My attitude is still, "show me what it can do."

So far as I can tell, it's still extremely good at...tracing things. Give it existing content and it can come up with some remarkable replicates.

But ask it to create something from scratch and it struggles.

So far as I know, the machines still struggle with "compositionality." Their detail level is incredible, because they've been trained extensively on detail, but they still don't understand what they're creating.

Which is why the physics of pie hits in AI videos, for example, still look so unbelievable even when the before and after are quite good.

Just because something has gotten better doesn't mean other problems it has will magically be solved. We've made some remarkable jumps, but many of the fundamental problems remain. It still struggles with hands, for example. You can get an extremely detailed hand with wrinkles and hair and whatnot, and then you realize there are six fingers, a mistake a human would never make, nor should a computer if it actually understood what it was making.

So, leaving aside the energy demands, environmental concerns, and the fact that most of the AI companies haven't turned a profit and have no idea how to short of a government bailout (Sam Altman floating that in the news lately fwiw), I'm still not convinced it will get better in ways that matter to me.

At this point, AI is already cranking out very short videos where you can't tell the difference. The issue at hand currently is the cost - the price of tokens to create endless prompts to make longer videos and even more real. As that cost comes down, over time, you will not be able to tell the difference between an AI generated video and real one. And the warnings from AI industry experts are out there and very alarming.

Well, so then I guess you do understand at least some of the people who "make the argument that it's absolutely terrible but also at the same time a threat to people who make real videos"


What I'm trying to do is to view AI as realistically as possible. And based on the evidence that I see, it seems realistic to conclude that AI material (of any sort--images, video, audio, whatever) has the potential to drown out natural material. I'm not saying anything about whether it should or should not do that, or whether it would be good or bad if it did that, or whatever. I don't think I have to make any such claims about my own preferences in this context, because the poll does that for me: the overwhelming majority of respondents seem to prefer natural material. So, to me, it's a good thing that the site admins have constructed an environment that (in this case, at least) supports the desires of the community. And part of the way they've done that is, yes, to build a kind of shield. I endorse that, and I would oppose any decision on the part of the admins that would undermine or defy a community preference that seems totally valid to me.


Yeah, I mean, I wouldn't be surprised if that's what some of them meant. But some of them may mean something more sophisticated, such as what you and I are now talking about. It seems kind of silly (or, at the very least, lacking in evidence) to attribute a single interpretation to a variety of people, imo ...especially since at least one person (besides me) had already provided another point of view.


I mean, insofar as "space" is an advertising opportunity, isn't it by definition a threat to real producers when something else takes up space? Like, we can quibble about the magnitude or seriousness of the threat, but this is basically the premise behind the entire SEO industry: whoever gets the best space wins.


Okay. But, like... this explains at most half of the correlation that you described above, right? If people are enjoying natural material less, that would give us a reason to expect them to buy less, which in turn would give producers a reason to produce less, and so on. But how would any of that give us any reason to expect an increase in, as you put it, "spammy videos taking up space on the forums"? Do you think that it's just pure coincidence that the AI stuff is taking up the space?


Both of those things can be true at once, no? People can spend money on something bad, causing it to become popular while also helping to drive good things away. I guess we would then have a question about why people are choosing to spend their money on something bad, but I don't think that question is unanswerable. Lots and lots of bad stuff is popular.

You definitely can, and if you're insinuating that creators not tell their audience that what they are viewing is AI, then we can go home right now because you already know the answer there.

I think ultimately, THAT is the concern. People producing AI at a level that can be incriminating or slanderous and we already seen what harm we can do with just simple headlines. Imagine now the ability to put pictures to words.

This seems like question begging. It reads like you are presupposing that AI will cause a reduction in what you call natural material (and what I call 'real videos'), and you are then going looking for possible logical reasons as to how this could happen.

Whereas I think there's more natural material than ever but that the overall quality or marketability of it is worse.


UMD users are lot more diverse and individual in their preferences than I think you assume.

MM wouldn't want to see producers suffer, but I'm fairly certain he wouldn't want to restrict the essentially harmless posts from several of the users here due to a few users' hypothetical concerns. At no point is MM going to prioritise AI over 'natural material', in the same way that MM doesn't prioritise mud videos over pie videos.

