Agreed. The saddest part isn't the art theft, it's that we're being conditioned to hate really cool aspects or styles of art because the people who fed the machine chose to feed it with the coolest styles and techniques they could find.
@HuntingSunder exactly. I hate that impressive rendering and time consuming details don't hit the same because my brain immediately goes "that's likely ai". My mind is poisoned by this shit
@@itsmeloart I mean, why hate it if you can't tell the difference ? It seems to me that people who know and care about those artstyle are much more able to tell. You'll probably be able to tell if you educate yourself about those artstyle and what make them good and if you get fooled you won't even be mad anymore but impressed ("damn, the IA even got that point correct and the consistency is there..."). Personally, looking the video made me grow an apreciation for abstract art where I am somehow able to tell despite the nonsensical aspect of it.
@@startnetworkbabinski3036 people spend literal decades of their life crafting their art style just to have it ripped off by ai within seconds. hating it is understandable, it's not something to be impressed by. its sad honestly
There's two things for me that are really frustrating about this that I just realized. One is that what we often recognize as "typical AI art" was at some point very much certain peoples' style (although it was different, better composed, less in-your-face etc.) Feels like AI is taking the spotlight away from these artists and gives it a bad rep. And the other is that inconsistencies are now considered "AI-like" when they could have been deliberate. A park near where I grew up had a statue of geese walking and one of them randomly had a human foot. You had to look twice to actually notice it. That was evidently very intentional and maybe even had a deeper meaning. If you just saw a photograph of this statue now, you'd think this is the AI getting confused, totally invalidating what the artist was going for. That sucks so much...
This! So much, everytime a human image was actually ai and the giveaway was " it being melty " I always felt like it is somewhat of an artistic choice. I probably just really suck at spotting mistakes but yeah, it really sucks that some of this more free-form or Exposed/Edited type of styling is getting a bad reputation for no reason.
Yeah someone with a glossy painting style would be @_stato_ though his paintings are way more detailed and sophisticated and less "in your face" as you describe but it is sad that there's a chance people might consider it AI just because it's stylistically similar.
I sometimes used to use an art style that would be reminiscent of certain AIs, and I noticed how more and more people started saying it was made by AI or AI inspired etc, when I've been doing it for years and it was just a style that came natural to me, but since then I'm much more considerate of not going close to things people might associate with AI imagery, as there is really a noticeable drop in appreciation any time an artwork is in that same field. Lots of people came to either think all such artstyles are done by AI, or just became so desensitized to it that it no longer sparks the same interest at all.
that's a fair observation. There are plenty people who understand this, and fewer who had seen this coming years ago. I fear that even if we do get to make genuine works of art, their inherent value will be that much lost on us due to technological advancement - watered down to put it in other words. I suspect that a decade or two from now on, once this becomes a rooted standard, hardly anything will surprise us in terms of art found in the digital world and real world (generated media used to make low-tier merchandise). There will surely be a need for art - as there had always been to an extent - but it won't be as thrilling anymore knowing that anyone can do it without much effort and knowledge. An item of luxury turns into a common good which seems fine on the outside, but becomes nothing special when you really think about it.
Yeah, some say AI will help people who have trouble learning art, but up and coming artists are exactly who will get screwed over the most. If you are a well-known amazing auteur, you already have an audience, people already know you don't use AI if they prefer that, and your art is probably just better than AI. As a new artist, you have effectively no way to get your stuff out to people.
Except that it's common for people to prompt to exclude hands entirely for the (now old) models which frequently muck them up. Funnily enough, it never occurred to me that humans would do the same and avoid drawing hands. I thought for sure a human would draw part of the upper hand rather than hiding it all behind the hat.
I have a big feeling those were basically painted over actual pictures. (like basically give the ai a picture as reference and make them paint that reference in that style) So usual tells like composition and environmental inconsistencies and such were hard to spot
I highly recommend looking for how many strokes there are: Impressionists outdoors worked VERY quickly back in their day because they had mere minutes before the vista they were trying to depict came and went. Pictures from that time will have brush strokes that are quick, rough and DELIBERATE, think like 10-25 minutes of time to work. A lot of AI art is unnecessarily refined, drowning the canvas in unnecessary strokes, taking away from the core spirit of that time period... However, this isn't without fault as contemporary artists taking inspiration from old Impressionists and adding refining strokes in the process will get caught in the crossfire. There's no winning.
Earth 5020: This artwork is worth 700B dollars! It's 100% made by a human hand! Artwork: *Poorly made stick figure family and sun drawn in the corner of the paper*
Having messed around with generative AI locally on my computer, 90% of it also sucks really bad, you just see the 10% (maybe even less) being shared online, that being said you can clearly see whether something that sucks was made by AI or a Human, and I much prefer to see something that sucks that was made by a human that put effort and actually tried than some garbage that was spit out in 3 seconds (sometimes less) by some algorithm.
@@SergioEduP shit that sucks won't be selling. stuff that's good won't be selling also because AI can do it cheaper. art will become a thing that only people that are SUPER INTO IT, will actually buy it from real artists but most people will just generate stuff and change whatever they need. it sucks now, of course... but you might have used a bad software or you didn't pay for it etc. and there are many different things. ChatGPT was working perfectly for me though, I have paid for it and the images i was getting, some of them, were very good... even as inspiration it's perfect. But one day it will become so unrecognizable that people that spent 90% of their life doing art with a vision of making a bag... nah man, that will not happen anymore... most of the stuff will be AI and people rather save money than spend 10x more to have it made by a human... (i mean, of course there will be people that will buy REAL art but the audience will be significantly smaller.)
@jaystergg It'll be like Hollywood Blockbusters; popularity without concern for originality, only profit. And the corporations are only saving money so they can fire skilled workers. It's what they've done for centuries in every single country. It's easier to do something from scratch than change a finished result that's only one layer but if most people don't care and are content with the cookie cutter results for cheap, this is the future. Eventually their money will dwindle because less of it is in circulation; it's being spread across fewer industries and companies. The growing population of homeless people will be criminalized for being homeless and sent into private prisons to become extremely cheap or slay ve labor. The working class that's left will be in the reduced service industry or live in factory villages or on farms and be unable to move. Others all over the world will also be forced into internment camps to produce raw or finished goods, which will be enforced by labeling these people as terrorists or war criminals. The only profitable market for human art will be for the people who are lucky enough to be born into a wealthy family and able to make art for other wealthy families for millions (actual art supplies, including programs and drawing tablets, will be reduced to 2-3 companies and very few will be produced.) Imprisoned homeless artists who are allowed to make art with free art programs or bulk goods because they're sickly or they're forced to produce this work that the prisons can sell for most of the profit; most likely to be fed into the Gen AI machines (there will be a lot of people forced into strip mining as well.) The same thing applies to every other creative hobby/job; the machines have to be fed for anything new to come out and occasionally if a wealthy person has passion then they'll shake up what's popular significantly. If some poor person is extremely lucky, their vision is extremely desirable for it's freshness, maybe they'll be retrieved as a personal art servant and kept in the big house. There won't be as many people in general but fewer are needed when the ruling class has machines boosting efficiency. Most people and their information are already the products being sold online. Most scams exist because people are finding it harder to survive or thrive and there aren't reasonable fall-backs to support them. In particular, everyone gets exploited in the current system as it is because that's all there's room for. The military will mostly have drones in countries that can afford them. The countries who can't afford them will be colonized: there will be a new age of empires. If a laughable nod to PR and guilt reigns, they won't call them that but this is where fasc ism, national ism, and other prejudices insist the wealthy are better than everyone else, regardless of which country they live in. Poor people will be deemed poor because they aren't skilled, intelligent, moral, or interesting enough, not because they could be all of those things and forced into this position by bad luck or social standards imposed on them by the 1% in power. Worse yet, they may be considered less human and therefore disposable. The higher ups don't need to interact with them so why would they care? And there will be a lot of homeless people because of all the wars across the world and because landlords make more money when they have a bunch of houses sitting empty for years. Overtime a lot of the landlords will also vanish and their property consolidated under companies owned by the wealthiest families. The Wealth Gap will increase exponentially. This is what happens when there isn't a Universal Income, and the wealthy have no empathy for those outside their social circle. As it stands, most consumer goods are already white label. Only a few factories across the world actually make what we buy, even for the big name brands that have existed for over a century (and this has been true for decades, research backs this up.) That's why there are so many products that seem so similar on market sites; it's just becoming more evident. Variety will diminish when fewer people need fewer unique goods and the mega-conglomerations are able to wreck their competition and consolidate further. And all of the Gen-AI companies will as well. This is the trajectory of that future. Or it could all be diverted if China finally accepts that the Mandate of Heaven has been revoked once again. We might have to look to Vietnam or some other country first though IDK.
