I think the company that owns the Eiffel Tower retains intellectual property rights over all images of it at night. So you don't see many real nighttime pictures of it.
Here's a prize for you! (That's how prizes work, right?) What I'm learning is that not only do I not know which ones are AI, but I also don't know what the Golden Gate bridge looks like. This is gonna make me second-guess my capabilities as a human? Can't wait for next episodes.
This. Also, eggs do have a small air bubble in them. Usually it's not loose and is attached to the bottom(the wider end) but sometimes it gets loose and moves to the side when boiling them causing those depressions as seen in the picture.
Love the game! kAN was also a perfect subject to kick it off with! Next time make it a gameshow with more contestants. They can lock their answers in, and then discuss on why they picked it for the content
Round 5 with the cars was really easy for me: Missaligned brake calipers =AI Picture one only has straight lines with absolutely no interuptions so four. (Also one had doubled side intakes)
You said this is the easiest it gets, referring to the following pictures were even harder. I took the meta meaning - it is only going to get harder and harder to detect AI pictures in the future.
I got almost all of them Checks back through... Round 5 I got wrong for the reason Kan got it right - I thought that there should be more than 4 lines for the aperture shape, but obviously wasn't thinking of the right camera lens/aperture. huh... was that it?
I am shocked that kAN got the "street in Sweden" wrong, there were so many signs pointing to number 4, the sidewalk/road being full of holes and being consistent in stone size and pattern, the signs, the building actually being made from brick without rebar poking through like in nr. 2, and the bike not having a weird red thing attached to it like in nr. 3.
I started out bad but then started getting almost all correct. My overall way was that some of the fakes were too crisp looking. If one wasn't as crisp that usually was it. Basically used my experience of taking photos and how it never seems to get the detail. Maybe bad logic but it worked
I got the forest image by the leafs. The leafs in the background look more like noise in the ai images. But Kan got the way better approach for this case
The ones that had reflections of actual things and not just light were pretty easy. eg - the last pic #1 had reflections of the trees outside the building on the wooden bench seat and the cabins #1 had a reflection of a tree that was out of frame in the window.
Round one is hard, because 1 2 and 4 look like they come from the same gen prompt, and there's not anything obviously fake about 3! but that dark brown seems so off
This is the best it is ever going to be, not the worst, right now the AI is still trained with training data that does not contain AI generated images.
Prize should be an AI generated picture of a trophy. But the last round is 3 real trophies, and one fake trophy and you have to pick the fake trophy to win😆
I felt car #4 was fake because of the horrible building behind it. Also last picture the table on picture #1 reflection doesn't match the background. How do we know the real pictures weren't AI to begin with.
The cat pictures were easy for me because cat whiskers signify to the cat whether their body will fit through a gap. Pictures 1, 2, and 3 have abnormally large whiskers on the sides and tops of their face, where picture 4 did not. This has been your daily cat fact. Responding STOP does nothing.
Adding another cat fact to the collection in this comment section. Cats cannot have 4 fur colors, the first three cats are black, grey, orange and white while cat #4 is just black, orange and white.
On the cats what I noticed was how numbers 1-3 the pupils on the eyes of the cats were all super contracted as if a light source was being shone directly into their eyes, however number 4 was a normal level of dilation for the lighting that was present
I need to know how kan thinks a "normal camera with six panel camera shutter" works. Also i am convinced that the building behind the car in round 3 is 99% AI.
yeah, i guess dapper intended on 15 sets of different images and accidently put 2 sets of cats and cut out the 2nd set and changed the picture number in post
The real Eiffel Tower image is actually a picture of the Tokyo Tower
I was gonna say the same. Initially I thought none of them were the real one and it was a trick round 😆
yep
oh good cus I tought no way THAT was the eiffel tower 😂
Yeah, my first thought was that 2 is definitely fake, since the Eiffel tower doesn't have that long of a spike on the top
I think the company that owns the Eiffel Tower retains intellectual property rights over all images of it at night. So you don't see many real nighttime pictures of it.
Here's a prize for you! (That's how prizes work, right?)
What I'm learning is that not only do I not know which ones are AI, but I also don't know what the Golden Gate bridge looks like. This is gonna make me second-guess my capabilities as a human? Can't wait for next episodes.
Thank you very much!!
4:06 That inconsistency between the halves of the egg is exactly why I chose 3 as real!
22:50 wait but #2 isn't the Eiffel tower???
This.
Also, eggs do have a small air bubble in them. Usually it's not loose and is attached to the bottom(the wider end) but sometimes it gets loose and moves to the side when boiling them causing those depressions as seen in the picture.
@@anteshell Yeah exactly, like often when I make hard-boiled eggs, they aren't the perfect oval "egg" shape, they're usually a little bumpy like that.
For the level of success detecting AI images, he should get Artificially Generated Money or Prizes (worth nothing in real money).
This is a great game. Make Kosmo play.
The prize can be an AI autographed NFT. 🤣
Love the game!
kAN was also a perfect subject to kick it off with!
