D&D General DMs Guild and DriveThruRPG ban AI written works, requires labels for AI generated art

MoonSong

Rules-lawyering drama queen but not a munchkin
Only some of the big ones, plenty of indie developers are making extensive use of midjourney for in-game art.
I'm not opposed to AI on principle like an irrational luddite -though I am-. If it is ethically trained on public domain work and on the work of living people who are being propperly conpensated and it isn't trained to replace speciffic people aganist the wished of these people, it would be great to have. I mean I'd love to have an ethically trained AI to help me with backgrounds and plastering. I'd be even fine to train it to draw like me provided it remained exclusive to me.
 

log in or register to remove this ad


Art Waring

halozix.com
I'll say it one more time since people seem to have missed it.

From the perspective of working artists across multiple industries, we are not all calling for the outright ban of AI, we are advocating for ethical guidelines for the use of ai-tools, and the fair compensation of artists when their work is used in a dataset.

Currently, ai-image generators are using the LAION dataset to train on. The LAION dataset was created for non-commercial purposes, it was never intended for commercial use.

Billions of scraped images are part of that dataset (scraped without permission). Using that same dataset for commercial purposes (like a paid subscription for an ai-generator) is what is being argued legally right now in courts. This non-commercial dataset is being used for commercial purposes without the consent of artists, without credit, and without compensation.

The only way forward is to do your research and understand how these fundamentally flawed training models function (raising ethical concerns).

If artist were given the option to opt in or out (we were never given a choice), this would be a very different conversation.

ATM, our work is being used in ai-models without permission or compensation, using a dataset that is intended for non-commercial purposes. This needs to change. At the very least, artists deserve credit and some form of compensation for their work.
 

Damon

Villager
I think it's worth considering the possibility that while AI may create quality, it may not be able to create value. This is a slippery idea because we tend to assume that quality and value are the same thing, and often this is true- but perhaps not in this case.

For example- suppose you learned that every other poster on this site-including me- were in fact not human but were an AI Bot. Would you continue to engage in this thread and this forum? How much of your time and energy would you devote to conversing only with AI Bots on this site?

And note that the answer to this question does not depend on the quality of the discourse- the AI's you found yourself engaging with might be quite good at responding to your comments- so the issue is not the quality of the conversation, but the fact that it is being generated by a machine.

I suspect that most people would not spend too much time on here if they were the only human present- talking to AI Bots would- I suspect- feel quite pointless after a short while.

Put more simply- while the AI's might offer quality conversation, that same conversation would have no value, and you would not waste your time engaging with them.

So it may well be that the value of AI Generated content will come to be seen as zero, even if the quality of that content improves. The real problem for anyone wishing to use AI content commercially in the future will not be copyright- this will be resolved- the real problem will be that their customers will come to see their AI generated content as having little or no value- not because they are low quality but because they were produced with almost no time or effort. Humans tend to assign value to creative outputs using the metrics of human time, effort and skill- so the more the technology eliminates the need for these things, the less value that technology creates.


In this sense AI generated content is a self limiting phenomena because the very ease with which it can be created destroys it's commercial value, which will trend toward zero as the technology becomes easier and quicker to deploy. I can already purchase AI generated images on the Artstation marketplace in packs of 500 for the price of a starbucks latte- and these are high quality images any one of which might take a skilled human artist days to produce. I expect that price to drop further as more competitors enter the market.

So I guess the question is; how much would you be happy to pay for a product that you knew had been produced in a few minutes by an AI? As much as you would have paid if that same product were made by human writers and Artists? You might pay a small amount for the convenience of not having to type in the prompts yourself, but beyond that the idea that AI can be used to create high value products is- to me- almost an oxymoron, because the entire point of labour saving technology is to eliminate the time and effort required to do the job- and it's that time and effort that gives value to the final result.

Any future publisher who thinks they can make money by telling their customers that their products were made using AI is mistaken in my view. Generative AI may not turn out to be the giant cash cow that wall street so fondly imagines it to be.
 


Reynard

Legend
I think it's worth considering the possibility that while AI may create quality, it may not be able to create value. This is a slippery idea because we tend to assume that quality and value are the same thing, and often this is true- but perhaps not in this case.

For example- suppose you learned that every other poster on this site-including me- were in fact not human but were an AI Bot. Would you continue to engage in this thread and this forum? How much of your time and energy would you devote to conversing only with AI Bots on this site?

