chat

I don't like AI at all and agree with the sentiment.

But uh the Blade Runners weren't the good guys in Blade Runner. They were escaped slave hunters. Deckard shoots a fleeing woman in the back in broad daylight and SA's another in his apartment.

This would be a good thread if the original image was actually AI but it's just a real photo taken with an iphone, they use AI upscaling now, which makes shit look like AI when it's actually real. This really fucks with identifying AI.

how bout no one uses ai here. it has no place in person to person communication

no more half measures walter

There is no reliable way for the average person to determine what is or is not LLM or Generative Image Model content.

I like the proposal. Not sure how to materially advance it but I agree something should be done to catch and remove this stuff.

I'm so scared of reaching the simulacrum comrades, I really don't know how to manage that eventuality. I'm still holding out hope it all goes down in flames but I'm losing confidence in that option.

If it's misrepresented as real, it absolutely should be removed and users who have a substantial history of making such posts should probably be assisted in re-evaluating their posting approach.

I picked Chat cause I dunno if anyone uses Feedback nowadays.

Chat seems fine to me personally, especially if you want non-hexbears to see it, but !feedback@hexbear.net is theoretically for technical site issues (and is pretty low traffic, yeah), while !hexbear@hexbear.net is for meta / moderation discussion (and also relatively low traffic, yeah)

I’m in favor of your proposal.

dang it. I was had

so it seems worth getting in front of

I want to get in front of a train

If it's purely for a shitpost, I personally don't think it's worth getting up in arms about

Edit: Also there have been updates in that thread: The exact pic may or may not be ai. but that specific meme car is in fact real

I think we should do your proposal because it would ultimately cause a massive amount of drama and hexbear hasn't had a moderation struggle session lately. It'll be like when hexbear renamed the_dunk_dank for the 4th time. Honestly it's a very very bad idea and hexbear should do it because of that. It's part of the hexbear character, making up moderation rules that don't exist on any other website and enforcing those rules haphazardly.

Requiring all AI images to be removed from the website would create a mechanism for any post to be removed on a whim if accused of being AI. The chance of false positives is too high.

In 1997, IBM Deep Blue was the first computer to defeat a world chess champion at chess. The modern chess engine called 'Stockfish' is virtually unbeatable by any human, since like 2014. This same principle applies to AI image generation. 2 years ago, it was very easy to spot an AI generated image. 2 years from now, it will be virtually impossible, if not already.

Hexbear already has rule 13 which prevents spreading misinformation. Any AI image which spreads misinformation can already be removed by that rule.

There are also many instances where it could be important to post an AI image. For example, Trump posts AI images on truthsocial all day, every day. What if Trump posts an image of an explosion in Iran and says "This is a photo of the US bombing Iran". Is hexbear not allowed to document the President of the imperial core using fake images to promote endless killing? This is not a hypothetical. This is something that Trump has done many times before. Trump posted AI images of the Iran women's soccer team and said that Iran would kill the women for not singing Iran's national anthem.

Forming an AI detection taskforce is an absurd proposition. Can you name a single website in the history of the internet which has created an AI image detection team? It doesn't exist because it is irrational. It would be less irrational for hexbear to ban all image posts than it would be for hexbear to try to determine which images are AI. At least if hexbear banned all images, then it would be consistent and not based on misinformed guessing.

It would be more reason to require that non-OC images include a link to the original source of that image and then determine whether then say whether the source is a credible source or a non-credible source.

The primary purpose of hexbear should be documenting real things that happen. If a powerful figure purports an AI generated image to be real, that should be documented.

pro genAI communists want to put genAI in our spaces because they know that other pro genAI spaces are deeply hostile to our ideology and the mainstream genAI companies have explicitly said they want us all dead. many will point to chinas use of AI but look me in the eye and tell me Xi Jinping is promoting AI anime girl chatbots.

if somebody goes to chatgpt and say "make an image of a horse riding Stalin" and posts that to /c/art they should probably get banned. But AFAICT that kind of thing almost never happens?

I fully agree, but also AI is hard for some people to spot, so if we do find out something is AI try not to just assume the poster is being deceptive unless they obviously are. Just be politely be like "Sorry dude I think this is AI"

Transparency is what matters. If a poster is conned as well though, they need to edit as soon as it's made known.

Trying to pass off slop as original content you created should be bannable imo.