3 ms·

But then why wouldn't AI generated art be able to stir me? Why is a human being in the loop so important as to be supposedly essential?

petralithic 1y ago
It's not. If one takes the fact that art is in the eye of the beholder [0], then yes, even AI art may stir you, especially as a human is the one generating at the end of the day, for a specific purpose and statement about what they want to convey. There is a good part of the series Remembrance of Earth's Past (of which The Three Body Problem is the first book) where the aliens are creating art and it shocks people to learn that the art they're so moved by was actually created by non-humans. This is exactly what this situation with AI feels like, and not even to the same extent because again AI is not autonomously making images, it's still a human at the end of the day picking what to prompt. [0] https://en.wikipedia.org/wiki/The_Death_of_the_Author https://en.wikipedia.org/wiki/The_Death_of_the_Author
- jacquesm 1y ago
  > it's still a human at the end of the day picking what to prompt I think that 'dutch people skating on a lake' or 'girl with a pearl earring' or 'dutch religious couple in front of their barn' without having an AI trained on various works will produce just noise. And if those particular works (you know the ones, right?) were not part of the input then the AI would never produce anything looking like the original, no matter how specific you made the prompt. It takes human input to animate it, and even then what it produces to me does not look original whereas any five year old is able to produce entirely original works of art, none of which can be reduced to a prompt. Prompts are instructions, they are settings on a mixer, they are not the music produced by the artists at the microphones.
  - petralithic 1y ago
    Have you actually used image generators today? It can produce things it's never seen if only you describe the constituent pieces. Prompts are a compressed version of the image one wants to create, and these days you don't even need "prompts" per se, you can say, make a woman looking towards the viewer, now add a pearl earing, now adjust this and that etc.
    - jacquesm 1y ago
      > Have you actually used image generators today? Why would you ask this? It sounds like a lead-up to some kind of put down. > It can produce things it's never seen if only you describe the constituent pieces. It can produce things it's never seen based on lots of things that it has seen. > Prompts are a compressed version of the image one wants to create They emphatically are not. They are instructions to a tool on what relative importance to assign to all of the templates that it was trained on. But it doesn't understand the output image any more than it understood any of the input images. There is no context available to it in the purest sense of the word. It has no emotion to express because it doesn't have emotions in the first place. > and these days you don't even need "prompts" per se, you can say, make a woman looking towards the viewer, now add a pearl earing, now adjust this and that etc. That's just a different path to building up the same prompt. It doesn't suddenly cause the AI to use red for a dress because it thinks it is a nice counterpoint to a flower in a different part of the image because it does not think at all.
      - petralithic 1y ago
        I think you're reading too much into my comment. It's not a put down, I'm genuinely asking because it seems many people still think anyone serious about AI just types prompts into Midjourney, but it's become a lot more complex than that, akin to electronic music production; producers haven't played every single note with a physical instrument their synths synthesize yet their arrangement of the notes is what makes them a producer, and so too with AI workflows such as those seen in ComfyUI. If one is not familiar then they might not understand where the field is today. Regarding prompts, I never said a computer "understands" or is "emotional" about an image, I don't think anyone actually thinks that, on either side of the debate so not sure why you're bringing that up. By "compressed" I just meant in the information theory way, in that if you have a specific series of words, and a given temperature and other settings for a given model, it will deterministically produce the same image, hence the set of those attributes can be thought of as a compressed representation of that image. I made no claims about it thinking whatsoever. > It can produce things it's never seen based on lots of things that it has seen. Yes, just like humans, as I had said in my initial comment about the same old arguments being said since 2021 when Stable Diffusion came out. But again that's tiresome so let's not repeat that here too.
jacquesm 1y ago
Because it is mimicking human input. Effectively you are getting a mixture of many pieces of artwork that humans made distilled down into some sloppy new one that was made without feeling, purpose or skill and that can be described by its prompt, a few kilobytes at best. Original human art can only be approximated but never captured with 100% fidelity regardless of the bitrate, that is what makes it unique to begin with. Even an imitation by another human (some of which can be very good) could stir you in the exact same way but they'd be copies, not original works. Anyway, this gets hairy quickly, that's why I chose to illustrate with a crappy recording of a magnificent piece that still captures that feeling - for me - whereas many others would likely disagree. Art is made by its creator because they want to and because they can, not because they are regurgitating output based on a multitude of inputs and a prompt. Paint me a Sistine Chapel is going to yield different results no matter how many times you would give that same prompt to Michelangelo depending on his mood, what happened recently, what he ate and his health as well as the season. That AI will produce the same result over and over again from the same prompt. It is a mechanistic transformation, not an original work, it reduces the input, it does not expand on it, it does not add its own feelings to it.
- perching_aix 1y ago
  I think this is a reasonable counter in some respects, although I do also think it's specific to the current iteration of AI art. It's a bit like when people describe how models don't have a will or the likes. Of course they don't, "they" are basically frozen in time. Training is way slower than inference, and even inference is often slower than "realtime". It just doesn't work that way from the get-go. They're also simply not very good - hence why they're being fed curated data. In that sense, and considering history, I can definitely see why it would (and should?) be considered differently. Not sure this is what you meant, but this is an interesting lens, so thanks for this.
- petralithic 1y ago
  Haven't these arguments been the same since Stable Diffusion came out? Someone (A) will say what you said, then someone else (B) will say, well humans remix as well, A: no that's different because we're humans not machines, B: there is no need to prefer a biological substrate over a silicon one; A: AI will produce the same result over and over, B: not if you change the temperature and randomize the seed. It's tiresome to read the same thing over and over again and at this point I don't think A's arguments will convince B and vice versa because both come from different initial input conditions in their thought processes. It's like trying to dig two parallel tunnels through a mountain from different heights and thinking they'll converge.

