@alisonw correct

I get the good intentions of the encyclopedia diagrams showing the different software packages, but I do not think it works at all.

@evan I'm with you on this one.

@evan it doesn't, just thought it sounded evocative, hah

I didn't answer this one. I usually do people as nodes, with some kind of boundary to indicate instances. Thanks all!

@alterelefant @anemone this is a person who's not only drunk but also on a videocall while driving. Only if they were also wearing roller skates could they do a worse job driving

@kkremitzki how does the rule apply here?

Regardless, it seems from the replies that most people answered the question as asked.

When I see AI-generated images on the Fediverse, it's one of two things. Either people seem to genuinely believe the photo is real, like with some fake deep-sea creatures a few months back. Or, the image is newsworthy on its own, like the Trump Jesus thing.

If I estimate, I think this happens maybe once or twice a week, so 1 out of 100 or maybe 200 images.

So, I'd say 0-25%.

Thanks everyone!

Second, because modern cameras use a lot of AI features. Lighting, autofocus, deblurring, all that processing.

Thank you everyone.

I came up with two reasons that this question could be misleading.

First, because I'm not sure it's fair to say that an LLM-generated image is a "photograph" in any meaningful way. So, maybe no photos are LLM-generated.