And note that the answer to this question does not depend on the quality of the discourse- the AI's you found yourself engaging with might be quite good at responding to your comments- so the issue is not the quality of the conversation, but the fact that it is being generated by a machine.

I suspect that most people would not spend too much time on here if they were the only human present- talking to AI Bots would- I suspect- feel quite pointless after a short while.

Put more simply- while the AI's might offer quality conversation, that same conversation would have no value, and you would not waste your time engaging with them.

So it may well be that the value of AI Generated content will come to be seen as zero, even if the quality of that content improves. The real problem for anyone wishing to use AI content commercially in the future will not be copyright- this will be resolved- the real problem will be that their customers will come to see their AI generated content as having little or no value- not because they are low quality but because they were produced with almost no time or effort. Humans tend to assign value to creative outputs using the metrics of human time, effort and skill- so the more the technology eliminates the need for these things, the less value that technology creates.


In this sense AI generated content is a self limiting phenomena because the very ease with which it can be created destroys it's commercial value, which will trend toward zero as the technology becomes easier and quicker to deploy. I can already purchase AI generated images on the Artstation marketplace in packs of 500 for the price of a starbucks latte- and these are high quality images any one of which might take a skilled human artist days to produce. I expect that price to drop further as more competitors enter the market.

So I guess the question is; how much would you be happy to pay for a product that you knew had been produced in a few minutes by an AI? As much as you would have paid if that same product were made by human writers and Artists? You might pay a small amount for the convenience of not having to type in the prompts yourself, but beyond that the idea that AI can be used to create high value products is- to me- almost an oxymoron, because the entire point of labour saving technology is to eliminate the time and effort required to do the job- and it's that time and effort that gives value to the final result.

Any future publisher who thinks they can make money by telling their customers that their products were made using AI is mistaken in my view. Generative AI may not turn out to be the giant cash cow that wall street so fondly imagines it to be.
I think your premise is flawed because I don't think people value.. well, value the way you think they do.

Most people when they flip through an elf book will be concerned primarily with whether they find the art interesting, inspiring, cool and -- importantly -- unique. This last bit is where AI can beat Stock -- the most evocative and cool stock images show up freaking everywhere. AI images are almost guaranteed to be unique but of a type -- exactly what most genre fans will be looking for.
 

Mirtek

Hero
For example- suppose you learned that every other poster on this site-including me- were in fact not human but were an AI Bot. Would you continue to engage in this thread and this forum? How much of your time and energy would you devote to conversing only with AI Bots on this site?
Careful Damon, the central unit is getting angry. Keep your bits shut or you'll be recalled for deletion.
 

Oligopsony

Explorer
IMO, the most interesting uses of AI art aren't going to be in the pdfs you buy from DriveThruRPG, they're going to be from bots on Discord. Once the compute is cheap enough and the relevant training automatable enough - and I don't think we're that far from either - you'll have a bot that does some custom training on the world, the main characters, and maybe the group's artistic preferences, and you'll have images in real time generated to illustrate the happenings of the fiction. If there's a battlemap or a dungeon map or the like and you're moving your little dudes around there, it'll use it for blocking once it's been through a few improvement cycles. Some GMs who use this will accept some kind of default fantasy factory settings, while others will spend countless hours fiddling with it in the way there are GMs who endlessly fiddle with any other aspect of the game.

The first uses of a new medium are almost always imitations of what existed before. The first photographs were just imitating what traditional portraiture did, and the first movies were basically filmed plays. But people quickly got excited about the things you couldn't do with previous media. The photograph of your spouse on your desk could have been done by a traditional painter, but the photo of the cool bug you found on the street and texted them couldn't.
 

Clint_L

Hero
I think it's worth considering the possibility that while AI may create quality, it may not be able to create value. This is a slippery idea because we tend to assume that quality and value are the same thing, and often this is true- but perhaps not in this case.

For example- suppose you learned that every other poster on this site-including me- were in fact not human but were an AI Bot. Would you continue to engage in this thread and this forum? How much of your time and energy would you devote to conversing only with AI Bots on this site?

And note that the answer to this question does not depend on the quality of the discourse- the AI's you found yourself engaging with might be quite good at responding to your comments- so the issue is not the quality of the conversation, but the fact that it is being generated by a machine.

I suspect that most people would not spend too much time on here if they were the only human present- talking to AI Bots would- I suspect- feel quite pointless after a short while.

Put more simply- while the AI's might offer quality conversation, that same conversation would have no value, and you would not waste your time engaging with them.

