Ok, ok. Title is slightly click bait but you gotta do what you gotta do to get those juicy upvotes more attention 😂
So, this really cool scientific paper just got published showcasing how much AI has progressed lately in generating realistic photos of people. It's so good that AI generated faces look more real to humans than actual human faces !
At least when it comes to white people. I know..Racist!
Recent evidence shows that AI-generated faces are now indistinguishable from human faces. However, algorithms are trained disproportionately on White faces, and thus White AI faces may appear especially realistic. White AI faces are judged as human more often than actual human faces—a phenomenon we term AI hyperrealism.
And surprise surprise... The people that were most confident in their judgement...Were the ones that made the most mistakes 😅
Paradoxically, people who made the most errors in this task were the most confident (a Dunning-Kruger effect)
Fucking Kruger. Always there to bite us in the ass.
Here's an example from one of the experiments:
Would you pass? I know I'd have a really hard time. Those AI photos do look realistic as fuck, with all the right shadows, blurs etc !
What about you?
But why was I people are tricked to think AI faces are more real than...real faces?
Well, to me it's because they look more..average?
Guess what
Critically, in line with our face-space theory prediction that AI faces would be more average than human ones, AI faces were significantly more average (less distinctive), familiar, and attractive, and less memorable than human faces.
One of the authors further explains
“It turns out that there are still physical differences between AI and human faces, but people tend to misinterpret them. For example, white AI faces tend to be more in-proportion and people mistake this as a sign of humanness,” Dr Dawel said.
Anyways, I just skimmed most of the paper and it was quite interesting, definitely worth to check it out.
This only makes wonder how long it will take till a quick prompt will be able to generate hyper realistic deep fakes with audio and everything... The implications of that are super scary! Propaganda of all kinds, identity theft, bulling, scams.. The sky is the limit!
Anyways, the paper is free to access if you want to give it a read. And if you are too lazy to do so...You can always ask Chatgpt for a quick summary, LMAO 🤣😆
References & Further Reading
Posted Using InLeo Alpha
