Why do people treat AI as sentient?

[poochyena]

I noticed this a lot, including in the recent wan show. They don't say people generated art, they said AI generated the art.

I just find it weird when someone uses AI to generate an image similar to another image, they blame the AI for it, and not the creator of the image.

Like, when taking a photo, no one says "sony camera took this photo" they say "I took this photo using my sony camera". AI art and text generators are tools, not sentient beings that produce things out of their own free will.

[Crunchy Dragon]

AI generates images based on prompts and what it was trained on. It's not sentient in the way that us humans are, but the AI is pretty much doing it as independently as it possibly can.

 

Feed the AI images to train it, give it a prompt, and it spits out stuff that fits the prompt.

 

Whether AI is or can be sentient is an entirely different discussion(imo) but I think it's fair to say that AI-generated content(be it movie scripts, images, music) is generated by the AI moreso than the human who wrote the AI. I might say "so and so's AI created this" or "my AI wrote this" when using it in a conversation, but ultimately the humans were not involved in the creation process a lot more than what it took to build, train, and prompt the AI.

 

12 minutes ago, poochyena said:

no one says "sony camera took this photo" they say "I took this photo using my sony camera".

#ShotOniPhone

 

While you are correct that the AI is not generating images or text purely of its own volition, the humans have no control over what the AI produces. Those "creative"(for lack of a better word) decisions are made entirely by the machine itself, influenced with the references that it was trained on, operating within whatever restrictions the developers impose.

[manikyath]

12 minutes ago, poochyena said:

they said AI generated the art.

just like "the printer produced the documents" or "the car rolled away, colliding with the mailbox" it's not because it's described as "performing the task" that it is considered sentient.

 

beyond that, in a sense, there's a person who started the image generation process, but the algorithm did all the work. "the voting machine counted all the votes" is probably a very good analogy here. yes, someone pressed the "count" button, but the machine performed the task.

 

as for directly addressing the camera analogy.

'i took this picture with my camera'

'my security camera recorded the thief'

both are the same 'action', but different details mean the sentence is built differently.

 

---

 

slightly off topic: however, people ascribing sentience to machines is a very "natural" thing to do, once something hits a certain treshold it triggers our instinct for empathy. this is something the folks over at boston dynamics could probably tell you a lot about, because every TV show that displays their 'spot' starts with someone petting it like a dog.

there's also at least one instance of someone going "mentally unstabile" and thinking a pseudo-random number generator is a deity talking to them directly.

[poochyena]

24 minutes ago, Crunchy Dragon said:

Those "creative"(for lack of a better word) decisions are made entirely by the machine itself

but its not. People have to give it prompts for it to work. Its not just creating works of out from its own free will, people are inputting what they want outputted.

31 minutes ago, Crunchy Dragon said:

#ShotOniPhone

short for "I shot this on an iphone" not "iphone shot this photo". The implication is heavily that a person took the photo using an iphone.

16 minutes ago, manikyath said:

just like "the printer produced the documents" or "the car rolled away, colliding with the mailbox" it's not because it's described as "performing the task" that it is considered sentient.

But people say that when those things do something unintended. When it happened without human interaction or directive. Generating images or text is explicitly through human interaction. If you pull up chatGPT and you do nothing, then chatGPT will also do nothing.

 

21 minutes ago, manikyath said:

"the voting machine counted all the votes" is probably a very good analogy here. yes, someone pressed the "count" button, but the machine performed the task.

The voting machine is probably the best other example of this phenomena.

[Crunchy Dragon]

1 minute ago, poochyena said:

People have to give it prompts for it to work. Its not just creating works of out from its own free will, people are inputting what they want outputted.

Correct, but the people are not creating the output. They tell the machine what they want, and the machine tries to create it.

 

3 minutes ago, poochyena said:

short for "I shot this on an iphone" not "iphone shot this photo". The implication is heavily that a person took the photo using an iphone.

Yes, that was a joke

4 minutes ago, poochyena said:

But people say that when those things do something unintended. When it happened without human interaction or directive. Generating images or text is explicitly through human interaction. If you pull up chatGPT and you do nothing, then chatGPT will also do nothing.

A car rolling away and hitting a mailbox still involves human interaction. The difference here, from our perspective, is that a car rolling into a mailbox takes human inaction(not parking the car properly) vs human action(putting a prompt into an image generator).