@@jaystergg I dont want money, im an artist. If i would only want money, i would work normal jobs. I only need enough money to be sustainable. Stop looking at things with that type of mindset. What is the point, people will never be happy. unless they can do something they enjoy. AI can NOT do it cheaper, pay a subscription or get those tokens. Oh you can do it for free? Oh it sucks, oh you can do it locally? Well time to get that new RTX Graphic card. I know how this space opperates. There are so many cheaper ways to make any kind of art. Physical or If you want to do digital drawing, you can even use free programms. (I hope i didnt missunderstand you and this was just a rant about, the future. But you seem to imply things...)
@@jaystergg Except AI art can't reach that top percentile that sells the most. The guweiz, the Shal.E, the Nixeuu, The _stato_, The riot artists, the Fortische artists. So no, Ai wont be able to replace them. But yes, for the average insta artist it will make comissions slightly harder to find but most people who comission already are into supporting creatives and locals regardless so it won't affect them as much either. I think it's better to have a more optimistic outlook on this.
Without watching the video, to answer the question: at this current point in time, if someone uploads an especially high resolution image, if you zoom in, none of the “strokes” makes sense. It looks good from a distance, but doesn’t hold up to scrutiny
The next biggest give away for me is things disappearing. Like a strand of hair going behind the arm and not showing up on the other side, because AI does not remember that the hair was supposed to be there. Or cutting off clouds suddenly because it crossed behind a tree etc.
what about people that "clean up/fix" the ai hands or whatever on photoshop and post it on twitter (formerly X)? sometimes the only way I can be certain it is AI is because there is an inhuman amount of drawings posted per week, like more than new 1 fully rendered drawing per day with background and everything? nah, red flag
Unfortunately couldn't submit the form because it was closed but it was fun to try. My main trick for the impressionist ones was the see the direction of the strokes around each object. E.g. Around the tree trunks normally in impressionism is wet-on-wet so to avoid the colours blending the strokes would draw lines around them. In AI images there would be strokes going perpendicular (into) the objects where it wouldn't make sense to paint that way.
Back in 2019, I asked an artist if a print was done with oil painting upon which she then told me it was all digital. I was incredibly impressed with her skills, even though I wouldn't say that I have an artist eye. She knew the fundamentals. And that's the difference between artists and AI-users. Artists are honest about the process and AI users want more recognition for less effort.
dont worry, it will be a thing of a past in couple of years... you will be amazed that something is AI made and not human... you won't be able to recognize that at all...
@jaystergg AI creates a finished image by learning what is expected to complete it; digital patterns. People will get fatigue from how cookie cutter it'll be.
@@DoveJS Once the database will be so huge enough (like 100 times more than now) that it won't look like a cookie cutter anymore (with blur / nonsense folds etc). mark my words. You might not believe it now but in couple of years it's going to be extremely polished and you won't be able to tell the difference. It's far from perfect now BUT give it like 10 years at max and artists will be struggling hard to pull of something more interesting or new.
@jaystergg I see you're in the 1% or else you think you're an exception because you're eager for the dystopian future I described. Edit: lol nevermind I didn't realize this was a different comment chain. Relevant but you haven't read that. You don't know what cookie cutter means though.
With some of the old classical art, a lot of "inconsistency" or "melting detail" is just dirty varnish. The photos were taken before restoration, or possibly the work was never restored. Also the impressionist paintings were a TRAP
honestly i think one of the easiest for me was in fact 'angry crosses'. i've studied a variety of mediums when it comes to art and the sort of different varietys in topics. im currently learning animation or a somewhat introduction to animation, though the school i go to is quite famous for its alumni. at my previous school i studied more traditional art over graphic and digital art. part of my previous lessons was on abstract art. and let me TELL you it was so hard. i think in a very 'literal' mindset making abstract art to me very hard to acomplish well. though id like to think thanks to that i found it the easiest to guess correctly. with a bit of doubt haha. it was a lot more little giveaways like the box on the left with the red dot having some white seep in. the circle above that (yellow) and how its coloring was a little off at the top of it. the way that some coloring from the lines would kind of seep into the white behind. as soon as i saw the first point i mentioned, i knew it was AI. could it be human error? yes. but what is the logic behind it? a human in my opinion would not keep that in despite it being abstract art. and like pikat said, "abstract art is honestly more intentional than people think". even in abstract art errors are going to be made. maybe a line looks too intentional. or it accidently smudged. there is, in my opinion, a difference between abstract art, and an inconsistent art piece. because even abstract has human 'touch ups'. and a human touch up would be to clean up the black box.
The most impressive part of this to me is how often we're dependent on a "tell" rather than the art as a whole - depending on things like a bar-code signature, or the way roots grow as the only way to identify whether it was made by a human or a machine...And how easily people can be mislead into thinking AI is human art or vice versa by including their "tells" in the counterpart.
Same man, didnt expect that jumpscare lmao. Thats one of the most popular art on Wallpaper engine too by Hangmoon. I have no doubt about that, since its older than OpenAI themselves.
Seems like its gotten better at figures and abstract things, but there’s still commonly issues with eyes, hands, patterns, blended shapes, inconsistent design details, and looking into the backgrounds for smaller details. I find additionally there’s a question about intention in the craft and why it was maybe created, y’know? Although for as good as us artists can see these things, the scarier part for me is how non artists are indifferent to whether it is or not. Seeing during Christmas both the ludicrous amount of amazon packages in recycling and the amount of AI decorations and ads was unnerving to say the least.
There were a couple that I didn’t judge and just said “it’s either AI or money laundering” of course that says more about my taste in art than the skills of the machine. But also what’s with people paying like a billion dollars for blocks that my 5 year old sister can draw?
Working with an AI upscaler for some time gave me a great idea of what to look for. AI has those weird lines that have blurred ends and often weirdly inconsistent width. It would be very hard to replicate, let alone accidentally do by a human. This is even more noticeable on surfaces like grids (fences, windows, etc), as well as branches.
9:11 Ferrari testarossa just has a freaky mirror. It’s easier to tell the difference when the art is of something very much in our real world and that you recognise not because that just means it’s human but because it’s easier to see what’s specifically wrong with it and then determine if that is a mistake a person or an ai would make.
Tbh I'm surprised 88% guessed "Paris scene" at 27:45 is human. I know someone could make something like this, but so many things just feel off. One tree is inside a bulding, one lamp post just gets thinner and vanishes without a lamp, the bike which is a tricicle on the left but the other wheel is missing, the umbrellas...
