=][= INTERIM GUIDANCE: AI/LLM GENERATED CONTENT =][=

[Joe]

There is no nuance when it comes to creative spaces - it simply isn't acceptable, and this was the wrong call to make even as an interim decision.

 

Anything else is chomping the AI-bros self-deprecating reasonings for why we should roll over and accept it.

[Cactus]

4 hours ago, Joe said:

There is no nuance when it comes to creative spaces - it simply isn't acceptable, and this was the wrong call to make even as an interim decision.

 

Anything else is chomping the AI-bros self-deprecating reasonings for why we should roll over and accept it.

Is this because a) anything that has had AI used in it's creation, even where that is declared, is not a 'proper' creative work of human expression or b) the AI tools available to us are built (in part) on large-scale unauthorised use of copyrighted works in their training data and we shouldn't reward wrongdoing?

 

I agree with the latter, but not the former.

 

I'm not trying to start a fight or put words in your mouth. I sense that both arguments have been made by various people in this thread but its not always clear why people are objecting.

[HeadlessCross]

30 minutes ago, Cactus said:

Is this because a) anything that has had AI used in it's creation, even where that is declared, is not a 'proper' creative work of human expression

It's not. It takes whatever already exists, without permission, mashes stuff together, and then some techbro is gonna pretend they're a writer or artist. 

 

What's there to admire there?

[Cactus]

3 minutes ago, HeadlessCross said:

It's not. It takes whatever already exists, without permission, mashes stuff together, and then some techbro is gonna pretend they're a writer or artist. 

 

What's there to admire there?

Nothing, but that's a straw man. The scenario I pose is one where the work is not 100% unfiltered gen-AI output, but one where some AI output was part of a human-driven creative process. Examples such as suggesting (not choosing) colour palettes, or creating an illustration based on the hobbyist's own model have been given already.

 

Before I'm accused of being a tech bro shill, I don't like generative AI or large-language models for reasons including creator rights, snake oil, jumped-up predictive text, the potential long-term effects on people who over-rely on AI, technical illiteracy preventing people in positions of power (small scale as well as national) from making informed decisions about its use, and the water-guzzling environmental impact of their data centres. However I also think that somebody who says in good faith "I made this in my hobby, using <AI tool> to help with X in the process" deserves politeness, not a fundamentalist anti-AI lobby.

 

My view is have a sense of proportion and allow some fuzzy grey areas around the edges. This thread should convince anyone that our community will definitely call out anybody passing off AI slop as their own work. I don't worry that we'll be overrun by it at all.

[Joe]

59 minutes ago, Cactus said:

Is this because a) anything that has had AI used in it's creation, even where that is declared, is not a 'proper' creative work of human expression or b) the AI tools available to us are built (in part) on large-scale unauthorised use of copyrighted works in their training data and we shouldn't reward wrongdoing?

 

I agree with the latter, but not the former.

 

I'm not trying to start a fight or put words in your mouth. I sense that both arguments have been made by various people in this thread but its not always clear why people are objecting.

 

Little of column a, little of column b - I'd also say it partly comes back round to what I said yesterday regarding internal theft on the forums (i.e. users stealing other users hobby material for training purposes, then re-uploading it). It also partly comes down to the fact that like, give an inch and they'll take a mile. We saw the exact same issue with the uptick in the tournament / competitive scene when it came to the game where it was begrudgingly tolerated by the community before it rapidly ran away with itself and left us in the position we are today.

 

I also frankly don't trust users - this includes our current moderation team - to police this adequately if AI / LLM content is allowed. 

 

That said, I am not surprised at some of the arguments being fielded by both sides in the thread, although it's a mite depressing that some people are as predictable as they are lol. Better that we hash this out in conversation now than allow it to fester however.

[LittlePlasticHomies]

19 hours ago, Mazer Rackham said:

Once you leave the room, I might smirk though, especially if you use the words 'glistening' and or 'cleavage' twice in the same paragraph.

That’s exactly how I described Commander Dante.

[drakheart]

AI has many uses, when it comes to out right AI creations such as text to image generation, but AI can also be used in more subtle ways to help improve and refine an artists work. I work as an architectural technician and we use AI packages to help improve our visuals for building projects. The design and rendering of the building itself are all our creation, but we use AI to improve the look of people and vegetation in our images and animations as the 3D models we use can often look fake, especially close up. The use of AI enhances the end product but doesn't alter the architects actual design that the visual is used to present. 

