The people mad about the tool, don't understand it, treating it like some boogeyman with agency. The so-called artists who cry about losing their jobs aren't good enough to be hired anyway. And the writers that wave their pitchforks and torches to give purity tests to their fellow creatives (IYKYK) are little better than the other list-makers of history.
When a house is built, do you give credit to the hammer and saw, or the craftsman?
If you are stabbed, are you angry at the knife, or the hand that wields it?
Being mad at the tool being used is equally as absurd as being mad at someone for using it.
What matters is what they do with it, not that it was used.
Cedar used Midjourney as a tool to develop my cover for my published novella, Cloak and Stola. Then I paid her to design my cover for my published novella, Down Pad. I paid another artist who uses AI tools (I think it's Midjourney, but I'm not sure) to design the image I use for my substack short story "Braids."
The issue here, as far as I can see it, is two-fold. Both revolve around copyright, of course, because copyright is the 800lb gorilla in every creative space.
Firstly, Cedar explains (very well, I might add) how generative image creation is trained uses copywritten images without the copyright-holders' permission. The argument can be made that regular artists do this sometimes as well, that an artist might be inspired by another artist's work and make their own work from said inspiration. This, as most artists will tell you, can cause problems if the created work hews too closely to the work that is causing inspiration. Just look at the 'Blurred Lines' lawsuit, where the estate of Marvin Gaye successfully sued because the song copied the 'sound' and 'feel' of Gaye's 'Gotta Give It Up'.
You might say that a generative image program doesn't directly copy pictures, but there is precedent for being sued because you were 'too' inspired by a source.
Also, of course, there is the whole 'not getting permission from the copyright holder to use the image in the training of the program' part. Cedar referred to Passive Guy, who said using a relatively small amount of protected material falls under fair use. With all due respect to Passive Guy, he is not my lawyer (and probably not yours either), nor would he be the judge at my trial, and some companies (*cough*Disney*cough*) seem to have limitless funds and no regard for the concept of 'fair use'. So even if PG's words prove true in the fullness of time, they won't keep me from being sued into homelessness by a large corporation now.
Secondly, there's the issue of who exactly owns the copyright of the generated image. Cedar runs through this with little more than a sentence, and seems to hold the belief that the program is a tool and therefore the user of the tool holds the copyright. This definitely needs to be decided in court, because this image-generation program isn't a chainsaw or paintbrush. It is not being wielded by the deft hand of a master in the trade, it is doing almost all of the heavy lifting in terms of image-generation on its own. It would be like if Captain Picard told the replicator to print a sculpture of a man playing a banjo and then proceeded to claim he created it himself.
I think a better illustration of my understanding (or lack thereof) of the issue is this: if entity 1 gives a prompt to entity 2, who owns the copyright for the resulting work? Well, if we replace 'entity 1' with 'someone representing Raconteur Press' and 'entity 2' with 'an image generating program', Cedar believes the copyright holder to be the Rac Press representative. But, if we replace 'entity 2' with 'a writer looking to write for a Rac Press anthology', then you damn well better believe the writer is keeping the copyright (unless Rac Press is planning on coughing up some serious cash, of course). If Rac Press believes they own a writer's work simply because they gave a prompt, Rac Press can pack sand.
Of course, I don't believe anyone at Raconteur Press believes they own an anthology writer's work simply because they gave an anthology prompt, that would be foolish. Along that same line of thinking, though, it seems hard to believe a promptmaker automatically owns the copyright of the resulting product of an image-generation program. And I think until the courts in this country rule on the legal definition of 'making something', it is best to err on the side of 'I know that I own the copyright of this image.'
An author who publishes with us in one of our anthologies gets the rights to their story returned to them after one year. This is spelled out in the contract the author signs.
You're comparing apples to oranges in your comment.
Using AI to achieve a particular goal is definitely creative. Try any of the basic free AI image creators and try to get everything just right. I tried several with the following prompt. "Female blacksmith forging an iron gate" Most of the tools shown were impossible, as were the accessories the smiths used, to say nothing of their dress. However they were all by default attractive. Obviously I was not proficient enough with the programs to achieve a successful edit.
I tend to use NightCafe for my character sketches. MS's autopilot could almost get what I wanted, and I still had to work on prompts. The other day, I tried Chatgpt's system, which uses Dalle 4. The system is getting closer to what I want. I want to be able to type a description of what I want, and the program takes that description and uses it to create an image.
I want to be able to write something like: The scene takes place in the cabin on a 16th-century sailing ship. A middle-aged man with white hair, wearing a black shirt, black pants, and black boots, sits in a chair behind a desk. A half-elven dryad kneels in front of the desk. She has long black hair and is wearing a crop-topped black silk blouse with a matching black skirt. She is looking up at the man in rapt attention.
When the AI can take all of that and give me the image I want, then it'll work.
Right now, I can ask for two different characters in the description and it still messes that up.
Eventually, AIart will get there, but it's not there right now.
"Lighting a candle" FTW.
In ten years time this won't even be controversial. Unfortunately we're not there yet.
