Can AI be used to produce art?

[nikos79]

There is a big debate if AI can produce art. I have the feeling not because it cannot understand or experience the relation to time and as a result death too.

Recently I came across some photos by an artist named Noemia Prada. They looked beautiful, nice compositions and emotional but … something was wrong. They looked as if they were soulless. After reading the artist statement I realized they were all made with the use of AI

 

[koraks]

Here's some samples of her work, on Instagram: https://www.instagram.com/noemiaprada/
I've had a quick look and frankly, for most of the images I had not guessed it would have been AI. Some look...let's say 'suspicious', but it's a similar look that can be achieved by starting with a photographic original and digital editing. Some feature elements or settings that are implausible/improbable and as such are a bit more of a giveaway. But most of them look like they might have been actual photographs.

Then what to make of them? The concept of 'art' I always find a tricky one. The images as such are in my view clichés: they're the kind of images we've been seeing in some shape or form for over a century. Does that make them more or less artistic than more original works? Maybe; I don't know. A lot of what we consider art (think of the Dutch masters etc.) was rooted in traditions of replication of the same thematic with similar compositions. So that in itself doesn't disqualify these images.

They're not really photographed, but instead made while sitting behind a desk. Does that make these images more or less artful? In the eyes of some, for sure. The skills involved are different ones, and presently, it seems that the general tendency at least on this forum is to regard the skills involved in AI-generation of imagery to be of a lesser kind than photographic skills. I'm not too sure about this; it's a different medium. Can we meaningfully compare them in a normative sense?

What stands out for me mostly is the fact that these images have a high "crying boy" content: they're a path well-trodden, although the technique used is relatively new. They still express a certain emotion - perhaps in a somewhat crude fashion. But the same can be said of a lot of pop music and perhaps also e.g. art from Christian traditions (think of the millions of crucifixion scenes).

Is this art? In my view not any more or less so than another guy with a camera walking up to Half Dome to come home with his own rendition of Adams' 1927 work. It can be pleasurable to look at, it is effective within the vernacular it's deeply rooted in, there's a certain skill involved.

There's no accounting for taste, one way or another.

 

[thinkbrown]

Broadly my view is that art is a way for a person to communicate something about their experience of the world. It's an expression of self. Computers aren't capable of that.

An interesting side effect of all the AI slop rapidly filling the internet is how much more of an appreciation I've gained for bad art. When it's so easy to produce an image based on a quick set of keywords, it seems all the more admirable to make art even if you're not quite so good at it.

 

[nikos79]

Here's some samples of her work, on Instagram: https://www.instagram.com/noemiaprada/
I've had a quick look and frankly, for most of the images I had not guessed it would have been AI.

I also got tricked. Moreover, because most of her photos are very close to the style of photography I personally like.

Does that make them more or less artistic than more original works? Maybe; I don't know.

I am also pondering on that. If it can moves us as viewers, why not?

They're not really photographed, but instead made while sitting behind a desk. Can we meaningfully compare them in a normative sense?

Now that I’m looking at more of her photographs, I think the most important thing I notice is that they exist “outside of time.”
They are outside the flow of time, as if they belong to a frozen world.
You can’t easily tell what happened before, what happened after, what lies outside the frame, or to which spatial or temporal moment what is outside the frame belongs.
They are all excessively staged and static, and the details that could convey time, even in the form of imperfections, are entirely absent.

Is this art?

I am not sure. But it is a different form of art. For a human being, photography is a way of seeing and feeling the world. For an artificial intelligence, photography is a way of organizing information visually.

Also I mentioned time in the beginning. Every photo or piece of art is a reminder of death. But for an AI what is time? Time is just a variable, not an experience. Therefore, if an AI would try to create art through photography it could not be moved by the passage of time, but it could perhaps study its structure.

I am a computer scientist so I am familiar with Information Theory. Perhaps for an AI balance and joy is not found in geometry as with HCB but in a different analogy I could see. Maximal visual information (high Shannon entropy) with low distortion (minimal noise).