 

In the same way a forum member on here could use AI to generate backdrops or environments for photos of models they have built/converted and painted to help present their work, as long it is made clear what part the AI took in the final piece. 

[HeadlessCross]

22 minutes ago, Cactus said:

Nothing, but that's a straw man. The scenario I pose is one where the work is not 100% unfiltered gen-AI output, but one where some AI output was part of a human-driven creative process. Examples such as suggesting (not choosing) colour palettes, or creating an illustration based on the hobbyist's own model have been given already.

 

Before I'm accused of being a tech bro shill, I don't like generative AI or large-language models for reasons including creator rights, snake oil, jumped-up predictive text, the potential long-term effects on people who over-rely on AI, technical illiteracy preventing people in positions of power (small scale as well as national) from making informed decisions about its use, and the water-guzzling environmental impact of their data centres. However I also think that somebody who says in good faith "I made this in my hobby, using <AI tool> to help with X in the process" deserves politeness, not a fundamentalist anti-AI lobby.

 

My view is have a sense of proportion and allow some fuzzy grey areas around the edges. This thread should convince anyone that our community will definitely call out anybody passing off AI slop as their own work. I don't worry that we'll be overrun by it at all.

You plugging in the info and letting AI do the work is not creative output. If you need to figure out color pallets, you can ask people in this forum with a good eye for color or talk to one of your artist friends. Use that as inspiration. You did zero work with the AI and it's to not be respected. 

 

And yes, we should be actively chasing out anyone trying to AI "creatively", and we should always be active in doing so. If you're not actively fighting it, you'll become part of the problem. 

[Doghouse]

I'm more old school, I was getting images of my models stolen and reposted back when the only AI we had was Skynet.

[lhg033]

Personal opinions on AI aside (and I do have those and get why people feel strongly about this) - I think this guidance on AI is a good starting point and I trust the Mods that if they feel this needs further review, with or without community input, later they will do so.

 

All I will say is thank you mods for the work that you do and leave it there.

 

[phandaal]

24 minutes ago, Joe said:

That said, I am not surprised at some of the arguments being fielded by both sides in the thread, although it's a mite depressing that some people are as predictable as they are lol.

 

Not expecting a response on this, given that your only interaction with me is putting the occasional dislike on my posts, but I would suggest that this is not depressing at all. People being consistent is just people being consistent.

 

We all have folks that make us roll our eyes (I am 100% on that list for some people ), but surely we are not expecting that people develop all new personalities just because we dislike them?

 

Anyway, back to the topic at hand - I don't expect to see a lack of policing by users. Based on the reactions here, I expect a bonanza of reports to be fired off if a submission gives even a hint of having used AI.

[Joe]

51 minutes ago, phandaal said:

 

Not expecting a response on this, given that your only interaction with me is putting the occasional dislike on my posts, but I would suggest that this is not depressing at all. People being consistent is just people being consistent.

 

We all have folks that make us roll our eyes (I am 100% on that list for some people ), but surely we are not expecting that people develop all new personalities just because we dislike them?

 

Anyway, back to the topic at hand - I don't expect to see a lack of policing by users. Based on the reactions here, I expect a bonanza of reports to be fired off if a submission gives even a hint of having used AI.

 

To be fair, I think this is more down to us often being on the opposite side of various conversations - which is in of itself fine, that's the beauty of communication. I should stress that I don't have a personal grievance with you (and I apologise if it's ever come across that way), I just don't necessarily agree with you a great deal of the time. But aye, absolutely agree on the eye-rolling and personalities side. 

 

The policing side does concern me more than a bit, really. We've all seen some of the conversations that've taken placed in the news forum in the last few months or so where it's gone on for pages and pages of increasingly argumentative, hostile or sometimes downright petty behaviour before anyone has asked a moderator to step in; assuming they aren't part of the problem themselves. There's something to be said about needing to shape up ourselves, really - and ultimately, it's better to step away from a conversation that's aggrieving us rather than carry it on out of spite.

[Mazer Rackham]

3 hours ago, LittlePlasticHomies said:

That’s exactly how I described Commander Dante.

 

Aha! You're the one who bought that Dante body-pillow I sold on Ebay! 

 

 Goldn_nips_luvr_8008!

 

Edited 14 hours ago by Mazer Rackham

[phandaal]

3 minutes ago, Joe said:

To be fair, I think this is more down to us often being on the opposite side of various conversations - which is in of itself fine, that's the beauty of communication. I should stress that I don't have a personal grievance with you (and I apologise if it's ever come across that way), I just don't necessarily agree with you a great deal of the time. But aye, absolutely agree on the eye-rolling and personalities side. 

