@h3artbl33d

Jee. Really ?
How can I recognize images being slop generated ?
I mean I don't use that crap myself so not sure how to do it.

@as400 @h3artbl33d It's not a bullet-proof method, but usually, I look at the texts. ChatGPT and co can't generate letters properly and there are often some spaces missing like in this screenshot (the second OpenBSD text at the bottom, the N is barely visible)

Also, another way is to look at the profile : the blog linked in this profile talk a lot about AI, so it helps to identify them

@as400 Or you could install openbsd-backgrounds from ports/packages, for photos taken by OpenBSD developers, not AI slop. ​

https://github.com/marcespie/openbsd-backgrounds

I am so basic, my background has been xsetroot -solid black since about 2004. I just can't handle the visual distractions.

@nuintari

@nuintari said in #OpenBSD wallpaper collection:

xsetroot -solid black

Sorry, I'm blind! (not really, just a "pure" joke)

My "artwork": https://huc.fr.eu.org/img/EBNH/OBSD/
Visible on: https://ebnh.fr.eu.org/tags/image/

@as400 @h3artbl33d

There's no one-size-fits-all method to recognizing slop, but consider that these images all feel very similar, as if someone described what they wanted and was given several dozen images to choose from. Also notice that the repo has a single commit, where all of the image were published, with absolutely no acknowledgement for the original artists. These in themselves do not say for certain that this is AI slop, but it heavily points in that direction.

Also, if you look at the repo's author, you'll see that they're very invested in AI, which is a clear giveaway. People who are publicly pro-AI are always posting slop,

@rdh @as400

Yeah that. And especially mind the misformex texts in some of the images - a clear giveaway.

Pigeonholing

@rdh@elekk.xyz said in #OpenBSD wallpaper collection:

… People who are publicly pro-AI are always posting slop,

I oppose pigeonholing of humans in that way.

Yesterday, for example, in yet another discussion that was spoilt by sloppy interruptions from irrational anti-AI hijackers, one of the comments was automatically removed by Reddit's abuse and harassment filter:

In the screenshot, a red alert:

Potential Harassment

Very few people will see the evidence, because Reddit automatically collapses unpopular comments. Old Reddit has the option to show, not collapse. I don't know whether new Reddit has the same flexibility.

Critically:

• people who are publicly anti-AI do not always write sloppily
• people who are publicly anti-AI do not always hijack other people's discussions
• people who are publicly anti-AI do not always abuse, or harass, when they don't like another person's point of view.

Overgeneralisations about individuals, or groups, when writing about subjects that are already highly contentious and not properly understood, can be unexpectedly divisive.

Jeebus so much crap - rehashed in the same style over and over again.

So this is what creative output will come down to.

I think I'll keep sticking to Breughel's "Triumph of death" for the time being, as my background wallpaper.

It gives me an eerie sense of comfort, and the skeletons are actually quite cute.

In this case, the repo maintainer was kind enough to tell you what AI was used in each file. Look at the file names. Most of them are DallE.