High entropy could mimic the amount of surprise, the visceral feeling that was mentioned in HCB thread. Low noise could mimic the economy of structure that is found in most powerful photos.

 

[koraks]

Broadly my view is that art is a way for a person to communicate something about their experience of the world. It's an expression of self. Computers aren't capable of that.

Sure, but these images are shaped on the basis of a person's explicit vision, which is encoded in the consecutive prompts and likely post-processing to realize that vision. There's no sign here of this imagery being randomly generated by a computer; it's very much directed by a personal and human intent.

Then there's the argument from the opposite direction - if we take a typical landscape photographer, they might 'communicate something about their experience of the world', but are likewise limited by that material world in expressing that experience per se. It's a different limitation from the limitations involved in computer-generating imagery - but limitations all the same.

So on the basis of these arguments, if I were to apply your criterion to this body of work, I couldn't dismiss it just on the basis that a computer (data center) was used in creating it.

When it's so easy to produce an image based on a quick set of keywords

It's easy to create what's referred to as 'slop'. It's not easy to create something interesting, regardless of the means used.
I can appreciate the skill that went into a mundane, cheesy gum bichromate print or so, because I'm aware of the menial work involved. Image-wise, it may still be the equivalent of AI slop and it's easily recognized as such. To the disadvantage of the AI crowd, at least on this forum there's generally not a lot of recognition, let alone appreciation of the skills involved in AI-generating art. I follow with a slanted eye the exploits of some people playing with this, and while I think what they produce is generally appalling in many ways (mostly due to the lack of any original artistic vision and a tendency towards misogynistic, cheesy quasi-porn), I do appreciate the fact that it's a specific skillset these people are developing. It's very much like watching people learning to paint and figure out brushes, paints etc. - the main difference is that the result looks far more polished in an objective sense than what a typical fledgling painter will produce.

Perhaps for an AI balance and joy is not found in geometry as with HCB but in a different analogy I could see. Maximal visual information (high Shannon entropy) with low distortion (minimal noise).

I doubt we could encode appreciation in theoretical constructs like these. They come into play, for sure. And then an artist comes along who does the exact opposite in utterly compelling work. Which is what the more interesting, powerful art seems to do much of the time anyway - breaking away from rules, conventions and set patterns.

For me, the defining element is still the set of brains that's controlling the show. I don't subscribe to the misconception that when it comes to AI-generated imagery, the computer is the brain. It's not. It's an extension of the hand; it's the digital equivalent of muscle memory or something along those lines.

There's a lot of slop around. Some of it was made by hand. It's easier to make it with a computer, so we'll end up with more AI slop than we did with oil paint slop. The latter does have the benefit of caloric value when it ends up in the incinerator.
Some slop we call art. Some art certainly isn't slop. I couldn't clearly draw a line between them; the line is sometimes diffuse as a result of the influence of time, preferences, cultural context and a whole lot of other things.

 

[runswithsizzers]

I think we should make a distinction between what you ask in your title, "Can AI produce art?" -- and what is actually happening in your example. Which is a human producing art, using AI as a tool. Strictly speaking, AI can not produce art, anymore than a paintbrush can, unless guided by human input.

I will admit that the degree of human intervention required to produce an AI work is several magnitudes less than what is required to produce a photograph or a painting. But I do think AI still falls into the category of "tool" rather than "creator". It remains to be seen if that will continue to be true in future.

As for your example...

I came across some photos by an artist named Noemia Prada. [...] After reading the artist statement I realized they were all made with the use of AI

When I do an online search for "artist Noemia Prada," the vast majority of the results require a social media account to view (Instagram, Facebook, Linkedin etc.), so I am unable to get past the first page. But in the few web pages I was able to view without an account, she describes herself as,
"I am an artist. I write, I paint, I take photos and I also like to be photographed. I’m a self-taught Painter/Photographer. My academic background is journalism and advertisement."

and, "I have a particular affinity for black and white photography—my world is defined by these tones."