 

Fair to say. It is one reason why I never use the "disagree" reaction here (no kidding, the history is there on my profile). My stance is: if it is important enough to me to make my disagreement public, it is important enough to write it out. If it is not important enough to write something down, then I might as well move on.

 

It will be interesting to see where this goes though, for sure. Generally, the places where people post their work here are not the same places people go to argue. Will the inclusion of AI-generated content change that? Part of me thinks it might, but I do not think it is likely, because the atmosphere of other pages is so much different than the NRBA and General Discussion pages.

[Frogian]

6 hours ago, HeadlessCross said:

, you can ask people in this forum with a good eye for color or talk to one of your artist friends. Use that as inspiration. You did zero work with the AI and it's to not be respected. 

 

 

what extra work are they done asking someone on a forum and taking their suggestions over asking an AI?

 

there is the value in art as a creative process and there value in art as a viewer of the end creation and i feel like there is quite a lot of confusing the two when this topic comes up

Edited 12 hours ago by Frogian

[kabaakaba]

Well trained model could easily "paint" miniature. Just try with you favorite llm. I think it could already do that. Do we really want THIS on b&c? We share and competing with each other. Trying out best. I don't want guy paid for AI post "his" miniatures here.

[Lay]

1 hour ago, Frogian said:

what extra work are they done asking someone on a forum and taking their suggestions over asking an AI?

 

Depending on one's personal stance on AI there might not be a difference. But I think your question raises an even better one: Why should the B&C, given its mission statement, tolerate a tool that offers an alternative to community engagement?

[ThaneOfTas]

9 hours ago, Cactus said:

Is this because a) anything that has had AI used in it's creation, even where that is declared, is not a 'proper' creative work of human expression or b) the AI tools available to us are built (in part) on large-scale unauthorised use of copyrighted works in their training data and we shouldn't reward wrongdoing?

A little of A) and a lot of B)

[Evil Eye]

The second point kinda falls apart given that any fan-made derivative work involves unauthorized use of copyrighted work to exist. Unless you're suggesting every single fan-artist, fanfic author, homebrew writer etc should go by Games Workshop to seek approval before making their work public in any way, which naturally would not be great.

 

...Arguably, any work of media at all relies on unauthorized use of copyrighted work to exist. No work of media exists in a vacuum; everything has been inspired by or derived from something else to some degree. GW's settings especially which started their lives as the creators' favourite sci-fi and fantasy works put in a blender and made into delicious, if rather legally dubious, smoothies. The difference between GW putting a million different fictional influences in their creation and an AI putting a million different works in its training set is that one is a bunch of nerds from Nottingham in the 80s and the other is a computer, but realistically it's much the same- other works go in, something else entirely comes out. You can argue the AI is more efficient at it, but from a "You didn't ask to put [X]'s work into a blender and inject the juice into your creation" angle, they're basically the same.

 

As I've harped on many times before, IP law is absolutely =][=CENSORED=][= in dire need of reform and barely equipped to deal with the existence of the internet, let alone AI, and I have a horrible feeling that laws intended to protect "creators" (read: corporations/states) against AI will instead just be used as another bludgeoning tool to throttle derivative works/parody/fair use out of existence, AI or not, but more to the point, you cannot argue against AI from the "it uses content without authorization" angle without condemning any unauthorized use of someone else's ideas, without which Warhammer would not exist.

 

Personally my issue with a lot of AI-produced work is it's just not very good. I can still see good uses for it (as a 3D modeller, I can see it being excellent for generating tiling texture alpha brushes for sculpting, which fundamentally is no different from using a pre-made brush from a pack or whatever- arguably if you generated it yourself it's more original given you at least made your own brush!), and I won't lie I think some of the absurdist comedy that utilizes the inherent nonsense of AI generation (including intentionally induced errors/hallucinations etc) is valid as a genre in its own right, given it's actually embracing the rampant ridiculousness of the "medium" as an end in and of itself.

 

But yes, most "press button, get art" stuff is, yeah, pretty bad. Not going to pretend otherwise there. As a technical tool/assistant it's definitely got uses (uses that are being sorely neglected in favour of "press button, get art")- trained generative algorithms could absolutely be used to make autosupports for 3D printing a lot better for example.

