I admit to having had him on my IG for several months together with other portrait photographers as an inspiration and a way to improve my own portrait skills.
I had no idea. And thinking about this now, I find what he's done brilliant. It shows in a very simple and concise way how far the generative AI has been for months now. And it shows (and proves) this to people who otherwise wouldn't be playing with any diffusion models otherwise.
In a way it blows your mind. If that's not art, I don't know what is.
It’s interesting because I always imagined that artists would be the last ones to be replaced by technology. I guess it really shows that we have no way to predict the future.
Artists won't be replaced by AI. Human art is about expression, about communicating thoughts and feelings. This current AI generation can only mimic the product, not the intent.
Even when we get AGI that can truly be emotional, it still won't replace or compete with human art. There will just be more of it, and we'll learn to appreciate it in different ways.
Art is not competitive. Just as new art styles and techniques made by humans didn't replace previous ones, AI won't either. It's just another tool or outlet.
I completely disagree with what you’re saying btw. I can just picture us 30 years from now: every time you’ll see art you’ll see a lot of AI-assisted stuff which will make you doubt everything you’ll see. There won’t be any “famous” painter anymore. The same way “famous photographers” aren’t really a thing anymore due to how good photography and editing software has become.
I’m a photographer and what’s funny is, the amount of work he put into producing these images is almost harder than just taking the damn photos or dare I say drawing what you’d like to see an enhancing it with photoshop.
It might be AI generated but there’s still a lot of human work gone into the images.
The sad news for me personally is taking photos and working with models and lighting people etc is fun, same for meeting people and taking their portrait, this guy got images, the hard way and didn’t seem to get the experience.
Maybe he enjoyed sifting through thousands of images and stitching them together, then editing with photo shop. I don’t know but it sounds much less interesting than photography but hey, to this guy it’s art, so good luck.
Right, I don't see this as inferior to traditional photography, just different. Technology is just another tool we can use to produce art, or assist artists in doing so.
Just as electronic music didn't replace instrumental music, or photography didn't replace paintings, we'll also find ways to incorporate AI as another tool that creates different forms of art. They will all have their audiences, and will broaden our senses in different ways.
The current problem is that we can't distinguish AI-generated/assisted works from others, but eventually we'll find ways to do that too.
> I’m a photographer
Judging by your username, I'm not so sure... :-P Quick check: are you alive?
There may be a dividing line somewhere between "people making art for the sake of it with no regard for its perception by others" and "people making art to be received by others" (whether they're making it only/primarily as a product for others is a different story). The latter ones will definitely compete with AI-generated art because human attention is limited. If there's more stuff to pay attention to (say, a million expressive portraits with a nice-sounding backstory a day), the attention will be divided among those, and the humans will receive less of it.
True, but that's always been the case with any product. There have always been imitations and knock-offs of any genuine product, whenever the technology or means to massively produce it cheaply and quickly becomes available. The effect of this is that it floods the market, a larger section of the population get to enjoy an inferior product (whether they care about that or not), but it doesn't take the genuine goods off the market. In some cases we create regulations around it if it becomes abusive, and companies can sort it out legally if they wish, but people who seek the original product will always have ways to find it.
In the case of art, I wouldn't label AI-generated works to be inferior, just different. Once we're able to distinguish them from traditional works done manually by humans, it will eventually settle into its own category, with its own supply and demand. The current problem is that it can pose as something that traditionally takes effort and skill to produce, but we'll find ways around that.
It turns goods into luxury items though. Today, you can still get manually crafted scissors, but they'll be expensive and not something most people can afford. Sure, not 100% of blacksmiths were replaced, but 99.9% of them. That's still a tough prospect if you see your job on the shortlist for getting replaced and you're not part of the 0.01% in that job today.
Can we find a way to tell apart human art from generated art? I'm not so sure in the long run. Today, we have a lot of pointers and with enough focus experts can tell them apart. But they can't in passing, and an average person probably can't without being told what exactly to look for and advanced tools to check those. We're only a few years into it though, so I find it hard to believe that we'll be able to tell whether a digital picture was produced with a digital camera or generated by an AI. You can probably make it much more expensive to fake it (e.g. film yourself with 3 different cameras which are positioned differently, so AI would need to also fake a multi-perspective video reliably), and maybe that'll be the thing (like "organic" diamonds vs industrially produced ones), but I'm not sure if you can put the cat back into the bag.