[manikyath]

1 minute ago, poochyena said:

But people say that when those things do something unintended. When it happened without human interaction or directive. Generating images or text is explicitly through human interaction. If you pull up chatGPT and you do nothing, then chatGPT will also do nothing.

2 minutes ago, poochyena said:

The voting machine is probably the best other example of this phenomena.

unless language is used VERY differently where you live, this is a perfectly normal way of describing things.

 

when a machine performs a task for you, it is said "the machine did the task", not "i did the task with the machine". i'm sorry, but i just really dont get your train of thought. there is no "phenomena", it is the english language, it is done in dutch too, and for my lack of french knowledge, the french actually make it even worse because of how 'reflexive' verbs work. if in french a computer catches fire, it "sets itself on fire". this does not imply that the computer is sentient and committing an act of massochism, it is just how the verb works.

[poochyena]

3 hours ago, Crunchy Dragon said:

Correct, but the people are not creating the output. They tell the machine what they want, and the machine tries to create it.

isn't that, like, everything that involves machines or computers?

[Crunchy Dragon]

4 minutes ago, poochyena said:

isn't that, like, everything that involves machines or computers?

Not nearly in the same sense.

 

Human input when typing words and sentences is pretty direct. We use the computer as a medium for communication, expression, creation. It's simply a more modern way of writing letters, visiting the public square, painting, drawing, or composing.

 

We don't communicate with each other on a forum like this by telling the computer to have a conversation with other users and letting it do whatever it wants. This isn't SubredditSimulator.

[LAwLz]

8 hours ago, Crunchy Dragon said:

ultimately the humans were not involved in the creation process a lot more than what it took to build, train, and prompt the AI.

Yeah, but the exact same argument can be said about for example cameras.

When someone asks "who took this picture?", in 9/10 cases, people will give you the photographers name, not the maker of the camera.

 

 

8 hours ago, Crunchy Dragon said:

While you are correct that the AI is not generating images or text purely of its own volition, the humans have no control over what the AI produces.

Yes, humans have full control over what the AI produces. No human input = no output.

If the argument is that the human sitting in the chair at the computer doesn't have any control because they just provide the input, not the process or output, then the same can be said about Photoshop, taking a picture with a camera, or any other software.

 

Do a human have control over the thousands of lines of code used in Photoshop to for example draw a line? No. A human "just" tells Photoshop "I want a line drawn from this point to this point", and then thousands upon thousands of calculations happens in the background to make that a reality.

[LAwLz]

8 hours ago, manikyath said:

beyond that, in a sense, there's a person who started the image generation process, but the algorithm did all the work. "the voting machine counted all the votes" is probably a very good analogy here. yes, someone pressed the "count" button, but the machine performed the task.

The problem is that a lot of people try and diminish the role a human plays in AI generated art (often because they don't understand it).

If we go with the voting analogy, a lot of the anti-AI crowd would be saying "you didn't vote, the machine did because it counted the vote. You just gave it some input and it calculated the result for you".

It would be pretty silly to claim that the voting machine "voted for you", and that "you didn't vote", right?

 

 

 

In the end I don't think it matters. We have used terms like "CGI" which stands for "computer generated imagery" for ages. You could argue that the computer generated those, not the humans who spent hours upon hours at the computer to make the computer create those images.

I think the difference now, with AI art, is that a lot of people are either very ignorant about how it works, are scared that they will lose their jobs and therefore demonize it, or a mix of the two, and as a result a lot of very stupid and bad arguments are being made against AI art. I also think there are a lot of emotions involved. People feel like AI art is different so they will try and rationalize why it's different, even if it doesn't make any logical sense.

 

"Oh yeah, someone using a camera is technically and mathematically just giving inputs as well, but it feels different to what AI art is".

 

I brought this up in a different thread, but the amount of AI used in cameras these days is insane. Google's camera will even fill in color it thinks belongs in the image if the camera wasn't able to capture it because for example the light was too low. Our hand movements doesn't even map to how the camera sensor moves in the real world either because we have image stabilization. If you move your camera 1cm to the left, your camera might automatically decide for you "no, you actually wanted to move the camera 9mm to the left, so I'll counteract your movement by 1mm for you, and you have no control over this decision". 

 

But because that change happened more gradually and to photography, something we have been doing for almost 200 years, people are more okay with it and treat it differently. Nobody would say that "you didn't take that picture, your camera did", even though the camera might have decided every single thing about the image except where roughly where the camera was pointed and when the shutter button was pressed. 99,99% of the work was done by the camera, a lot of which was AI, and people still claim the human took the picture.