Before starting the video I got an ad for tourbox elite plus creative console, it had Pikat in it so it took me a moment to realise it was a skippable ad, I was so bamboozled
Only got 2 wrong: Colourful Town and Cherub. The shading and cloth on Cherub should've been obvious to me as a 3D artist. The last two were a bit tough, had to inspect them pretty closely. I feel like most AI can be recognized - like Pikat mentioned - by the lines, consistency, shading, detail and purpose. If something seems to hold no artistic purpose, it's likely AI. Consistency and lines can usually be recognized by following longer sections. Small details often morph together. Below is a list of what I personally noticed first, for the curious. This was fun. 1, 2, 3: Clear detail, realistic birds, and it's Warhammer. The deer are very realistic, good cloth. Intentional folds, no morphing hair, a few very human mistakes. 4, 5, 6: Hands, buildings and eyes. Very symmetrical and no errors in spots where AI would make them. Eyes are abysmal, and zero intent to be seen. X, 8, 9: [Cherub wrong]. The line of people is consistent, no morphing between the people sleeping. Clear intent, nothing is placed without purpose, and the path stays consistent. 10, 11, 12: Bottom right "stairs". Buildings morph with the bushes/trees. Anatomically correct animals, people are low detail but not morphing because of it. 13, 14, 15: Abysmal hair, blends together and morphs all over the place. I do 3D and recognize the look easily, and the inside is normal. No people morphing, text is legible, and bottom right. 16, 17, 18: Obvious. Weird morphing, things trying to be text when it makes no sense for there to be any. No hair morphing, goes in between other strands, very beginner look. 19, 20, 21: Very famous and obviously intentional. Shadows make no sense with the light direction, "windows" are just weird lines. Eyes and clothing detail, shading, very obvious. 22, 23, 24: Everything is intentional, very human, super sick colouring. The leg is not the size of an infant's. Blending and morphing in the middle, the hair and ear make no sense. 25, 26, 27: Very symmetrical and intentional, flag shaded properly. Eyes and hands, obvious. So much morphing and contradictory information, should again be obvious. 28, 29, 30: Very intentional, very nice circle. Almost thought AI, but bucket shine's detail cleared it. Dress folds are nice (sorry Fuwa), building windows are consistent, solid perspective. 31, 32, 33: Chair missing a leg. Very photo-esque, nice reflections, people look real even though little detail. Obvious. 34, 35, 36: Leaves don't look like that, and wax doesn't float or suddenly jump from candle to wall. Cloth is basically flawless, also famous. Very consistent and clear. 37, 38, 39: Obvious. Pillars, also obvious. Consistent and feels purposeful but chaotic (tough one). 40, 41, 42: Hands/knuckles, horns, very obvious. Again, people farther back still look normal and very well captured, even with little detail. The cracks are consistent. 43, 44, 45: Windows morphing, the birds are just splotches. Lamp posts have no lamps, 3rd story windows are inconsistent. Inconsistent trees, bad cloud depth, 85% sure of AI. 46, 47, X: Hand morphing into leg, asymmetrical landing gear, 100% sure of AI. Looks very real, negative faces and hands at the top, 95% sure of human. [Colourful Town wrong]. 49, 50: Shadow makes an odd turn, roof on left stops midway, roof on right extrudes rather than slopes down, 70% sure of AI. Leg and hand morphing, inconsistent head, 90% sure of AI.
The ones that got me: Woman in field: I went back and forth on this one, and I erred for human specifically because I could easily see someone making the main mistake I saw which was the eyes Woman in jumble: Another erred human, but specifically because I am *very* ignorant of this general style's traits, and I err human when I can't tell Cherub: I went back and forth over this one, and erred human only because the wings in the back continued *mostly* consistently, and while they looked a little off to me I, as stated err on the side of human if there's enough doubt. It was enough that I flipped quite a few times, and if the wings hadn't gone behind the arms and come back fairly aligned on the other side I'd have erred as AI though. Fancy Car: This one got me, but for very human errors that tipped me out of erring in the way of human. There's something Off with the placement of where the windows can be seen through in the back. Greek Temple: This one got me because of the difference in clarity of various characters' faces as well as how some of the clothing feels 'newer' than the clothing in the rest. Honestly feels like a shitpost though, which definitely had me flip-flopping, but I couldn't see enough details to see it wasn't AI. Angry crosses: I flipped on this SO much. I'd see something that felt really intentional, and then I'd see things that looked *super* AI, and I erred human. Mind you, abstract styles are not my forte, but it did fool me into thinking several areas were intentional. Mother and child: Erred human as usual, but had the halo not been Very AI I'd have been completely fooled. Riverside Cafe: It got me. Hook line and sinker. Woman unicorn: There were some weird tangents that convinced me it could be, and then I just said "why not" and kept my guess. Paris scene: This one ALSO got me look line and sinker. Landing craft: Got me because it reminded me of something I 100% knew was real because it predated AI. Colourful town: Got me because a lot of the very intentional things (like 2/3 human figures practically blendiing into the BG thanks to colour choices) seeemed like BAD ideas Medeterranian town: Erred human. Punk robot: Erred human, but mostly because I just didn't want to bother. Ones I got instantly: Ancient gate - It had the Vibes but I did double check Anime girl in black - Lol String doll - Rainbow girl - You summed it up Creepy skill - I wish I knew how to explain why this one dinged me as human immediately Ice Princess - Celestial display - the little boats in the water and how they were a specific colour is what made me think 'human'. Muscular man - immediate glance had me considering human, but on even a *vaguely* closer look I was all "Nah, no way" Minaret boat - Purple squares - Turtle house -Mfer couldn't even get the same style of skin on the two front legs Wounded christ - The folds on some of the wrists were just too good, also it looked vaguely familiar White Blob - Misc: Green Hills - Funny enough it was the extra shading behind the roof and how some areas of the hill that should still be hill were Roof instead. It was NOT easy though. Bucolic scene - Had me tripping a little, because I wasn't as zoomed in as I needed to be to see some details. I should watc this in full screen. [does not] Leafy Lane - This one almost got me, but something felt Off and then I noticed the shadows were WEIRD Fractured lady - This one *almost* got me, but the fact that there were things that never repeated when things in other equivalent areas convinced me. Giant ship - I couldn't make my mind up on this one. The nets all being straight instead of with some 'give' actually was what made me go 'human' so much that even my AI senses were "... Not sure, even though there's some of the same issues as AI" Girl in white - I kept thinking 'I want to go AI out of SPITE', but something was just Too Human for me to do so. Funny enough, your "It didn't feel ike an older painting" is the *exact* opposite of what I felt. It felt late 1800's to me. Serene River - the reflections were too consistent to be AI to me. Incuding how they were all adapted to the same angle. Still life: I only got this one just barely. Something felt Off.
"Haven't gotten a single one wrong yet" I'M DOING SIGNIFICANTLY WORSE THAN RANDOM CHANCE. WTF!? I reeeeeeally hope the timer has been shortened for the video.
I will say, if you get really familiar with certain periods of art history, the AI trying to mimic those eras stands out so much because it throws in stuff that doesn’t fit the era at all. I notice sometimes it has style inconsistencies in general, but sometimes people just draw things that way too, so it’s not a catch all
15:14 looked like a painting only bougureau would paint but I’ve never seen it before and the skin had way too few cool tones to be an actual painting by bougureau
The fancy car I got right because I looked at the consistency of the wheels and the size of the grills on the side and front and thought "yeah, that's human"
31:34 i dont know if my phone screen is just ass but i paused to zoom into the skeleton here and as i moved the screen around the eyes started jittering i got caught so off guard
The cherub threw me off. The thing you said about the folds being very messy and inconsistent. I didn’t feel like it was. Even now I’m looking at it and it makes sense to me…. 4:33
27:30 I only got that its AI because at a time when Tricycles like the one in the picture (which only has 2 wheels) were commonly used, the women would not have been wearing skirts that short, as you see on the left. That is not a style of dress that fits with the timeperiod of the rest of the image.
