As AI Art Debate Continues, Artists Find Themselves Victims of Witch Hunts

We recently spent some time in an “Artists Against Ai” Facebook group, to get a better understanding as to why so many artists are against Ai generated works. As the debate over AI generated art continues, many artists are now turning on each other in “witch hunts” to try and identify AI. Much like the original Salem Witch Trials, this is resulting in artists being falsely accused of utilizing AI, often through relying on unreliable tools.

First, some background. There is a growing community of artists which believe that “AI is theft”, since many training models were built upon publicly available, but copyrighted works. The argument is that since the AI learned how to render images utilizing publicly available images which did not give permission for that usage, that the AI is stealing from other artists. Of course, this thinking becomes problematic because many artists say their art styles were “inspired” by others’, as well as the significant community of artists which not only utilize copyrighted works without permission in “fan art”, but even sell these derivative works without paying royalties to the copyright holder. For example, it’s very common to see “fan art” being sold featuring Marvel, Disney, or Star Trek/Star Wars characters.

The argument that AI is theft because it learns its art style from others without their permission is a dangerous precedent which, if successful, could result in severe crackdowns against human artists who have done the same in their own learning of how to create art.

Regardless of the legal arguments surrounding if AI use in learning is “fair use” under copyright laws, even more concerning is now artists efforts to publicly attack (and potentially ruin the careers of) artists whose works remotely look like they may be AI generated. The problem with this of course is that these detectors are very unreliable, and leading to many false accusations.

These witch hunts appear motivated by a burning hatred for all things AI. Sometimes even resulting in the very artists who are against AI finding themselves on the receiving end of these witch hunts, even when the works are actual physical works and not digital art.

In fact, this hatred has grown so much that artists are now utilizing Nightshade, a tool designed to protect art from unauthorized usage in AI training models, to “poison” public domain works. At this point, this is no longer about protecting their own works from unauthorized usage, but purely about sabotaging the AI training data.

So just how reliable are these detectors? We decided to put a few to the test.

First up, we tested ZeroGPT, a text AI detection tool. We decided to feed ZeroGPT text which we knew couldn’t possibly be AI generated – the Constitution of the United States of America.

Well that’s concerning. Either we’re living in a simulation, or ZeroGPT is just simply not very accurate at detecting text.

Alright, but surely AI image detection works better, right? We used “Is It AI?”‘s image detector with something else we’re confident was not AI generated – the Mona Lisa.

Well, either we just confirmed we’re living in a simulation, or these “AI detector” tools are completely bunk.

Of course, the real victims in all of this will be the actual artists and writers who put in hours upon hours, only to have false accusations that their works are AI generated. These accusations have the potential to destroy careers, and ruin incomes.

By the way, we here at Radio Free Hub City absolutely use AI as a starting point for some of our articles, and will also use AI for some of our cover graphics. We do this because we’re on a shoe-string budget, and want to utilize AI as a starting point to help us provide quality information at an accelerated rate. But that’s just it, a starting point. We don’t use AI to write a finished article, because AI is simply not advanced enough to completely understand the intricacies of human speech and writing, just like it can do an excellent job at generating images we ask it to generate, but these images lack emotion or soul. It’s fine for cover graphics, but we’d never try to sell AI generated art – because it simply lacks the creativity of human artists.

AI is a great building block, or starting point, for helping people jumpstart their creativity. And by engaging in these witch-hunts, artists are doing themselves a disservice, potentially pushing away other artists who genuinely enjoy creating on their own without AI, but become discouraged by false accusations. We hope that the “anti-AI” community sees the errors in their ways soon, and corrects course before too many careers are ruined.

Like it or not, AI is being significantly adopted by multiple industries. Programs such as Microsoft Word are now utilizing AI to help draft emails faster, and programs like Adobe Photoshop are utilizing AI to help edit photos. At the end of the day, AI is just another tool in the toolbox, and it’s how you utilize that tool that matters.

Article by multiple RFHC contributors.