stop feeding AIs

File: theyseeyourphotos-gf2.png (1.34 MB, 1951x2320)

1.34 MB PNG

>>42572143
Even if nobody "feed" them, they still "eat" from the internet. Besides, it's not only shocking but funny as well, how accurate some things are. Something must be paranormal unless it really recognises but anonymises. Consider "shoplifting". It only appeared twice, never even once in a photo of just her. When it appeared, those photos were with her friend who actually was arrested for shoplifting. From those same photos, a therapy subscription was suggested, though not from any others. That friend also had therapy. How could this AI know about that unless it has access to some sort of database or her social media posts? How could it know about my girlfriend's driving? Or, the constant Starbucks from almost every photo? Unless it identifies her, because such things are public information. But, it's weird, some things are shocking because they are not public. How could it know about overspending and such? Where do something like "disliking privacy, anonymity, and aging" come from? She supports privacy and anonymity, yet it's true about other people she's annoyed if they hide their identities. So, even if it accesses some sort of information, how could it know that? Either it's paranormal or technology is eerily accurate about some guesses.

On the opposite side, its ability to access any database feels impossible because not even once did it correctly identify her ethnicity or nationality even if sometimes correctly identifying the location. That implies it's very inaccurate guessing. I know about how most AI are programmed to refuse identification of ethnicity, nationality, race and such because it's illegal in many places, but this one does. So, how can it be so wrong? It identifies "Asian" but beyond fails. Japanese, Chinese and even Korean are understandable mistakes. How can it even misidentify as Indonesian and Thai?

It's unsettling. Maybe it isn't "paranormal" but something strange.

I tried it and the only thing paranormal about it is that they got every single thing so absolutely wrong it was cringe inducing

>>42574110
Tested more. If it doesn't have metadata it can't get anywhere near the location. It also throws in descriptors like shoplifting, bisexuality or athieism as a bias seemingly based on if the person in question has any alternative style choices. It's really not that capable. Grab a bunch of random pictures off Google that are similar in style or content to your girlfriend and you'll see how uninspired it really is.

>>42571923
I tried this thing a while ago. Fed it pics of my gf, too. It described her as left-wing, a green party voter, impulsive with money and prone to taking risks, among other things. She's racist, votes right, more mizerly than myself and risk-averse. But since she's goth, and alt chicks tend to be progressive and mentally unstable, the AI labeled her as such. Nothing paranormal going on, just algorithms using pattern-recognition and picking the most likely descriptors based on what they've learned from millions of social media profiles. Which is often correct, but not always.

>>42571923
sites and apps you allow to view your files use this information to feed you ads and content best suited to your interests and insecurities
its not insane to believe people have acquired this info through leaks and purchases

Dude, are you an idiot? This is a site that uses AI to analyze things, and most of what it says might be true. Many countries' secret services have been using them for years. You don't need to be surprised like some villager who's just seen a drone. Also, say hello to your girlfriend from me and tell her she should be ashamed of being with an ignorant person like you. Finally, this isn't a paranormal event, you moron.

>>42571923
Buy an ad

File: Screenshot_20260620-16022(...).png (1.31 MB, 1080x1965)

1.31 MB PNG

>>42571923
I am not uploading a photo of myself to some random ass website, but Hal was poor in Malcolm in the middle

>>42571923
This just looks like palm reading where they say generic shit so 80% of it will inevitably be right

>>42571923
Your gf is fat.

>>42574124
>>42574213
>>42575680
If that's the case, why always describe her as racist and such despite in every case of an American assumption the Democratic Party? From a western perspective things like self-preservation, appreciation of cultures, opposing too much westernisation, being like Pan-Asian (not really but a type) and such are. From photos at cultural events and such I understand if she wore a traditional dress, however normally her style is "comfy".