30:51 honestly this one to me was prettyyy easy. The blotchy pattern that’s on the street and the building is very off putting. Like it’s too seamless for something that’s supposed to be painted especially from the dark section of the door to the walls of the building it just doesn’t look natural at all
Hello! I have been learning and working with pixel vtubers a lot and I see all the mouth forms you use on your avatar look very good and smooth, how did you rig all of them?
One trick i learned to tell Ai apart, is that since it's all based on noise functions, AI images always end up averaging out in terms of light. Human art can have unbalanced values, but AI is mostly a gray middle ground.
I immediately clocked the Mediterranean Town as AI because the perspective lines where all out of whack, I did have a bit of doubt that it could have been a human that had done no research into the basics of art tho
It’s just kind of sad that this test even exists. In everyday life, you’re not actively looking for ai so it’s much harder to notice. I don’t like looking at details so I feel like after while, I genuinely won’t be able to tell.
Hello pikat im new to this channel but i draw like you but im a traditional artist draw in paper i like your explanation its easy for me to understand but im have little problem drawing the hair is few meters away to the basic head shape and i use your tutorial how to use box method but i say its not align with the box i draw can you have advice for me?
With the blue hair anime girl, my figured it was human because of the eyes. A lot of ai anime art does not have both eyes looking the same when they're super shiny.
2:30 what threw me for loop was that the OTHER arm is also seems like it bends at the upper arm. That being said I’m not sure there’s anything more human than the making sure you can avoid having to draw hands lol.
I just got recommended this to me, I don't watch her streams or anything but I assume she's an art streamer and her viewers are far more familiar with art than the average person. So, the fact that so many polls were as close as they were, and that people were actively trying to figure out if it's AI or not, really says something.The average person will never spot the things she's pointing out when casually looking at any kind of art, and it's only going to get harder to spot them...
i'm not an artist, i'm trying to learn how to draw, so my guesses were quite dodgy, although they were extremly accurate the moment buildings or other structures got introduced, but in exchange i just threw in the towel when it came to the abstract art. I feel like buildings add something to an art piece that make it a lot easier for me to tell if something is AI or not (for now).
I didnt ever think something ai made was human, but yes got some wrong bc i thought human art was actually A.I. like the Serene River. The impresisonists didnt kick my ass as much as some art which had made decisions i wouldn't have made like the weird tree in the middle or the 1st boat which had what seemed like boat on the bottom left corner that seemed cropped.
I found the impressionist ones to be particularly easy. While the AI can mimic somewhat the impressionist brushstrokes, it fails to capture that impressionism doesn't just mean blurry painting. The lighting in the AI examples is consistently very drab compared to how a real impressionist would paint. And the subject matter is far too detail focused. They have no atmosphere or purpose, and just look like a photograph with a filter applied.
1. Human 2. Human 3. Human 4. AI 5. Human 6. AI 7. AI 8. AI - Wrong - Thought its AI due to fading people in the background and the foreground having an odd colorvariation. Plus the rock on the left side looked awfully like a miniature. On top the bridges started to feel miniaturized as well. 9. Human 10. AI I will leave it here as I have to head to sleep but will update this tomorow. I think my biggest issue is that I am trying to sus out the mistakes made by AI to the point where I sit here and just take art as it is. AI usually has a very distinct look to it. Its usually quite easy to tell through fairly obvious mistakes that AI makes.
I'm SO bad at this but I got green hills right because of those super wonky houses in the front on the right side. but man if it wasn't for those houses I would have guessed human. I wish I was better at this especially as an artist. Omg i also got fancy car right, I got it because (I really had to pause the video lol) the reflections made perfect sense, the details were all consistent, like the lines were really well done and not weird or wobbly or different shapes. The headlight and the rear view threw me off tho. I had a hard time telling that was a rear view mirror at first. Leafy lane seemed off to me because of the window things on the building. Idk how I'm getting these impressionist ones right when I do very stylized cartoony "anime inspired" art lmaoo. Riverside Cafe, there are no trees for the leaves to be attached to on that one overhead the cafe itself, the lamp post is kinda floating off the side its base lol. And yeah the tables are weird but like you said it could be human.
This was an intresting exercise, the impressionist/abstract ones got me and the only human i mistook was the boat one. I think its going to become harder to tell AI in the future not because AI gets better but because humans are copying AI. I've been seeing this trend more and more in traditional paintings.
With the Greek Temple, I at first thought it was human because of the people in the crowd, but then I looked at the hands and severaly of them look cursed, so, like, wtf? XD
its honestly sad that some styles can be mistaken for ai now because of how much they've been fed into the machine
Agreed. The saddest part isn't the art theft, it's that we're being conditioned to hate really cool aspects or styles of art because the people who fed the machine chose to feed it with the coolest styles and techniques they could find.
@HuntingSunder exactly. I hate that impressive rendering and time consuming details don't hit the same because my brain immediately goes "that's likely ai". My mind is poisoned by this shit
@@itsmeloart details that make sense are indicative of human art.
@@itsmeloart I mean, why hate it if you can't tell the difference ? It seems to me that people who know and care about those artstyle are much more able to tell. You'll probably be able to tell if you educate yourself about those artstyle and what make them good and if you get fooled you won't even be mad anymore but impressed ("damn, the IA even got that point correct and the consistency is there...").
Personally, looking the video made me grow an apreciation for abstract art where I am somehow able to tell despite the nonsensical aspect of it.
@@startnetworkbabinski3036 people spend literal decades of their life crafting their art style just to have it ripped off by ai within seconds. hating it is understandable, it's not something to be impressed by. its sad honestly
There's two things for me that are really frustrating about this that I just realized. One is that what we often recognize as "typical AI art" was at some point very much certain peoples' style (although it was different, better composed, less in-your-face etc.) Feels like AI is taking the spotlight away from these artists and gives it a bad rep. And the other is that inconsistencies are now considered "AI-like" when they could have been deliberate. A park near where I grew up had a statue of geese walking and one of them randomly had a human foot. You had to look twice to actually notice it. That was evidently very intentional and maybe even had a deeper meaning. If you just saw a photograph of this statue now, you'd think this is the AI getting confused, totally invalidating what the artist was going for. That sucks so much...
This! So much, everytime a human image was actually ai and the giveaway was " it being melty " I always felt like it is somewhat of an artistic choice. I probably just really suck at spotting mistakes but yeah, it really sucks that some of this more free-form or Exposed/Edited type of styling is getting a bad reputation for no reason.
Yeah someone with a glossy painting style would be @_stato_ though his paintings are way more detailed and sophisticated and less "in your face" as you describe but it is sad that there's a chance people might consider it AI just because it's stylistically similar.
I sometimes used to use an art style that would be reminiscent of certain AIs, and I noticed how more and more people started saying it was made by AI or AI inspired etc, when I've been doing it for years and it was just a style that came natural to me, but since then I'm much more considerate of not going close to things people might associate with AI imagery, as there is really a noticeable drop in appreciation any time an artwork is in that same field. Lots of people came to either think all such artstyles are done by AI, or just became so desensitized to it that it no longer sparks the same interest at all.
that's a fair observation. There are plenty people who understand this, and fewer who had seen this coming years ago. I fear that even if we do get to make genuine works of art, their inherent value will be that much lost on us due to technological advancement - watered down to put it in other words.
I suspect that a decade or two from now on, once this becomes a rooted standard, hardly anything will surprise us in terms of art found in the digital world and real world (generated media used to make low-tier merchandise). There will surely be a need for art - as there had always been to an extent - but it won't be as thrilling anymore knowing that anyone can do it without much effort and knowledge. An item of luxury turns into a common good which seems fine on the outside, but becomes nothing special when you really think about it.