I don't think producers need defending in particular from AI wam videos. Apart from a couple of exceptions, almost all wam producers are doing this as a side business, or as a hobby. Almost no-one is making serious money from this or doing it as their main source of income. 'Wam models' are just fetish models who do other modelling, and wam isn't necessarily their main source of income either (again, apart from a couple of exceptions). People are fretting about livelihoods being ruined as if most people don't make their main money elsewhere. Producers will just make the videos they want to make and release them for sale. People will either buy them or they won't. Producers can't (or don't want to) produce to every niche, and that's why a lot of the prolific AI users move to AI to fill their niche. What's sad for me (and many other UK members), is that the straightforward but well-made 'put a good looking woman on a white chair and gunge her for 10 minutes' type of video has died off over the last decade. Standards have fallen for us. That was my point about not buying videos much lately, and it has nothing to do with AI and everything to do with a lack of quality material that fits our preference.

But anyway. Lots of the discourse on this seems overblown to me. Doom-and-gloom. AI clips as an existential threat. It's all just wank material, both fake and natural. It's not fine art. It shouldn't matter that much.


Exactly, and we shouldn't need to differentiate between different types of popular bad stuff on a weird fetish website.

Anything people can jerk off to for free harms producers. How producers still make sales is guys who want to see specific types of videos that are not available for free. Once AI get advanced enough to generate WAM videos that are jerkable, it'll cause great harm to producers and this forum won't matter since AI WAM will be all over the net; youtube, reddit, twitter...

None of this is free. You either pay for the hardware yourself (like, about $8k for a usable GPU for AI video), or you rent the server space, or you take the consumer option and pay like 75c per video generation or whatever it is.

Almost everyone with a WAM fetish started out jerking off to TV show clips, for free.

Some of us can already produce AI videos that are jerkable. I've already made some. I've even sold some. UMD is still here. As long as there is a knowledge barrier to making good quality stuff, it will remain uncommon and therefore of value - and the same applies to real videos as it does to AI.

Did I say "will"? Or did I say "has the potential to"? And if I said the latter, could you please explain to me how "has the potential to" reads as "will"? Last time I checked, words read as themselves, not as other, completely different words.


Yeah, I mean, I understand your point of view. But are you disagreeing with me or trying to change the subject before saying whether you agree or not? Because, as far as I can tell, "natural material got worse" is not actually a counter-argument to "AI has the potential to drown out natural material." Unless there's something very big that I'm missing, both of those things can be true at once.


By "assume," what do you mean? Because what I did was more like "conclude on the basis of evidence." As I said before, I'm going off of the poll results, which are empirical evidence and not, like, some kind of sinister bias of mine. Again, I really need you to read carefully and respond to what I'm actually saying instead of doing whatever this is.


Can you define the word "hypothetical" for me, please?


Yes, that's coming through very clearly.


Have you spoken with a lot of producers about this, out of curiosity?


Then can I ask you to please stay on topic? Again: I understand your personal preferences. I hear you when you say that you can't find the stuff you like anymore. But this literally a thread about "AI attitudes." So to say that your opinion "has nothing to do with AI" sure makes it seem like your opinion is irrelevant to this thread.


But then why are you yourself acting like Mr. Doom-And-Gloom about how our "standards" have fallen and all that stuff? I'm very confused here, because it seems like you're saying that other people shouldn't care about precisely the same things that you yourself care about. Why are their preferences or values unimportant? Why don't they matter just as much as you do?


Huh? Why not? Where is this idea coming from?

Well you haven't been around long enough to see technology put producers out of business. In 1999 I was buying VHS upskirt tapes from TNV video. The owner of TNV and Rob Blaine were best friends, hence I used TNV models for my Pie Wedgies custom with Rob. But I digress. Then came the ability to host videos online, then can "free-tubes" which flooded the internet with free upskirt videos, literally putting TNV out of business. And yes, free WAM AI fill flood the net.

Apologies, allow me to correct myself. It reads like you are presupposing that AI has the potential to... etc


Dude, this isn't a legal cross-examination and I'm not trying to "catch you out" or imply that you are sinister. The poll is evidence, but it's evidence of a self-selected collation of opinions, it's not empirical or experimental evidence. What I was trying to get at is that there are all sorts of people that use UMD and that decisions don't need to be made by referendum. We are not a singular block, we are a loose collection of individuals.

Where we disagree, can we make it about the meat of the disagreement rather than try to dissect each other's exact words? If you take issue with the way I've worded something, please ask for clarification. My intentions are not malicious.