 

AI is a technology. It's not even a new technology, machine learning and procedural generation are as old as computers themselves (older still considering the basic principle of "randomly generated input informs output" describes anything that uses dice to a T), it's just a particularly refined and currently massively hyped-up version of said tech. I have no doubt the AI boom will deflate eventually, albeit probably not as catastrophically as everyone is expecting/fearing/hoping it will, and in all likelihood the market will probably transition to more specialized, less hardware-reliant uses of the same basic tech but refined for specific usecases (I predict "smart manuals" for helping people learn software or coding languages could be a real market). But it's not going to give us Deep Thought, nor is it going to bring about the rise of AM. It's just a technology, and like any technology it can be used for good or ill.

 

To end on a positive note, I know of one inarguably good use of it; the girl who hosts a massive archive of scientific data and records for anyone to access for free is currently developing an AI "librarian" trained on that archive that with any luck will be able to help scientists out, either by retrieving the exact information needed from the immense archive (as opposed to having to spend potentially hours trawling through multiple documents) or by answering questions that can be covered by the contents therein. By her own admission the project is a work in progress and needs to be developed to be reliable, but to me at least a programme designed to help forward the cause of science by quickly gathering information for researchers that would take ages to find otherwise sounds about as noble a use for AI tech as it gets.

[Wormwoods]

It's a forum for a creative hobby, something we all engage with because something about this setting speaks to us, makes us want to explore and create. It was an act of human creation, drawing influences, combining ideas from elsewhere, and using that to make something cool enough that we're all still mucking about in it some 40-odd years later. All art is based on previously created art, and none of us were born with our creative output pre-loaded into our brains, ready to spring forth.

 

But that isn't what the bloody plagiarism collage machine is. Art isn't about the end product, it's about the process, and a set of programs that 'do that for you' removes any interest I could have in engaging with something. The whole point is that you do it. That you experiment, and practice, and learn how to do this as part of a community of people doing the same.

 

Pandora's Box still had hope in the bottom. the LLMs are just full of stolen art and [expletive]. 

 

I don't suppose there's a way to hide a specific users blogs or posts from view, is there? Might come in handy going forward.

[Lord_Ikka]

9 hours ago, Joe said:

The policing side does concern me more than a bit, really. We've all seen some of the conversations that've taken placed in the news forum in the last few months or so where it's gone on for pages and pages of increasingly argumentative, hostile or sometimes downright petty behaviour before anyone has asked a moderator to step in; assuming they aren't part of the problem themselves. 

We as mods tend to take a light hand as long as arguments are constructive- it is only when things go so far as to be on the line of breaking the forum rules that we will step in. Regardless of anyone's personal take on the B&C's moderation style, as moderators we prefer to let conversations flow and members regulate themselves, as everyone here is an adult and should know how to act appropriately even in a heated discussion. 

 

Could we be more active/heavy with our moderation- definitely. Would the members like it- probably not. There is a fine line between active moderation, disregard, and crushing of opinions. If any member has an issue with how the site/a specific thread is moderated, it is very easy to PM a mod and start a conversation. We are here to curate the site and expand the hobby, and we all do this because we have a love of the hobby. 

 

10 hours ago, Joe said:

There's something to be said about needing to shape up ourselves, really - and ultimately, it's better to step away from a conversation that's aggrieving us rather than carry it on out of spite.

Mod hat off - This is something that needs to be encouraged more, no matter how hard it is for some to do. Personally, there is a reason I don't spend much time talking/offering my opinion on various topics anymore, or letting my responses dry on the vine- there is enough drama and stress in my normal day-to-day life that I would rather not get worked up in a space that I go to for relaxation. It is one reason I backed away from this topic, as those who don't have the same opinion as I do will most likely not change their mind from anything I say and I would rather not get into an increasingly pointed, and probably personal, argument. 

[HeadlessCross]

7 hours ago, Frogian said:

what extra work are they done asking someone on a forum and taking their suggestions over asking an AI?

 

there is the value in art as a creative process and there value in art as a viewer of the end creation and i feel like there is quite a lot of confusing the two when this topic comes up

 

One requires active engagement and discussion which leads to inspiration, vs "Please do this for me, I'm creatively bankrupt and thinking is bad :("

 

no, there's no value for AI as part of a "creative" process, because you're not creating anything to begin with the moment you use AI to steal other ideas and pass them off as your own, credit or no credit. 

Edited 5 hours ago by HeadlessCross

[Teetengee]

6 hours ago, Wormwoods said:

I don't suppose there's a way to hide a specific users blogs or posts from view, is there? Might come in handy going forward.

you can block users which hides and minimizes anything they post (so you have to click in to see them)