of generative AI entertainment media that is a) novel, as in it did not exist before,

So, the limitation is simple, and you already noted it: It may be that it did not exist, but that doesn't make it really "novel". The way generative AI works, all its products are derivative, by definition. You feed in already extant works at one end, and they get put through a blender and come out the other. In general, you'll get works that follow the popular themes, structures, formats, and formulae that you fed into it as examples. No new ideas, no new styles or approaches, and so on.

Now, there's a lot of room in the world for formulaic works - there's big money in formulae in, say Romance novels. So it isn't that bad.

I think this can have a lot of legs in... videogames, actually. If you need something that's basically a first person shooter with a smattering of story to tie together all the shoot-em-up, this is fine. The player doesn't know the details of the maps and plots, so each game is like a new section of DLC for their game.

And... put a set of 3d displays on your face, and this is great for interactive fiction - your own personal holodeck adventure.

And, such a thing will put a bullet in the head of Hallmark Channel holiday rom-coms. :)
 

bloodtide

Legend
I wouldn't overestimate the value of truly creative ideas for the sort of casual entertainment I'm talking about. Derivative works are very popular, and not for no reason. There are thousands of police procedural television episodes, for example, and very few of them surprising or innovative.
There are whole groups of people that don't want much from content, and this has always been true. For books romance and mystery stand out as the two big types with a set simple content fan base. TV shows has soap operas, westerns and cop/fire/hospital shows. Movies have romantic comedies and action hero stories. And then adding cartoons, tabloids and reality content.

Plenty of people fit this fan base. They watch only the 'cop' shows. Read the romance books. And so on. This type of fan is not looking for anything other then bland content. The only "variety" they want, is within the tight limit of the focused content. The killer was the "daughter" not the "son" and such.

Can AI give that content......maybe. Even the worst such content written by a human will have some variety. Even just by pure randomness a human might think of any random thing at any random time and add it to some content. Sure the content will be 99% the same all all the other focused content. Though that 1% might stand out.

And people might notice. AI, when given data and told to write, can only write from that data. It can't randomly add new things all that well. The AI can't add what it does not know. Of course a human can't either, but a human has life experiences plus content experience.

I'd guess at least 50% of the simple content mill zombies will notice the loss of that 1%....
 

(I know it's a bit of nerco)

I think sturgeon law is gonna apply heavily as the models go on to be better than pervious ones. As with the painting/video generations you can take the time to get the image in your head by doing tweaks or even getting an new idea by having the model show you what your writing would if you go from 1st person to say 3rd person.
 

945f9399f6a193926df4914b6f526397.jpg
 

Stalker0

Legend
Similar to how video tools and web site tools have allowed normal people to become video editors and website owners, I imagine these tools will allow a lot more "amateur authors".

AI will take care of the writing and prose. It will make your writing look good, even if your not a great writer. So it will come down tot he story itself, your imaginative work on the page.

A good story will shift away from an author's ability to wordsmith creative language and go more into the ideas and characters of the story itself. The stories that contain interesting, nuanced characters, bold creative ideas rarely explored in writing.... those will be the stories that would likely rise to the top in this new world.

In short, the people that are good but not particularly creative writers will fall by the wayside. Whereas the people that have some amazing ideas but never could put on the page may suddenly find the tools needed to do just that.
 

Li Shenron

Legend
It would be detrimental to human society. We'd never have anything new again.
I wonder if there is a niche area of entertainment where it would actually be beneficial to human society... an area where:

  • nobody cares about shared experience
  • people often want specific things
  • plagiarism is of no concern
  • nothing is truly original and originality doesn't really matter

An area of entertainment where real people are actually exploited and possibly generates revenue to criminal parties. Can you guess? Hint: it's a 4 letter word and starts with P. If AI could be asked to generate accurate on-demand media for this type of entertainment, perhaps we could free a lot of people from exploitation.
 

What do you think? Assuming this is a thing that can and will occur, what impacts do you think it will have on culture, fandom and individuals? And would you employ such a service?
I think this will occur, and assuming for the purposes of the + thread that it's neither dreadful quality nor reliant on half-inched material, I think it will probably see a faddish period of success, before the pendulum swings back rather strongly to human-authored material (which might be increasingly AI assisted, but in a far more human-driven and complex-to-create way than prompts etc.).

Two things which I think will contribute to it being a fad and not retaining long-term popularity:

1) It can't be copyrighted. I don't expect that to change without major tech changes and fundamental changes to how training data is gathered. This means there's little incentive for a company to promote any given work.

2) It's not repeatable. Because of the nature of generative AI, the same exact prompts and selections will not produce the same or necessarily even similar work. Even if you can record it and transmit it to others (and I expect the gen AI companies to attempt to restrict/limit this, for various reasons), that it's so personalized is likely to limit the appeal. If it's reactive it won't even really be recordable in a meaningful way.

3) The ultra-derivative, trope-laden stories it will produce will be initially extremely funny/amusing, but because they'll have nothing to say to anyone, operating solely as distorted mirrors to our culture (which is interesting but not compelling), they won't have staying power.

I don't think it'll go away entirely - because of point 3, I think it'll kind of stick around as a way of mocking contemporary media for whatever period, of illustrating the most excessively dominant tropes and so on.

Also, being real, the main success this tech could have is porn. People don't generally want to share that, and people have very specific likes/dislikes/etc. None of the three issues really applies to porn. Further, people will positively argue that it's more ethical to make porn this way than using real people (philosophically questionable, but an easy claim), which will help it er... penetrate the market. Whoever develops fully-personalized, especially interactive porn, no matter how derivative or trope-y (indeed those could be assets) that works very reliably, is going to make a lot of money. Note: I very much doubt this will us gen-AI for the images, they're error-prone and hallucinatory - more likely it'll use gen-AI for the behaviours, scenarios, interactions and so on, and some kind of proc-gen (rather than gen-AI) 3D graphics for the imagery.
 

In short, the people that are good but not particularly creative writers will fall by the wayside. Whereas the people that have some amazing ideas but never could put on the page may suddenly find the tools needed to do just that.
The most successful writers are absolutely not the most creative writers, nor about "rarely explored" concepts, so this is an extremely strange idea.

Just look at the fiction best-sellers lists.

These aren't ultra-creative books, nor the best-written books in the sense of prose - they are, by and large, sharply written books about subjects that have been covered by many other authors, that are easy to read, compelling, tell a tight story and so on. So I don't really see why that fundamental would suddenly change.

And gen-AI simply can't tell a tight story because it can't understand it - that's not going to change until AGI.
 

Remove ads

Top