Oh come on, you know what it means. It's the adjective form of hypothesis. You could say it's something that has the potential to happen.


I've certainly read enough posts from various producers over the years about how no-one makes any money and everyone does this out of love. And I also know that a 'top selling' video on UMD can be fewer sales than I had previously assumed. Show me who's paying their mortgage and feeding their family solely from wam videos (apart from Brianna Olivia who maybe is) and I'll gladly admit I was wrong.


It is on topic: 1. It's a potential reason why people turn using AI. 2. It's a reason for reduced sales (which is one of the possible consequences of AI that are being discussed here). It explains the context of my viewpoint.


I'm not saying anyone's preferences or values are unimportant. But I am implying that no-ones preferences or values should be more important than anyone else's. And I wasn't talking about 'our standards', I didn't mean moral standards or ethical standards, I mean the literal physical production quality of the gunge videos for sale now compared to the past. You seem to be saying that natural material is under threat, and I'm kind of saying that a lot of the damage has already been done.


See above. There is bad 'natural material' just as there is bad AI, and if it's popular, it's popular. Who are either of us to say that one has more value than the other?

I'm all up for an enjoyable discussion where we disagree on points, but I'm not particularly interested in another round of replies like this one. If you want to continue this, can we make it single block of text rather than fragmenting the quotes?

I bought my first wam video in 2006 in my twenties, and while I never bought a wam VHS, I'm old enough that I could have bought one.

Of course piracy is an issue. I'm not an expert in upskirt videos but I'm guessing the subject matter might have made it awkward to make a copyright claim?

AI WAM is already all over YouTube, people are making it for attention and kudos but they have paid money to make it. Maybe some of them are hoping to monetise their channel.

I think that good stuff will always sell, as long as it's good and it's not possible for literally anyone to make it, and I think the same principle applies equally to real video and to AI. I think good AI WAM will probably always take a certain level of knowledge to make and will remain rare, because the low-effort consumer AI is restricted from making fetish stuff, and it isn't trained on wam. It won't magically get better at wam without being trained on it, and it never will get trained on it because it's not intended for fetish use.

Crappy WAM AI is all over youtube. When it gets to the point where you literally can't tell the difference, that'll be when it starts making a noticeable dent.

Once more: I did not presuppose. I cited sources. Did you not read that part? Did you read it and then forget? Are you going to tell me that those sources are also invalid for some reason? What's going on here?


So then, not an assumption, yes? Just like how my claim about the potential of AI material to swamp natural material is not a presupposition but rather an evidence-based conclusion, howsoever reliable or accurate? So can we please both agree at least that these claims of mine are backed by some kind of evidence, even if we then go on to disagree about the quality, relevance, or accuracy of the evidence that I've used?


Okay, so, first of all, if you want to say "there are all sorts of people and decisions don't need to be made by referendum," please just say that. If all you say is, "you're making an assumption" or "this sounds like a presupposition," I have no earthly way of knowing that you mean to talk about diversity of opinion or decision-making procedure (i.e., what you see as the meat of the disagreement). I really would like to try to address your real concerns, but the more that you express your concerns as accusations of malfeasance or irrationality on my part, the harder it is for me to know what you're trying to say. Fair?

Second, yeah, I fully agree. People here have different tastes. This poll needn't be, like, the end of the process of making a decision. Although I'm not sure what decision you think we're making, so I'm not yet sure why you feel that this is meat, but I guess you can explain that in time.


Ah, so do you mean to say that people's real concerns are addressing a hypothetical event? And not that the concerns themselves are merely hypothetical (as in, unvoiced, imagined, etc.)? Because if so, I guess I don't understand why we would want to wait until there's a real, non-hypothetical problem before we do something about it. Surely you see some value to preventative measures in other areas of life. So surely you must be able to see at least an abstract case for taking preventative measures in this instance.


So, correct me if I'm wrong, but there's nothing in here that says that you've directly spoken to a lot of producers about their own perceptions of AI or their own ideas about how they would respond if AI impinged more upon their work. I get your point about the finances here being low-stakes in many cases, but in the absence of direct testimony from producers I'm a little leery of trying to say what they would or would not do. Typically I find that it goes badly if I try to read the minds of strangers over the internet, so I try to avoid doing that.