[Arika]

It's generally because people don't really think about how GANs work.

 

8 hours ago, Crunchy Dragon said:

While you are correct that the AI is not generating images or text purely of its own volition, the humans have no control over what the AI produces. Those "creative"(for lack of a better word) decisions are made entirely by the machine itself, influenced with the references that it was trained on, operating within whatever restrictions the developers impose.

ok, let's take a somewhat extreme example, but an example none the less that i have found will generally change people's minds about it:

Let's say someone uses stable diffusion to generate something objectively Illegal, for example realistic looking CP with very deliberate prompts, who is liable?

Do we arrest the model creator? the Stable Diffusion team? or the person that had the direct hand in creating said images though what they have typed?

[Spindel]

I think people need to learn the difference between sentience and sapience. 

 

Heck I think most people need to learn what define sentience in the first place. 

[LAwLz]

21 minutes ago, Arika S said:

ok, let's take a somewhat extreme example, but an example none the less that i have found will generally change people's minds about it:

Let's say someone uses stable diffusion to generate something objectively Illegal, for example realistic looking CP with very deliberate prompts, who is liable?

Do we arrest the model creator? the Stable Diffusion team? or the person that had the direct hand in creating said images though what they have typed?

My guess, based on how people reacted to the "ChatGPT coded malware!" thread is that people will put the responsibility on the Stable Diffusion developers.

 

Not because it makes any logical sense, but because they see it as a way of destroying a technology they deem a threat to their job. If they can make it so that the creator of the program is responsible for what the users to use it for, they hope that the technology goes away because of all the legal issues, and they will be able to keep their jobs.

 

Either that, or everyone gets blamed, for the same reason (see it as an opportunity to destroy the technology by labeling anyone and everyone involved with it a potential criminal).

 

But maybe the people in this thread will have different opinions than the people in the other thread.

 

 

Of course, the logical thing would be to blame the person deliberately using the technology for bad things. We don't blame knife manufacturers if someone gets stabbed. We don't blame car manufacturers if someone flees from a crime scene. We don't blame browser developers if someone pirates something. We don't blame camera manufacturers if someone takes a picture of a naked child, nor do we blame Adobe if someone edits said picture in Photoshop.

We shouldn't blame AI if someone deliberately misuses it for bad things.

[manikyath]

40 minutes ago, LAwLz said:

 

If we go with the voting analogy, a lot of the anti-AI crowd would be saying "you didn't vote, the machine did because it counted the vote. You just gave it some input and it calculated the result for you".

that is entirely the wrong way to interpret that. the function of the voting machine is to count the votes, that's LITERALLY what it does.

 

in the same way, an image generating 'AI' literally generates images (sometimes more unique than others.. but that's not the topic.), and a random number generator generates random numbers.

 

yes, a human programmed it at the start, but the very concept of automation is that we set up a 'machine', so that the machine can do work for us. again.. that's HOW IT WORKS, and the exact description of "the voting machine counted the votes" is how language works.

 

there is no pro- or anti- anything argument here, my point is that this is how language works around the concept of machines performing a task. stop trying to pull my words in ways they CLEARLY werent written.

 

your analogy on the voting machine is essentially the "car rolled down a hill into a mailbox" example being interpreted as "the car disengaged it's handbrake and started steering for mailboxes". it becomes a very different view, because you've changed the parameters of the universe the example exists in.

[Sauron]

It's actually not that uncommon to see images described as "phone camera picture" or "DSLR photo" when talking in general terms... similarly "you generated this with AI?" seems to be used in the same way as "you took this with an iphone?". I don't think anyone is thinking or implying that an AI spontaneously generated these images with no prompting...

[mr moose]

Saying "AI generated the art" is not the same thing as saying "AI is sentient".  Unless someone specifically calls AI sentient then the problem is not a debate about sentience, but about who you think should get the credit for a picture.

[LAwLz]

2 hours ago, manikyath said:

that is entirely the wrong way to interpret that. the function of the voting machine is to count the votes, that's LITERALLY what it does.

 

in the same way, an image generating 'AI' literally generates images (sometimes more unique than others.. but that's not the topic.), and a random number generator generates random numbers.

 

yes, a human programmed it at the start, but the very concept of automation is that we set up a 'machine', so that the machine can do work for us. again.. that's HOW IT WORKS, and the exact description of "the voting machine counted the votes" is how language works.

