Some of the real pictures are using filters that make them look like AI.
Yeah, one of them definitely had the fake, post-processed bokeh effect added to it that a lot of phones with “portrait mode” use. Which, to be completely fair, makes that technically an AI-generated image.
I was looking for artifacts of AI generation, and I found them, but I’m still wrong. I can’t win.
It makes it an AI-edited image, not AI-generated.
But yeah I agree, these sure seem like they were cherry-picked to fool you.
I gave up after the 2nd heavily edited photo. What’s the point if the games rigged. Are either real faces? No.
Yeah what the fuck was that about? Clearly rigged for sensationalism.
Yeah, but I found there are telltale AI signs in the fake ones.
But you’re absolutely right. They made it harder by editing the real faces to make them look fake anyway.
the real photos just looked like professional photography, that’s how your photos will look like when you hire a photographer
This is the sort of thing that I like to send to people who assure me that “all AI generated art looks wrong” or whatever.
No, the AI generated art that looks wrong is the only AI generated art that you notice. The rest slips by.
9/10.
AI sucks at reflections, so pay attention to the pupils to see if both eyes reflect similar shapes.
Also a few had odd lines where their neck was like surgically reattached to another body.
Also, pupils are often not regular circles in AI images. The only one I got wrong was the real picture of the guy wearing dirty glasses.
I think the point still holds. AI generation has gotten very very good which requires you to look for minute details most people won’t know to look for. The small issues you point out are probably easily solved if you really wanted to make them even harder to detect with a post processing model or just eventually improvements. This is just some random blog post too, so it’s unlikely they even put that much work into it. I’m sure experts will emerge that will have all those details in mind to make them even harder to detect.
I can see a few glaring mistakes, but that’s not too bad.
As always, it’s hard to determine what is AI and what is a filter. The guy whose entire face was edited to be flat and tilted 10 degrees toward the camera got me. That said, 8/10. The first two clued me into what the author was going for and I got the rest right.
Modern phone cameras are passing pretty much all photos through AI filters now to add detail or upscale, so filtered photos are more the norm than exception now, so that doesn’t really help people to filter out AI images. Point still stands that it’s already gotten extremely good and now requires recognizing tiny details. It doesn’t have to go far to get to the point where it’ll be next to impossible to tell.
Yep, same with 8/10. Real faces are less symmetrical than AI
I failed the first 6 guesses and quit. Years before this recent AI boom, there was thispersondoesnotexist and even back then, AI was generating extremely convincing faces.
This thread is being used to train AI, no?
Our words might be scrapped, but it won’t help with facial generation.
So reading why we can tell the difference won’t train something to work around it?
Not with current methods.
I think this would require something closer to an actual AI an not the pattern recognition machines we use at the moment.
Fuck, I got 50%. If others had the same result, you can guess the implications.
So I figured out the difference here, and it’s gamed to make you think they’re backward.
The real photos appear to be professionally edited with the light, filters, and bokeh being perfect while the AI generated are more like candids you’d take with your phone. This is an intentional move by the author to make this scarier than it actually is, imo. Next to each other we expect the AI to be more ‘perfect’ than the real photos. Once I figured that out I got them all right.
If they’d put candids next to these AI generated images I’m willing to bet you’d have done better.
Same, so no better than random chance
Removed by mod
And those small tiny issues can be fixed with a Lora or post processing to refine those specific issues that the general model missed.
Same. We’re as good as a coin toss… 😭
6/10. Of the ones I got wrong 3 of them seemed heavily photoshopped.
I got the same score, though I got the dark haired woman. I don’t remember the one I missed instead.
I feel like trying to identify unedited images and AI isn’t as hard, but when they’re edited it gives many of the same details that AI generated ones have. Mostly it’s the smoothing out of details. This could be just that the AI was trained on a dataset pulled from images available online (which are almost all edited to some extent) or just because the way it generates it makes details not quite clear. Probably a little of both.
5/10 on my first try. Basically we’re all just guessing. That is scary as fuck.
Does the ninth person have two different earpieces !? Common !
Got 6/10, for at least 2 pictures I looked at the outside corners of the eyes, if they match it’s likely AI, most real humans likely do not have symmetry there.
Well since it’s a bunch of real faces combined it’s not that difficult to make a realistic face.
It’s perfectly ok to admit you got 0/10. No judgement here!
That’s not how modern AI image generation works. AI no longer just mix-and-matches various assets from a library. It’s creating its own unique images based on what it knows about the shapes and colors of things its been trained on.