Yeah, some say AI will help people who have trouble learning art, but up and coming artists are exactly who will get screwed over the most. If you are a well-known amazing auteur, you already have an audience, people already know you don't use AI if they prefer that, and your art is probably just better than AI. As a new artist, you have effectively no way to get your stuff out to people.
Fun enough this actually trains one of the most important skill of any artist, their observation skills
“The design is very human”
the design is very human
@@Afunnylilguy the design is very human
Unfortunately so
"The design is very human"
You can tell the Blue hair anime girl was drawn by a human because the hands are covered
I thought it was ai bc the hand looked like a 2nd little forearm
Except that it's common for people to prompt to exclude hands entirely for the (now old) models which frequently muck them up.
Funnily enough, it never occurred to me that humans would do the same and avoid drawing hands. I thought for sure a human would draw part of the upper hand rather than hiding it all behind the hat.
The arm behind the hat was very human, but I was thrown for a loop with the massively long arm going down, that just seemed so AI.
The blue hair anime girl was used in some nightcore covers back in the day.
that one I knew it was human because they hidden her hands from the frame xD just to avoid suffering when drawing hands
Looks like hatsune miku to me but aight
Man the impressionist ones kept kicking my ass. Those were the major ones I kept guessing incorrectly on
I have a big feeling those were basically painted over actual pictures. (like basically give the ai a picture as reference and make them paint that reference in that style)
So usual tells like composition and environmental inconsistencies and such were hard to spot
I highly recommend looking for how many strokes there are: Impressionists outdoors worked VERY quickly back in their day because they had mere minutes before the vista they were trying to depict came and went. Pictures from that time will have brush strokes that are quick, rough and DELIBERATE, think like 10-25 minutes of time to work. A lot of AI art is unnecessarily refined, drowning the canvas in unnecessary strokes, taking away from the core spirit of that time period...
However, this isn't without fault as contemporary artists taking inspiration from old Impressionists and adding refining strokes in the process will get caught in the crossfire. There's no winning.
Earth 5020: This artwork is worth 700B dollars! It's 100% made by a human hand!
Artwork: *Poorly made stick figure family and sun drawn in the corner of the paper*
My art sucks so much anyone would think it's made by child, but atleast it's human
Having messed around with generative AI locally on my computer, 90% of it also sucks really bad, you just see the 10% (maybe even less) being shared online, that being said you can clearly see whether something that sucks was made by AI or a Human, and I much prefer to see something that sucks that was made by a human that put effort and actually tried than some garbage that was spit out in 3 seconds (sometimes less) by some algorithm.
@@SergioEduP shit that sucks won't be selling. stuff that's good won't be selling also because AI can do it cheaper. art will become a thing that only people that are SUPER INTO IT, will actually buy it from real artists but most people will just generate stuff and change whatever they need. it sucks now, of course... but you might have used a bad software or you didn't pay for it etc. and there are many different things. ChatGPT was working perfectly for me though, I have paid for it and the images i was getting, some of them, were very good... even as inspiration it's perfect. But one day it will become so unrecognizable that people that spent 90% of their life doing art with a vision of making a bag... nah man, that will not happen anymore... most of the stuff will be AI and people rather save money than spend 10x more to have it made by a human... (i mean, of course there will be people that will buy REAL art but the audience will be significantly smaller.)
@jaystergg It'll be like Hollywood Blockbusters; popularity without concern for originality, only profit. And the corporations are only saving money so they can fire skilled workers. It's what they've done for centuries in every single country. It's easier to do something from scratch than change a finished result that's only one layer but if most people don't care and are content with the cookie cutter results for cheap, this is the future. Eventually their money will dwindle because less of it is in circulation; it's being spread across fewer industries and companies.
The growing population of homeless people will be criminalized for being homeless and sent into private prisons to become extremely cheap or slay ve labor. The working class that's left will be in the reduced service industry or live in factory villages or on farms and be unable to move. Others all over the world will also be forced into internment camps to produce raw or finished goods, which will be enforced by labeling these people as terrorists or war criminals.
The only profitable market for human art will be for the people who are lucky enough to be born into a wealthy family and able to make art for other wealthy families for millions (actual art supplies, including programs and drawing tablets, will be reduced to 2-3 companies and very few will be produced.) Imprisoned homeless artists who are allowed to make art with free art programs or bulk goods because they're sickly or they're forced to produce this work that the prisons can sell for most of the profit; most likely to be fed into the Gen AI machines (there will be a lot of people forced into strip mining as well.) The same thing applies to every other creative hobby/job; the machines have to be fed for anything new to come out and occasionally if a wealthy person has passion then they'll shake up what's popular significantly. If some poor person is extremely lucky, their vision is extremely desirable for it's freshness, maybe they'll be retrieved as a personal art servant and kept in the big house.
There won't be as many people in general but fewer are needed when the ruling class has machines boosting efficiency. Most people and their information are already the products being sold online. Most scams exist because people are finding it harder to survive or thrive and there aren't reasonable fall-backs to support them. In particular, everyone gets exploited in the current system as it is because that's all there's room for.
The military will mostly have drones in countries that can afford them. The countries who can't afford them will be colonized: there will be a new age of empires. If a laughable nod to PR and guilt reigns, they won't call them that but this is where fasc ism, national ism, and other prejudices insist the wealthy are better than everyone else, regardless of which country they live in. Poor people will be deemed poor because they aren't skilled, intelligent, moral, or interesting enough, not because they could be all of those things and forced into this position by bad luck or social standards imposed on them by the 1% in power. Worse yet, they may be considered less human and therefore disposable. The higher ups don't need to interact with them so why would they care?
And there will be a lot of homeless people because of all the wars across the world and because landlords make more money when they have a bunch of houses sitting empty for years. Overtime a lot of the landlords will also vanish and their property consolidated under companies owned by the wealthiest families. The Wealth Gap will increase exponentially.
This is what happens when there isn't a Universal Income, and the wealthy have no empathy for those outside their social circle. As it stands, most consumer goods are already white label. Only a few factories across the world actually make what we buy, even for the big name brands that have existed for over a century (and this has been true for decades, research backs this up.) That's why there are so many products that seem so similar on market sites; it's just becoming more evident. Variety will diminish when fewer people need fewer unique goods and the mega-conglomerations are able to wreck their competition and consolidate further. And all of the Gen-AI companies will as well. This is the trajectory of that future.
Or it could all be diverted if China finally accepts that the Mandate of Heaven has been revoked once again. We might have to look to Vietnam or some other country first though IDK.
@@jaystergg I dont want money, im an artist. If i would only want money, i would work normal jobs. I only need enough money to be sustainable.
Stop looking at things with that type of mindset. What is the point, people will never be happy. unless they can do something they enjoy.
AI can NOT do it cheaper, pay a subscription or get those tokens. Oh you can do it for free? Oh it sucks, oh you can do it locally? Well time to get that new RTX Graphic card. I know how this space opperates.
There are so many cheaper ways to make any kind of art. Physical or If you want to do digital drawing, you can even use free programms.
(I hope i didnt missunderstand you and this was just a rant about, the future. But you seem to imply things...)
@@jaystergg Except AI art can't reach that top percentile that sells the most. The guweiz, the Shal.E, the Nixeuu, The _stato_, The riot artists, the Fortische artists. So no, Ai wont be able to replace them. But yes, for the average insta artist it will make comissions slightly harder to find but most people who comission already are into supporting creatives and locals regardless so it won't affect them as much either. I think it's better to have a more optimistic outlook on this.
