genZ: a note from genX(we survived the uncanny valley of polar express) dont tell them HOW you can tell. theyll just take it as notes to try again. stay disillusioned and let them keep failing. you owe them nothing. no enemy more vicious than apathy and anyone plaigerizing deserves as much. this is what drives our nihilism.
Rick Beato mentioned that his teenaged son is able to hear AI-generated music with 100% accuracy. I (dude in his late 50s) can’t do the same. It’s weird because I write software synthesis applications and I’ve always been able to tell (with 100% accuracy) the difference between compressed and uncompressed audio (even when it’s “lossless” compression), but whatever it is that kids are hearing with AI, it’s not that.
Excuse me good sir, are you saying you can tell FLAC and WAV apart?
I’m calling bullshit on compressed audio claim. They produce the same sound by definition
I can read the difference between a pure uncompressed .txt file, and a nasty compressed .txt file that was inside a (yuck!) .zip folder
I’ve always been able to tell (with 100% accuracy) the difference between compressed and uncompressed audio (even when it’s “lossless” compression)
???
The data coming out of your DAC is identical with lossless compression
Maybe their computer just lags hard enough to skip samples whenever decompressing.
Or he does what most audiophiles do and look at the label/bitrate/price tag and say “Hmm, yes, that is of far higher quality”. Not saying that differences don’t exist, but I’ve noticed audio tends to be especially rife with people who claim to be able to hear differences where they should not be humanly perceptible and will fail blind tests time and time again.
I’ve taught my gen alpha kids to accurately identify slop. I love watching corridor crew debunking videos with them.
Lol “Gen z”.
We all have eyes, don’t we?
The problem is, that we’re already at a point where some images require experienced eyes to be spotted. Someone working in graphics can tell, someone who doesn’t will have more trouble to differentiate. At it’ll get worse, because you just know, they’ll double down on improving the AI up until a point where fiction and reality becomes indistinguishable when it comes to photos.
Most older people I know IRL can’t spot generated images, my Gen Z coworkers always can.
Thank got they don’t have a product created for intelligent people. Revenue would drop by 100%.
I probably can’t but skip it anyway. Even if it’s real, it’s normally some cherry-picked race-bait bullshit.
You can tell when it just looks too perfect or uniform imo
And the lighting is just a bit too vibrant, the colors have just a bit too much of the Lisa Frank HDR look…
yeah, real artists express their emotion, which is depression
The problem is that as they learn how people can tell the difference, they’ll improve the models.
“The 600 series had rubber skin. We spotted them easy, but these are new. They look human… sweat, bad breath, everything. Very hard to spot. I had to wait till he moved on you before I could zero him.” -Kyle Reese
This statment assumes human’s learning capability is slower or static compared to AI.
I think there will be a point where it will be almost impossible to tell the difference, I already fell for some pictures only for the comments told me to check the face of someone in the background, it is a matter of time, until AI improves, so those details are fixed.
And I am likely blind to many cases where I fell for it and did not see the comments.
And on top of that, imagine a real picture with someone in the background is exceptionally ugly, or has some deformity, and suddenly that picture is no longer deemed real.
I am paranoid that we now live in a world where “Pics or didn’t happen” is a thing of the past. States already push AI propaganda, objective reality is a thing of the past.
They can’t improve the models in that way. They are non-determainistic and don’t listen to instructions. They can tinker with training data but straight up forcing a model to work a certain way take all of the magic away. Without the magic, it would be worse and everything would look more uniform.
Yup. This kind of feedback is super easy to scrape & condense with LLMs and feed back into training.
There is absolutely no way we are winning this war. It is just a matter of time, unless datacentres are blocked but even then, it will only slow it down. The pandora box of generative AI has been opened and there is no way of undoing that.
Yes. Nobody has ever stopped the march of technological progress, even if as now it is a detriment to society and people’s mental health. It is unfortunately a mathematical inevitibility.
I’m not sure how survivorship bias applies here?
You are aware only of those generated images that you did recognize. It’s not like every image on the internet comes with disclamer for you to compare with and calculate your hit and miss percentage.
I guess it applies more to some comments than original post though.
Guy is aware of his engagement falling 40% ,thats not survivorship bias LOL Its not a “oh these planescomewith holes here” problem, but a “sir we are losing almost half our planes”.
