The current trend of Lensa, Midjourney, etc is rent seeking ghoulish behavior that seeks to destabilize artistic labor even more than it already is.
This is NFT shit all over again and theres needs to be a zero tolerance of it. This is automated theft and exploitation to be stopped.
Putting words in people's mouth by using imaginary quotes conceding your absurd premises makes it clear that the threat you are upset with exists only in your imagination, Kaden.
The best part of Cohost are its extensive controls to curate your experience, and a userbase eager to accommodate people who require material that upsets them tagged. You already have all the tools you need to curate your dashboard and limit your exposure to this material.
I personally believe that there's little value that could be added in repeatedly attempting to make this entire debate about one single argument based purely on shock value. It makes the arguments against AI art look simpler than they actually are, and makes the actual hard-hitting arguments take a back seat. Not to mention the article itself that this info came from (which I definitely think hasn't helped here with its clickbait title) also goes into detail on the ways they're already trying to combat stuff like this - how they're trying to completely filter out extreme materials like that.
Finally, I personally believe it's not even relevant unless you're actively trying to produce materials like that, in which case, you're a sicko (derogatory). If you tell it to generate stuff like that, I feel the fault lies more with you than whoever trained the model - which also included things like safety filters that you likely had to disable to make it output such things. Depending on what horrible things you chose to create, you might even be violating the license you had to agree to before downloading the model. Regardless, I don't exactly think any good model is referring to revenge porn or ISIS beheadings when you tell it to create a picture of a cute kitten.
I think the biggest argument against AI art has to do with its capacity to displace artists, and the impact it will have on artists looking to make a living if its existence is normalized. This is a problem that can't be solved by a model from someone who actually completely filters out the bad material - it has to be solved outside of that.
I like to think that moves that places such as the US Copyright Office are making helps (in the US, AI-generated art can't be copyrighted as it lacks "human authorship", and multiple AI-generated works have already been rejected for copyright due to that reason), but not all governments are the same in that manner, unfortunately.
My opinions on this topic are a bit mixed, admittedly. I think, in some regards, there has been a Pandora's Box that has been opened - AI art won't cease to exist, even if it were completely and totally outlawed. I think the ideal solution would be making it completely unviable to use in any situation that involves the exchange of money - but that involves some massive systemic change, and I'm sure companies that want to profit off of these models won't like that.
As far as if it should be banned from Cohost, though? I think that'd cause some problems - mainly, how can you tell what art is AI-produced and what isn't? Sure, some models invisibly watermark their output, but then there's folks who will remove those functions, or will destroy the watermark in some way through manipulating the image. Even worse, some folks know the watermarking is so that the generated works won't be included in newer datasets, so there might be a few artists that deliberately watermark their own works so they won't get trained on.
Sure, some images might be more obviously AI-generated by just how badly they fudge fingers or other artifacts, but not every AI-generated image is like that, and what if the poster made a very minor edit to fix such details? Even worse - what if someone's art style involves some level of simplification of features that looks similar *enough* to how AI fudges those features?
I think a blanket ban on AI art will just lead to people obscuring the fact they generated their works through a neural network, and will lead to people trying to witch hunt folks that they *think* are making AI generated art - even if they aren't using neural networks in their art. I think it should be a situation where people are encouraged to tag AI-generated works as such, so that tag muffling can do its job.
If there was some way to identify, with perfect accuracy, which images are and aren't AI generated, I think a ban would be a better idea. As is right now, however, I feel like it'd just lead to more toxicity, with AI art wordsmiths trying to obscure the fact their works are generated by neural networks, and some folks choosing to try and hunt those people down and perhaps not always coming up with the right answer.
This is extremely nit-picky and not very related to the general topic of discussion that has previously occurred: How is a ban a feature? This seems more like a request for site policy?
I think I generally agree with sentiments that have already been shared: a blanket ban on AI-generated art would most likely result in people not mentioning that the art they post or share is AI-generated.
Perhaps staff could write in scripts that scan posted art for certain informational tags (invisible watermarks of known AI artists, or other specific info that might be found within the images' metadata) that would generally indicate an image was a result from an AI generator, but removing those information tags can be as easy as taking a screenshot before posting. At what point do we start asking staff to go snipe hunting for us?
If staff cannot reliably and effectively ban AI art, then it would fall to the users of cohost to report it. Some images are obviously generated by AI while others are not. Many artists have a period of fine-tuning in their creation process; there is nothing stopping an AI-generated image from being run through photoshop before posting. How many times have we fallen for a slick photo edit?
I am not a fan of AI-generated art and never really have been. In its current state I especially do not want to interact with it. The decision for me comes down to 1) being able to knowingly block/ignore users who post AI-generated art or 2) spending time trying to guess whether or not any image I come across has been generated by a program. The latter option seems inherently toxic to artists as a group. I would rather have AI-generated art allowed on cohost so that I can better ignore it.
The conversation in this thread has ceased to be useful or constructive. I'm locking this for now. We do not have any pending community guidelines changes re: AI Art to announce at this time.