Oooooooh wait. Wait wait wait. If I understand you correctly, what you're telling me is:

-Over time, you've found that it's harder and harder to find material you like.
-Because it's harder and harder to find material you like, you yourself have spent less and less money on naturally produced material.
-At the same time, you (and others like you) may decide (or, in your case, have already decided? I think?) to use AI to generate the material you like.

Is that accurate? Did I get that right? Because if so, it sounds like you're explaining precisely one of the dynamics that the people in this thread are concerned about. Specifically, if I'm reading the series of events correctly, you chose to use AI. It wasn't forced upon you. No one held a gun to your head. You made a choice to use a low-cost, readily available option instead of saving up for custom work or joining with other people here to advocate for the kind of work you like or producing the work you like yourself or, indeed, any other option.

And, look, do I understand why you did that? Yes, sure. Money is scarce. Time and effort are scarce. Why not just do the cheap and easy thing? I myself am often cheap and lazy. But do you now understand that the choice you're describing is a choice that will take money, attention, and support from producers of natural material and give that support to AI and AI producers? In other words, do you understand that you are *specifically* describing the exact type of dynamic that many others in this thread are worried about--and which you yourself have repeatedly tried to dismiss as overblown or not real?


Okay--is this not a reason to have precisely the type of shield that you and I were discussing earlier? So that one group's preference for AI doesn't overrun another group's preference for non-AI in exactly the way I described? Or am I missing something.


Oh, but hold on. The fact that you're describing this as a matter of "standards" at all rather than a matter of, say, "tastes" or "preferences" is important. I know that newer work doesn't meet *your* standards, but *your* standards are not the universal measure of "quality." That's why I see this as a doom-and-gloom comment and not just, like, normal everyday griping. It's one thing to say "I don't like this." It's another thing entirely to say "This is worse."


Except, again, unless I've misunderstood something you yourself have also laid out step-by-step how AI is (capable of) doing new and unique kinds of damage. So if you already believe that AI is doing (or, at least, is capable of doing) new and unique kinds of damage, why are you putting so much effort into arguing with other people who are saying exactly the same thing?


Uhhhh aren't you the one who talked about declining standards and all that? When you say that "the literal physical production quality of the gunge videos for sale now" is worse, do you not mean "to say that one [thing] has more value than the other"?

Also, for the second or third time, *I'm* not saying that anything is better or worse. What I'm saying is that we have evidence of a strong community preference and that others in this thread have described AI as bad or inferior. I know you've tried to dismiss or cast doubt on their preferences, but I really have to insist that you recognize the difference between me citing a source and me deciding by fiat that something is the way I say it is.


Well, I mean, *you* can stop doing that whenever you want. If that's not how you like to communicate, then I'm really not sure why you started doing it in the first place. Just because I do it that way doesn't mean you have to.

As for myself, I don't yet see how changing my text formatting is going to help with anything. Like, seriously and not jokingly, unless that's going to help you to stop rushing to mischaracterize my beliefs as a weird sideways way of expressing what you're actually trying to say, then no I'm going to keep doing this because this is what works for me.

Now ask yourself: Is that good for AI or bad for AI?

Here is the bigger picture: If you login to YouTube now, and search for any old thing, about 50% (maybe more) of the "original" content is either recycled crap or cruddy AI. It's to the point where I don't even browse on there anymore and I know I am not alone.

Less people using YouTube = Less monetizing through Ad revenue. Sure, Google has a stake in AI but it's kinda like an investment property. If they prove more losses over gains, sooner or later they will drop or limit support for it. Look at the many big names who were behind 3D HD and look how fast that fad dropped off long before the advent of VR.

Now does this mean AI is a fad? No but it could be just a transitional technology into something better or commercially viable. There are tons of examples of those those. HD DVD, Beta and Laser disc are fine examples that comes to mind.

No need to be so defensive Larry. Anyway, I'm out.

I think there's a bit more to it than that, although I might be misunderstanding your point.

Have you seen this year's (AI generated) Coca-Cola Christmas advert? It's only 1 minute long, but it's effectively a montage of short clips. There are a few video responses on YouTube which point out the problems with it, e.g. the trucks don't stay consistent from scene to scene (different number of wheels) and each type of animal only appears for a few seconds. In other words, the AI generator doesn't have any kind of memory to handle internal consistency, and I don't think buying extra tokens will solve that problem.