 

there is no pro- or anti- anything argument here, my point is that this is how language works around the concept of machines performing a task. stop trying to pull my words in ways they CLEARLY werent written.

 

your analogy on the voting machine is essentially the "car rolled down a hill into a mailbox" example being interpreted as "the car disengaged it's handbrake and started steering for mailboxes". it becomes a very different view, because you've changed the parameters of the universe the example exists in.

The function of a voting machine is to calculate a result based on the input it is given, and produce an output. 

The function of Photoshop is to calculate the result based on the input it is given, and produce an output.

The function of a camera is to calculate the result based on the input it is given, and produce an output.

The function of an AI art program is to calculate the result based on the input it is given, and produce an output.

 

 

A voting machine "literally generates the result", based on the input it is given. The "result" is the final number it spits out at the end, and while humans gave it inputs, the process of counting and creating the final number was out of human hands. A computer did those things.

Same as an AI art program which generates an image based on the input it is given, and the same goes for Photoshop.

Humans gives inputs, and a computer interprets/processes those inputs and produces an output. In all three cases, there are three distinct and separate stages

1. Input
2. Processing
3. Output

Humans are only involved in the first one (input). But that's why we generally put the most attention and focus on that aspect of the creation process, even though most of the actual work (when looking at it from a mathematical POV) is done in the processing and output stages.

 

The inconsistency is that when it comes to AI art, some people puts all the focus on the processing and output stage and think that the input is not important. But when the same people start discussing other types of art then all of a sudden the input is the most important aspect.

"You didn't create that because you just gave a computer some inputs. The computer did all the work with processing and output". That sentence is to some people completely fine to say when it comes to AI art, but for some reason invalid when said for photography, photoshop, 3D modeling, etc.

 

 

 

1 hour ago, Sauron said:

It's actually not that uncommon to see images described as "phone camera picture" or "DSLR photo" when talking in general terms... similarly "you generated this with AI?" seems to be used in the same way as "you took this with an iphone?". I don't think anyone is thinking or implying that an AI spontaneously generated these images with no prompting...

There are a lot of people who will claim that someone using an AI program to create art didn't create it, that the AI did.

In those cases, it's the same as saying "your iPhone took this picture?" instead of "you took this picture with an iPhone?".

 

I wouldn't have any issue with people saying "you created this with AI?" or "you generated this with AI?", just like I don't have any issue with people saying "you photoshopped this?" or "you took this picture with an iPhone?".

My issue is with people who say "you didn't make this. The AI did." or "an AI created this", or any equivalence.

 

I think a lot of that stems from traditional artists not wanting to be associated with Ai artists because they are two very different fields, but we use the same verb for it.

If we hadn't come up with the verb "photograph" and instead kept using the word "paint" to describe using a camera, then I am sure painters would have been mad too.

 

In an alternative world (or maybe this world), I wouldn't be surprised if some painter would say: "You didn't paint that image, your camera did! What I do is real painting!" when shown a photograph.

[Sauron]

5 minutes ago, LAwLz said:

There are a lot of people who will claim that someone using an AI program to create art didn't create it, that the AI did.

I guess there could be degrees of this being correct depending on how much prompting went into it... if you just press a "make art" button and the AI spits something out with random prompts I could see the argument that your contribution is minimal. You could probably apply a similar reasoning with photos, after all while all photos are technically art there's a reason only some end up in museums and expositions. Personally I'm not that interested in this debate because frankly I hate the concept of intellectual property and therefore I really don't care much if someone should or shouldn't be considered the author of an art piece. So far it really doesn't look like anyone is about to lose their job to AI art generators anyway, save perhaps the most bottom of the barrel twitter hentai artists...

[LAwLz]

4 hours ago, Sauron said:

I guess there could be degrees of this being correct depending on how much prompting went into it... if you just press a "make art" button and the AI spits something out with random prompts I could see the argument that your contribution is minimal.

I can understand that reasoning but at the same time, we have never made this distinction for anything else.

We have never judged photographs by how many settings were tuned manually. A picture taken in "auto mode" on a camera has not been judged more harshly than one taking in all manual mode. Taking a picture with auto-focus turned on does not make the picture less valued, nor has it ever been an argument for why "the camera took that picture, not the photographer".

I have only ever seen this distinction be brought up when discussing AI art.