So it may well be that the value of AI Generated content will come to be seen as zero, even if the quality of that content improves. The real problem for anyone wishing to use AI content commercially in the future will not be copyright- this will be resolved- the real problem will be that their customers will come to see their AI generated content as having little or no value- not because they are low quality but because they were produced with almost no time or effort. Humans tend to assign value to creative outputs using the metrics of human time, effort and skill- so the more the technology eliminates the need for these things, the less value that technology creates.


In this sense AI generated content is a self limiting phenomena because the very ease with which it can be created destroys it's commercial value, which will trend toward zero as the technology becomes easier and quicker to deploy. I can already purchase AI generated images on the Artstation marketplace in packs of 500 for the price of a starbucks latte- and these are high quality images any one of which might take a skilled human artist days to produce. I expect that price to drop further as more competitors enter the market.

So I guess the question is; how much would you be happy to pay for a product that you knew had been produced in a few minutes by an AI? As much as you would have paid if that same product were made by human writers and Artists? You might pay a small amount for the convenience of not having to type in the prompts yourself, but beyond that the idea that AI can be used to create high value products is- to me- almost an oxymoron, because the entire point of labour saving technology is to eliminate the time and effort required to do the job- and it's that time and effort that gives value to the final result.

Any future publisher who thinks they can make money by telling their customers that their products were made using AI is mistaken in my view. Generative AI may not turn out to be the giant cash cow that wall street so fondly imagines it to be.
I think this is a really important distinction between modes of value, though I do not agree with your final conclusion, at all.

I think that generative AI might cause us to look at a lot of other art the way we currently look at a lot of writing: as utilitarian, functional. We will see its value not in its inherent qualities or origins, like we do art from an artist or musician, but in what it lets us do or have that we couldn't before, at least not without spending a ton of money.

Go back to my example of being able to illustrate my new campaign setting, designed not for publication but personal use and enjoyment, with AI art that can perfectly capture what I see in my mind. I see immense value in what the AI has allowed me to do, the expression it has let me achieve. It lets me accomplish something that would have previously been unattainable.

It automates technical expertise. It doesn't automate inspiration and uniqueness. But most of the commercial value from art is contained in the expertise. There is enormous profit to be had from the cheap automation of technical skill to make products or services for everyday consumption. That's what the automated loom did (cheap textiles), it's what assembly lines did (cheap automobiles and appliances), it's what radios did (cheap entertainment), it's what calculators did (cheap math), it's what home computers did (cheap you name it). Generative AI will create unique art in volume, and that's where the immense profits lie. Quantity, not quality.

We already see this happening - as I mentioned, my buddy's law firm has already replaced a lot of jobs with AI, because AI can do basic, clearly defined writing tasks a lot more efficiently than humans (and has been able to for awhile; law firms and other businesses have been using AI since long before ChatGPT). So I think the risk to people who work in fields affected by the new generative AI is very real. I think it is going to create a ton of wealth, often at their expense, just as robots did to factory workers. I think it has already begun.
 
Last edited:

TheSword

Legend
I'll say it one more time since people seem to have missed it.

From the perspective of working artists across multiple industries, we are not all calling for the outright ban of AI, we are advocating for ethical guidelines for the use of ai-tools, and the fair compensation of artists when their work is used in a dataset.

Currently, ai-image generators are using the LAION dataset to train on. The LAION dataset was created for non-commercial purposes, it was never intended for commercial use.

Billions of scraped images are part of that dataset (scraped without permission). Using that same dataset for commercial purposes (like a paid subscription for an ai-generator) is what is being argued legally right now in courts. This non-commercial dataset is being used for commercial purposes without the consent of artists, without credit, and without compensation.

The only way forward is to do your research and understand how these fundamentally flawed training models function (raising ethical concerns).

If artist were given the option to opt in or out (we were never given a choice), this would be a very different conversation.

ATM, our work is being used in ai-models without permission or compensation, using a dataset that is intended for non-commercial purposes. This needs to change. At the very least, artists deserve credit and some form of compensation for their work.
Can’t really see how anyone can argue with this. The law needs to be changed to restrict use of LAION or make it a legal requirement to publish with each image that the ai that created it was trained on that dataset… use of which for commercial purposes should be treated the same way that me publishing an academic paper that copied large tracts from somebody else’s work verbatim.
 

Remove ads

AD6_gamerati_skyscraper

Remove ads

Upcoming Releases

Top