The problem isn’t the use of AI. It’s the lack of editorial effort to use AI well.
Modern image generation models can handle text fine. Or the author could have left those artefacts blank and added the text themselves in “post production”
The problem is the use of AI. It’s a reliable indicator that the author doesn’t actually care about the quality of the work, so I shouldn’t bother to read the text.
Stock photos are (generally) made by professionals, memes impart some social context. Can't say either about an AI image. So i absolutely hold an AI image in less esteem, especially when someone clearly just used the first draft spit out by the model where all of the text is wrong and the image doesn't even resemble what it purports to illustrate.
> Stock photos are (generally) made by professionals
GenAI models are made by professionals. But in both our cases, it’s a different set of professionals to the ones who chose which images to use in their articles.
> memes impart some social context
The “social context” here is just regurgitating an unrelated viral joke. I’d hardly say that requires more respect of the readers than a purpose generated image that is intended specifically for context of that article.
Plus you can use GenAI to generate a meme too if you really wanted.
> Can't say either about an AI image
I literally just did ;)
> So i absolutely hold an AI image in less esteem, especially when someone clearly just used the first draft spit out by the model where all of the text is wrong and the image doesn't even resemble what it purports to illustrate.
And this is exactly why I said “The problem isn’t the use of AI. It’s the lack of editorial effort to use AI well.”
What you’re complaining about isn’t how the image came to be, it’s the shoddy workmanship. And that’s a different problem.
Lazy people will choose images lazily regardless of the tech. The problem here isn’t the use of AI, it’s the lack of editorial care.
Or to put it another way, if the author used a bad stock photo or offensive meme, then we wouldnt be blaming the tech. But when they use GenAI to produce the same shoddy output we do blame the tech. That screams like our own biases affecting who we blame the for crappy image rather than an inherent problem with the technology.
And if we then acknowledge that the problem isn’t inherent in the tech, then we are back to agreeing with my original point.
No if they used a bad stock photo it wouldn't be worth noticing and if it were an offensive meme it would have been chosen with intent. A bad AI image is so bad to be worth noticing and serves no other purpose than to fill up space hoping nobody notices how bad it is (which, yes, is what stock image are for, but those are made by IMAGE MAKING PROFESSIONALS so even the bad ones are inoffensive. Doctors and mechanics are both professionals so you'll get Dialysis from a Jiffy Lube?)
> No if they used a bad stock photo it wouldn't be worth noticing
Surely that depends on how bad the stock photo is? I’ve seen plenty of people comment on bad stock photos in the past. And plenty more moan about the use of memes in articles.
> and if it were an offensive meme it would have been chosen with intent.
…which is better why? It’s still offensive. Except now you saying the author being a dick is better than an author failing to use a decent image generation model?
Personally I’d rather scroll past a crappy image than an offensive one.
> A bad AI image is so bad to be worth noticing and serves no other purpose than to fill up space hoping nobody notices how bad it is
Exactly. Bad images. BAD!!
You’re comparing good stock photos with bad AI images and arguing that AI is the problem. And thus we are right back to my original point where I said it’s the editorial process that failed by allowing a bad image, regardless of the source.
> Doctors and mechanics are both professionals so you'll get Dialysis from a Jiffy Lube?)
Doctors don’t build dialysis machines so your analogy is wrong.
Your whole argument here ignores the fact that gen ai has been used in reporting for a while now; and used well.
This image in this specific article is crappy. But it’s also not representative of all AI images in all articles. Its crappiness is an outlier rather than an example that proves your argument about the ethics of AI generated images.
Modern image generation models can handle text fine. Or the author could have left those artefacts blank and added the text themselves in “post production”