 

And I mean, how do we even measure "contribution"? For a camera, is it measured by how many settings are set to "auto"?

For AI art, is it measured by how many settings you change or how much you write into the prompt?

For animation, is it based on how many frames you drew, vs how many were generated by the program? If I make two frames "by hand", and then have a program generate a smooth transition between the two at 60 FPS and the sequence takes 2 seconds, I will only have made 2 out of 120 frames. Does that mean that I contributed ~0.8%? Even if those 0.8% served as the basis for all the other frames?

 

I think it is very weird that we have never had any conversation like this before AI art started appearing. Nobody had any issue with animators saying "they created an animation" even though they might "only" have given a computer some inputs and the computer did the rest of the work. Most people didn't have any issue saying that photographs were art, even if they were taking by someone with their camera set to auto and all they did was point in a general direction and click a single button (while the camera did thousands upon thousands of calculations, including those using AI, to figure out how to generate an image).

 

 

 

4 hours ago, Sauron said:

You could probably apply a similar reasoning with photos, after all while all photos are technically art there's a reason only some end up in museums and expositions.

I think that has more to do with the quality of the work rather than some technical reason. I don't think museums has rejected photographs because they used auto focus and auto white balance rather than set those things manually. They have cared about the picture itself and possibly the story of the photographer. Not the model of the camera or which settings it used.

 

AI art has only existed for a little while, and I think that there will be AI art that would belong in a museum. However, there is a risk that the art world will treat it differently just like it seems like some people here treat it differently. That it being AI art will automatically make it ineligible to some museums in order to gatekeep it. I mean, you don't have to look very far to find really strange things in the art world that seem entirely based on social status. For example Robery Ryman who was famous for paitning "white-on-white" paintings. And I don't mean he painted beautiful art while limiting himself to just the color white. I mean he literally just put white paint on a white canvas so the entire thing is white. 

This is what an art critic said about one of his pieces:

Quote

The only truly white thing in this universe is light. But even white light is actually polychromatic, and Ryman’s whites bristle with other colours, often toward the cooler and more metallic end of the spectrum. In a masterpiece – no other word will do – from around 1960, icy blue underpainting recalls the snow in Monet’s Magpie, and the strokes tumble down from the top left corner into a beige void. The pigment in a smaller work from the same year is nearly green, daubed in an irregular quadrilateral on an untreated and unstretched piece of linen, as matter-of-fact as the face of Jesus on the Shroud of Turin. A larger painting sees white transform from a romantic near-teal to a sallow, more stifling tone. They open up if you give them time, though really, it’s worth making multiple visits to this show to see the paintings in different light environments. I went three times – in sunny conditions, cloudy ones, and then (against the rules) at night – and each time was a different show.

It might be a bit hard to see, but here is a picture from the museum where this art piece hangs. From what I can tell it's valued at somewhere around 20 million dollars. It's the one to the left:

Spoiler

I think this is the same painting:

Spoiler

 

 

When you see stuff like that then it makes you, or at least me, question if museums really should be the arbiters of what is or isn't art.

 

 

 

5 hours ago, Sauron said:

So far it really doesn't look like anyone is about to lose their job to AI art generators anyway, save perhaps the most bottom of the barrel twitter hentai artists...

I think that is a bit short sighted. We have only had AI art programs for like a year, and they have already developed a ton. Progress might slow down the more mature the technology becomes, but I wouldn't be surprised if it will cause quite a few people to lose their jobs. Or rather, I don't think people will lose their jobs in the sense that they get fired, but freelance artists might find it harder and harder to find work. I also think that companies will use AI as tools to make things easier and faster for their current employees.

Gearbox have also said that they use AI tools to assist in creating game art and assets for their games. These tools are used by their already existing artists so nobody has lost their job because of it, but if their current employees can use these tools to work faster then it might have prevented Gearbox from hiring another artist. Instead of them having 10 artists that are begging for an 11th to be hired, the work burden might be reasonable with their 10 artists so no further requirement would be necessary.

 

Plus, I think the vast majority of people calling themselves artists might be in that "twitter hentai artist" category. For every Robery Ryman there are probably 100 "I'll draw your D&D character for 5 dollars" artists.

On fiverr alone there are 811 people advertising "I'll draw X in the style of Disney" right now. If we just look for people advertising illustration services, we find over 100,000 artists on fiberr alone. That's probably a larger population than all famous artists that have ever existed in the history of mankind. 