Okay

Personally, I'm perfectly willing to carry on this conversation and make whatever decisions we end up making without your continued input. But thank you for helping us to better understand the concerns and limitations of an AI proponent.

So if AI gets banned from the site it'll be all my fault, huh? Aw, shucks!

I didn't realise you were part of the steering committee for the site but ok

Im shocked that im on the short side of this one. Then again ive found interest in cartoons and am a big fiction fan so I guess that checks out why I like AI. I've been watching it pretty regularly for the last 4 months and the improvement in that short of time is astounding. For a group that has so many members with such specific tastes I think AI will surprise everyone.

By the point AI makes anything that satisfies a large enough consumer base for this specific very niche fetish (unlikely for the reasons various people mentioned here) environmental destruction, job loss, the creation of other dogshit AI media that is force-fed to become the new norm or the creation of some Nick Land style "nothing human makes it out of the near future" will be much more pressing concerns whether than how "lifelike" an AI wam video looks. Also a lot of presumptions that this tech will just increase and increase in capacity until no one can tell anything apart (if that happens we're all fucked anyway so this'll be the least of anyone's concerns)

Or the trend will buckle govs will force limitations, the AI bubble bursts, backlash to potential loss will spark hostility, Butlerian Jihad might even take place! In that case we can get on with things as usual and empower y'know human creativity which is actually fulfilling and has connections with actual people. No robot will have that emotionality and like i said if it does we're fucked anyway.

If it makes us Luddite so be it. Them folks were right to be angry their livelihoods were destroyed by tech increases. If ours gets fucked by all this not just in terms of our eroticism and human connection but also the wider scale losses of our industries and jobs then I'd gladly take the sledgehammer to them data centres like a Red Guard in Maoist China.

Okay I was staying out of this conversation but seriously this entire thread is a sausage fest of men talking about women's images, not even their own images. Which makes the entire conversation completely tone deaf. Only one other woman has commented on this thread and this conversation needs more women's input and perspective. So here it is.

AI is definitely the new technology that we need to learn to accept as a reality. It's being implemented and it's already replacing workers in the workplace and negatively impacting the job market. People's livelihoods are being affected by this. So there's a human element here that needs to be respected and I think this is what makes AI such a touchy subject.

My biggest issue with AI is the same issue with piracy. It's very easy for people who watch porn to separate the image of a woman from the living human person she actually is. So because piracy is an issue, so is a tool like AI that makes pirating much easier. As far as celebrities, their entire career is actually based on their image and the use of it. Women who would never make porn getting cut and pasted into porn is an age old practice that all women fight against, it violates the rule of consent.

Okay, here's where I get really serious. You're raping a woman of her control over her own body, and the narrative of her own sexual identity and experiences when you steal her image. This is a form of rape, it needs to be called out for what it is. Forcing anyone to participate on any level in your sexual desires against their consent is a form of rape. It's not just an image of a person, it's a real person's image. And it's negatively impacting the person psychologically, which affects a person physically.

It's not okay to use a celebrity's image without consent for public display on a fetish or adult site. It's not okay to use any person's image in this way who doesn't consent to it. And because social media has created an environment specifically for people to display their image, it's really important to respect and protect every person's image as a living entity. It's not just an image, it's a real person.

There is obvious benefit to certain groups of individuals using it, but as a whole, I don't see the benefit of selling AI content. Everyone has access to AI generators. Anyone can create AI. Anyone can create from it and create what they want, they don't need to pay others for it. So I don't see it as a need that only specialized people can provide.

There's nothing wrong with creating a fantasy if AI is used respectfully. But that's the biggest glaring problem with AI, the technology itself isn't as dangerous as the people using it. And that's where I oppose the use of it here on this website. There is an epidemic in this community of piracy, and it's those same people who are looking at AI to make stealing easier for them.

When I read how anything doesn't think AI is really going to disrupt producers this is what I think:

Blockbuster CEO: Streaming movies on TV? Well that'll never affect us
Blackberry CEO: Internet on your phone? That'll never affect us
Kodak: Digital cameras? We're all good.

Technology, since forever, has obliterated entire categories of jobs. And that the current state of AI is where it will end up is beyond silly. It's progressing at a geometric rate. It's already replacing so many jobs that they are bills introduced in Congress to address it.

AI industry insiders have already rang alarm bells that in the not too distant future no one will be able to tell the difference between AI and reality regarding pics and movies.