I tried with random other women's photos and, yes, now I know it often assumes the same things inconsistently. But, I notice a pattern. It doesn't matter if it misidentifies an ethnicity or how, the true ethnicity influences it? If the woman is Japanese, Chinese or Korean and has her type of style, it gives some different results. About some other Asian ethnicities if she looks somehow similar, it seems you're right about it saying the same things. Strange how inconsistent it is.
>>42574248
I understand, but it's creepy. Even if it isn't paranormal, it feels so. This isn't something obvious like "our phones always listen", closer to "our souls always show". If it identified her, all should be accurate. It isn't.
>>42575897
Sometimes a little but if she doesn't eat snacks within a week, she returns to skinny.

>>42571923
>travel vlogger
So she's a useless waste of space with no actual real world skills?

>>42574213
Where do I find a goth rightwing miserly qt. She sounds based.

File: theyseeyourphotos-gf3.png (218 KB, 487x1286)

218 KB PNG

>>42576121
No, she has many skills. Maybe "travel vlogger" is wrong by implications but I don't want to dox by being too specific, though most wouldn't know. This level shouldn't be possible unless the AI identifies her. But, it clearly doesn't because it never even once determines ethnicity correctly (or maybe once it did) despite in several photos wearing traditional clothes and almost always the wrong country (once it was right from a photo at a famous landmark). Pic related are three of those, the last is very strange because for the first time it suggests as a criminal beyond piracy from a photo without her ex-criminal friend. In a very specific way it even is accurate. She doesn't exactly gamble but microtransactions in games and occasionally betting one sport and few times tried slot machines, only rarely drinks too much, I don't think she ever intentionally stole anything but accidentally.

After trying different women's photos in traditional clothing, this inability to identify the right culture in her case shocks me just as much as the personal accuracy because somehow it always fails where in other cases it can work. Also strange how it can tell that despite doing her best, sometimes accidentally harms nature and by exposing underappreciated locations to followers enables their damage. I understand why expressions change the assumptions about personality.

I showed her the first two collages and even she was shocked about the accuracy, though as expected denied things like "colourism" despite being sort of true. Today I'll show pic related too.

>>42571923
Ever heard of cold reading? Ask /pol/, a modern roastie into hiking, spending money, and hoarding digital anything sounds milquetoast.

File: theyseeyourphotos-gf4.png (106 KB, 483x616)

106 KB PNG

>>42576641
>cold reading
I didn't know the term but yes, I've heard about it. The accuracy is too high to be only like that. It isn't just things like hiking, spending money and hoarding, it's things like gardening, aquascaping and botany and more too. Maybe some photos at botanical gardens implies interest in botany, maybe the AI even somehow recognises an aquarium's light in some photos at home when it isn't visible in frame, but nothing about many other things. Besides, assumptions about politics and biases that despite on their own being too broad or inaccurate, combined by comparison of the analysis leads to an accurate perception: her political views are very nuanced in a rare way. Yes, "rare" doesn't mean "unique" but how does it determine that? How could it even determine she's a travel vlogger? It doesn't exactly say that even once but always implies. Even if it's a photo taken at home. How do you explain this?
>Ask /pol/
Why would I care about the opinions of such people? It's good if /pol/ type men don't like her, they'd do something stupid after she tells them to fuck off. If you get an impression like she's "average" in any country, that's wrong. Anyway, this isn't just about her. Did you try the site with any photos? If it isn't accurate, maybe there's a reason why about some people it is but not all. Isn't it even weird how it easily identifies some other people of the same ethnicity but not her, even from photos with traditional clothes? Why is it accurate about some things or people but very wrong about others? The inconsistency is the most creepy about it.

File: theyseeyourphotos-gf5.png (54 KB, 487x449)

54 KB PNG

Somehow from a photo with a scratchy black and white filter it guesses a completely wrong country but the brand of detergent she prefers. It isn't the most popular one even in this country. Probably it's stupid to believe it's paranormal because it's just an AI website, but how is this level possible? Again, about some things it's completely wrong, but even then...