Without watching the video, to answer the question: at this current point in time, if someone uploads an especially high resolution image, if you zoom in, none of the “strokes” makes sense. It looks good from a distance, but doesn’t hold up to scrutiny
The next biggest give away for me is things disappearing. Like a strand of hair going behind the arm and not showing up on the other side, because AI does not remember that the hair was supposed to be there. Or cutting off clouds suddenly because it crossed behind a tree etc.
The problem is that AI doesn't always do these things, or at least doesn't always do them in a way that is distinguishable from a human.
one of the rules of the form was that you can't zoom into any of the images
what about people that "clean up/fix" the ai hands or whatever on photoshop and post it on twitter (formerly X)?
sometimes the only way I can be certain it is AI is because there is an inhuman amount of drawings posted per week, like more than new 1 fully rendered drawing per day with background and everything? nah, red flag
Distance? From a normal distance it usually looks fine.
Hiding the hands on that first anime girl felt like such a human thing to do. AI would just go "Hands? Have 5 of them!"
Unfortunately couldn't submit the form because it was closed but it was fun to try. My main trick for the impressionist ones was the see the direction of the strokes around each object.
E.g. Around the tree trunks normally in impressionism is wet-on-wet so to avoid the colours blending the strokes would draw lines around them. In AI images there would be strokes going perpendicular (into) the objects where it wouldn't make sense to paint that way.
7:19 not me telling that it wasn’t a real painting becasue Van Gogh wouldn’t have painted in that aspect ratio in Aurves
Back in 2019, I asked an artist if a print was done with oil painting upon which she then told me it was all digital. I was incredibly impressed with her skills, even though I wouldn't say that I have an artist eye. She knew the fundamentals.
And that's the difference between artists and AI-users. Artists are honest about the process and AI users want more recognition for less effort.
dont worry, it will be a thing of a past in couple of years... you will be amazed that something is AI made and not human... you won't be able to recognize that at all...
@jaystergg AI creates a finished image by learning what is expected to complete it; digital patterns. People will get fatigue from how cookie cutter it'll be.
@@DoveJS Once the database will be so huge enough (like 100 times more than now) that it won't look like a cookie cutter anymore (with blur / nonsense folds etc). mark my words. You might not believe it now but in couple of years it's going to be extremely polished and you won't be able to tell the difference. It's far from perfect now BUT give it like 10 years at max and artists will be struggling hard to pull of something more interesting or new.
@jaystergg I see you're in the 1% or else you think you're an exception because you're eager for the dystopian future I described. Edit: lol nevermind I didn't realize this was a different comment chain. Relevant but you haven't read that. You don't know what cookie cutter means though.
@@jaystergg they literally pulled images from the whole internet wtf do you mean 100 times more than now there just isn't enough data
With some of the old classical art, a lot of "inconsistency" or "melting detail" is just dirty varnish. The photos were taken before restoration, or possibly the work was never restored.
Also the impressionist paintings were a TRAP
honestly i think one of the easiest for me was in fact 'angry crosses'. i've studied a variety of mediums when it comes to art and the sort of different varietys in topics. im currently learning animation or a somewhat introduction to animation, though the school i go to is quite famous for its alumni. at my previous school i studied more traditional art over graphic and digital art. part of my previous lessons was on abstract art. and let me TELL you it was so hard. i think in a very 'literal' mindset making abstract art to me very hard to acomplish well. though id like to think thanks to that i found it the easiest to guess correctly. with a bit of doubt haha. it was a lot more little giveaways like the box on the left with the red dot having some white seep in. the circle above that (yellow) and how its coloring was a little off at the top of it. the way that some coloring from the lines would kind of seep into the white behind.
as soon as i saw the first point i mentioned, i knew it was AI. could it be human error? yes. but what is the logic behind it? a human in my opinion would not keep that in despite it being abstract art. and like pikat said, "abstract art is honestly more intentional than people think". even in abstract art errors are going to be made. maybe a line looks too intentional. or it accidently smudged. there is, in my opinion, a difference between abstract art, and an inconsistent art piece. because even abstract has human 'touch ups'. and a human touch up would be to clean up the black box.
The overall aspect of the picture is the best way of telling it rather than looking at specific things like the hair, folds, shading, hands etc
thats insane how much eye opening this video was!
31:21 is a crazy take because the only person ik who would paint like that is dali.
hes like the total opposite of a boring human
The most impressive part of this to me is how often we're dependent on a "tell" rather than the art as a whole - depending on things like a bar-code signature, or the way roots grow as the only way to identify whether it was made by a human or a machine...And how easily people can be mislead into thinking AI is human art or vice versa by including their "tells" in the counterpart.
Since AI is a bit of a hassle, let's make it a tour guide for the beginners
the jumpscare @ 14:19 when my background is one of the picture and i was so hoping it was human (and it was)
Same man, didnt expect that jumpscare lmao. Thats one of the most popular art on Wallpaper engine too by Hangmoon.
I have no doubt about that, since its older than OpenAI themselves.
That's my mousemat, which i bought before Dalle even existed, so was easy for me 😂
27:24 This AI art is VERY similar to another artwork I’ve seen…I cannot remember its name
it reminds me a bit of boulevard montmarte
@@jjfan8540 YEAH!!! Something like that. I knew it was nighttime on the painting 🤔
I agree, that one threw me off hard because I felt like I had seen it or something similar somewhere else
I'm at a loss for words
Seems like its gotten better at figures and abstract things, but there’s still commonly issues with eyes, hands, patterns, blended shapes, inconsistent design details, and looking into the backgrounds for smaller details.
I find additionally there’s a question about intention in the craft and why it was maybe created, y’know?
Although for as good as us artists can see these things, the scarier part for me is how non artists are indifferent to whether it is or not. Seeing during Christmas both the ludicrous amount of amazon packages in recycling and the amount of AI decorations and ads was unnerving to say the least.
There were a couple that I didn’t judge and just said “it’s either AI or money laundering” of course that says more about my taste in art than the skills of the machine. But also what’s with people paying like a billion dollars for blocks that my 5 year old sister can draw?
3:55 the eyes are a giveaway honestly. The right eye looks effed up
Working with an AI upscaler for some time gave me a great idea of what to look for. AI has those weird lines that have blurred ends and often weirdly inconsistent width. It would be very hard to replicate, let alone accidentally do by a human. This is even more noticeable on surfaces like grids (fences, windows, etc), as well as branches.
Early🙏 your art has low-key helped me a shit ton😢
19:58 this AI stole from Van Gogh
9:11 Ferrari testarossa just has a freaky mirror. It’s easier to tell the difference when the art is of something very much in our real world and that you recognise not because that just means it’s human but because it’s easier to see what’s specifically wrong with it and then determine if that is a mistake a person or an ai would make.
All the impressionist has been replaced by AI 😂
thing is, impressionism is very easy to train AI for and often conceals a lot of the inconsistencies that ai make
Tbh I'm surprised 88% guessed "Paris scene" at 27:45 is human. I know someone could make something like this, but so many things just feel off.
One tree is inside a bulding, one lamp post just gets thinner and vanishes without a lamp, the bike which is a tricicle on the left but the other wheel is missing, the umbrellas...
Wait, it's AI, right? That's what Pikat said
@@bigkspicy8257 yeah it is AI, most guessed it was human for some reason
@blobfish4693 I misread your comment, my bad
Before starting the video I got an ad for tourbox elite plus creative console, it had Pikat in it so it took me a moment to realise it was a skippable ad, I was so bamboozled
Only got 2 wrong: Colourful Town and Cherub. The shading and cloth on Cherub should've been obvious to me as a 3D artist. The last two were a bit tough, had to inspect them pretty closely. I feel like most AI can be recognized - like Pikat mentioned - by the lines, consistency, shading, detail and purpose. If something seems to hold no artistic purpose, it's likely AI. Consistency and lines can usually be recognized by following longer sections. Small details often morph together. Below is a list of what I personally noticed first, for the curious. This was fun.