[Fluph]

17 hours ago, poochyena said:

I noticed this a lot, including in the recent wan show. They don't say people generated art, they said AI generated the art.

I just find it weird when someone uses AI to generate an image similar to another image, they blame the AI for it, and not the creator of the image.

Like, when taking a photo, no one says "sony camera took this photo" they say "I took this photo using my sony camera". AI art and text generators are tools, not sentient beings that produce things out of their own free will.

They blame the AI program as the makers of it trained the Ai on said images.
At no point is anyone treating Ai as sentient.

[Sauron]

3 hours ago, LAwLz said:

And I mean, how do we even measure "contribution"? For a camera, is it measured by how many settings are set to "auto"?

It's hard to make a 1:1 comparison to a camera because ultimately the most important thing is what's in frame and no matter how many settings you set to "auto" the camera won't move to find a good shot for you. On the other hand an AI could literally just have you press a single button and generate whatever. Maybe the closest equivalent is you accidentally pressing the shutter button and accepting whatever the camera gives you? It could still be a decent picture but I don't know how meaningful the term "author" would be in that case.

3 hours ago, LAwLz said:

I think that has more to do with the quality of the work rather than some technical reason. I don't think museums has rejected photographs because they used auto focus and auto white balance rather than set those things manually. They have cared about the picture itself and possibly the story of the photographer. Not the model of the camera or which settings it used.

Of course, I'm just saying it usually takes specific intent to take a good photo regardless of what camera and settings you use. The same seems to be the case with generative AI, if you don't know what you're looking for and how to get it you'll probably get an underwhelming result.

3 hours ago, LAwLz said:

AI art has only existed for a little while, and I think that there will be AI art that would belong in a museum. However, there is a risk that the art world will treat it differently just like it seems like some people here treat it differently. That it being AI art will automatically make it ineligible to some museums in order to gatekeep it.

Meh, I wouldn't worry too much about it. There was outrage when cameras came on the scene, particularly from portrait painters, but that faded quickly and now we have photographic art while still having paintings - they're just less likely to be a dime-a-dozen portrait because a camera can do it better and without years of training. Similarly good AI art will be recognized and maybe low quality fanart will be done through AI generators rather than adobe illustrator.

3 hours ago, LAwLz said:

When you see stuff like that then it makes you, or at least me, question if museums really should be the arbiters of what is or isn't art.

You'd actually be surprised how technically hard to make some of these paintings are. Getting a perfectly uniform white isn't that easy. Not that it matters that much, because yes, art isn't just valuable because of the skill behind it and yes, context does matter - museums will often have other paintings from the same period, or maybe from right before or after, showing the progression of art in those times and why a seemingly absurd painting might have more meaning than you'd assume. Not that museums are the ultimate arbiters of anything, but if it's there it's likely it's well regarded by the art community in general, not just by the curators at that specific museum.

3 hours ago, LAwLz said:

We have only had AI art programs for like a year, and they have already developed a ton.

Not quite - we only had decent AI art programs for about a year. It took a while to get here and there's no guarantee further progress will be swift or even possible. Or it could be the opposite, but for now I don't see it causing any real issue. If that changes then I might start worrying about it - not that you can really do much once the tech is developed anyway.

 

I'm always a little skeptical of these developments because there are plenty of past examples of a breakthrough tech apparently being right around the corner, ready to revolutionize our lives, and then getting swamped for decades. VR comes to mind, as well as self driving tech and AR glasses. Neural networks by their nature are hard to improve upon reliably.

3 hours ago, LAwLz said:

These tools are used by their already existing artists so nobody has lost their job because of it, but if their current employees can use these tools to work faster then it might have prevented Gearbox from hiring another artist.

I guess, but then we could blame Illustrator or Flash for millions of lost jobs... let alone computers as a whole. It's one thing to slightly and progressively lower demand for a profession due to automation, it's another to entirely replace a job with a machine. Besides, in my experience if work is done faster it just leads to more work needing to be done

3 hours ago, LAwLz said:

Plus, I think the vast majority of people calling themselves artists might be in that "twitter hentai artist" category. For every Robery Ryman there are probably 100 "I'll draw your D&D character for 5 dollars" artists.

Which goes back to my portrait painter example. I'm sorry if some people's livelihoods depend on this but I don't think what's being lost is worth holding back technological development...

[Crunchy Dragon]

14 hours ago, Arika S said:

very deliberate prompts

This would fall on the human, in my mind due to the prompts.