petralithic 1y ago

It's not. If one takes the fact that art is in the eye of the beholder [0], then yes, even AI art may stir you, especially as a human is the one generating at the end of the day, for a specific purpose and statement about what they want to convey. There is a good part of the series Remembrance of Earth's Past (of which The Three Body Problem is the first book) where the aliens are creating art and it shocks people to learn that the art they're so moved by was actually created by non-humans. This is exactly what this situation with AI feels like, and not even to the same extent because again AI is not autonomously making images, it's still a human at the end of the day picking what to prompt. [0] https://en.wikipedia.org/wiki/The_Death_of_the_Author https://en.wikipedia.org/wiki/The_Death_of_the_Author

jacquesm 1y ago
> it's still a human at the end of the day picking what to prompt I think that 'dutch people skating on a lake' or 'girl with a pearl earring' or 'dutch religious couple in front of their barn' without having an AI trained on various works will produce just noise. And if those particular works (you know the ones, right?) were not part of the input then the AI would never produce anything looking like the original, no matter how specific you made the prompt. It takes human input to animate it, and even then what it produces to me does not look original whereas any five year old is able to produce entirely original works of art, none of which can be reduced to a prompt. Prompts are instructions, they are settings on a mixer, they are not the music produced by the artists at the microphones.
- petralithic 1y ago
  Have you actually used image generators today? It can produce things it's never seen if only you describe the constituent pieces. Prompts are a compressed version of the image one wants to create, and these days you don't even need "prompts" per se, you can say, make a woman looking towards the viewer, now add a pearl earing, now adjust this and that etc.
  - jacquesm 1y ago
    > Have you actually used image generators today? Why would you ask this? It sounds like a lead-up to some kind of put down. > It can produce things it's never seen if only you describe the constituent pieces. It can produce things it's never seen based on lots of things that it has seen. > Prompts are a compressed version of the image one wants to create They emphatically are not. They are instructions to a tool on what relative importance to assign to all of the templates that it was trained on. But it doesn't understand the output image any more than it understood any of the input images. There is no context available to it in the purest sense of the word. It has no emotion to express because it doesn't have emotions in the first place. > and these days you don't even need "prompts" per se, you can say, make a woman looking towards the viewer, now add a pearl earing, now adjust this and that etc. That's just a different path to building up the same prompt. It doesn't suddenly cause the AI to use red for a dress because it thinks it is a nice counterpoint to a flower in a different part of the image because it does not think at all.
    - petralithic 1y ago
      I think you're reading too much into my comment. It's not a put down, I'm genuinely asking because it seems many people still think anyone serious about AI just types prompts into Midjourney, but it's become a lot more complex than that, akin to electronic music production; producers haven't played every single note with a physical instrument their synths synthesize yet their arrangement of the notes is what makes them a producer, and so too with AI workflows such as those seen in ComfyUI. If one is not familiar then they might not understand where the field is today. Regarding prompts, I never said a computer "understands" or is "emotional" about an image, I don't think anyone actually thinks that, on either side of the debate so not sure why you're bringing that up. By "compressed" I just meant in the information theory way, in that if you have a specific series of words, and a given temperature and other settings for a given model, it will deterministically produce the same image, hence the set of those attributes can be thought of as a compressed representation of that image. I made no claims about it thinking whatsoever. > It can produce things it's never seen based on lots of things that it has seen. Yes, just like humans, as I had said in my initial comment about the same old arguments being said since 2021 when Stable Diffusion came out. But again that's tiresome so let's not repeat that here too.

jacquesm 1y ago

Because it is mimicking human input. Effectively you are getting a mixture of many pieces of artwork that humans made distilled down into some sloppy new one that was made without feeling, purpose or skill and that can be described by its prompt, a few kilobytes at best. Original human art can only be approximated but never captured with 100% fidelity regardless of the bitrate, that is what makes it unique to begin with. Even an imitation by another human (some of which can be very good) could stir you in the exact same way but they'd be copies, not original works. Anyway, this gets hairy quickly, that's why I chose to illustrate with a crappy recording of a magnificent piece that still captures that feeling - for me - whereas many others would likely disagree. Art is made by its creator because they want to and because they can, not because they are regurgitating output based on a multitude of inputs and a prompt. Paint me a Sistine Chapel is going to yield different results no matter how many times you would give that same prompt to Michelangelo depending on his mood, what happened recently, what he ate and his health as well as the season. That AI will produce the same result over and over again from the same prompt. It is a mechanistic transformation, not an original work, it reduces the input, it does not expand on it, it does not add its own feelings to it.

perching_aix 1y ago
I think this is a reasonable counter in some respects, although I do also think it's specific to the current iteration of AI art. It's a bit like when people describe how models don't have a will or the likes. Of course they don't, "they" are basically frozen in time. Training is way slower than inference, and even inference is often slower than "realtime". It just doesn't work that way from the get-go. They're also simply not very good - hence why they're being fed curated data. In that sense, and considering history, I can definitely see why it would (and should?) be considered differently. Not sure this is what you meant, but this is an interesting lens, so thanks for this.
petralithic 1y ago
Haven't these arguments been the same since Stable Diffusion came out? Someone (A) will say what you said, then someone else (B) will say, well humans remix as well, A: no that's different because we're humans not machines, B: there is no need to prefer a biological substrate over a silicon one; A: AI will produce the same result over and over, B: not if you change the temperature and randomize the seed. It's tiresome to read the same thing over and over again and at this point I don't think A's arguments will convince B and vice versa because both come from different initial input conditions in their thought processes. It's like trying to dig two parallel tunnels through a mountain from different heights and thinking they'll converge.