1, 2, 3: Clear detail, realistic birds, and it's Warhammer. The deer are very realistic, good cloth. Intentional folds, no morphing hair, a few very human mistakes.
4, 5, 6: Hands, buildings and eyes. Very symmetrical and no errors in spots where AI would make them. Eyes are abysmal, and zero intent to be seen.
X, 8, 9: [Cherub wrong]. The line of people is consistent, no morphing between the people sleeping. Clear intent, nothing is placed without purpose, and the path stays consistent.
10, 11, 12: Bottom right "stairs". Buildings morph with the bushes/trees. Anatomically correct animals, people are low detail but not morphing because of it.
13, 14, 15: Abysmal hair, blends together and morphs all over the place. I do 3D and recognize the look easily, and the inside is normal. No people morphing, text is legible, and bottom right.
16, 17, 18: Obvious. Weird morphing, things trying to be text when it makes no sense for there to be any. No hair morphing, goes in between other strands, very beginner look.
19, 20, 21: Very famous and obviously intentional. Shadows make no sense with the light direction, "windows" are just weird lines. Eyes and clothing detail, shading, very obvious.
22, 23, 24: Everything is intentional, very human, super sick colouring. The leg is not the size of an infant's. Blending and morphing in the middle, the hair and ear make no sense.
25, 26, 27: Very symmetrical and intentional, flag shaded properly. Eyes and hands, obvious. So much morphing and contradictory information, should again be obvious.
28, 29, 30: Very intentional, very nice circle. Almost thought AI, but bucket shine's detail cleared it. Dress folds are nice (sorry Fuwa), building windows are consistent, solid perspective.
31, 32, 33: Chair missing a leg. Very photo-esque, nice reflections, people look real even though little detail. Obvious.
34, 35, 36: Leaves don't look like that, and wax doesn't float or suddenly jump from candle to wall. Cloth is basically flawless, also famous. Very consistent and clear.
37, 38, 39: Obvious. Pillars, also obvious. Consistent and feels purposeful but chaotic (tough one).
40, 41, 42: Hands/knuckles, horns, very obvious. Again, people farther back still look normal and very well captured, even with little detail. The cracks are consistent.
43, 44, 45: Windows morphing, the birds are just splotches. Lamp posts have no lamps, 3rd story windows are inconsistent. Inconsistent trees, bad cloud depth, 85% sure of AI.
46, 47, X: Hand morphing into leg, asymmetrical landing gear, 100% sure of AI. Looks very real, negative faces and hands at the top, 95% sure of human. [Colourful Town wrong].
49, 50: Shadow makes an odd turn, roof on left stops midway, roof on right extrudes rather than slopes down, 70% sure of AI. Leg and hand morphing, inconsistent head, 90% sure of AI.
Pikat when is came too choosing between human or ai: "The design is very human"
Great video moetly go them right during the stream. This give greag insight on how to spot them.
14:58 can't stop seeing AI artworks that are stiff and lifeless portraits
didnt get a straight answer but I guess it is ai then, thx
Anime Girl in Black, i honestly could care less if a human or Ai created that one, to me it's a beautiful work. Looks gorgeous
7:31 look at the buildings
Yeah, the forms are so wobbly and unclear
Oo when I catch whoever "made" those impressionist paintings. They were the ones I got the most wrong on
21:18 not to mention the turtle fingers.
After watching this, you can tell that the design is very human
Leafy lane stuffed up as the lighting was going right. But shadow was going towards the light sores (to the left).
AI is getting scarier, at some point I couldn't tell the difference between AI or human
The ones that got me:
Woman in field: I went back and forth on this one, and I erred for human specifically because I could easily see someone making the main mistake I saw which was the eyes
Woman in jumble: Another erred human, but specifically because I am *very* ignorant of this general style's traits, and I err human when I can't tell
Cherub: I went back and forth over this one, and erred human only because the wings in the back continued *mostly* consistently, and while they looked a little off to me I, as stated err on the side of human if there's enough doubt. It was enough that I flipped quite a few times, and if the wings hadn't gone behind the arms and come back fairly aligned on the other side I'd have erred as AI though.
Fancy Car: This one got me, but for very human errors that tipped me out of erring in the way of human. There's something Off with the placement of where the windows can be seen through in the back.
Greek Temple: This one got me because of the difference in clarity of various characters' faces as well as how some of the clothing feels 'newer' than the clothing in the rest. Honestly feels like a shitpost though, which definitely had me flip-flopping, but I couldn't see enough details to see it wasn't AI.
Angry crosses: I flipped on this SO much. I'd see something that felt really intentional, and then I'd see things that looked *super* AI, and I erred human. Mind you, abstract styles are not my forte, but it did fool me into thinking several areas were intentional.
Mother and child: Erred human as usual, but had the halo not been Very AI I'd have been completely fooled.
Riverside Cafe: It got me. Hook line and sinker.
Woman unicorn: There were some weird tangents that convinced me it could be, and then I just said "why not" and kept my guess.
Paris scene: This one ALSO got me look line and sinker.
Landing craft: Got me because it reminded me of something I 100% knew was real because it predated AI.
Colourful town: Got me because a lot of the very intentional things (like 2/3 human figures practically blendiing into the BG thanks to colour choices) seeemed like BAD ideas
Medeterranian town: Erred human.
Punk robot: Erred human, but mostly because I just didn't want to bother.
Ones I got instantly:
Ancient gate - It had the Vibes but I did double check
Anime girl in black - Lol
String doll -
Rainbow girl - You summed it up
Creepy skill - I wish I knew how to explain why this one dinged me as human immediately
Ice Princess -
Celestial display - the little boats in the water and how they were a specific colour is what made me think 'human'.
Muscular man - immediate glance had me considering human, but on even a *vaguely* closer look I was all "Nah, no way"
Minaret boat -
Purple squares -
Turtle house -Mfer couldn't even get the same style of skin on the two front legs
Wounded christ - The folds on some of the wrists were just too good, also it looked vaguely familiar
White Blob -
Misc:
Green Hills - Funny enough it was the extra shading behind the roof and how some areas of the hill that should still be hill were Roof instead. It was NOT easy though.
Bucolic scene - Had me tripping a little, because I wasn't as zoomed in as I needed to be to see some details. I should watc this in full screen. [does not]
Leafy Lane - This one almost got me, but something felt Off and then I noticed the shadows were WEIRD
Fractured lady - This one *almost* got me, but the fact that there were things that never repeated when things in other equivalent areas convinced me.
Giant ship - I couldn't make my mind up on this one. The nets all being straight instead of with some 'give' actually was what made me go 'human' so much that even my AI senses were "... Not sure, even though there's some of the same issues as AI"
Girl in white - I kept thinking 'I want to go AI out of SPITE', but something was just Too Human for me to do so. Funny enough, your "It didn't feel ike an older painting" is the *exact* opposite of what I felt. It felt late 1800's to me.
Serene River - the reflections were too consistent to be AI to me. Incuding how they were all adapted to the same angle.
Still life: I only got this one just barely. Something felt Off.
"Haven't gotten a single one wrong yet"
I'M DOING SIGNIFICANTLY WORSE THAN RANDOM CHANCE. WTF!?
I reeeeeeally hope the timer has been shortened for the video.
I will say, if you get really familiar with certain periods of art history, the AI trying to mimic those eras stands out so much because it throws in stuff that doesn’t fit the era at all. I notice sometimes it has style inconsistencies in general, but sometimes people just draw things that way too, so it’s not a catch all
15:14 looked like a painting only bougureau would paint but I’ve never seen it before and the skin had way too few cool tones to be an actual painting by bougureau
20:34 I knew it was human cause the water reflection gave it away.
