Invasive Diffusion: How one unwilling illustrator found herself turned into an AI model – Waxy.org
Invasive Diffusion: How one unwilling illustrator found herself turned into an AI model – Waxy.org
Stop reading this and go read Andy’s post here. It’s fascinating stuff and Andy’s been following the subject for the past year or two, exploring all the ethical questions around AI generated art.
I also want to highlight how scary good his generated artwork from selfies turned out. I talked to Andy about this and he gathered a couple dozen portrait photos of himself and fed it into an AI program and what he got out of it was this:
They’re all amazing but especially so the Pixar one and the Dali one, I mean, I could see Pixar the company using this tool internally to do rough first drafts based on whatever their voice actors look like, before they dive into really making a film look perfect.
Even if you’re not an artist this subject is endlessly fascinating and everyone’s got a different opinion on what’s legal and ethical about it. Andy did all the legwork to track down and interview both the artist involved as well as the person that added tools meant to copy the artist.
A Whole Lotta Nothing 
I find this really concerning and discouraging. One thing I did as I began creating the world where my indie-tabletop-roleplaying-game-in-development Our Magic is set was to hire an artist who was not yet able to fully make their living from their art. The money I paid them was a fair and respectful wage—I even told them when they were undercharging me and we corrected for that—and helped them grow their experience and portfolio. It also gave me a creative, thoughtful, human collaborator who enhanced the project beyond where my initial creative prompts began.
If I could have just had an AI make some stuff based on already existing stuff instead:
• that artist wouldn’t have gotten the boost toward the illustrator job they now have
• the initial people designs for my world would have been much more derivative of existing art and ideas
• I might have given up and gone for something ‘good enough’ instead of iterating and receiving an infusion of new ideas that took things to a better point
• I would not have been able to be confident that the illustrations I got were actually fully licensed to me (without incorporating art owned by someone else or used without the artist’s permission)
• that artist wouldn’t have been able to be involved at the ground floor in my project, with pieces that will carry on to a wider audience receiving proper attribution to them as the artist.
LikeLiked by 1 person
All great points Dinah!
I think this field is so new and moving so fast it’s hard even for lawyers to wrap their heads around it. Hopefully this all shakes out soon in a good direction but for now I’m still kind of shocked at the quality coming out of these projects.
Everything really does happen slowly, then all at once.
LikeLiked by 1 person
A thing that really jumps out at me at the end of all of this is that we are going to need to nurture artists through funding models other than abundant commercial publishing/animation work. It’s very clear that production companies, agencies and mixed-media creators who are not illustrators would, in the majority of cases, vastly prefer cheap work from an AI rather than costly professional commissions. We cannot change or shame that motive enough to turn the tide. I think it’s very important to address this before there are serious consequences.
Taking artwork for an AI model without getting the licensing lined up is stealing. If anyone is frustrated by this statement, I can reassure you that, based on all of my prior experience with design work, you will have no problem finding artists willing to supply licensed work for cheap for your AI models. There is always someone willing to undercut the whole world.
LikeLike