I believe her <"Hands" project> is primarily traditional photography -- but I have no way to be certain, which is, of course, a significant issue with AI.

So when you say, her photos are "all made with the use of AI" I believe it would be more accurate to say "some" of her work is made with AI.

On <this page>, she says, "These images are not photographs. They are AI-designed artworks created using MidJourney, shaped by my artistic vision — a fusion of technology, emotion, and a deeply humanistic way of seeing."

Many of the examples of her AI work prominently feature text -- signs, placards, labels -- often one word like, "life" or "hope" or "fragile" -- somewhat like René Magritte's painting, The Treachery of Images (This is Not a Pipe), but without the irony.

Some might argue, if you need to add a text label to a visual work, then either you have no confidence in your work -- or no confidence in your viewers ability to get your meaning. On the other hand, there is a long tradition of mixed media, collages, and that sort of thing, so don't I question the validity of what she is trying to do. But some of it reminds me of those inspirational / motivational posters they hang in office workspaces, more that it does of fine art photography.

Question for discussion: If someone is making motivational posters, and makes no pretense of creating original art, then is AI more acceptable in the "graphics/illustrator" environment -- as compared to a "fine arts" environment?

 

[koraks]

I think we should make a distinction between what you ask in your title, "Can AI produce art?" -- and what is actually happening in your example.

I apologize for the thread title; I put it in and I did so a little too quickly. I hope you don't mind if I slightly adjust it to be more representative of the example @nikos79 posted.

 

[warden]

Bearing in mind the new thread title: Can AI be used to produce art? I think the answer is yes. AI is a tool that artists can and do use to create art. 2025 was probably the first banner year for artists leveraging AI in the creation of commercially and/or artistically successful music, sculpture, literature, cinema, and other arts including photography. It's hard to imagine the advancements, for better and worse, that are headed our way.

 

[runswithsizzers]

I apologize for the thread title; I put it in and I did so a little too quickly. I hope you don't mind if I slightly adjust it to be more representative of the example @nikos79 posted.

That is a different question -- and having changed the question, my answer changes from no to yes.

 

[loccdor]

Yes it can. Just look at this book cover from renowned musical artist and author Jim E. Brown:

 

[Alan Edward Klein]

My definition of art is something that inspires the viewer emotionally and/or spiritually. If no one is inspired, then the work is not art, and the maker is not an artist. So, art could be generated from AI if viewers feel inspired. OF course, the question is raised, since AI uses photos, paintings, and other art created by humans as the basis of creating its own, we can argue that at least AI is humanlike, sort of like the Tin Man in the Wizard of Oz. No real heart.

 

[Cholentpot]

AI is a tool like any other.

But what OP didn't ask is if AI can produce good art. Generative AI, - that is - using AI to generate something from Ex nihilo is still using human inspiration and human input via phrases. This then becomes the age old question of what makes art good. Is it the idea or the perceived effort or a combination of both.

In my opinion punching in prompts to a black box that goes and skims the collective thought of the internet accomplishes neither. There is no inspired thought going on and even less effort. Generative AI art falls below mass produced Velvet Elvis stuff in my ranking. You can hang it on your walls but you know I'll be a'judging you. If you don't care then I don't care. If you ask my opinion then I'll give it.

 

[runswithsizzers]

But what OP didn't ask is if AI can produce good art.

Where does the imperitave come from to disingush between "good art" and "bad art"? How does that make the world a better place?

Why not just declare, "Let there be art!" and let it go at that?

 

[Cholentpot]

Where does the imperitave come from to disingush between "good art" and "bad art"? How does that make the world a better place?

Why not just declare, "Let there be art!" and let it go at that?