I recognized "Celestial Display" from the song Brainpower by Noma lol
The fancy car I got right because I looked at the consistency of the wheels and the size of the grills on the side and front and thought "yeah, that's human"
man i was so sure rainbow girl was ai because her eyebrows and her hair blends into her neck in a spot and her nose seems off.
what tipe of app u use for drawing? i love you're drawings
Using the pieces I fail to tell apart as a checklist of stuff to study later (like clothing folds dang I don’t get those)
31:34 i dont know if my phone screen is just ass but i paused to zoom into the skeleton here and as i moved the screen around the eyes started jittering i got caught so off guard
9:35 there is a Franklin in the bottom right corner and he looks suspicious
Maybe it’s like a collage?
I never seen this painting
I did graduate in an art school but I gave up on all the impressionists backround ones.
11:55, i was looking at highlights in the eyes, they should be symmetrical afaik because of how reflections in spherical objects work.
The cherub threw me off. The thing you said about the folds being very messy and inconsistent. I didn’t feel like it was. Even now I’m looking at it and it makes sense to me…. 4:33
14:30 was not expecting my desktop wallpaper to appear lol
If you took an art history class and spent a lot of time online in the 2000s a lot of these were pretty easy
27:30 I only got that its AI because at a time when Tricycles like the one in the picture (which only has 2 wheels) were commonly used, the women would not have been wearing skirts that short, as you see on the left. That is not a style of dress that fits with the timeperiod of the rest of the image.
ai focuses on faces a lot, thats probably the best giveaway
I only got the impressionist and 1 of the early abstract pieces wrong, so I'd say I still did pretty well
On the Cherub, I could tell the correct answer because the neck was funky
The Riverside Café looked a little too similar to Van Gogh's Café Terrace in the Evening in a way I thought an AI would do.
i feel like if you're an artist you can get most of them right (aside from the more abstract ones)
30:51 honestly this one to me was prettyyy easy. The blotchy pattern that’s on the street and the building is very off putting. Like it’s too seamless for something that’s supposed to be painted especially from the dark section of the door to the walls of the building it just doesn’t look natural at all
Hello!
I have been learning and working with pixel vtubers a lot and I see all the mouth forms you use on your avatar look very good and smooth, how did you rig all of them?
One trick i learned to tell Ai apart, is that since it's all based on noise functions, AI images always end up averaging out in terms of light. Human art can have unbalanced values, but AI is mostly a gray middle ground.
I immediately clocked the Mediterranean Town as AI because the perspective lines where all out of whack, I did have a bit of doubt that it could have been a human that had done no research into the basics of art tho
It’s just kind of sad that this test even exists. In everyday life, you’re not actively looking for ai so it’s much harder to notice. I don’t like looking at details so I feel like after while, I genuinely won’t be able to tell.
The impressionist ones were the easiest for me to tell, probably because that’s the way I paint, but the anime ones were hard for me to tell lol
8:51 THE EARRING IS ATTACHED TO THE HAIR
Hello pikat im new to this channel but i draw like you but im a traditional artist draw in paper i like your explanation its easy for me to understand but im have little problem drawing the hair is few meters away to the basic head shape and i use your tutorial how to use box method but i say its not align with the box i draw can you have advice for me?
An easy way to tell the abstract art is AI is to look at the plain, blank areas. If there are arbitrary, really thin lines. Its AI... every time
It's crazy to think that "survey" style videos like these are just going to be used for AI improvement research
With the blue hair anime girl, my figured it was human because of the eyes. A lot of ai anime art does not have both eyes looking the same when they're super shiny.
2:30 what threw me for loop was that the OTHER arm is also seems like it bends at the upper arm. That being said I’m not sure there’s anything more human than the making sure you can avoid having to draw hands lol.
Artists nowadays must be terrified about getting hands wrong in fear that they'll be accused of using AI.
I just got recommended this to me, I don't watch her streams or anything but I assume she's an art streamer and her viewers are far more familiar with art than the average person. So, the fact that so many polls were as close as they were, and that people were actively trying to figure out if it's AI or not, really says something.The average person will never spot the things she's pointing out when casually looking at any kind of art, and it's only going to get harder to spot them...
I love artists who art looks ai
i'm not an artist, i'm trying to learn how to draw, so my guesses were quite dodgy, although they were extremly accurate the moment buildings or other structures got introduced, but in exchange i just threw in the towel when it came to the abstract art. I feel like buildings add something to an art piece that make it a lot easier for me to tell if something is AI or not (for now).
I managed to get them all right except the giant gate near the start where I reverse psyched myself out and changed my answer last second, rip LMAO
I didnt ever think something ai made was human, but yes got some wrong bc i thought human art was actually A.I. like the Serene River. The impresisonists didnt kick my ass as much as some art which had made decisions i wouldn't have made like the weird tree in the middle or the 1st boat which had what seemed like boat on the bottom left corner that seemed cropped.
I pretty much got most of it right but a couple did threw me off the loop like the Green hills one and the Fancy car art....
21:56 the candle wax was so ai
The blue hair anime girl's left hand feels so damn AI. 😬
14:18 1700pp type painting
painting from the new era
I found the impressionist ones to be particularly easy. While the AI can mimic somewhat the impressionist brushstrokes, it fails to capture that impressionism doesn't just mean blurry painting. The lighting in the AI examples is consistently very drab compared to how a real impressionist would paint. And the subject matter is far too detail focused. They have no atmosphere or purpose, and just look like a photograph with a filter applied.
I got 5 wrong and I’m actually proud of myself 😮 guess I do kinda know. Those impressionist stuff got me the most
6:46 i thought it was ai because of tilted houses
9:24 The fancy car literally looks like a fridge with LED wheels wha
1. Human
2. Human
3. Human
4. AI
5. Human
6. AI
7. AI
8. AI - Wrong - Thought its AI due to fading people in the background and the foreground having an odd colorvariation. Plus the rock on the left side looked awfully like a miniature. On top the bridges started to feel miniaturized as well.
9. Human
10. AI
I will leave it here as I have to head to sleep but will update this tomorow. I think my biggest issue is that I am trying to sus out the mistakes made by AI to the point where I sit here and just take art as it is. AI usually has a very distinct look to it. Its usually quite easy to tell through fairly obvious mistakes that AI makes.
I got 12 wrong, I really started to trip up later in the quiz
I'm unsure how well the images were actually checked to be AI. Because I'm not convinced that some of the human ones are made entirely without AI
I'm SO bad at this but I got green hills right because of those super wonky houses in the front on the right side. but man if it wasn't for those houses I would have guessed human. I wish I was better at this especially as an artist.
Omg i also got fancy car right, I got it because (I really had to pause the video lol) the reflections made perfect sense, the details were all consistent, like the lines were really well done and not weird or wobbly or different shapes. The headlight and the rear view threw me off tho. I had a hard time telling that was a rear view mirror at first.
Leafy lane seemed off to me because of the window things on the building. Idk how I'm getting these impressionist ones right when I do very stylized cartoony "anime inspired" art lmaoo.
Riverside Cafe, there are no trees for the leaves to be attached to on that one overhead the cafe itself, the lamp post is kinda floating off the side its base lol. And yeah the tables are weird but like you said it could be human.
wow paris scene got me that one was the best one yet imo
This was an intresting exercise, the impressionist/abstract ones got me and the only human i mistook was the boat one.
I think its going to become harder to tell AI in the future not because AI gets better but because humans are copying AI. I've been seeing this trend more and more in traditional paintings.
With the Greek Temple, I at first thought it was human because of the people in the crowd, but then I looked at the hands and severaly of them look cursed, so, like, wtf? XD