“Im seeing a lot less AI generated images on the internet lately”
Wait…
[Insert survivorship bias here]
whenever I see a company using AI art for their campaigns, I automatically ignore that company
First, GenZ haven’t been “kids” for quite some time now, granpa lol. Its Gen Alpha, and even then they’re getting older as well.
Second, GenZ and subsequent ones are completely disassosiated from the commercial world. A huge chunk of them don’t like “commercial perfect”, pre-baked vibes. That shows with everything: the garbagy/shitposty anti-mainstream meme culture, the “ugly” haircuts, minimally branded clothing (only area where this doesn’t apply much is in music, since its kinda the other way around with how “mainstreamy” most music has become).
It’s pretty normal to refer to twenty-somethings as kids if you’re old.
Yeah, the older most people get the more broad their definition of “kid” becomes. I’m almost 40 but I’d wager someone north of 60 would still consider me a kid, relative to their own perspective.
Am in my 20s, and still feel like we’re kids. We just happen to pay taxes now…
Okay, I wanted to know if I could do this and had a look if I could find a game or something like that. Turns out, yes, I could find one. And yes, I can see it. I got 19 of 20 right on medium difficulty for a start. I’ll try some different combinations but it surprised me how fast I was.
I’m definitely not gen z
Edit: and the site is a wild mix of languages for me. But it’ll do.
deleted by creator
On easy mode. Damn, just the accuracy of random clicking. I hate this BS, AI is indistinguishable nowadays. Luckily i already assume everything is faked.
Though that makes me wonder. How much attention are people giving to random images regularly?
Like within the 20sec window on easy mode, i couldn’t really find telltale signs, excluding one pic with an anomaly. Maybe with few minutes i could find some signs, but that’s already beyond the the attention and effort that a random picture deserves.
And like the gen X in this thread said. Ad’s should be already automatically blocked, giving an AD enough attention to distinguish whatever it’s AI or human made, is already giving it too much attention. The AD already won then.
This is fascinating to me. Like, I got 90% every time, and I wasn’t really analyzing at all, just going by feel. Most of the time I answered within the first 2 or 3 seconds. And I’m not saying that to brag because I genuinely have no idea what we were doing differently. Like if you asked me to teach you I’d just be vaguely gesturing and shrugging helplessly. It’s just vibes.
I want to see someone do a study on this. Why is it that some people instantly discern most AI images as feeling wrong somehow and other people don’t? What is the common factor here?
I do agree that it’s fascinating and I’m rather curious why is it that way. Lacking this ability seems to make me especially vulnerable to AI manipulation.
Like for me there isn’t any different vibe or feeling between AI or human made pictures. Those are all the same. I have to carful scrutiny every picture to find stereotypical AI anomalies. For example text on the background being in alien language or random lines, some object on the picture not making sense aka straight line disappearing after going behind some object (continuity) or items melting into one or the old one of too many fingers, but with image compression algorithms, the background anomalies can sometimes become indistinguishable from just background blur and anomalies are getting rarer and rarer.
I did hard portraits first and got 15/20 Then I did medium and got 17/20
What is going on with that language mix? Did they vibe code this using different libraries and not set the global language variable?
That link. They won’t let us play it unless we accept cookies and enter our personal information.
I didn’t have to create an account and could decline cookies
Lucky you. My go button was grayed out and unclickable unless I accepted cookies or give them my information so I left.
That website is vibecoded slop, it’s also using AI detection to determine if things are AI or not.
Images that are clearly AI get weird explanations that have nothing to do with the image, like I got an AI image of an owl sitting in a hole in a plaster wall, and it said it was AI because the snow was too uniform? There was no snow. Absolute slop.
After your description I’m not clicking the link, but it sounds like using that website is failing a bigger picture fake or real test.
deleted by creator
My 6 year old habitually asks me “is this AI or just Photoshop?”.
Im gen Z and AI imagery has always triggered an instunctual disgust response in me, but some are really hard to tell these days. I’m afraid we’re already starting to overcome the uncanny valley at least for those cases where people put effort into making it look real
Username does not check out
I honestly think derealization has an impact on how we view AI. Like, AI images are almost upsetting in nature for me now because it’s just so wrong looking.
I know people who cannot tell AI images from reality anymore if they’re photorealistic. These people have never experienced dissociation or derealization before. I tried to express what I saw in the images that was so fucked and they did not see it.
Really needs a source, sounds plausible but could easily be made up.