If you entered your works into a prestigious gallery and got bumped by AI slop how would that make you feel? What if it derailed your entire career? What if you now have to compete against data farms powered by nuclear plants harvesting the work of billions with no permission or copyright conditions? Or better you, you've established your work and specific vision and are making traction and you get derailed by someone using your specific body of work via AI?

Does 'Let there be Art!' still sound like a good thing?

 

[MattKing]

Most Art creation is essentially the result of choosing what to include and what not to include.
So if AI creates the thing, the choice made to include and present it as Art is what makes it Art.

 

[Cholentpot]

Most Art creation is essentially the result of choosing what to include and what not to include.
So if AI creates the thing, the choice made to include and present it as Art is what makes it Art.

And the choice of the public to accept it as art plays a part too.

It comes down to the individual. Can you yourself articulate why it is or isn't art? I've said my piece and I think I've put up a few good points. Aside from labeling it as art does anyone have a valid defense of Generative AI as a legitimate artform?

 

[nikos79]

Here is an image generated by AI without any specific prompt in the form “use a car or a man sitting by a river etc.”

Instead, I just told it to use as a prompt my generic reply before on the post mentioning the Shannon Information Theory and Entropy. I was curious, what kind of photo would it generate?
How would it take my ideas about portraying the structure of time and visual information into an image?

Happy new year

 

[runswithsizzers]

If you entered your works into a prestigious gallery and got bumped by AI slop how would that make you feel? What if it derailed your entire career? What if you now have to compete against data farms powered by nuclear plants harvesting the work of billions with no permission or copyright conditions? Or better you, you've established your work and specific vision and are making traction and you get derailed by someone using your specific body of work via AI?

Does 'Let there be Art!' still sound like a good thing?

Well, I have no artistic ambitions, so I would never find myself in any of those situations.

I guess I am forgetting, there are people actually trying to make a living from art. Nobody I know is.

And when faced with the inevitable, I tend to be somewhat fatalistic. Yes, I feel bad for the buggy whip makers, but cars are coming, so what are you going to do? Guess I am getting too old and tired to rage against the machine. Or at least, getting a lot more selective about which fights to fight.

 

[Cholentpot]

Well, I have no artistic ambitions, so I would never find myself in any of those situations.

I guess I am forgetting, there are people actually trying to make a living from art. Nobody I know is.

And when faced with the inevitable, I tend to be somewhat fatalistic. Yes, I feel bad for the buggy whip makers, but cars are coming, so what are you going to do? Guess I am getting too old and tired to rage against the machine. Or at least, getting a lot more selective about which fights to fight.

This isn't buggywhip or plastic couch covers. I don't think that applies.

The closest is when photography took off and the classic artists felt threatened until they understood that photography was more than pushing a shutter. Or when TV came around and threatened Hollywood.

Art is one of the things that makes us human. If we throw that away we lose a major piece of our collective soul. One more step to being lazy lumps in a Wall-E world.

 

[Andrew O'Neill]

Yes, AI can be used to make Art.
There was a time when people were asking the same question about photography.

 

[Cholentpot]

Yes, AI can be used to make Art.
There was a time when people were asking the same question about photography.

Can AI as a tool be used to assist artists? Of course.

However, does putting a prompt into GPT and sitting back count as art?

 

[runswithsizzers]

This isn't buggywhip or plastic couch covers. I don't think that applies.

The closest is when photography took off and the classic artists felt threatened until they understood that photography was more than pushing a shutter. Or when TV came around and threatened Hollywood.

Art is one of the things that makes us human. If we throw that away we lose a major piece of our collective soul. One more step to being lazy lumps in a Wall-E world.

I am concerned about AI, especially the negative effect on climate change due to the energy required to power the AI computers.

And I do appreciate the role of the arts in a civilized society. But change happens, and right now it looks like AI is here to stay. When disruptive technologies destroy the status quo, there will always be some people who suffer greatly and others who adapt and survive. Given the apparent momentum of AI -- and the enormous investment being made by very powerful people -- it's hard for me to imagine any but the few most influential artists being able to do very much to stop it. It might work out better for fine art photographers to try to find strategies to adapt to AI, or ways work around it, rather than trying to stop it?? I don't know.

