The Generative AI Battle
At this point we cannot un-ring the bell, so how do we handle generative AI ethically and address the elephant in the room?
Generative AI, for those not in the know, is AI that “creates” things. ChatGPT or Midjourney are examples of this. I put “create” in quotes because it’s really just remixing things it’s found elsewhere and not capable of genuine creation on their own. All they can do is mix and match and have no unique ideas.
This issue has been a really hot-button issue in creative communities. Artists furious that AI was trained on their artwork without their consent and is then being used by big companies to replace hiring artists. It’s a thing that’s been actually happening, though that seems to be cooling down as companies hit the wall of the fact that anything AI creates is not covered by copyright. As such, anything used in a video game, novel, television show, etc. isn’t the property of the company. That means character designs, scripts, content… none of it could be owned by the company, and anyone could grab it and use it.
To anyone who knows the industry, the ownership of an idea and the exclusivity of the thing is a huge part of where you get the money. If anybody could make a Marvel movie, it wouldn’t be something anybody cared about. If you don’t have copyright of a book, you can’t sell it exclusively, and someone could upload the whole thing somewhere legally. The same thing with music or anything else. Ownership of the intellectual property is how you make money off a thing.
That immediately knocks out Disney/Pixar firing their artists and writers and using AI exclusively to design characters and make movies and write scripts, to be clear.
However, that’s not the only place generative AI is used. In my case, I have been looking for ways to market my books. To do so, you need eye-catching images that draw attention to get folks to click on links to buy. That’s the basic flow. Cool picture, read the text, click the link, buy the book. (It’s a very, very simple version of a “sales funnel” for people in the know.) The image used to catch attention doesn’t need to be unique or specific to my book. It doesn’t need to do anything but get someone to go, “That’s cool.”
Prior to generative AI, folks in my position had a few avenues of doing this. We could:
Hire a graphic designer (which is very expensive) to create a slate of marketing materials for us to use and use those marketing materials exclusively.
Create our own marketing materials by buying stock photos or stock art and putting things together on our own (also often very expensive and requires knowledge of graphic design).
Create our own marketing materials out of royalty-free assets from Pixabay and other such sources (also requires knowledge of graphic design and has an extremely limited selection).
Not market and cry.
That last one is only sort of tongue-in-cheek.
Now, as someone who is an artist in my own right but is extremely poor, I have been using the third option. It’s extremely time intensive, requires me to use stock images that are just the best I have available, so my marketing images have been pretty generic and lackluster because, frankly, I can’t afford to pay someone.
That, and finding a non-binary model who looks anything even remotely resembling my main character has been almost impossible. Cassiel isn’t a waif-like enby. She looks like a football player but has softer, more feminine facial features while still very much resembling a man. It’s complicated. Suffice to say, she isn’t an easy person to find a model for. Let alone find angel wings that don’t look like they were purchased from Spirit Halloween.
I’ve been looking longingly at the cool images generated by AI, but until now I had held off because Midjourney et al. had been training their AI on artists who hadn’t consented to the process and whose work had, essentially, been stolen. While folks using it for personal D&D campaigns or something doesn’t ruffle my feathers (because most gamers just use Google image search and grab a picture that way anyhow), using it for marketing materials in a professional sphere made my teeth itch. I didn’t want to contribute to the theft of art from artists.
Enter Canva.
Canva recently unveiled its generative AI, but it’s doing so the right way. Its data set is limited to the images it has licensed from artists who are consenting to have their work used in that way. The artists and photographers whose work is being used are being fairly compensated. While I still would never have copyright of the images created (which is fine with me in these cases), I can now generate images for marketing materials, splash for my website, and other such things that look cool. I can also paint over them some and edit them for my own purposes, so I can create a Cassiel who looks pretty close to how Cassiel looks in my head.
Now, why not hire an artist to just… draw or paint Cassiel the way I see them? Frankly, I believe in fair compensation, and I can’t drop $500 on a single art piece right now. My goal is to use these generative AI images to create marketing materials that will help me sell books and allow me to eventually hire artists to do art like that. I feel pretty comfortable with that decision, too.
The argument around generative AI has primarily been the payment and consent of the artists whose works are being used to create the images and protecting the role of artists and art in the world. For anyone who might be unclear, I fully and wholeheartedly support artists and art. I don’t believe generative AI should replace them for a second. However, I also am extremely poor and trying to market and need things to do that with. Asking an artist to do something for free would be unfair and unreasonable. They can’t eat that way. But that’s about what I can afford.
That had put me in the impossible position of having to either become a professional artist myself (which I am, just not of characters — I do maps) or don’t market, which means I would never be able to hire an artist anyway because… well, I imagine that’s obvious.
This whole situation is sticky, but honestly, Canva’s attitude toward this, and their moves give me hope that there is a spot for generative AI that can cover the gap between the position I’m in and the place where I can hire artists to do custom art for me. Because that is the goal here.