I was originally going to put this into the Log, but it might be unwelcome.
You want a way to rattle image-generation Boosters? Most of the arguments they use can be used to defend Googling an image and putting a filter over it.
- “All forms of media take inspiration from one another, so that means it’s fine to Google another image, download it, and apply a filter to call it mine!”
- “Artists are really privilieged, so it’s morally OK to take their art and filter it!”
- “Using filtered images I downloaded from Google for game sprites will help me finish my game faster!”
- “I suck at drawing, so I have to resort to taking images from people who can draw and filtering them!”
- “People saying that my filtered images aren’t art are tyrannical! I deserve to have my filtered images be seen as equal to hand-drawn ones!”
AI Boosters use a standard motte-and-bailey doctrine to assert the right to steal art and put it into a dataset, yet entice people to buy their generated images. When Boosters want people to invest in AI, they occupy the bailey and say that “AI is faster and better than drawing by hand”. When Boosters are confronted with their ethical problems, as shown above, they retreat into the motte and complain that “it takes tons of time and work to make the AI do what I want”. Remember this when you find Boosters. Or don’t, since I doubt the sites where they lurk are worth your time.
One argument I’ve heard from the more ancap types is that these things are okay because intellectual property is a scam the stifles innovation. Let’s take that at face value.
Another concept that ancaps often support is time-preference behavior. In a nutshell, high time preference behavior is rewarded by immediate consequences (say, stealing the item gets it in your hands immediately for free), and low time preference behavior rewards long-term consequences (buying the item costs money, but you don’t go to jail or contribute to the closing of the business). They argue that societies that primarily operate on low time preference behavior are more prosperous than those that operate on high time preference behavior.
I would argue that the use of AI image generators in particular is short time preference behavior. Sure, you get the picture very fast and for very little money, but the widespread use of AI will discourage more and more people from either becoming or remaining artists. One funny thing AI researchers have found is that AI image gens can’t be trained on AI images or else they’ll produce objectively useless garage. So taking this to its logical conclusion, the society that relies on AI for all its images will eventually run out of human artists and thus be unable to produce any new art. AI fans are taking the short-term cost savings without considering the long term consequences of eliminating an entire industry.
One argument I’ve heard from the more ancap types is that these things are okay because intellectual property is a scam the stifles innovation. Let’s take that at face value.
IP abolitionism isn’t exclusive to Ancaps. And you don’t even address it again in the remainder of your post, instead attacking a concept ancaps typically use to justify racism.
My intention wasn’t to debunk either position. My intention was to show that, from their perspective, they too should oppose the proliferation of AI. It’s not a communist-vs-capitalist thing here. It’s a question of, does the use of this stuff even make sense to begin with?
