ai image detector website that seems pretty decent in my experience:
All of the details look pretty consistent tho, idk how else you would make it for such a quick joke but i’ve never seen an AI image get so many small details absolutely identical before
I would guess they’re just really good at photoshop and have a lot of time to throw away but I don’t really know
ok nvm i found a higher quality uncropped version and now it looks a lot more ai both to me and the website
I don’t hate AI, I’m mostly concerned about it’s impact to artists. This is an ai detector, so it’s probably beneficial to artists. Ofc energy usage and whatnot is a concern too, but 2 inferences that this thing is doing is probably barely any energy usage compared to like toasting a piece of bread since it only takes a second or two to provide results
When all this tech was new I was following it very closely actually and installing a lot of local image models, but now that things have settled it seems pretty straightforwardly net negative outside of maybe DLSS. (also, the last big improvement in local image models was like a year ago, so it’s not really interesting to follow anymore)
ai image detector website that seems pretty decent in my experience:
All of the details look pretty consistent tho, idk how else you would make it for such a quick joke but i’ve never seen an AI image get so many small details absolutely identical before
I would guess they’re just really good at photoshop and have a lot of time to throw away but I don’t really know
ok nvm i found a higher quality uncropped version and now it looks a lot more ai both to me and the website
Using AI to detect if an image is AI because you hate AI is a special breed of hypocrisy.
I don’t hate AI, I’m mostly concerned about it’s impact to artists. This is an ai detector, so it’s probably beneficial to artists. Ofc energy usage and whatnot is a concern too, but 2 inferences that this thing is doing is probably barely any energy usage compared to like toasting a piece of bread since it only takes a second or two to provide results
When all this tech was new I was following it very closely actually and installing a lot of local image models, but now that things have settled it seems pretty straightforwardly net negative outside of maybe DLSS. (also, the last big improvement in local image models was like a year ago, so it’s not really interesting to follow anymore)
I don’t know how AI detectors work, seems like it would just be the more classical machine learning classification approach, but maybe not.
All AI is just rebranded machine learning, is it not?
Being able to reasonably generate understandable language is an achievement. Trying to pretend it’s can reason because of that is a mistake.
dis i agree with-
it’s so obviously ai to me without tools but hey glad you came to the same conclusion via ur own methods
It gives ai vibes, but I can’t really point out anything specific that’s off in the low quality version
And it is definitely possible to make basically the same image from photoshop