This was the issue I had with the Photoslop situation and this is why I felt it wasn't a small situation. We objectify women if we feel we can just take their image and manipulate it and use it anyway we feel.

May I offer a slightly different perspective, about words, rather than pictures?
I've enjoyed posting stories here for years, and been delighted by a few others who enjoy reading them and post comments.
AI has opened up a whole new world on this.
I've been fascinated to find AI sites that will discuss my writing, suggest improvements and analyse the characters and situation. All for free! (So far).
I've certainly found this a lot of fun.
Words are Better Than Pictures.
Peter

Stealing someone's image or publicly mimicking someone's likeness for non-consensual gratification is never acceptable.

I don't think some people here realise how serious an offence it is.

Yeah, Nostalgic Erotica and CKCP brought up this same point, and I definitely think it's worth highlighting. There is definitely a population of AI users who are so fixated on their own desires that they feel that they're entitled to have whatever they want regardless of little things like ethics or respect for the people around them. It's not *all* AI users, but it's a large enough (and noticeable enough) population that it's definitely worth talking about.


I agree with this, too. There are still a lot of unresolved questions about whether the current AI technology is viable (financially, environmentally, legislatively...), but it sure does feel glib or maybe even purposefully ignorant for people to insist that it could never even possibly cause problems. As you say, we've seen this many many many MANY times before, so I feel like someone would have to have extremely powerful evidence in order to argue that it couldn't possibly happen with AI.

All.
Of.
This.

I agree with most everything said. Just realize we're getting into unchartered waters. Up to now we've mainly had to deal with posting one girl's face on another's body; cue in deep fake porn. But to actually produce WAM videos you need to hire models.

Coming shortly, anyone with a modest amount of money (a mere fraction of what it would cost to hire models) will be able to type in their dream script and prompt "turn this into a video" - then sell it to anyone willing to buy.

That is the funniest AI-generated photo I can remember seeing. Thank your very human intelligence for having the idea for it, comrade.

There are 2 areas that I think will have different levels of impact in the near term.

The quality of Deepfaking is getting better and will continue to do so. For imaging, you would previously need to train a local model lora to produce something convincing, it takes time, effort and knowledge to this well but once trained can be applied to any number of images. The quality of this done right is astounding with new models like Qwen, HiDream and Wan2.2 and I've no reason to think this won't continue to get better.
What has happened in the last 6 months is image-edit models have become available that makes it almost trivial for anyone to upload a starting image and ask the model to add mess, change poses, undress etc. Nano banana was the first but qwen-edit is arguably far more powerful from a deepfaking point of view as it is also trainable on any concept or person. By the end of the month, I expect nano banana pro to be released running on gemini 3 pro and this will be be far more powerful than the current model and will support much higher resolution/detail.

This for me is the immediate and most concerning aspect of AI and where we as community need to establish strong guidelines of what is acceptable.

From a general content production point of view and the impact on producers and/or whether one person or another enjoys it.. well I think we've a way to go at least for quality long form video.
I've spent a lot of time training image models over the last year or 2. The quality of trained open source models surpasses commercial models but even if the quality of a single image is indistinguishable from "real material" it is a oneshot affair. There is no temporal progression and consistency, I can't create an image set that is consistently and progressively messy. Edit models may help that as they improve but i have my doubts this is the correct solution.

Video, I disagree that any wam video produced by AI is "good enough" yet. I've mentioned the constraints and limitations of commercial models already elsewhere but they've just not been trained on enough messy content to be good. Grok and VEO3 are quite woeful even outside the guardrail constraints. Sora2 pro is probably the best model for wam content generation but even that struggled with artifacts and inconsistencies even before the guardrails were raised.

I'm not naive enough to say video won't get better but the cost and resources needed are still massive constraints and these models are only producing 8-15sec clips.
Opensource models have the same strengths as open image models in that they are trainable but the resources and skills to train a good model, let alone generate a short 5-8 sec clip without audio is beyond most people. Wan2.5 is so far beyond consumer resources it's laughable and good luck training that on cloud without massive bills.
The nature of technology is that hardware becomes cheaper over time but the models also need to become bigger and better which negates any hardware benefits. There may be a paradigm shift in transformer AI which flattens this out but I would be genuinely surprised to see anything like 1-2min clips in a long time and even then so constrained by guardrails or quality they should have little to no impact on producer Wam content.