All human art is the collection of human experiences at the time. It is not individual as all interactions we have that shape us as individuals arise out of our interactions with other people.
Kropotkin has something to say about the individual and the influence of others, but in short, you can not construct anything individually without resting on the shoulders of everyone that came before you.
The individual struggle is a struggle because of everyone around you, the way they behave towards you, and the way they don’t. Simultaneously if you were the last person on earth, your struggle would be the absence of everyone else.
Art is, by and large, an expression of the cumulative experiences of a person at that point in time, which come out of their interactions with everyone else. White implies black, and self implies other.
The impressionists, think Degas and Monet, went against the Salon and forged their own path, and that influenced their art. Their counter culture was by influenced by the very existence of mainstream culture.
To look the art in isolation and without context is to ignore their struggle of the individual against the broader society.
And in this view, AI art captures the cumulative experiences of all humanity, and with the right direction can show intent by an individual.
Artists gaslight non-artists and say their work has soul and gatekeep "art" saying it can't be replicated. However this is not soul it is egoism and pointless toil. In reality the end product is just paint on a canvas that has specific patterns and style applied to it and this can be replicated and improved upon by an ai artist.
Sure, but I'm not endorsing gatekeeping. I don't see these works as inferior, just different. Once we have ways to distinguish them, they will become a different category of art we can also enjoy.
The problem is when images can be produced effortlessly by anyone to resemble anything else, which current AI tools allow. This makes them capable of posing as something that traditionally took a lot of skill and effort to generate. We need to be able to distinguish between these two, so that we can determine their relative value (both emotional and monetary). Art is tricky in that sense, since it's mostly based on personal enjoyment, but I think we can agree that things like NFTs are scams posing as high value art, while traditional paintings should be valued much more highly in every way.
>We need to be able to distinguish between these two, so that we can determine their relative value (both emotional and monetary).
Why?
Why should the nature of a work of art's creator have any bearing on its emotional impact? If the emotional impact is the same, why should the nature of the creator have any bearing on a piece's monetary value?
I suppose you can argue that part of what you feel when you look at artwork is empathy with the artist, and that empathizing with a human feels better than empathizing with software. I think that argument sucks.
I think if you felt something looking at these photos (fauxtos?) when you thought they were taken by a human, and now they make you feel something different because you know that's not true, that's not a judgement of the artwork, that's a reflection of you. I'd also argue that it's art on that merit alone.
> I guess it really shows that we have no way to predict the future.
I don't think one failed prediction invalidates all predictions?
We have already partially replaced artists with technology. Have a look at how photography changed painting (eg painting portraits is no longer the staple breadwinner it used to be). Or see how recorded music removed the need for each bar to have their own performer.
In the past technology just meant that we re-defined art and what it means to be an artist. We will continue to do so in the future.
> Have a look at how photography changed painting (eg painting portraits is no longer the staple breadwinner it used to be).
Has it, though? You can make a point that photorealism in painting has been technically superseded by photography, but we still find such paintings enjoyable and impressive. Sure, there's a smaller market demand from a purely practical perspective, but photorealistic paintings are still popular, and demonstrate the artist's skill and craftsmanship.
Technology can never replace art. It can just enhance it, broaden our abilities, and give us new perspectives to enjoy it.
In the past technology just meant that we re-defined art and what it means to be an artist. We will continue to do so in the future.
Similar: cars have not completely replaced horses. Horses still exist. People just have a very different relationship with them these days, and we also have much fewer horses (especially per capita) than 100 years ago.
I also find it curious there is still so much demand for photography when pretty much every photo you could ever want has already been taken. I can buy anything from Getty images.
Have you seen the level of photography on popular photography social media over the last ten years ? It absolutely blows your mind the amount of good content in there. I’m not sure “more” content will necessarily matter?
Then there’s authenticity. A photo of a real object, person or event actually matters to people for all sorts of reasons.
I’m still going to guess people will still pay artists to paint there portraits in 500 years, and just as we geeks like to argue computers are just better at everything, I’m not sure that’s what matters to people when it comes to art.
> I guess it really shows that we have no way to predict the future.
IDK, I think that in the AI circles the idea that this would happen was clear long before it actually became possible.