Considering all the other disruptive changes humanity is facing, I fear AI taking the jobs of artists may end up being less than a top priority in our hierarchy of needs.

 

[Andrew O'Neill]

Can AI as a tool be used to assist artists? Of course.

However, does putting a prompt into GPT and sitting back count as art?

Well, one could say the same thing about pressing the print button on an inkjet printer... At the end of the day, Art is in the eye of the beholder. Personally, I prefer to make my own Art. From scratch. By hand. If there is an AI tool for compositing my own images together, I'd probably try it out. I guess my AI tool is Photoshop...

 

[Cholentpot]

I am concerned about AI, especially the negative effect on climate change due to the energy required to power the AI computers.

And I do appreciate the role of the arts in a civilized society. But change happens, and right now it looks like AI is here to stay. When disruptive technologies destroy the status quo, there will always be some people who suffer greatly and others who adapt and survive. Given the apparent momentum of AI -- and the enormous investment being made by very powerful people -- it's hard for me to imagine any but the few most influential artists being able to do very much to stop it. It might work out better for fine art photographers to try to find strategies to adapt to AI, or ways work around it, rather than trying to stop it?? I don't know.

Considering all the other disruptive changes humanity is facing, I fear AI taking the jobs of artists may end up being less than a top priority in our hierarchy of needs.

Well, one could say the same thing about pressing the print button on an inkjet printer... At the end of the day, Art is in the eye of the beholder. Personally, I prefer to make my own Art. From scratch. By hand. If there is an AI tool for compositing my own images together, I'd probably try it out. I guess my AI tool is Photoshop...

I think I'm not getting my point through. Or maybe the crowd isn't immersed in the AI world as much as I am.

AI as an assisting tool is a powerful force that can and should be harnessed by anyone with the ability and will to learn it. This is not the AI I take issue with. The AI I have problems with is Generative AI. Take Andrew's example. Is the guy running the Xerox machine an artist? I think we all can agree that running off copies of artwork may take technical knowhow but is in no way art. But if Andy creates a piece of art and then prints it from the printer it hasn't diminished anything.

Generative AI is the worlds most powerful copy machine. You ask it to make something and it spits a copy of something out. You're just commissioning a copy and pushing the button on a machine to make that copy.

AI is not the same as crafting something in Photoshop. Even using AI assisted tools in photoshop to create something is not the same as Generative AI.

 

[runswithsizzers]

Or maybe the crowd isn't immersed in the AI world as much as I am.

If you don't mind me asking, how are you involved with AI? Have you spent much time using prompts to generate AI images?

I have generated only a few AI images, maybe a dozen, so I am only very briefly familiar with the process. But it seems to me, creating AI images can involve a certain amount of creativity and skill.

First, one must conceive the image. It requires a certain amount of imagination to come up with an idea -- a concept -- and then create a mental image of that concept. Someone with a vivid imagination will "see" a lot of details -- what particular objects will be in the image, their colors and textures, composition (how the objects should be arranged), framing, and lighting. This part of the process would be no different from a painter visualizing the concept for a new painting, and it is inherently a creative process.

Next, one must come up with the right prompts which lead to a generative image which matches the image in your head. In my experience (very limited), it can take some time and multiple additional promts before an acceptable result is obtained. Each modified prompt creates a different version of the scene; sometimes the changes are favorable towards your goal, sometimes not. I am guessing that with practice, more carefully chosen prompts will get better results?

Certainly, the process of creating an AI generated image is quicker and easier than making a photograph or a painting. Maybe it is not art. Maybe it's more of a craft? I don't know. But I do think some creativity is required to make an AI image, and I do think it is possible to make AI images which are compelling, enjoyable, and/or thought provoking in ways that are similar to other visual arts.