Next time make it a gameshow with more contestants. They can lock their answers in, and then discuss on why they picked it for the content
Golden Gate Bridge #4 is a long exposure photo (the cars become blurred lines), but has clear clouds in the sky.
and on the bottum right the lines are missing a car, if you hadnt noticed yet, also i suspected ai to not show construction on the bridge
WHERE"S THE PRIZE DAPPER?!
Cool!
Round 5 with the cars was really easy for me:
Missaligned brake calipers =AI
Picture one only has straight lines with absolutely no interuptions so four. (Also one had doubled side intakes)
You said this is the easiest it gets, referring to the following pictures were even harder. I took the meta meaning - it is only going to get harder and harder to detect AI pictures in the future.
give him half a hot wheels car lol
Kan Deserves a Zeepkist Track with Cheese. Nothing more. Because he didn't win, he delivered a subpar guessing game.
What happened to 6 days of zeepkist?😂😂😂
I got almost all of them
Checks back through...
Round 5 I got wrong for the reason Kan got it right - I thought that there should be more than 4 lines for the aperture shape, but obviously wasn't thinking of the right camera lens/aperture.
huh... was that it?
I am shocked that kAN got the "street in Sweden" wrong, there were so many signs pointing to number 4, the sidewalk/road being full of holes and being consistent in stone size and pattern, the signs, the building actually being made from brick without rebar poking through like in nr. 2, and the bike not having a weird red thing attached to it like in nr. 3.
I started out bad but then started getting almost all correct. My overall way was that some of the fakes were too crisp looking. If one wasn't as crisp that usually was it. Basically used my experience of taking photos and how it never seems to get the detail. Maybe bad logic but it worked
I got the forest image by the leafs. The leafs in the background look more like noise in the ai images. But Kan got the way better approach for this case
The ones that had reflections of actual things and not just light were pretty easy. eg - the last pic #1 had reflections of the trees outside the building on the wooden bench seat and the cabins #1 had a reflection of a tree that was out of frame in the window.
Round one is hard, because 1 2 and 4 look like they come from the same gen prompt, and there's not anything obviously fake about 3! but that dark brown seems so off
Aww I got 700pts... I did well on the first 4 then only got a few from there on.
In my opinion the fact that AI can't spell or read it's more of a feature than a bug. That is the separating line.
Very good video, would love to see more like this.
Yeah
if we all guess less than 50% than ai is now good
Send him half a toy car😂
It's called the Golden Gate Bridge because it spans the Golden Gate Strait
Well, I tried to play along, and I did worse than Kan 😅
It’s Called a soft boiled egg
Kan at the end is going a bit crazy there
Early for this noicee
Day 3 of saying cheese: cheese (no cheese today😢)
Is it just me or are the real images a tiny bit smaller than the ai generated images
The prize should be CHEESE!
#4 of the cats is edited and done poorly. This makes it look AI generated.
This is the best it is ever going to be, not the worst, right now the AI is still trained with training data that does not contain AI generated images.
I beat Kan by 2. Sorry Kan.
16th
WHERES THE CHEESE DAPPER
I think its in the salads
@@SullySadface Or in a stand at the tower
I think the prize should be an image where you asked the AI to make a self-portrait.
Friends? Isn't it just Kan?
Prize should be an AI generated picture of a trophy. But the last round is 3 real trophies, and one fake trophy and you have to pick the fake trophy to win😆
Kan should get a cheese wheel for participating
I felt car #4 was fake because of the horrible building behind it. Also last picture the table on picture #1 reflection doesn't match the background. How do we know the real pictures weren't AI to begin with.
The cat pictures were easy for me because cat whiskers signify to the cat whether their body will fit through a gap. Pictures 1, 2, and 3 have abnormally large whiskers on the sides and tops of their face, where picture 4 did not. This has been your daily cat fact. Responding STOP does nothing.
Adding another cat fact to the collection in this comment section.
Cats cannot have 4 fur colors, the first three cats are black, grey, orange and white while cat #4 is just black, orange and white.
On the cats what I noticed was how numbers 1-3 the pupils on the eyes of the cats were all super contracted as if a light source was being shone directly into their eyes, however number 4 was a normal level of dilation for the lighting that was present
as a person living in france i feel offended but am also laughing so hard at the fact that dapper said the tokyo tower was the real Eiffel tower
The prize should be a physical framed picture. Get him the fastest correct answer or the worst answer.
Is it me or do the fake AI pictures give off "Almost real but I'm dreaming" vibes
I need to know how kan thinks a "normal camera with six panel camera shutter" works. Also i am convinced that the building behind the car in round 3 is 99% AI.
It also doesnt help that some digital cameras are shutterless
building looks funky indeed
@kANGaming "F stands for effort"
I'm surprised by how hard this was I only got 7
Imagine this will be the new reCaptcha xD
The second you mentioned color tone it was gg.
The cheese is they all AI
I scored 700 :D
Cool
Dapper! I'd love to use these photos for a team building game. Do you have the images in a zip file of some sort?
Love you❤❤❤❤
Caesar dressing has anchovy in it.
anyone else notice the 2 frames of different AI cats at 12:03
That's way weird, it took a bit of finessing to get it in freeze frame. Wonder why that happened?
yeah, i guess dapper intended on 15 sets of different images and accidently put 2 sets of cats and cut out the 2nd set and changed the picture number in post