I recall a university workshop probably 20 years ago with a discussion about generative art of various forms (music, visual, etc) - as some genres of computer-generated art were a thing even back the - and the main conclusion about genres of "mainstream" art (which were far beyond our reach at the time) was that the difficult part is "judging taste"; in essence, if we got a reasonably accurate way of evaluating that one song/picture/sculpture/whatever is "better" than the other, then the rest is "just" relatively straightforward engineering. So the consensus prediction already two decades ago was that it's going to happen.
And while we don't have that "magic taste function", what we have made is an "indistinguishability" function, which can fulfil a similar role, and is effectively what the "photographer" in this story has used - if a model generates items that are (almost) indistinguishable from "good portrait photography art", however you define that, then it can make "art" that is as good as the rest of it. It can't really go beyond that, because if it accidentally generated something better then the model wouldn't be able to recognize that it's better, only that it's different, and discard that option, so improving on the status quo would require curation from someone with "good taste", but it apparently can beat mediocre photographers by matching the results of good ones.
> [...] I always imagined that artists would be the last ones to be replaced by technology.
A better guess would probably be any job that requires the dexterity, strength, versatility, and sensitivity of human hands (not just one of those, but their combination). Think electricians, or plumbers.
I think art/performances are about what they mean to you. You can listen to a recording of a classical orchestra with a headset at home, but it's not the same experience as going to a live concert. And I don't see a handmade woodwork the same as an industrial one. Context matters a lot.
I'm certain artists will find a way to use AI, but it will not remove the need for the artist to make art. Just like I can take photos with my camera, but that does not make them art.
If, in an alternative universe, we had no cameras, there would be a lot more painters for portraits, news pieces, books, travel guides, etc. A notable wedding or a summit would have 5-10 artists capturing key moments. So yes, cameras did replace painters in the same way cars replaced horses. Not a total elimination, but close by comparison to what would be.
It replaced "utilitarian" (if that's a word) paintings, right? Not "artistic" paintings. Of course, maybe the same people were doing both and had to adapt.
My point was about art. I don't think that cameras made it harder to make artistic paintings, did they?
Cameras paved the way to Hollywood and YouTube. The word “artist” also has become much broader in the wake of that fractal innovation, but I posit that if we had no movies or digital media, there would be many more Art Basels everywhere.
Yeah this is what I meant. The job will become much more tool-assisted which will lower the barrier for entry and we then won’t celebrate “artists” like we used to
Duchamp and Warhol would emphatically, EMPHATICALLY agree with you there.
All this is very Warhol. Andy would have loved this. I'm sure some of the surviving Pop artists are trying to think of something cool to do with it. In a way, the coolest Pop thing they can do with it is to sit back and watch literally all the world become artists (in a sense).
Everybody will be an artist for fifteen minutes. Warhol had it all figured out.
It's certainly good, but I find zero point in portraits of nonexistent people. Why should they exist? Isn't the point of a portrait to document and expose a facet of real life?
question here is "why does a photograph exist". Problem with this kind of question is that the answer is either personal and subjective, or it's an enumeration and an attempt to unite all (or subset) of the subjective reasons into some coherent theory.
What You're referring to reminds me of something like event photography. Photos that serve as a souvenir of an event / person / thing. Pictures that can be described as having "moment captured in time" as their raison d'etre. These are indeed unlikely to be replaced anytime soon.
But that's only a small subset of what we use photographs for. Lots of photography is about capturing an emmotion, rather than reality. Think of all the marketing (to name the biggest economic factor). If You're making an ad for watches, You don't give a damn if that cool guy wearing them is real or not. If You're a hotel or some travel agency, You are probably going to need "smiling happy couples / families around an infinity pool with a sunny beach as a background". Oh, hej, looks like that's actually a Stable Diffusion prompt...
If you believe photography is only a technic for creating a pretty picture to hang on the wall, you’ll be impressed with AI pictures. No one who understands the intent behind photography will be impressed by randomized data crafted to deceive as the real thing.
For instance, portraiture is to document people, who they are, their history, culture, etc. What’s so impressive about people that don’t exist? You’re correct.
I had no idea. And thinking about this now, I find what he's done brilliant. It shows in a very simple and concise way how far the generative AI has been for months now. And it shows (and proves) this to people who otherwise wouldn't be playing with any diffusion models otherwise.
In a way it blows your mind. If that's not art, I don't know what is.