 

14 hours ago, LAwLz said:

Yeah, but the exact same argument can be said about for example cameras.

When someone asks "who took this picture?", in 9/10 cases, people will give you the photographers name, not the maker of the camera.

Correct, however in this sense, the human is doing "all the work". They found what they wanted a picture of, adjusted all the necessary settings, and pushed all the necessary buttons. Human input and action is necessary through every step of the process.

 

14 hours ago, LAwLz said:

Yes, humans have full control over what the AI produces. No human input = no output.

If the argument is that the human sitting in the chair at the computer doesn't have any control because they just provide the input, not the process or output, then the same can be said about Photoshop, taking a picture with a camera, or any other software.

 

Do a human have control over the thousands of lines of code used in Photoshop to for example draw a line? No. A human "just" tells Photoshop "I want a line drawn from this point to this point", and then thousands upon thousands of calculations happens in the background to make that a reality.

My argument here is that with Photoshop, photography, drawing, painting, composing, etc, every single decision is made through human action. With an AI, all the human does is give it a prompt, then the AI has full control from that point on.

 

If I asked you to draw me a picture of "a reindeer riding a can of Miller Lite across the Milky Way", then walked away and left you to draw a picture based on my prompt, I have no creative input or control beyond just giving you the input and waiting for the output. Every creative decision is made by you, with no influence from me other than the prompt I gave you.

[tikker]

23 hours ago, poochyena said:

but its not. People have to give it prompts for it to work. Its not just creating works of out from its own free will, people are inputting what they want outputted.

I don't think that that is very different from paying an artist to draw something you tell them to draw. Apart from choosing the job of an artist, they are not really drawing it of their own free will and output what people input to them.

[Sauron]

9 hours ago, Crunchy Dragon said:

If I asked you to draw me a picture of "a reindeer riding a can of Miller Lite across the Milky Way", then walked away and left you to draw a picture based on my prompt, I have no creative input or control beyond just giving you the input and waiting for the output. Every creative decision is made by you, with no influence from me other than the prompt I gave you.

Wouldn't this be a little like a movie director telling the cameramen and VFX artists what they'd like out of a given shot? We still credit directors with authorship of their movies and they can be more or less directly involved with the camerawork and effects (as you might be with defining stricter and stricter prompts) depending on what they're looking to achieve. You could imagine a "good" artist being so proficient at prompting AI generators that they can tweak the output to perfectly match their vision, much like a good photographer would know how to get the shot they want out of a camera, while an amateur might instead get something that looks "good enough" and move on.

[Blue4130]

16 hours ago, LAwLz said:

I can understand that reasoning but at the same time, we have never made this distinction for anything else.

We have never judged photographs by how many settings were tuned manually. A picture taken in "auto mode" on a camera has not been judged more harshly than one taking in all manual mode. Taking a picture with auto-focus turned on does not make the picture less valued, nor has it ever been an argument for why "the camera took that picture, not the photographer".

I have only ever seen this distinction be brought up when discussing AI art.

 

And I mean, how do we even measure "contribution"? For a camera, is it measured by how many settings are set to "auto"?

For AI art, is it measured by how many settings you change or how much you write into the prompt?

For animation, is it based on how many frames you drew, vs how many were generated by the program? If I make two frames "by hand", and then have a program generate a smooth transition between the two at 60 FPS and the sequence takes 2 seconds, I will only have made 2 out of 120 frames. Does that mean that I contributed ~0.8%? Even if those 0.8% served as the basis for all the other frames?

 

I think it is very weird that we have never had any conversation like this before AI art started appearing. Nobody had any issue with animators saying "they created an animation" even though they might "only" have given a computer some inputs and the computer did the rest of the work. Most people didn't have any issue saying that photographs were art, even if they were taking by someone with their camera set to auto and all they did was point in a general direction and click a single button (while the camera did thousands upon thousands of calculations, including those using AI, to figure out how to generate an image).

 

 

 

I think that has more to do with the quality of the work rather than some technical reason. I don't think museums has rejected photographs because they used auto focus and auto white balance rather than set those things manually. They have cared about the picture itself and possibly the story of the photographer. Not the model of the camera or which settings it used.

 

 

I don't know how much of a photographer you are, but I am a college educated, former professional photographer and photojournalist. People in the photo world absolutely do judge and value photos taken on auto/af differently. Or film vs digital. Ego's are big and many will find and way to knock someone else down.