Remember when CGI in movies was cheating? Remember the controversies about "ugly tellers" (early arms)?
The people mad about the tool, don't understand it, treating it like some boogeyman with agency. The so-called artists who cry about losing their jobs aren't good enough to be hired anyway. And the writers that wave their pitchforks and torches to give purity tests to their fellow creatives (IYKYK) are little better than the other list-makers of history.
When a house is built, do you give credit to the hammer and saw, or the craftsman?
If you are stabbed, are you angry at the knife, or the hand that wields it?
Being mad at the tool being used is equally as absurd as being mad at someone for using it.
What matters is what they do with it, not that it was used.
Cedar used Midjourney as a tool to develop my cover for my published novella, Cloak and Stola. Then I paid her to design my cover for my published novella, Down Pad. I paid another artist who uses AI tools (I think it's Midjourney, but I'm not sure) to design the image I use for my substack short story "Braids."
The issue here, as far as I can see it, is two-fold. Both revolve around copyright, of course, because copyright is the 800lb gorilla in every creative space.
Firstly, Cedar explains (very well, I might add) how generative image creation is trained uses copywritten images without the copyright-holders' permission. The argument can be made that regular artists do this sometimes as well, that an artist might be inspired by another artist's work and make their own work from said inspiration. This, as most artists will tell you, can cause problems if the created work hews too closely to the work that is causing inspiration. Just look at the 'Blurred Lines' lawsuit, where the estate of Marvin Gaye successfully sued because the song copied the 'sound' and 'feel' of Gaye's 'Gotta Give It Up'.
You might say that a generative image program doesn't directly copy pictures, but there is precedent for being sued because you were 'too' inspired by a source.
Also, of course, there is the whole 'not getting permission from the copyright holder to use the image in the training of the program' part. Cedar referred to Passive Guy, who said using a relatively small amount of protected material falls under fair use. With all due respect to Passive Guy, he is not my lawyer (and probably not yours either), nor would he be the judge at my trial, and some companies (*cough*Disney*cough*) seem to have limitless funds and no regard for the concept of 'fair use'. So even if PG's words prove true in the fullness of time, they won't keep me from being sued into homelessness by a large corporation now.
Secondly, there's the issue of who exactly owns the copyright of the generated image. Cedar runs through this with little more than a sentence, and seems to hold the belief that the program is a tool and therefore the user of the tool holds the copyright. This definitely needs to be decided in court, because this image-generation program isn't a chainsaw or paintbrush. It is not being wielded by the deft hand of a master in the trade, it is doing almost all of the heavy lifting in terms of image-generation on its own. It would be like if Captain Picard told the replicator to print a sculpture of a man playing a banjo and then proceeded to claim he created it himself.
I think a better illustration of my understanding (or lack thereof) of the issue is this: if entity 1 gives a prompt to entity 2, who owns the copyright for the resulting work? Well, if we replace 'entity 1' with 'someone representing Raconteur Press' and 'entity 2' with 'an image generating program', Cedar believes the copyright holder to be the Rac Press representative. But, if we replace 'entity 2' with 'a writer looking to write for a Rac Press anthology', then you damn well better believe the writer is keeping the copyright (unless Rac Press is planning on coughing up some serious cash, of course). If Rac Press believes they own a writer's work simply because they gave a prompt, Rac Press can pack sand.
Of course, I don't believe anyone at Raconteur Press believes they own an anthology writer's work simply because they gave an anthology prompt, that would be foolish. Along that same line of thinking, though, it seems hard to believe a promptmaker automatically owns the copyright of the resulting product of an image-generation program. And I think until the courts in this country rule on the legal definition of 'making something', it is best to err on the side of 'I know that I own the copyright of this image.'
An author who publishes with us in one of our anthologies gets the rights to their story returned to them after one year. This is spelled out in the contract the author signs.
You're comparing apples to oranges in your comment.
-R.O.
Good article. Getting real sick of these militant luddites.
Using AI to achieve a particular goal is definitely creative. Try any of the basic free AI image creators and try to get everything just right. I tried several with the following prompt. "Female blacksmith forging an iron gate" Most of the tools shown were impossible, as were the accessories the smiths used, to say nothing of their dress. However they were all by default attractive. Obviously I was not proficient enough with the programs to achieve a successful edit.
I tend to use NightCafe for my character sketches. MS's autopilot could almost get what I wanted, and I still had to work on prompts. The other day, I tried Chatgpt's system, which uses Dalle 4. The system is getting closer to what I want. I want to be able to type a description of what I want, and the program takes that description and uses it to create an image.
I want to be able to write something like: The scene takes place in the cabin on a 16th-century sailing ship. A middle-aged man with white hair, wearing a black shirt, black pants, and black boots, sits in a chair behind a desk. A half-elven dryad kneels in front of the desk. She has long black hair and is wearing a crop-topped black silk blouse with a matching black skirt. She is looking up at the man in rapt attention.
When the AI can take all of that and give me the image I want, then it'll work.
Right now, I can ask for two different characters in the description and it still messes that up.
Eventually, AIart will get there, but it's not there right now.